Skip to Content
Rule

Rule To Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule); Revisions to Acid Rain Program; Revisions to the NOX

Action

Final Rule.

Summary

In today's action, EPA finds that 28 States and the District of Columbia contribute significantly to nonattainment of the national ambient air quality standards (NAAQS) for fine particles (PM 2.5) and/or 8-hour ozone in downwind States. The EPA is requiring these upwind States to revise their State implementation plans (SIPs) to include control measures to reduce emissions of sulfur dioxide (SO 2) and/or nitrogen oxides (NO X). Sulfur dioxide is a precursor to PM 2.5 formation, and NO X is a precursor to both ozone and PM 2.5 formation. Reducing upwind precursor emissions will assist the downwind PM 2.5 and 8-hour ozone nonattainment areas in achieving the NAAQS. Moreover, attainment will be achieved in a more equitable, cost-effective manner than if each nonattainment area attempted to achieve attainment by implementing local emissions reductions alone.

Based on State obligations to address interstate transport of pollutants under section 110(a)(2)(D) of the Clean Air Act (CAA), EPA is specifying statewide emissions reduction requirements for SO 2 and NO X. The EPA is specifying that the emissions reductions be implemented in two phases. The first phase of NO X reductions starts in 2009 (covering 2009-2014) and the first phase of SO 2 reductions starts in 2010 (covering 2010-2014); the second phase of reductions for both NO X and SO 2 starts in 2015 (covering 2015 and thereafter). The required emissions reductions requirements are based on controls that are known to be highly cost effective for electric generating units (EGUs).

Today's action also includes model rules for multi-State cap and trade programs for annual SO 2 and NO X emissions for PM 2.5 and seasonal NO X emissions for ozone that States can choose to adopt to meet the required emissions reductions in a flexible and cost-effective manner.

Today's action also includes revisions to the Acid Rain Program regulations under title IV of the CAA, particularly the regulatory provisions governing the SO 2 cap and trade program. The revisions are made because they streamline the operation of the Acid Rain SO 2 cap and trade program and/or facilitate the interaction of that cap and trade program with the model SO 2 cap and trade program included in today's action. In addition, today's action provides for the NO X SIP Call cap and trade program to be replaced by the CAIR ozone-season NO X trading program.

Unified Agenda

Clean Air Interstate Rule; Formerly Titled Interstate Air Quality Rule

4 actions from January 30th, 2004 to November 2004

  • January 30th, 2004
  • June 2004
    • Supplemental NPRM
  • August 2004
    • Notice of Data Availability
  • November 2004
    • Final Action
 

Table of Contents Back to Top

Tables Back to Top

DATES: Back to Top

The effective date of today's action, except for the revisions to 40 CFR parts 72, 73, 74, and 77 of the Acid Rain Program regulations, is July 11, 2005. States must submit to EPA for approval enforceable plans for complying with the requirements of this rule by September 11, 2006. The effective date for today's revisions to 40 CFR parts 72, 73, 74, and 77 of the Acid Rain Program regulations is July 1, 2006.

ADDRESSES: Back to Top

The EPA has established a docket for this action under Docket ID No. OAR-2003-0053. All documents in the docket are listed in the EDOCKET index at http://www.epa.gov/edocket. Although listed in the index, some information is not publicly available, i.e., Confidential Business Information (CBI) or other information whose disclosure is restricted by statute. Certain other material, such as copyrighted material, is not placed on the Internet and will be publicly available only in hard copy form. Publicly available docket materials are available either electronically in EDOCKET or in hard copy at the EPA Docket Center, EPA West, Room B102, 1301 Constitution Avenue, NW., Washington, DC. The Public Reading Room is open from 8:30 a.m. to 4:30 p.m., Monday through Friday, excluding legal holidays. The telephone number for the Public Reading Room is (202) 566-1744, and the telephone number for the Air Docket is (202) 566-1742.

FOR FURTHER INFORMATION CONTACT: Back to Top

For general questions concerning today's action, please contact Carla Oldham, U.S. EPA, Office of Air Quality Planning and Standards, Air Quality Strategies and Standards Division, Mail Code C539-02, Research Triangle Park, NC, 27711, telephone (919) 541-3347, e-mail at oldham.carla@epa.gov. For legal questions, please contact Sonja Petersen, U.S. EPA, Office of General Counsel, Mail Code 2344A, 1200 Pennsylvania Avenue, NW., Washington, DC, 20460, telephone (202) 564-4079, e-mail at petersen.sonja@epa.gov. For questions regarding air quality analyses, please contact Norm Possiel, U.S. EPA, Office of Air Quality Planning and Standards, Emissions Monitoring and Analysis Division, Mail Code D243-01, Research Triangle Park, NC, 27711, telephone (919) 541-5692, e-mail at possiel.norm@epa.gov. For questions regarding the EGU cost analyses, emissions inventories, and budgets, please contact Roman Kramarchuk, U.S. EPA, Office of Atmospheric Programs, Clean Air Markets Division, Mail Code 6204J, 1200 Pennsylvania Avenue, NW., Washington, DC, 20460, telephone (202) 343-9089, e-mail at kramarchuk.roman@epa.gov. For questions regarding statewide emissions inventories, please contact Ron Ryan, U.S. EPA, Office of Air Quality Planning and Standards, Emissions Monitoring and Analysis Division, Mail Code D205-01, Research Triangle Park, NC, 27711, telephone (919) 541-4330, e-mail at ryan.ron@epa.gov. For questions regarding emissions reporting requirements, please contact Bill Kuykendal, U.S. EPA, Office of Air Quality Planning and Standards, Emissions Monitoring and Analysis Division, Mail Code D205-01, Research Triangle Park, NC, 27711, telephone (919) 541-5372, e-mail at kuykendal.bill@epa.gov. For questions regarding the model cap and trade programs, please contact Sam Waltzer, U.S. EPA, Office of Atmospheric Programs, Clean Air Markets Division, Mail Code 6204J, 1200 Pennsylvania Avenue, NW., Washington, DC, 20460, telephone (202) 343-9175, e-mail at waltzer.sam@epa.gov. For questions regarding analyses required by statutes and executive orders, please contact Linda Chappell, U.S. EPA, Office of Air Quality Planning and Standards, Air Quality Strategies and Standards Division, Mail Code C339-01, Research Triangle Park, NC, 27711, telephone (919) 541-2864, e-mail at chappell.linda@epa.gov.For questions regarding the Acid Rain Program regulation revisions, please contact Dwight C. Alpern, U.S. EPA, Office of Atmospheric Programs, Clean Air Markets Division, Mail Code 6204J, 1200 Pennsylvania Avenue, NW., Washington, DC, 20460, telephone (202) 343-9151, e-mail at alpern.dwight@epa.gov.

SUPPLEMENTARY INFORMATION: Back to Top

Regulated Entities Back to Top

Except for the revisions to the Acid Rain Program regulations, this action does not directly regulate emissions sources. Instead, it requires States to revise their SIPs to include control measures to reduce emissions of NO X and SO 2. The emissions reductions requirement assigned to the States are based on controls that are known to be highly cost effective for EGUs.

Entities potentially regulated by the revisions to the Acid Rain Program regulations in this action are fossil-fuel-fired boilers, turbines, and internal combustion engines, including those that serve generators producing electricity, generate steam, or cogenerate electricity and steam. Regulated categories and entities include:

Category 1NAICS code Examples of potentially regulated entities
1North American Industry Classification System.
2Federal, State, or local government-owned and operated establishments are classified according to the activity in which they are engaged.
Industry 221112 and others Electric service providers, boilers, turbines, and internal combustion engines from a wide range of industries.
Federal government 221122 Fossil fuel-fired electric utility steam generating units owned by the Federal government.
State/local/Tribal government 221122 921150 Fossil fuel-fired electric utility steam generating units owned by municipalities. Fossil fuel-fired electric utility steam generating units in Indian Country.

This table is not intended to be exhaustive, but rather provides a guide for readers regarding entities likely to be regulated by the revisions to the Acid Rain Program regulations in this action. This table lists the types of entities that EPA is aware could potentially be regulated. Other types of entities not listed in the table could also be regulated. To determine whether your facility is regulated, you should carefully examine the applicability criteria in 40 CFR 72.6 and 74.2 and the exemptions in 40 CFR 72.7 and 72.8. If you have questions regarding the applicability of the revisions to the Acid Rain Program regulations in this action to a particular entity, consult persons listed in the preceding FOR FURTHER INFORMATION CONTACT section.

Web Site for Rulemaking Information Back to Top

The EPA has also established a Web site for this rulemaking at http://www.epa.gov/cleanairinterstaterule/ or http://www.epa.gov/cair/ (formerly at http://www.epa.gov/interstateairquality/) which includes the rulemaking actions and certain other related information that the public may find useful.

Outline Back to Top

I. Overview

A. What Are the Central Requirements of this Rule?

B. Why Is EPA Taking this Action?

1. Policy Rationale for Addressing Transported Pollution Contributing to PM 2.5 and Ozone Problems

a. The PM 2.5 Problem

b. The 8-hour Ozone Problem

c. Other Environmental Effects Associated with SO 2 and NO X Emissions

2. The CAA Requires States to Act as Good Neighbors by Limiting Downwind Impacts

3. Today's Rule Will Improve Air Quality

C. What was the Process for Developing this Rule?

D. What Are the Major Changes Between the Proposals and the Final Rule?

II. The EPA's Analytical Approach

A. How Did EPA Interpret the Clean Air Act's Pollution Transport Provisions in the NO X SIP Call?

1. Clean Air Act Requirements

2. The NO X SIP Call Rulemaking

a. Analytical Approach of NO X SIP Call

b. Regulatory Requirements

c. SIP Submittal and Implementation Requirements

3. Michigan v. EPA Court Case

4. Implementation of the NO X SIP Call

B. How Does EPA Interpret the Clean Air Act's Pollution Transport Provisions in Today's Rule

1. CAIR Analytical Approach

a. Nature of Nonattainment Problem and Overview of Today's Approach

b. Air Quality Factor

c. Cost Factor

d. Other Factors

e. Regulatory Requirements

f. SIP Submittal and Implementation Requirements

2. What Did Commenters Say and What Is EPA's Response?

a. Aspects of Contribute-Significantly Test

III. Why Does This Rule Focus on SO 2 and NO X, and How Were Significant Downwind Impacts Determined?

A. What Is the Basis for EPA's Decision to Require Reductions in Upwind Emissions of SO 2 and NO X to Address PM 2.5 related transport?

1. How Did EPA determine which pollutants were necessary to control to address interstate transport for PM 2.5?

a. What Did EPA propose regarding this issue in the NPR?

b. How Does EPA address public comments on its proposal to address SO 2 and NO X emissions and not other pollutants?

c. What Is EPA's Final Determination?

2. What Is the role for local emissions reduction strategies?

a. Summary of analyses and conclusions in the proposal

b. Summary and Response to Public Comments

B. What Is the Basis for EPA's Decision to Require Reductions in Upwind Emissions of NO X to Address Ozone-Related Transport?

1. How Did EPA Determine Which Pollutants Were Necessary to Control to Address Interstate Transport for Ozone?

2. How Did EPA Determine That Reductions in Interstate Transport, as Well as Reductions in Local Emissions, Are Warranted to Help Ozone Nonattainment Areas to Meet the 8-hour Ozone Standard?

a. What Did EPA Say in its Proposal Notice?

b. What Did Commenters Say?

C. Comments on Excluding Future Case Measures from the Emissions Baselines Used to Estimate Downwind Ambient Contribution

D. What Criteria Should Be Used to Determine Which States

1. What Is the Appropriate Metric for Assessing DownwindPM 2.5 Contribution?

a. Notice of Proposed Rulemaking

b. Comments and EPA's Responses

c. Today's Action

2. What Is the Level of the PM 2.5 Contribution Threshold?

a. Notice of Proposed Rulemaking

b. Comments and EPA's Responses

c. Today's Action

E. What Criteria Should Be Used to Determine Which States are Subject to this Rule Because They Contribute to Ozone Nonattainment?

1. Notice of Proposed Rulemaking

2. Comments and EPA Responses

3. Today's Action

F. Issues Related to Timing of the CAIR Controls

1. Overview

2. By Design, the CAIR Cap and Trade Program Will Achieve Significant Emissions Reductions Prior to the Cap Deadlines

3. Additional Justification for the SO 2 and NO X Annual Controls

4. Additional Justification for Ozone NO X Requirements

IV. What Amounts of SO 2 and NO X Emissions Did EPA Determine Should Be Reduced?

A. What Methodology Did EPA Use to Determine the Amounts of SO 2 and NO X Emissions That Must Be Eliminated?

1. The EPA's Cost Modeling Methodology

2. The EPA's Proposed Methodology to Determine Amounts of Emissions that Must be Eliminated

a. Overview of EPA Proposal for the Levels of Reductions and Resulting Caps, and their Timing

b. Regulatory History: NO X SIP Call

c. Proposed Criteria for Emissions Reduction Requirements

3. What Are the Most Significant Comments that EPA Received about its Proposed Methodology for Determining the Amounts of SO 2 and NO X Emissions that Must Be Eliminated, and What Are EPA's Responses?

4. The EPA's Evaluation of Highly Cost-Effective SO 2 and NO X Emissions Reductions Based on Controlling EGUs

a. SO 2 Emissions Reductions Requirements

b. NO X Emissions Reductions Requirements

B. What Other Sources Did EPA Consider when Determining Emission Reduction Requirements?

1. Potential Sources of Highly Cost-Effective Emissions Reductions

a. Mobile and Area Sources

b. Non-EGU Boilers and Turbines

c. Other Non-EGU Stationary Sources

C. Schedule for Implementing SO 2 and NO X Emissions Reduction Requirements for PM 2.5 and Ozone

1. Overview

2. Engineering Factors Affecting Timing for Control Retrofits

a. NPR

b. Comments

c. Responses

3. Assure Financial Stability

D. Control Requirements in Today's Final Rule

1. Criteria Used to Determine Final Control Requirements

2. Final Control Requirements

V. Determination of State Emissions Budgets

A. What Is the Approach for Setting State-by-State Annual Emissions Reductions Requirements and EGU Budgets?

1. SO 2 Emissions Budgets

a. State Annual SO 2 Emission Budget Methodology

b. Final SO 2 State Emission Budget Methodology

c. Use of SO 2 budgets

2. NO X Annual Emissions Budgets

a. Overview

b. State Annual NO X Emissions Budget Methodology

c. Final Annual State NO X Emission Budgets

d. Use of Annual NO X Budgets

e. NO X Compliance Supplement Pool

B. What Is the Approach for Setting State-by-State Emissions Reductions Requirements and EGU Budgets for States with NO X Ozone Season Reduction Requirements?

1. States Subject to Ozone-season Requirements

VI. Air Quality Modeling Approach and Results

A. What Air Quality Modeling Platform Did EPA Use?

1. Air Quality Models

a. The PM 2.5 Air Quality Model and Evaluation

b. Ozone Air Quality Modeling Platform and Model Evaluation

c. Model Grid Cell Configuration

2. Emissions Inventory Data

3. Meteorological Data

B. How Did EPA Project Future Nonattainment for PM 2.5 and 8-Hour Ozone?

1. Projection of Future PM 2.5 Nonattainment

a. Methodology for Projecting Future PM 2.5 Nonattainment

b. Projected 2010 and 2015 Base Case PM 2.5 Nonattainment Counties

2. Projection of Future 8-Hour Ozone Nonattainment

a. Methodology for Projecting Future 8-Hour Ozone Nonattainment

b. Projected 2010 and 2015 Base Case 8-Hour Ozone Nonattainment Counties

C. How did EPA Assess Interstate Contributions to Nonattainment?

1. PM 2.5 Contribution Modeling Approach

2. 8-Hour Ozone Contribution Modeling Approach

D. What Are the Estimated Interstate Contributions to PM 2.5 and 8-Hour Ozone Nonattainment?

1. Results of PM 2.5 Contribution Modeling

2. Results of 8-Hour Ozone Contribution Modeling

E. What Are the Estimated Air Quality Impacts of the Final Rule?

1. Estimated Impacts on PM 2.5 Concentrations and Attainment

2. Estimated Impacts on 8-Hour Ozone Concentrations and Attainment

F. What Are the Estimated Visibility Impacts of the Final Rule?

1. Methods for Calculating Projected Visibility in Class I Areas

2. Visibility Improvements in Class I Areas

VII. SIP Criteria and Emissions Reporting Requirements

A. What Criteria Will EPA Use to Evaluate the Approvability of a Transport SIP?

1. Introduction

2. Requirements for States Choosing to Control EGUs

a. Emissions Caps and Monitoring

b. Using the Model Trading Rules

c. Using a Mechanism Other than the Model Trading Rules

d. Retirement of Excess Title IV Allowances

3. Requirements for States Choosing to Control Sources Other than EGUs

a. Overview of Requirements

b. Eligibility of Non-EGU Reductions

c. Emissions Controls and Monitoring

d. Emissions Inventories and Demonstrating Reductions

4. Controls on Non-EGUs Only

5. Use of Banked Allowances and the Compliance Supplement Pool

B. State Implementation Plan Schedules

1. State Implementation Plan Submission Schedule

a. The EPA's Authority to Require Section 110(a)(2)(D) Submissions in Accordance with the Schedule of Section 110(a)(1)

b. The EPA's Authority to Require Section 110(a)(2)(D) Submissions Prior to Formal Designation of Nonattainment Areas under Section 107

c. The EPA's Authority to Require Section 110(a)(2)(D) Submissions Prior to State Submission of Nonattainment Area Plans Under Section 172

d. The EPA's Authority to Require Section 110(a)(2)(D) Submissions Prior to Completion of the Next Review of the PM 2.5 and 8-hour Ozone NAAQS

e. The EPA's Authority to Require States to Make Section 110(a)(2)(D) Submissions within 18 Months of this Final Rule

C. What Happens If a State Fails to Submit a Transport SIP or EPA Disapproves the Submitted SIP?

1. Under What Circumstances Is EPA Required to Promulgate a FIP?

2. What Are the Completeness Criteria?

3. When Would EPA Promulgate the CAIR Transport FIP?

D. What Are the Emissions Reporting Requirements for States?

1. Purpose and Authority

2. Pre-existing Emission Reporting Requirements

3. Summary of the Proposed Emissions Reporting Requirements

4. Summary of Comments Received and EPA's Responses

5. Summary of the Emissions Reporting Requirements

VIII. Model NO X and SO 2 Cap and Trade Programs

A. What Is the Overall Structure of the Model NO X and SO 2 Cap and Trade Programs?

B. What Is the Process for States to Adopt the Model Cap and Trade Programs and How Will It Interact with Existing Programs?

1. Adopting the Model Cap and Trade Programs

2. Flexibility in Adopting Model Cap and Trade Rules

C. What Sources Are Affected under the Model Cap and Trade Rules?

1. 25 MW Cut-off

2. Definition of Fossil Fuel-fired

3. Exemption for Cogeneration Units

a. Efficiency Standard for Cogeneration Units

b. One-third Potential Electric Output Capacity

c. Clarifying “For Sale”

d. Multiple Cogeneration Units

D. How Are Emission Allowances Allocated to Sources?

1. Allocation of NO X and SO 2 Allowances

a. Required Aspects of a State NO X Allocation Approach

b. Flexibility and Options for a State NO X Allowance Allocations Approach

E. What Mechanisms Affect the Trading of Emission Allowances?

1. Banking

a. The CAIR NPR and SNPR Proposal for the Model Rules and Input from Commenters

b. The Final CAIR Model Rules and Banking

2. Interpollutant Trading Mechanisms

a. The CAIR NPR Proposal for the Model Rules and Input from Commenters

b. Interpollutant Trading and the Final CAIR Model Rules

F. Are There Incentives for Early Reductions?

1. Incentives for Early SO 2 Reductions

a. The CAIR NPR and SNPR Proposal for the Model Rules and Input from Commenters

b. SO 2 Early Reduction Incentives in the Final CAIR Model Rules

2. Incentives for Early NO X Reductions

a. The CAIR NPR and SNPR Proposal for the Model Rules and Input from Commenters

b. NO X Early Reduction Incentives in the Final CAIR Model Rules

G. Are There Individual Unit “Opt-In” Provisions?

1. Applicability

2. Allowing Single Pollutant

3. Allocation Method for Opt-Ins

4. Alternative Opt-In Approach

5. Opting Out

6. Regulatory Relief for Opt in Units

H. What Are the Source-Level Emissions Monitoring and Reporting Requirements?

I. What is Different Between CAIR's Annual and Seasonal NO X Model Cap and Trade Rules?

J. Are There Additional Changes to Proposed Model Cap and Trade Rules Reflected in the Regulatory Language?

IX. Interactions with Other Clean Air Act Requirements

A. How Does this Rule Interact with the NO X SIP Call?

B. How Does this Rule Interact with the Acid Rain Program?

1. Legal Authority for Using Title IV Allowances in CAIR Model SO 2 Cap and trade Program

2. Legal Authority for Requiring Retirement of Excess Title IV Allowances if State Does Not Use CAIR Model SO 2 Cap and trade Program

3. Revisions to Acid Rain Regulations

C. How Does the Rule Interact With the Regional Haze Program?

1. How Does this Rule Relate to Requirements for Best Available Retrofit Technology (Bart) under the Visibility Provisions of the CAA?

a. Supplemental Notice of Proposed Rulemaking

b. Comments and EPA's Responses

c. Today's Action

2. What Improvements did EPA Make to the BART Versus CAIR Modeling, and What are the New Results?

a. Supplemental Notice of Proposed Rulemaking

b. Comments and EPA Responses

c. Today's Action

D. How Will EPA Handle State Petitions Under Section 126 of the CAA?

E. Will Sources Subject to CAIR Also Be Subject To New Source Review?

X. Statutory and Executive Order Reviews

A. Executive Order 12866: Regulatory Planning and Review

1. What Economic Analyses Were Conducted for the Rulemaking?

2. What Are the Benefits and Costs of this Rule?

a. Control Scenario

b. Cost Analysis and Economic Impacts

c. Human Health Benefit Analysis

d. Quantified and Monetized Welfare Benefits

3. How Do the Benefits Compare to the Costs of This Final Rule?

4. What are the Unquantified and Unmonetized Benefits of CAIR Emissions Reductions?

a. What are the Benefits of Reduced Deposition of Sulfur and Nitrogen to Aquatic, Forest, and Coastal Ecosystems?

b. Are There Health or Welfare Disbenefits of CAIR That Have Not Been Quantified?

B. Paperwork Reduction Act

C. Regulatory Flexibility Act

D. Unfunded Mandates Reform Act

E. Executive Order 13132: Federalism

F. Executive Order 13175: Consultation and Coordination with Indian Tribal Governments

G. Executive Order 13045: Protection of Children from Environmental Health and Safety Risks

H. Executive Order 13211: Actions that Significantly Affect Energy Supply, Distribution, or Use

I. National Technology Transfer Advancement Act

J. Executive Order 12898: Federal Actions to Address Environmental Justice in Minority Populations and Low-Income Populations

K. Congressional Review Act

L. Judicial Review

CFR Revisions and Additions (Rule Text)

Part 51

Part 72

Part 73

Part 74

Part 77

Part 78

Part 96

I. Overview Back to Top

By notice of proposed rulemaking dated January 30, 2004 and by notice of supplemental rulemaking dated June 10, 2004, EPA proposed to find that certain States must reduce emissions of SO 2 and/or NO X because those emissions contribute significantly to downwind areas in other States that are not meeting the annual PM 2.5 NAAQS or the 8-hour ozone NAAQS. [1] Today, EPA takes final action requiring 28 States and the District of Columbia to adopt and submit revisions to their State implementation plans (SIPs), under the requirements of CAA section 110(a)(2)(D), that would eliminate specified amounts of SO 2 and/or NO X emissions.

Each State may independently determine which emissions sources to subject to controls, and which control measures to adopt. The EPA's analysis indicates that emissions reductions from electric generating units (EGUs) are highly cost effective, and EPA encourages States to adopt controls for EGUs. States that do so must place an enforceable limit, or cap, on EGU emissions (see section VII for discussion). The EPA has calculated the amount of each State's EGU emissions cap, or budget, based on reductions that EPA has determined are highly cost effective. States may allow their EGUs to participate in an EPA-administered cap and trade program as a way to reduce the cost of compliance, and to provide compliance flexibility. The cap and trade programs are described in more detail in section VIII.

The EPA estimates that today's action will reduce SO 2 emissions by 3.5 million tons [2] in 2010 and by 3.8 million tons in 2015; and would reduce annual NO X emissions by 1.2 million tons in 2009 and by 1.5 million tons in 2015. [2] (These numbers are for the 23 States and the District of Columbia that are affected by the annual SO 2 and NO X requirements of CAIR.) If all the affected States choose to achieve these reductions through EGU controls, then EGU SO 2 emissions in the affected States would be capped at 3.6 million tons in 2010 and 2.5 million tons in 2015 [4] ; and EGU annual NO X emissions would be capped at 1.5 million tons in 2009 and 1.3 million tons in 2015. The EPA estimates that the required SO 2 and NO X emissions reductions would, by themselves, bring into attainment 52 of the 79 counties that are otherwise projected to be in nonattainment for PM 2.5 in 2010, and 57 of the 74 counties that are otherwise projected to be in nonattainment for PM 2.5 in 2015. The EPA further estimates that the required NO X emissions reductions would, by themselves, bring into attainment 3 of the 40 counties that are otherwise projected to be in nonattainment for 8-hour ozone in 2010, and 6 of the 22 counties that are projected to be in nonattainment for 8-hour ozone in 2015. In addition, today's rule will improve PM 2.5 and 8-hour ozone air quality in the areas that would remain nonattainment for those two NAAQS after implementation of today's rule. Because of today's rule, the States with those remaining nonattainment areas will find it less burdensome and less expensive to reach attainment by adopting additional local controls. The Clean Air Interstate Rule (CAIR) will also reduce PM 2.5 and 8-hour ozone levels in attainment areas, providing significant health and environmental benefits in all areas of the eastern US.

The EPA's CAIR and the previously promulgated NO X SIP Call reflect EPA's determination that the required SO 2 and NO X reductions are sufficient to eliminate upwind States' significant contribution to downwind nonattainment. These programs are not designed to eliminate all contributions to transport, but rather to balance the burden for achieving attainment between regional-scale and local-scale control programs.

The EPA conducted a regulatory impact analysis (RIA), entitled “Regulatory Impact Analysis for the Final Clean Air Interstate Rule (March 2005)” that estimates the annual private compliance costs (1999$) of $2.4 billion for 2010 and $3.6 billion for 2015, if all States make the required emissions reductions through the power industry. Additionally, the RIA includes a benefit-cost analysis demonstrating that substantial net economic benefits to society will be achieved from the emissions reductions required in this rulemaking. For determination of net benefits, the above private costs were converted to social costs that are lower since transfer payments, such as taxes, are removed from the estimates. The EPA analysis shows that today's action inclusive of the concurrent New Jersey and Delaware proposal will generate annual net benefits of approximately $71.4 or $60.4 billion in 2010 and $98.5 or $83.2 billion in 2015. [5] These alternate net benefit estimates reflect differing assumptions about the social discount rate used to estimate the benefits and costs of the rule. The lower estimates reflect a discount rate of 7 percent and the higher estimates a discount rate of 3 percent. In 2015, the total annual quantified benefits are $101 or $86.3 billion and the annual social costs are $2.6 or $3.1 billion—benefits outweigh costs in 2015 by a ratio of 39 to 1 or 28 to 1 (3 percent and 7 percent discount rates, respectively). These estimates do not include the value of benefits or costs that we cannot monetize.

In 2015, we estimate that PM-related annual benefits include approximately 17,000 fewer premature fatalities, 8,700 fewer cases of chronic bronchitis, 22,000 fewer non-fatal heart attacks, 10,500 fewer hospitalization admissions (for respiratory and cardiovascular disease combined) and result in significant reductions in days of restricted activity due to respiratory illness (with an estimate of 9.9 million fewer minor restricted activity days) and approximately 1,700,000 fewer work loss days. We also estimate substantial health improvements for children from reduced upper and lower respiratory illness, acute bronchitis, and asthma attacks.

Ozone health-related benefits are expected to occur during the summer ozone season (usually ranging from May to September in the Eastern U.S.). Based upon modeling for 2015, annual ozone-related health benefits are expected to include 2,800 fewer hospital admissions for respiratory illnesses, 280 fewer emergency room admissions for asthma, 690,000 fewer days with restricted activity levels, and 510,000 fewer days where children are absent from school due to illnesses.

In addition to these significant health benefits, the rule will result in ecological and welfare benefits. These benefits include visibility improvements; reductions in acidification in lakes, streams, and forests; reduced eutrophication in water bodies; and benefits from reduced ozone levels for forests and agricultural production.

Several other documents containing detailed explanations of other key elements of today's rule are also included in the docket. These include a detailed explanation of how EPA calculated the State-by-State EGU emissions budgets, and a detailed explanation of the air quality modeling analyses which support this rule. [6] Responses to comments that are not addressed in the preamble to today's rule are included in a separate document. [7]

The remaining sections of the preamble describe the final CAIR requirements and our responses to comments on many of the most important features of the CAIR. Section II, “EPA's Analytical Approach,” summarizes EPA's overall analytical approach and responds to general comments on that approach. Section III, “Why Does This Rule Focus on SO 2 and NO X, and How Were Significant Downwind Impacts Determined?,” outlines the rationale for the CAIR focus on SO 2 and NO X, which are precursors that contribute to PM 2.5 (SO 2, NO X) or ozone (NO X) transport, and the analytic approach EPA used to determine which States had large enough downwind ambient air quality impacts to become subject to today's requirements. Section IV, “What Amounts of SO 2 and NO X Emissions Did EPA Determine Should Be Reduced?,” describes EPA's methodology for determining the amounts of SO 2 and NO X emissions reductions required under today's rule. Section V, “Determination of State Emissions Budgets,” describes how EPA determined the State-by-State emissions reductions requirements and, in the event States elect to control EGUs, the State-by-State EGU emissions budgets. Section VI, “Air Quality Modeling Approach and Results,” describes the technical aspects of the air quality modeling and summarizes the numerical results of that modeling. Section VII, “SIP Criteria and Emissions Reporting Requirements,” describes the SIP submission date and other SIP requirements associated with the emissions controls that States might adopt. Section VIII, “NO X and SO 2 Model Cap and Trade Programs,” describes the EPA administered cap and trade programs that States electing to control emissions from EGUs are encouraged to adopt. Section IX, “Interactions with Other Clean Air Act Requirements,” discusses how this rule interacts with the acid rain provisions in CAA title IV, the NO X SIP Call, the best available retrofit technology (BART) requirements, and other CAA or regulatory requirements. Finally, section X, “Statutory and Executive Order Reviews,” describes the applicability of various administrative requirements for today's rule and how EPA addressed these requirements.

A. What Are the Central Requirements of This Rule?

In today's action, we establish SIP requirements for the affected upwind States under CAA section 110(a)(2). Clean Air Act section 110(a)(2)(D) requires SIPs to contain adequate provisions prohibiting air pollutant emissions from sources or activities in those States that contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to a NAAQS. Based on air quality modeling analyses and cost analyses, EPA has concluded that SO 2 and NO X emissions in certain States in the eastern part of the country, through the phenomenon of air pollution transport, [8] contribute significantly to downwind nonattainment, or interfere with maintenance, of the PM 2.5 and 8-hour ozone NAAQS. The EPA is requiring SIP revisions in 28 States and the District of Columbia to reduce SO 2 and/or NO X emissions, which are important precursors of PM 2.5 (NO X and SO 2) and ozone (NO X).

The 23 States along with the District of Columbia that must reduce annual SO 2 and NO X emissions for the purposes of the PM 2.5 NAAQS are: Alabama, Florida, Georgia, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Texas, Virginia, West Virginia, and Wisconsin.

The 25 States along with the District of Columbia that must reduce NO X emissions for the purposes of the 8-hour ozone NAAQS are: Alabama, Arkansas, Connecticut, Delaware, Florida, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Massachusetts, Michigan, Mississippi, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia, and Wisconsin. In addition to making the findings of significant contribution to nonattainment or interference with maintenance, EPA is requiring each State to make specified amounts of SO 2 and/or NO X emissions reductions to eliminate their significant contribution to downwind States. The affected States and the District of Columbia are required to adopt and submit the required SIP revision with the necessary control measures by 18 months from the signature date of today's rule.

The emissions reductions requirements are based on controls that EPA has determined to be highly cost effective for EGUs. However, States have the flexibility to choose the measures to adopt to achieve the specified emissions reductions. If the State chooses to control EGUs, then it must establish a budget—that is, an emissions cap—for those sources. Today's rule defines the EGU budgets for each affected State if a State chooses to control only EGUs. The rule also explains the emission reduction requirements if a State chooses to achieve some or all of its required emission reductions by controlling sources other than EGUs. Due to feasibility constraints, EPA is requiring emissions reductions be implemented in two phases. The first phase of NO X reductions starts in 2009 (covering 2009-2014) and the first phase of SO 2 reductions starts in 2010 (covering 2010-2014); the second phase of reductions for both NO X and SO 2 starts in 2015 (covering 2015 and thereafter). For States subject to findings of significant contribution for PM 2.5, EPA is establishing annual emissions budgets. For States subject to findings of significant contribution for 8-hour ozone, the CAIR specifies ozone-season NO X emissions budgets. States subject to findings for both PM 2.5 and ozone will have both an annual and an ozone season NO X budget.

The EPA is providing, as an option to States, model cap and trade programs for EGUs. The EPA will administer these programs, which will be governed by rules provided by EPA that States may adopt or incorporate by reference.

With respect to federally recognized Indian Tribes, the applicability of this rule is governed by three factors: The flexible regulatory framework for Tribes provided by the CAA and the Tribal Authority Rule (TAR); the absence of any existing EGUs on Tribal lands in the CAIR region; and the existence of reservations within the geographic areas which we determined to contribute significantly to nonattainment areas.

Under CAA section 301(d) as implemented by the TAR, eligible Indian Tribes may implement all, but are not required to implement any, programs under the CAA for which EPA has determined that it is appropriate to treat Tribes similarly to States. Tribes may also implement “reasonably severable” elements of programs (40 CFR 49.7(c)). In the absence of Tribal implementation of a CAA program or programs, EPA will utilize Federal implementation for the relevant area of Indian country as necessary or appropriate to protect air quality, in consultation with the Tribal government.

The TAR contains a list of provisions for which it is not appropriate to treat Tribes in the same manner as States (40 CFR 49.4). The CAIR is based on the States' obligations under CAA section 110(a)(2)(D) to prohibit emissions which would contribute significantly to nonattainment in, or interfere with maintenance by, other States due to pollution transport. Because CAA section 110(a)(2)(D) is not among the provisions we determined to be inappropriate to apply to Tribes in the same manner as States, that section is applicable, where necessary and appropriate, to Tribes.

However, among the CAA provisions not appropriate for Tribes are “[s]pecific plan submittal and implementation deadlines for NAAQS-related requirements * * *” (40 CFR 49.4(a)). Therefore, Tribes are not required to submit implementation plans under section 110(a)(2)(D). Moreover, because no Tribal lands in the CAIR region currently contain any of the sources (EGUs) on which we based the emissions reductions requirements applicable to States, there are no emission reduction requirements applicable to Tribes.

At the same time, the existence of the CAIR cap and trade program in some or all of the affected States will have implications for any future construction of EGUs on Tribal lands. The geographic scope of the CAIR cap and trade program is being determined by a two step-process: the EPA's determination of which States significantly contribute to downwind areas, and the decision by those affected States whether to satisfy their emission reduction requirement by participating in the CAIR cap and trade program.

With respect to the first step of this process (significant contribution test), notwithstanding the political autonomy of Tribes, we view the zero-out modeling as representing the entire geographic area within the State being considered, regardless of the jurisdictional status of areas within the State. Therefore, any EGU constructed in the future on a reservation within a CAIR-affected State would be located in an area which we have already determined to significantly contribute to downwind nonattainment. [9]

With respect to decisions by States to participate in the CAIR cap and trade program, because Tribal governments are autonomous, such a decision would not be directly binding for any Tribe located within the State.

Nonetheless, as a matter of a policy, cap and trade programs by their nature must apply consistently throughout the geographic region of the program in order to be effective. Otherwise, the existence of areas not covered by the cap could create incentives to locate sources there, and thereby undermine the environmental goals of the program. [10]

In light of these considerations, in the event of any future planned construction of EGUs on Tribal lands within the CAIR region, EPA intends to work with the relevant Tribal government to regulate the EGU through either a Tribal implementation plan (TIP) or a Federal implementation plan (FIP). We anticipate that at a minimum, a proposed EGU on a reservation within a State participating in the CAIR cap and trade program would need to be made subject to the cap and trade program. In the case of a new EGU on a reservation in a CAIR-affected State which chose not to participate in the cap and trade program, the new EGU might also be required, through a TIP or FIP, to participate in the program. This would depend on the potential for emissions shifting and other specific circumstances (e.g., whether the EGU would service the electric grid of States involved in the cap and trade program.) Again, EPA will work with the relevant Tribal government to determine the appropriate application of the CAIR.

Finally, as discussed in the SNPR, Tribes have objected to emissions trading programs that allocate allowances based on historic emissions, on the grounds that this rewards first-in-time emitters at the expense of those who have not yet enjoyed a fair opportunity to pursue economic development. Comments on the CAIR proposal from Tribes requested a Federal set-aside of allowances for Tribes, or other special Tribal allowance provisions. The few comments received from States on the issue generally opposed allocations based on Indian country status. One State expressed a willingness to share its emissions budget with Tribes in the event an EGU locates in Indian country.

The EPA does not believe there is sufficient information to design Tribal allocation provisions at this time. A program designed to address concerns which remain largely speculative is likely to create more problems through unintended consequences than it solves. Therefore, rather than create a Federal allowance set-aside for Tribes, EPA will work with Tribes and potentially affected States to address concerns regarding the equity of allowance allocations on a case-by-case basis as the need arises. The EPA may choose to revisit this issue through a separate rulemaking in the future.

B. Why Is EPA Taking This Action?

Emissions reductions to eliminate transported pollution are required by the CAA, as noted above. There are strong policy reasons for addressing interstate pollution transport.

1. Policy Rationale for Addressing Transported Pollution Contributing to PM 2.5 and Ozone Problems

Emissions from upwind States can alone, or in combination with local emissions, result in air quality levels that exceed the NAAQS and jeopardize the health of residents in downwind communities. Control of PM 2.5 and ozone requires a reasonable balance between local and regional controls. If significant contributions of pollution from upwind States that can be abated by highly cost-effective controls are unabated, the downwind area must achieve greater local emissions reductions, thereby incurring extra clean-up costs. Requiring reasonable controls for both upwind and local emissions sources should result in achieving air quality standards at a lesser cost than a strategy that relies solely on local controls. For all these reasons, addressing interstate transport in advance of the time that States must adopt local nonattainment plans, will make it easier for States to develop their nonattainment plans because the States will know the degree to which the pollution flowing into their nonattainment areas will be reduced.

The EPA addressed interstate pollution transport for ozone in the NO X SIP Call rule published in 1998. [11] Today's rulemaking is EPA's first attempt to address interstate pollution transport for PM 2.5. The NO X SIP Call is substantially reducing ozone transport, helping downwind areas meet the 1-hour and 8-hour ozone standards. The EPA has reassessed ozone transport in this rulemaking for two reasons. First, several years have passed since promulgation of the NO X SIP Call and updated air quality and emissions data are available. Second, some areas are expected to face substantial difficulty in meeting the 8-hour ozone standards. As a result, EPA has determined it is important to assess the degree to which ozone transport will remain a problem after full implementation of the NO X SIP Call, and to assess whether further controls are warranted to ensure continued progress toward attainment. The modeling for the CAIR includes the NO X SIP Call in the baseline and examines later years than the NO X SIP Call analyses.

a. The PM 2.5 Problem

By action dated July 18, 1997, we revised the NAAQS for particulate matter (PM) to add new standards for fine particles, using as the indicator particles with aerodynamic diameters smaller than a nominal 2.5 micrometers, termed PM 2.5 (62 FR 38652). We established health- and welfare-based (primary and secondary) annual and 24-hour standards for PM 2.5. The annual standards are 15 micrograms per cubic meter, based on the 3-year average of annual mean PM 2.5 concentrations. The 24-hour standard is a level of 65 micrograms per cubic meter, based on the 3-year average of the annual 98th percentile of 24-hour concentrations. The annual standard is generally considered the most limiting.

Fine particles are associated with a number of serious health effects including premature mortality, aggravation of respiratory and cardiovascular disease (as indicated by increased hospital admissions, emergency room visits, absences from school or work, and restricted activity days), lung disease, decreased lung function, asthma attacks, and certain cardiovascular problems such as heart attacks and cardiac arrhythmia. The EPA has estimated that attainment of the PM 2.5 standards would prolong tens of thousands of lives and would prevent, each year, tens of thousands of hospital admissions as well as hundreds of thousands of doctor visits, absences from work and school, and respiratory illnesses in children.

Individuals particularly sensitive to fine particle exposure include older adults, people with heart and lung disease, and children. More detailed information on health effects of fine particles can be found on EPA's Web site at: http://www.epa.gov/ttn/naaqs/standards/pm/s_pm_index.html.

At the time EPA established the PM 2.5 primary NAAQS in 1997, we also established welfare-based (secondary) NAAQS identical to the primary standards. The secondary standards are designed to protect against major environmental effects caused by PM such as visibility impairment—including in Class I areas which include national parks and wilderness areas across the country—soiling, and materials damage.

As discussed in other sections of this preamble, SO 2 and NO X emissions both contribute to fine particle concentrations. In addition, NO X emissions contribute to ozone problems, described in the next section. We believe the CAIR will significantly reduce SO 2 and NO X emissions that contribute to the PM 2.5 and 8-hour ozone problems described here.

The PM 2.5 ambient air quality monitoring for the 2001-2003 period shows that areas violating the standards are located across much of the eastern half of the United States and in parts of California, and Montana. Based on these nationwide data, 82 counties have at least one monitor that violates either the annual or the 24-hour PM 2.5 standard. Most areas violate only the annual standard; a small number of areas violate both the annual and 24-hour standards; and no areas violate just the 24-hour standard. The population of these 82 counties totals over 56 million people.

Only two States in the western part of the U.S., California and Montana, have counties that exceeded the PM 2.5 standards. On the other hand, in the eastern part of the U.S., 124 sites in 69 counties (with total population of 34 million) violated the annual PM 2.5 standard of 15.0 micrograms per cubic meter (μg/m 3) over the 3-year period from 2001 to 2003, while 469 sites met the annual standard. No sites in the eastern part of the United States exceeded the daily PM 2.5 standard of 65 μg/m 3. The 69 violating counties are located in a region made up of 16 States (plus the District of Columbia), extending eastward from St. Louis County, Missouri, the western-most violating county and including the following States: Alabama, Delaware, Georgia, Illinois, Indiana, Kentucky, Maryland, Missouri, Michigan, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Tennessee, West Virginia, and the District of Columbia. The EPA published the PM 2.5 attainment and nonattainment designations on January 5, 2005 (70 FR 944). The designations will be effective on April 5, 2005.

Because interstate transport is not believed to be a significant contributor to exceedances of the PM 2.5 standards in California or Montana, today's final CAIR does not cover these States.

b. The 8-Hour Ozone Problem

By action dated July 18, 1997, we promulgated identical revised primary and secondary ozone standards that specified an 8-hour ozone standard of 0.08 parts per million (ppm). Specifically, under the standards, the 3-year average of the fourth highest daily maximum 8-hour average ozone concentration may not exceed 0.08 ppm. In general, the revised 8-hour standards are more protective of public health and the environment and more stringent than the pre-existing 1-hour ozone standards. All areas that were violating the 1-hour ozone standard at the time of the 8-hour ozone designations were also designated as nonattainment for the 8-hour ozone standard. More areas do not meet the 8-hour standard than do not meet the 1-hour standard. The EPA published the 8-hour ozone attainment and nonattainment designations in the Federal Register on April 30, 2004 (69 FR 23858). The designations were effective on June 15, 2004. Pursuant to EPA's final rule to implement the 8-hour ozone standard (69 FR 23951; April 30, 2004), EPA will revoke the 1-hour ozone standard on June 15, 2005, 1 year after the effective date of the 8-hour designations.

Short-term (1- to 3-hour) and prolonged (6- to 8-hour) exposures to ambient ozone have been linked to a number of adverse health effects. Short-term exposure to ozone can irritate the respiratory system, causing coughing, throat irritation, and chest pain. Ozone can reduce lung function and make it more difficult to breathe deeply. Breathing may become more rapid and shallow than normal, thereby limiting a person's normal activity. Ozone also can aggravate asthma, leading to more asthma attacks that require a doctor's attention and the use of additional medication. Increased hospital admissions and emergency room visits for respiratory problems have been associated with ambient ozone exposures. Longer-term ozone exposure can inflame and damage the lining of the lungs, which may lead to permanent changes in lung tissue and irreversible reductions in lung function. A lower quality of life may result if the inflammation occurs repeatedly over a long time period (such as months, years, a lifetime).

People who are particularly susceptible to the effects of ozone include children and adults who are active outdoors, people with respiratory diseases, such as asthma, and people with unusual sensitivity to ozone.

In addition to causing adverse health effects, ozone affects vegetation and ecosystems, leading to reductions in agricultural crop and commercial forest yields; reduced growth and survivability of tree seedlings; and increased plant susceptibility to disease, pests, and other environmental stresses (e.g., harsh weather). In long-lived species, these effects may become evident only after several years or even decades and have the potential for long-term adverse impacts on forest ecosystems. Ozone damage to the foliage of trees and other plants can also decrease the aesthetic value of ornamental species used in residential landscaping, as well as the natural beauty of our national parks and recreation areas. The economic value of some welfare losses due to ozone can be calculated, such as crop yield loss from both reduced seed production (e.g., soybean) and visible injury to some leaf crops (e.g., lettuce, spinach, tobacco), as well as visible injury to ornamental plants (i.e., grass, flowers, shrubs). Other types of welfare loss may not be quantifiable (e.g., reduced aesthetic value of trees growing in heavily visited national parks). More detailed information on health effects of ozone can be found at the following EPA Web site: http://www.epa.gov/ttn/naaqs/standards/ozone/s_o3_index.html.

Almost all areas of the country have experienced some progress in lowering ozone concentrations over the last 20 years. As reported in the EPA's report, “The Ozone Report: Measuring Progress Through 2003,” [12] national average levels of 1-hour ozone improved by 29 percent between 1980 and 2003 while 8-hour levels improved by 21 percent over the same time period. The Northeast and West regions have shown the greatest improvement since 1980. However, most of that improvement occurred during the first part of the period. In fact, during the most recent 10 years, ozone levels have been relatively constant reflecting little if any air quality improvement. For this reason, ozone has exhibited the slowest progress of the six major pollutants tracked nationally.

Although ambient ozone levels remained relatively constant over the past decade, additional control requirements have reduced emissions of the two major ozone precursors, VOC and NO X, although at different rates. Emissions of VOCs were reduced by 32 percent from 1990 levels, while emissions of NO X declined by 22 percent.

Ozone remains a significant public health concern. Presently, wide geographic areas, including most of the nation's major population centers, experience unhealthy ozone levels, that is, concentrations violating the NAAQS for 8-hour ozone. These areas include much of the eastern part of the United States and large areas of California. More specifically, 297 counties with a total population of over 124 million people currently violate the 8-hour ozone standard. Most of these ozone violations occur in the eastern half of the United States: 268 counties with a population of over 93 million.

When ozone and PM 2.5 are examined jointly, 322 counties with 131 million people are violating at least one of the standards while 57 counties nationwide have concentrations violating both standards with a total population of over 49 million people. Of these, 46 counties with a population of over 28 million are in the Eastern United States.

c. Other Environmental Effects Associated With SO 2 and NO X Emissions

Today's action will result in benefits in addition to the enumerated human health and welfare benefits resulting from reductions in ambient levels of PM 2.5 and ozone. Reductions in NO X and SO 2 will contribute to substantial visibility improvements in many parts of the Eastern U.S. where people live, work, and recreate, including Federal Class I areas such as the Great Smoky Mountains. Reductions in these pollutants will also reduce acidification and eutrophication of water bodies in the region. In addition, reduced mercury emissions are anticipated as a result of this rule. Reduced mercury emissions will lessen mercury contamination in lakes and thereby potentially decrease both human and wildlife exposure to mercury-contaminated fish.

2. The CAA Requires States To Act as Good Neighbors by Limiting Downwind Impacts

The CAA includes the “good neighbor” provision of section 110(a)(2)(D), which requires that every SIP prohibit emissions from any source or other type of emissions activity in amounts that will contribute significantly to nonattainment in any downwind State, or that will interfere with maintenance in any downwind State. In today's action, EPA is determining that 28 States and the District of Columbia, all in the eastern part of the United States, have emissions of SO 2 and/or NO X that will contribute significantly to nonattainment, or interfere with maintenance, of the PM 2.5 NAAQS and/or the 8-hour ozone NAAQS in another State. Under EPA's general authority to clarify the applicability of CAA requirements, as provided in CAA section 301(a)(1), EPA is establishing the amount of SO 2 and NO X emissions that each affected State must prohibit by submitting appropriate SIP provisions to EPA. The improvements in air quality will assist downwind States in developing their SIPs to provide for attainment and maintenance in those nonattainment areas.

3. Today's Rule Will Improve Air Quality

The EPA has estimated the improvements in emissions and air quality that would result from implementing the CAIR. These improvements, which are substantial, are summarized earlier in this section.

C. What Was the Process for Developing This Rule?

By action dated January 30, 2004, EPA issued a proposal that included many of the components of today's action. “Rule to Reduce Interstate Transport of Fine Particulate Matter and Ozone (Interstate Air Quality Rule); Proposed Rule,” (69 FR 4566). The Administrator signed the proposed rule—termed, at that time, the Interstate Air Quality Rule—on December 17, 2003, and EPA posted it on its Web site for this rule on that date. The Web site address at that time was http://www.epa.gov/interstateairquality. (The address has since changed to http://www.epa.gov/cleanairinterstaterule/ or http://www.epa.gov/cair/.)

The EPA held public hearings on the proposal, in conjunction with a proposed rulemaking concerning mercury and other hazardous air pollutants from EGUs, on February 25-26, 2004, in Chicago, Illinois; Philadelphia, Pennsylvania; and Research Triangle Park, North Carolina. The comment period for the NPR closed on March 30, 2004. The EPA received over 6,700 comments on the proposal.

By action dated June 10, 2004, EPA issued a supplemental notice of proposed rulemaking (SNPR), “Supplemental Proposal for the Rule to Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule); Proposed Rule,” (69 FR 32684). The Administrator signed the SNPR for this rule—now called the Clean Air Interstate Rule—on May 18, 2004, and EPA placed it on the Web site on that date. The SNPR included, among other things, proposed regulatory language for the rule, revised proposals concerning State-level emissions budgets, proposed State reporting requirements and SIP approvability criteria, and proposed model cap and trade rules. The SNPR also proposed that under certain circumstances the CAIR requirements could replace the BART requirements of CAA sections 169A and 169B. The EPA held a public hearing on the SNPR on June 3, 2004, in Alexandria, Virginia. The comment period for the SNPR closed on July 26, 2004. The EPA received over 400 comments on the SNPR.

By a notice of data availability (NODA) dated August 6, 2004, EPA announced the availability of additional documents for this action. “Availability of Additional Information Supporting the Rule To Reduce Interstate Transport of Fine Particulate Matter and Ozone (Clean Air Interstate Rule),” (69 FR 47828). The documents had been placed on the website on or about July 27, 2004, and in the EDOCKET on that date, or shortly thereafter. The EPA allowed public comment on those additional documents until August 27, 2004. Around 30 comments were received on the NODA.

The EPA has responded to all significant public comments either in this preamble or in the response to comment document which is contained in the docket.

Comments on Rulemaking Process: Some commenters expressed concerns about certain aspects of this process. One concern was that EPA did not allow sufficient time to comment on the SNPR. Commenters noted that important program elements—including regulatory language—appeared for the first time in the SNPR, but EPA held a public hearing on the SNPR 7 days before the SNPR was published in the Federal Register and only 16 days after the SNPR had been posted on the website. The EPA believes that the 16-day period preceding the public hearing, and the total of 45 days to comment on the SNPR following its publication in the Federal Register, constituted an adequate opportunity for members of the public to comment on the SNPR.

Commenters also expressed concern that certain technical documents were not made available in sufficient time to comment. However, EPA had placed all technical support documents for the NPR in the EDOCKET as of the date of publication of the NPR, and all technical support documents for the SNPR had been placed in the EDOCKET as of the date of publication of the SNPR.

Commenters also expressed concern that in the SNPR, EPA proposed significant changes to other regulatory programs. The EPA agrees that the SNPR did include proposed changes to certain regulatory programs, i.e., the requirements for BART under CAA sections 169A and 169B (concerning visibility), certain provisions (primarily concerning the allowance-holding requirement) in the title IV (Acid Rain Program) rules, and certain emissions reporting rules under the NO X SIP Call (40 CFR 51.122) and Consolidated Emissions Reporting Rule (CERR) (title 40, part 51, subpart A). The EPA believes that to the extent the requirements for BART and emissions reporting rule revisions are tied to the CAIR, affected members of the public had adequate notice of those revisions. (These revisions are described in section VII.) However, the SNPR contained some revisions to the emissions reporting rules that were not tied to the transport provisions. The EPA is not taking final action today on the proposal for the emissions reporting rules that were not tied to the transport provisions and instead is issuing a new proposal for them, which will provide additional notice and opportunity to comment.

Further, the Acid Rain Program rule revisions, although connected to the CAIR, apply to all persons subject to the Acid Rain Program, including persons who are not affected by the CAIR. (These revisions are described in section IX.) Specifically, as explained in section IX, the revisions to the Acid Rain Program rules are aimed at facilitating coordination of the Acid Rain Program and the CAIR model SO 2 cap and trade rule and/or are being adopted on their own merits, independently of the need to coordinate with the CAIR. Most of the proposed revisions involve changing from unit-by-unit to source-by-source compliance with the allowance-holding requirement of the Acid Rain Program and therefore affect every source subject to the Acid Rain Program, whether or not the source is also in a State covered by the CAIR. The change to source-by-source compliance increases a source's flexibility to use—in meeting the allowance-holding requirement—allowances held by any unit at the source. This flexibility reduces the likelihood that sources will incur large excess emissions penalties from inadvertent, minor errors (e.g., in how allowances are distributed among the units at the source), while preserving the environmental goals of the Acid Rain Program. The remaining revisions to the Acid Rain Program rules similarly cover all Acid Rain Program sources. Indeed, none of the comments on the proposed Acid Rain Program rule revisions stated that the revisions would apply only to certain Acid Rain Program sources, but rather seemed to treat the revisions as applying program-wide. As discussed in section IX, EPA is finalizing, with minor modifications, the Acid Rain Program rule revisions.

Commenters also expressed concern that between the NPR and the SNPR, EPA had proposed program elements in a piecemeal fashion, which made it more difficult to comprehend and comment on the rule, and that the SNPR's comment period was too short to allow the public adequate opportunity to comment on the numerous and complex issues raised in that proposal. The EPA recognizes the challenges faced by commenters in this rulemaking, however, we believe that the comment periods for the NPR and SNPR were adequate, and note that we did receive extensive and highly detailed, technical comments on both proposals.

D. What Are the Major Changes Between the Proposals and the Final Rule?

The EPA is finalizing a number of revisions to the proposed elements of the CAIR. These revisions are in response to information received in public comments and new analyses conducted by EPA. The following is a summary list of those changes:

  • The first phase of NO X reductions starts in 2009 (covering 2009-2014) instead of 2010. The first phase of the SO 2 reductions still starts in 2010 (covering 2010-2014).
  • The emissions inventories used for PM 2.5 and 8-hour ozone air quality modeling have been updated and improved; we modeled PM 2.5 using the Community Multiscale Air Quality Model (CMAQ) and meteorology for 2001 instead of the Regional Model for Simulating Aerosols and Deposition (REMSAD) and meteorology for 1996.
  • The final CAIR does not cover Kansas based on new analyses of its contribution to downwind PM 2.5 nonattainment.
  • Arkansas, Delaware, Massachusetts, and New Jersey are not subject to the CAIR based on their contribution to PM 2.5 nonattainment and maintenance. However, they remain subject to NO X emissions reductions requirements on the basis of their contribution to downwind 8-hour ozone nonattainment. This requirement is for the ozone season rather than the entire year. The EPA is issuing a new proposal to include Delaware and New Jersey for the PM 2.5 NAAQS based on additional considerations.
  • The change in States covered by the rule necessitates a re-analysis of the NO X budgets for all covered States. This changes the amount of the budget, but not the procedure EPA used to calculate it.
  • The SIP approval criteria have been changed to no longer exclude measures otherwise required by the CAA from being included in the State's compliance with CAIR.
  • A 200,000 ton compliance supplement pool was added for NO X. Allowances from this pool can either be awarded to sources that make early reductions or to sources that demonstrate need.
  • All States for which EPA has made a finding with respect to ozone are subject to an ozone season cap. In order to implement this ozone season cap, EPA has finalized an ozone season NO X trading program in addition to the annual NO X and SO 2 trading programs that were proposed.
  • A number of changes were made to the trading rule including: changes to the model NO X allocation methodology (to fuel weight allocations) and the addition of opt in provisions.
  • The EPA is not finalizing some of the emissions reporting requirements in response to public comments indicating we gave inadequate notice of the changes that were proposed to be applicable to all States, not just those affected by the CAIR emission reduction requirements. These are being reproposed, with modifications, in a separate action to allow additional opportunity for public comment by all affected States and other parties.

II. The EPA's Analytical Approach Back to Top

Overview: Today's rulemaking is based on the “good neighbor” provision of CAA section 110(a)(2)(D), which requires States to develop SIP provisions assuring that emissions from their sources do not contribute significantly to downwind nonattainment, or interfere with maintenance, of the NAAQS. The EPA interpreted this provision, and developed a detailed methodology for applying it, in the NO X SIP Call rulemaking, which concerned interstate transport of ozone precursors.

Today's rule requires upwind States to submit SIP revisions requiring their sources to reduce emissions of certain precursors that significantly contribute to nonattainment in, or interfere with maintenance of, the PM 2.5 and 8-hour ozone national ambient air quality standards in downwind States. The EPA developed today's rule relying heavily on the NO X SIP Call approach.

This section of the preamble outlines the key aspects of today's approach, some of which are described in greater detail in other sections of the preamble. The EPA received comments on today's approach that we respond to either in this section or in the other sections of the preamble. This section also describes how today's approach varies from the NO X SIP Call, which variations result from, among other things, the fact that today's action regulates a different pollutant (PM 2.5) with a different precursor (SO 2).

A. How Did EPA Interpret the Clean Air Act's Pollution Transport Provisions in the NO X SIP Call?

1. Clean Air Act Requirements

The central CAA provisions concerning pollutant transport, for purposes of today's action, are found in section 110(a)(2)(D). Under these provisions, each SIP must—

(D) Contain adequate provisions

(i) Prohibiting * * * any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will—

(I) Contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to any * * * national primary or secondary ambient air quality standard * * *.

2. The NO X SIP Call Rulemaking

Promulgated by action dated October 27, 1998, the NO X SIP Call was EPA's principal effort to reduce interstate transport of precursors for both the 1-hour ozone NAAQS and the 8-hour ozone NAAQS. (See “Finding of Significant Contribution and Rulemaking for Certain States in the Ozone Transport Assessment Group Region for Purposes of Reducing Regional Transport of Ozone; Rule,” (63 FR 57356).) In that rulemaking, EPA imposed seasonal NO X reduction requirements on 22 States and the District of Columbia in the eastern part of the country.

a. Analytical Approach of NO X SIP Call

In the NO X SIP Call, EPA interpreted section 110(a)(2)(D) to authorize EPA to determine the amount of emissions in upwind States that “contribute significantly” to downwind nonattainment or “interfere with” downwind maintenance, and to require those States to eliminate that amount of emissions. The EPA recognized that States must retain full authority to choose the sources to control, and the control mechanisms, to achieve those reductions.

The EPA set out several criteria or factors for the “contribute significantly” test, and further indicated that the same criteria should apply to the “interfere with maintenance” provision: [13]

* * * EPA determined the amount of emissions that significantly contribute to downwind nonattainment from sources in a particular upwind State primarily by (i) evaluating, with respect to each upwind State, several air quality related factors, including determining that all emissions from the State have a sufficiently great impact downwind (in the context of the collective contribution nature of the ozone problem); and (ii) determining the amount of that State's emissions that can be eliminated through the application of cost-effective controls. Before reaching a conclusion, EPA evaluated several secondary, and more general, considerations. These include:

  • The consistency of the regional reductions with the attainment needs of the downwind areas with nonattainment problems.
  • The overall fairness of the control regimes required of the downwind and upwind areas, including the extent of the controls required or implemented by the downwind and upwind areas.
  • General cost considerations, including the relative cost-effectiveness of additional downwind controls compared to upwind controls.

63 FR 57403

i. Air Quality Factor

The first factor concerns evaluating the impact on downwind air quality of the upwind State's emissions. As EPA stated in the NO X SIP Call:* * *

EPA specifically considered three air quality factors with respect to each upwind State * * *.

  • The overall nature of the ozone problem (i.e., “collective contribution”).
  • The extent of the downwind nonattainment problems to which the upwind State's emissions are linked, including the ambient impact of controls required under the CAA or otherwise implemented in the downwind areas.
  • The ambient impact of the emissions from the upwind State's sources on the downwind nonattainment problems.

63 FR 57376

The EPA explained the first factor, collective contribution, by noting,

[V]irtually every nonattainment problem is caused by numerous sources over a wide geographic area* * *[. This] factor suggest[s] that the solution to the problem is the implementation over a wide area of controls on many sources, each of which may have a small or unmeasureable ambient impact by itself.

63 FR 57377

The second air quality factor—the extent of downwind nonattainment problems—concerns whether downwind areas should be considered to be in nonattainment. This determination took into account the then-current air quality of the area, the predicted future air quality (assuming the implementation of required controls, but not the transport requirements that were the subject of the NO X SIP Call), and the boundaries of the area in light of designation status (63 FR 57377).

The EPA applied the third air quality factor—the ambient impact of emissions from the upwind sources—by projecting the amount of the upwind State's entire inventory of anthropogenic emissions to the year 2007, and then quantifying, through the appropriate air quality modeling techniques, the impact of those emissions on downwind nonattainment. [14] Specifically, (i) EPA determined the minimum threshold impact that the upwind State's emissions must have on a downwind nonattainment area to be considered potentially to contribute significantly to nonattainment; and then (ii) for States with impacts above that threshold, EPA developed a set of metrics for further evaluating the contribution of the upwind State's emissions on a downwind nonattainment area (63 FR 57378). The EPA considered a State with emissions that had a sufficiently great impact to contribute significantly to the downwind area (depending on application of the cost factor). In general, EPA established the thresholds at a relatively low level, which reflected the collective contribution phenomenon. That is, because the ozone problem is caused by many relatively small contributions, even relatively small contributors must participate in the solution.

ii. Cost Factor

The cost factor is the second major factor that EPA applied to determine the significant contribution to nonattainment: “EPA * * * determined whether any amounts of the NO X emissions may be eliminated through controls that, on a cost-per-ton basis, may be considered to be highly cost effective.” (See 63 FR 57377.)

(I) Choice of Highly Cost-Effective Standard

The EPA selected the standard of highly cost effective in order to assure State flexibility in selecting control strategies to meet the emissions reduction requirements of the rulemaking. That is, the rulemaking required the States to achieve specified levels of emissions reductions—the levels achievable if States implemented the control strategies that EPA identified as highly cost effective—but the rulemaking did not mandate those highly cost-effective control strategies, or any other control strategy. Indeed, in calculating the amount of the required emissions reductions by assuming the implementation of highly cost-effective control strategies, EPA assured that other control strategies—ones that were cost effective, if not highly cost effective—remained available to the States.

(II) Determination of Highly Cost-Effective Amount

The EPA determined the dollar amount considered to be highly cost effective by reference to the cost effectiveness of recently promulgated or proposed NO X controls. The EPA determined that the average cost effectiveness of controls in the reference list ranged up to approximately $1,800 per ton of NO X removed (1990$), on an annual basis. The EPA considered the controls in the reference list to be cost effective.

The EPA established $2,000 (1990$) in average cost effectiveness for summer ozone season emissions reductions as, at least directionally, the highly cost-effective amount. Identifying this amount on an ozone season basis was appropriate because the NO X SIP Call concerned the ozone standard, for which emissions reductions during only the summer ozone season are necessary. This level of costs reflected the fact that in general, States with downwind ozone nonattainment areas had already implemented extensive controls. Accordingly, it was evident that the level of upwind controls EPA selected would prove necessary for the downwind areas to reach attainment.

(III) Source Categories

The EPA then determined that the source categories for which highly cost-effective controls were available included EGUs, large industrial boilers and turbines, and cement kilns. At the same time, EPA determined, for those source categories, the level of controls that would cost an amount consistent with the highly cost-effective amount and that would be feasible. The EPA considered other source categories, but found that highly cost-effective controls were not available from them for various reasons, including the size of the sources, the relatively small amount of emissions from the sources, or the control costs.

iii. Other Factors

The EPA also relied on several other, secondary considerations before concluding that the identified amount of emissions reductions were required. The first concerned the consistency of regional reductions with downwind attainment needs. The EPA ascertained the ozone air quality impacts of the required emissions reductions, and determined that those impacts improved air quality downwind, but not to the point that would raise questions about whether the amount of reductions was more than necessary (63 FR 57379).

The second general consideration was “the overall fairness of the control regimes” to which the downwind and upwind areas were subject. The EPA explained:

Most broadly, EPA believes that overall notions of fairness suggest that upwind sources which contribute significant amounts to the nonattainment problem should implement cost-effective reductions. When upwind emitters exacerbate their downwind neighbors' ozone nonattainment problems, and thereby visit upon their downwind neighbors additional health risks and potential clean-up costs, EPA considers it fair to require the upwind neighbors to reduce at least the portion of their emissions for which highly cost-effective controls are available.

In addition, EPA recognizes that in many instances, areas designated as nonattainment under the 1-hour NAAQS have incurred ozone control costs since the early 1970s. Moreover, virtually all components of their NO X and VOC inventories are subject to SIP-required or Federal controls designed to reduce ozone. Furthermore, these areas have complied with almost all of the specific control requirements under the CAA, and generally are moving towards compliance with their remaining obligations. The CAA's sanctions and FIP provisions provide assurance that these remaining controls will be implemented. By comparison, many upwind States in the midwest and south have had fewer nonattainment problems and have incurred fewer control obligations.

(63 FR 57379.)

The third general consideration was “general cost considerations.” The EPA noted that “in general, areas that currently have, or that in the past have had, nonattainment problems * * * have already incurred ozone control costs.” The next set of controls available to these nonattainment areas would be more expensive than the controls available to the upwind areas. The EPA found that this cost scenario further confirmed the reasonableness of the upwind control obligations (63 FR 57379).

In the NO X SIP Call, EPA considered all of these factors together in determining the level of controls considered to be highly cost effective. This level of controls reflected the then-present state of ozone controls: Within the region, the nonattainment areas were already required to—and had already implemented—VOC and NO X controls that covered much of their inventory. However, the upwind States in the region generally had not done so (except to the extent of their ozone nonattainment areas). In this context, EPA considered it reasonable to impose an additional control burden on the upwind States. Air quality modeling showed that even with this additional level of upwind controls, residual nonattainment remained, so that further reductions from downwind and/or upwind areas would be necessary.

b. Regulatory Requirements

After ascertaining the controls that qualified as highly cost effective, EPA developed a methodology for calculating the amount of NO X emissions that each State was required to reduce on grounds that those emissions contribute significantly to nonattainment downwind. The total amount of required NO X emissions reductions was the sum of the amounts that would be reduced by application of highly cost-effective controls to each of the source categories for which EPA determined that such controls were available (63 FR 57378).

The largest of these source categories was EGUs. The EPA determined the amount of reductions associated with EGU controls by applying the control rate that EPA considered to reflect highly cost-effective controls to each State's EGU heat input. That heat input, in turn, was adjusted to reflect projected growth.

Each affected State retained the authority to achieve the required level of reductions by implementing whatever controls on whatever sources it wished, and EPA determined that there were other source categories for which cost-effective, if not highly cost-effective, controls were available (63 FR 57378). If the States chose to control EGUs, then the NO X SIP Call mandated certain requirements—including a statewide cap on EGU NO X emissions—but also made available an EPA-administered regionwide EGU allowance trading program that the States could choose to adopt.

c. SIP Submittal and Implementation Requirements

At the time EPA promulgated the NO X SIP Call, States already had SIPs for the 1-hour ozone NAAQS in place. In the NO X SIP Call, EPA determined that the 1-hour SIPs for the affected States were deficient, and EPA called on these States, under CAA section 110(k)(5), to submit, within 12 months of promulgation of the NO X SIP Call, SIP revisions to cure the deficiency by complying with the NO X SIP Call regulatory requirements. The EPA further required that the NO X SIP Call-required controls be implemented as expeditiously as practicable. The EPA determined this date to be within 3 years of the SIP submittal date (with that period extended to the beginning of the next ozone season), in light of the various constraints that EGUs would confront in implementing controls.

For the SIPs due under the 8-hour ozone NAAQS, in the NO X SIP Call, EPA did not incorporate a section 110(k)(5) SIP call, but instead required States to submit, under section 110(a)(1)-(2), SIP revisions to fulfill the requirements of section 110(a)(2)(D). The EPA required these 8-hour ozone SIPs to be submitted—and the controls mandated therein to be implemented—on the same schedule as the 1-hour SIPs.

However, EPA stayed the 8-hour ozone requirements of the NO X SIP Call, due to litigation concerning the 8-hour ozone NAAQS. To date, EPA has not lifted that stay.

3. Michigan v. EPA Court Case

Petitioners brought legal challenges to various components of the NO X SIP Call's analytical approach in the United States Court of Appeals for the District of Columbia Circuit, in Michigan v. EPA, 213 F.3d 663 (DC Cir., 2000), cert. denied, 532 U.S. 904 (2001). The Court upheld the essential features of the air quality modeling part of EPA's approach, id. at 673; as well as EPA's definition of “contribute significantly” to include the factor of highly cost-effective controls, id. at 679. The Court did vacate or remand certain specific applications of EPA's approach, and delayed the implementation date to May 31, 2004. See, e.g., id. at 67, 681-85, 692-94. In addition, in a subsequent case that reviewed separate EPA rulemakings making technical corrections to the NO X SIP Call, the DC Circuit remanded for a better explanation EPA's methodology for computing the growth component in the EGU heat input calculation. Appalachian Power Co. v. EPA, 251 F.3d 1026 (DC Cir., 2001). [15]

4. Implementation of the NO X SIP Call

The court decisions left intact most of the NO X SIP Call requirements. All States subject to those requirements—which EPA has termed the NO X SIP Call Phase I requirements—submitted SIPs incorporating them, and requiring control implementation by May 31, 2004 or earlier. The EPA has approved those SIPs.

The EPA responded to the DC Circuit's EGU growth remand decisions through a Federal Register action that provided a more detailed explanation and other supporting information for the EGU growth methodology (67 FR 21868; May 1, 2002). The Court subsequently upheld that explanation. West Virginia v. EPA, 362 F.3d 861 (DC Cir. 2004). In addition, by action dated April 21, 2004, EPA promulgated a rulemaking that responded to other remanded and vacated issues, and included the remaining requirements—termed the NO X SIP Call Phase II requirements—for the affected States (69 FR 21604).

B. How Does EPA Interpret the Clean Air Act's Pollution Transport Provisions in Today's Rule?

1. CAIR Analytical Approach

Today, EPA adopts much the same interpretation and application of section 110(a)(2)(D) for regulating downwind transport of precursors of PM 2.5 and 8-hour ozone as EPA adopted for the NO X SIP Call. We are adjusting some aspects of the NO X SIP Call analytic approach for various reasons, including the need to account for regulation of a different pollutant (PM 2.5) with an additional precursor (SO 2).

a. Nature of Nonattainment Problem and Overview of Today's Approach

As described in section I, above, the interstate transport component of current nonattainment of the PM 2.5 and 8-hour ozone NAAQS is primarily confined to the eastern part of the country, although in an area that is larger, by several States, than the area that EPA focused on in the NO X SIP Call for only ozone. As described in section III, it is evident that local controls alone cannot be counted on to solve the nonattainment problems, although uncertainties remain in the state of knowledge of these nonattainment problems as well as the precise role interstate and local controls should play. As in the case of the NO X SIP Call, it is not reasonable to expect a local area to bear the entire burden of solving the air quality problems, even if doing so were technically possible.

Turning to the interstate component of the nonattainment problems, as discussed in section III below, for PM 2.5, we find sufficient information is available to address the adverse downwind impacts caused by SO 2 and NO X, and to develop emissions reductions requirements for SO 2 and NO X. However, we do not have sufficient information to address other precursors. As discussed in section III below, for 8-hour ozone, we reiterate the finding of the NO X SIP Call that NO X emissions, and not VOC emissions, are of primary importance for interstate transport purposes.

We interpret CAA section 110(a)(2)(D) to require SIPs in upwind States to eliminate the amounts of emissions that contribute significantly to downwind nonattainment or interfere with downwind maintenance. As described below, in today's rule, EPA determines that upwind States' emissions contribute significantly to nonattainment or interfere with maintenance of the PM 2.5 NAAQS.

To quantify the amounts of those emissions that contribute significantly to nonattainment, we primarily focus on the air quality factor reflecting the upwind State's ambient impact on downwind nonattainment areas, and the cost factor of highly cost-effective controls. However, as with the NO X SIP Call, EPA also considers other factors, which serve to establish the broad context for applying the air quality and cost factors. Today, we adopt the formulation of those factors as described in the CAIR NPR, which has little conceptual difference from EPA's application of those factors in the NO X SIP Call.

Discussion of issues relating to maintenance are found in section III below.

b. Air Quality Factor

i. PM 2.5

With respect to the PM 2.5 NAAQS, as described in section VI, we employed air quality modeling techniques to assess the impact of each upwind State's entire inventory of anthropogenic SO 2 and NO X emissions on downwind nonattainment and maintenance. For air quality and technical reasons described below, EPA determined that upwind SO 2 and NO X emissions contribute significantly to nonattainment as of the year 2010. Therefore, EPA projected SO 2 and NO X emissions to the year 2010, assuming certain required controls (but not controls required under CAIR), and then modeled the impact of those projected emissions (termed the base case inventory) on downwind PM 2.5 nonattainment in that year.

As discussed in section III, we adopt today a threshold air quality impact of 0.2 μg/m 3, so that an upwind State with contributions to downwind nonattainment below this level would not be subject to regulatory requirements, but a State with contributions at or higher than this level would be subject to further evaluation.

Because of the inherent differences between the PM 2.5 and ozone NAAQS, this threshold necessarily differs from the threshold chosen for the NO X SIP Call in terms of: (i) The metrics selected to evaluate the threshold, and (ii) the specific level of the threshold. Even so, the threshold EPA proposed for PM 2.5 is generally consistent with the approach taken in the NO X SIP Call for the threshold level for ozone in that both are relatively low. This level reflects the fact that PM 2.5 nonattainment, like ozone, is caused by many sources in a broad region, and therefore may be solved only by controlling sources throughout the region. As with the NO X SIP Call, the collective contribution condition of PM 2.5 air quality is reflected in the proposed relatively low threshold. [16]

The EPA determined that as of 2010, 23 upwind States and the District of Columbia will have contributions to downwind PM 2.5 nonattainment areas that are sufficiently high to meet the air quality factor of the transport test.

ii. 8-Hour Ozone

With respect to the 8-hour ozone NAAQS, we also employed, as described in section VI, air quality modeling techniques to assess the impact of each upwind State's entire inventory of NO X and VOC emissions on downwind nonattainment. The EPA determined that upwind NO X emissions contribute significantly to 8-hour ozone nonattainment as of the year 2010. Therefore, EPA projected NO X emissions to the year 2010, assuming certain required controls (but not controls required under CAIR), and then modeled the impact of those projected emissions (termed the base case inventory) on downwind 8-hour ozone nonattainment in that year.

For the 8-hour ozone air quality factor, EPA employs the same threshold amounts and metrics that it used in the NO X SIP Call. That is, as described in section VI, emissions from an upwind State contribute significantly to nonattainment if the maximum contribution is at least 2 parts per billion, the average contribution is greater than one percent, and certain other numerical criteria are met.

The EPA determined that as of 2010, 25 upwind States and the District of Columbia will have contributions to downwind nonattainment areas that are sufficiently high to meet the air quality factor of the transport test.

c. Cost Factor

The second major factor that EPA applies is the cost factor. As in the case of the NO X SIP Call, EPA interprets this factor as mandating emissions reductions in amounts that would result from application of highly cost-effective controls. We ascertain the level of costs as highly cost effective by reference to the cost effectiveness of recent controls. As we stated in the CAIR NPR, in determining the appropriate level of controls, we considered feasibility issues—as we did in the NO X SIP Call—specifically, “the applicability, performance, and reliability of different types of pollution control technologies for different types of sources; * * * and other implementation costs of a regulatory program for any particular group of sources.” (See CAIR NPR, 69 FR 4585.)

As described in section IV, today we conclude that at present, EGUs are the only source category for which highly cost-effective SO 2 and NO X controls are available. In making this determination, we examined what information is available concerning which source categories emit relatively large amounts of emissions, and what difficulties sources have in implementing controls. These criteria are similar to those considered in the NO X SIP Call.

As discussed in section IV, for PM 2.5, today's action finalizes our proposal to identify as highly cost effective the dollar amount of cost effectiveness that falls near the low end of the reference range for both annual SO 2 controls and annual NO X controls. We identify this level based on the overall context of the PM 2.5 implementation program, discussed below.

For upwind States affecting downwind 8-hour ozone nonattainment areas, we apply the cost factor for ozone-season NO X controls in much the same manner as for the NO X SIP Call, although some aspects of the analysis have been updated. The level of NO X control identified as highly cost effective is more stringent than in the NO X SIP Call.

d. Other Factors

As with the NO X SIP Call, EPA considers other factors that influence the application of the air quality and cost factors, and that confirm the conclusions concerning the amounts of emissions that upwind States must eliminate as contributing significantly to downwind nonattainment. Specifically, as we stated in the CAIR NPR, “We are striving in this proposal to set up a reasonable balance of regional and local controls to provide a cost effective and equitable governmental approach to attainment with the NAAQS for fine particles and ozone.” (See 69 FR 4612.) In this manner, we broadly incorporate the fairness concept and relative-cost-of-control (regional costs compared to local costs) concept that we generally considered in the NO X SIP Call.

i. PM 2.5 Controls

For PM 2.5, we promulgated the NAAQS in 1997, we issued designations of areas in December 2004 (70 FR 944; January 5, 2005), and we intend to promulgate implementation requirements during 2005. We project that by 2010, without CAIR or other controls not already adopted, 80 counties in the CAIR region would be in nonattainment of the annual standard.

Our state of knowledge is incomplete as to the best control regime to achieve attainment and maintenance of this NAAQS in individual areas, but we do know that transported SO 2 and NO X emissions are important contributors to PM 2.5 nonattainment. In addition, we have concluded that available controls for at least the portion of these emissions from EGUs are feasible and relatively inexpensive on a cost-per-ton basis, and generate significant ambient benefits. These ambient benefits include bringing many areas into attainment and decreasing PM 2.5 levels in the rest of the nonattainment areas. Moreover, available information indicates that local controls are likely to be relatively more expensive on a per-ton basis, and will not reduce emissions sufficiently to bring many areas into attainment.

In light of this information, we plan to proceed by requiring the level of regulatory control specified today on upwind SO 2 and NO X emissions. We consider today's action to be both prudent and effective within the circumstances of the developing PM 2.5 implementation program. This action is one of the initial steps in implementing the PM 2.5 NAAQS. States, localities, and Tribes, as well as EPA, will continue to evaluate the efficacy of local controls. Finally, as discussed in section VI, air quality modeling confirms that these regional controls are not more than is necessary for downwind areas to attain.

This overall plan is well within the ambit of EPA's authority to proceed with regulation on a step-by-step basis. The time frame for section 110(a)(2)(D) SIPs, described in section VII, makes clear that EPA has the authority to establish the upwind reduction obligations before having full information about how best to achieve attainment goals, including having full information about downwind control costs and the efficacy of downwind control measures.

ii. Ozone Controls

The EPA determined the level of required NO X reductions for purposes of 8-hour ozone transport through much the same process as for purposes of PM 2.5 transport.

e. Regulatory Requirements

i. Annual SO 2 and NO X Emissions Reductions

Although EPA determined that upwind emissions will contribute significantly to both PM 2.5 nonattainment and 8-hour ozone nonattainment in 2010, the amount of requisite emissions controls cannot feasibly be implemented by 2009 for NO X, or 2010 for SO 2. Instead, EPA has determined to implement the reductions in two phases for each pollutant: 2009 for NO X, and 2010 for SO 2 initially, with lower caps for both in 2015.

As described in section IV, EPA evaluated the cost of emissions reductions under consideration against the level of highly cost-effective controls. Through a multi-year process involving studies and other regulatory and legislative efforts, as well as involvement with citizen, industry, and State stakeholders, EPA arrived at an amount of SO 2 emissions reductions for evaluation purposes for the CAIR region. The EPA ascertained the costs of these reductions and today determines that they should be considered highly cost effective. These amounts correspond to reducing Title IV SO 2 allowances for utilities by 65 percent in 2015 and 50 percent in 2010 in CAIR States.

As described in section V, EPA further determined that these emissions reductions requirements should be allocated to the States in proportion to the title IV SO 2 allowances allocated under the CAA to their EGUs. This approach is consistent with the system Congress established for allocating title IV allowances and facilitates implementation of the SO 2 interstate trading program.

For annual NO X emissions, EPA determined a target regionwide amount of both emissions reductions and the EGU budget by multiplying current heat input by emission rates of 0.125 lb/mmBtu and 0.15 lb/mmBtu for 2015 and 2010, respectively. The EPA then evaluated those amounts through the Integrated Planning Model (IPM), which indicated the associated amounts of heat input and emission rates projected for those years. The IPM indicated that the amounts of heat input for 2015 and 2010 were higher than current heat input (in light of the increased electricity demand for 2015 and 2010), and that the emissions rates were lower than 0.125 lb/mmBtu (2015) and 0.15 lb/mmBtu (2010). The IPM calculated the costs to achieve those emissions reductions and EGU budget (assuming EGU controls) by 2015 and 2009, which costs EPA determined were highly cost effective and feasible, respectively. The EPA used this same approach to determine the seasonal budget for NO X reductions for purposes of the ozone standard.

As described in section V, we allocated this regionwide amount to the individual States in accordance with their average heat input from EGUs both subject to and not subject to title IV. We adjusted heat input for type of fuel used. The EPA believes that this method is a reasonable indicator of each State's appropriate share of the requirements. This method differs from what EPA used in the NO X SIP Call, which relied on State-specific projections of growth in heat input.

We require implementation of the PM 2.5 and 8-hour ozone reductions in two phases, in 2009 and 2015. As discussed in section IV, these dates are the most expeditious that are practicable—the same standard for the implementation period in the NO X SIP Call—based on engineering and financial factors; the performance and applicability of control measures; and the impact of implementation on, in the case of EGUs, electricity reliability. The EPA considered these same factors in determining the implementation period for the NO X SIP Call requirements, but factual differences lead to the two-phase approach adopted in today's action.

As discussed in section VII, each upwind State may achieve the required reductions by regulating any sources of SO 2 or NO X that it wishes. However, if the State chooses to regulate certain source categories (such as EGUs), it must comply with certain requirements (such as capping EGU emissions), and it may take advantage of certain opportunities (such as participation in the EPA-administered EGU cap and trade program). Some aspects of these requirements and the cap and trade program differ from those in the NO X SIP Call, as explained in section VIII. However, like the NO X SIP Call, the State may allow sources to opt in to the CAIR trading program, as described in section VIII.

f. SIP Submittal and Implementation Requirements

Today EPA requires that the PM 2.5 and 8-hour ozone SIPs be submitted within 18 months of promulgation of today's action. This period is 6 months longer than the SIPs due under the NO X SIP Call. This difference is due to the fact that PM 2.5 implementation is only now beginning, and it makes sense to keep the NO X SIPs due under the 8-hour ozone requirements on the same schedule as the NO X and SO 2 SIPs due under the PM 2.5 requirements.

2. What Did Commenters Say and What Is EPA's Response?

Many of the comments on today's action concern various aspects of EPA's analytical approach. Most of those comments are discussed elsewhere in today's action. Comments on the most basic elements of EPA's approach are discussed here.

a. Aspects of Contribute-Significantly Test

i. Date for Evaluation of Downwind Impacts

Comment: Some commenters took issue with EPA's approach of determining the upwind State's air quality impact on downwind areas by modeling only the State's 2010 base case emissions (that is, projected 2010 emissions before the 2010 CAIR controls). These commenters stated that although evaluating the upwind State's base case emissions in 2010 might indicate whether that State's air quality impact on downwind areas is sufficiently high to justify imposition of the 2010 (Phase I) controls, it does not justify imposition of the 2015 (Phase II) controls. Rather, according to the commenters, EPA should conduct further air quality modeling that evaluates the upwind State's 2015 base case emissions—taking into account the CAIR 2010 controls but not the CAIR 2015 controls—to determine whether the State continues (even after imposition of the CAIR 2010 controls) to have a sufficient downwind ambient impact to justify the 2015 controls.

Commenters added that, in their view, PM 2.5 precursors generally were decreasing after 2010, the PM 2.5 nonattainment problem was generally diminishing as well, and the contribution of some upwind States to downwind areas was relatively small. These facts, according to the commenters, indicated that some upwind States should not be subject to the 2015 reductions requirement.

Some commenters stated, more broadly, that the threshold contribution level selected by EPA should be considered a floor, so that upwind States should be obliged to reduce their emissions only to the level at which their contribution to downwind nonattainment does not exceed that threshold level.

Response: The EPA views the CAIR emission reduction requirements as a single action, but one that cannot be fully implemented in 2009 (for NO X) or 2010 (for SO 2), and must instead be partially deferred until 2015, solely for reasons of feasibility. Under these circumstances, EPA does not believe it appropriate to re-evaluate the 2015 component, as commenters have suggested.

Under EPA's approach, which mirrors that of the NO X SIP Call, EPA projects, for each upwind State, SO 2 and NO X inventories, as of 2010, taking into account controls required under other CAA provisions and controls adopted by State and local agencies. The EPA then uses air quality modeling techniques to determine the impact of these emissions on downwind air quality. The EPA then requires upwind States whose emissions have a sufficiently high impact to eliminate the amount of their emissions that could be eliminated through application of highly cost-effective controls. These emissions reductions must be implemented as expeditiously as practicable. Were it feasible to implement all the reductions by 2009 (for NO X) or 2010 (for SO 2), EPA would so require. Because part of the emissions reductions cannot feasibly be implemented until 2015, EPA is requiring today's two-phase approach. This analytic method is the same as for the NO X SIP Call, except that in that rulemaking all of the required emissions reductions could feasibly be implemented in one phase.

As in the case of the NO X SIP Call, EPA takes the view that once a State's emissions are determined to contribute to downwind nonattainment by at least a threshold amount, then the upwind State should reduce its emissions by the amount that would result from implementation of highly cost-effective controls. This approach is justified by the benefits of reducing the upwind contribution to downwind nonattainment, coupled with the relatively low costs. However, EPA does consider the ambient impacts of the required emissions reductions. For today's action, air quality modeling indicates that the regionwide emissions reductions do not reduce PM 2.5 levels beyond what is needed for attainment and maintenance. (See also section III below.) Most important for present purposes, as long as the controls yield downwind benefits needed to reduce the extent of nonattainment, the controls should not be lessened simply because they may have the effect of reducing the upwind State's contribution to below the initial threshold.

The DC Circuit, in upholding the NO X SIP Call, rejected similar arguments to those raised by commenters (Michigan v. EPA, 213 F.3d at 679). In the NO X SIP Call rulemaking, commenters argued that EPA's analytic approach to the “contribute significantly” test was flawed because it meant that States with different impacts downwind would nevertheless have to implement the same level of controls (i.e., those that were highly cost effective). Commenters urged EPA to recast its approach by limiting an upwind State's emissions reductions to the point at which the remaining emissions no longer caused a downwind ambient impact above the threshold level for significance. (“Responses to Significant Comments on the Proposed Finding of Significant Contribution and Rulemaking for Certain States in the Ozone Transport Assessment Group (OTAG) Region for Purposes of Reducing Regional Transport of Ozone (62 FR 60318; November 7, 1997 and 63 FR 25902; May 11, 1998),” U.S. E.P.A. (September 1998), Docket Number A-96-56-VI-C-1, at 213-16.)

Petitioners challenging the NO X SIP Call in Michigan v. EPA used the same arguments to contend that EPA's analytic approach in the NO X SIP Call was arbitrary and capricious. The Court dismissed these arguments, stating:

* * * EPA required that all of the covered jurisdictions, regardless of amount of contribution, reduce their NO X by an amount achievable with “highly cost-effective controls.” Petitioners claim that EPA's uniform control strategy is irrational. * * * [T]hey observe that where two states differ considerably in the amount of their respective NO X contributions to downwind nonattainment, under the EPA rule even the small contributors must make reductions equivalent to those achievable by highly cost-effective measures. This of course flows ineluctably from the EPA's decision to draw the “significant contribution” line on a basis of cost differentials. Our upholding of that decision logically entails upholding this consequence.

(Michigan v. EPA, 213 F.3d at 679.)

Thus, the Court approved EPA's approach of requiring the same control level on all affected States, without concern as to the arguably inconsistent ambient impacts that may result. By the same token, in today's action, EPA's approach should be accepted notwithstanding that the upwind controls could, at least in theory, result in an ambient impact that is below the initial threshold. For this reason, there is no basis to conduct a separate evaluation of the 2015 controls.

ii. Residual Nonattainment

Comment: A commenter expressed concern that too many areas will remain out of attainment for the PM 2.5 and 8-hour ozone NAAQS even after implementation of the CAIR rule.

Response: Section 110(a)(2)(D) of the CAA requires upwind States to prohibit the amount of emissions that contribute significantly to downwind nonattainment, but does not require the upwind States to prohibit sufficient emissions to assure that the downwind areas attain. Rather, downwind areas continue to bear the responsibility of addressing remaining nonattainment.

iii. Relationship of Reductions to Attainment Dates

Comment: Some commenters, who viewed the CAIR as imposing unduly light obligations on upwind States, argued that because States with nonattainment areas must develop SIPs that provide for attainment regardless of the cost of the requisite controls, and because the courts have viewed attainment deadlines as central to the CAA, EPA should require that upwind emissions contributing to downwind nonattainment must be eliminated by the downwind attainment dates, and not later.

Other commenters, who viewed the CAIR as imposing unduly heavy obligations on upwind States, argued that EPA had no authority to require upwind emissions reductions after the downwind attainment dates because by that time, the upwind emissions were no longer contributing to nonattainment. These commenters further argued that EPA has no authority to accelerate the emissions reductions because the controls could not feasibly be implemented by an earlier date.

Response: We note first that part of this issue is moot since EPA is requiring NO X controls in 2009, within the statutory time periods for attainment. With respect to remaining issues, EPA's interpretation and application of the “contribute significantly to nonattainment” standard of section 110(a)(2)(D) is not necessarily constrained by the downwind area's attainment date in either manner suggested by the commenters.

First, although it is true that the nonattainment area requirements and deadlines in CAA title I, part D, mean that the downwind area must achieve attainment by its attainment date without regard to the feasibility of emissions reductions from sources in that nonattainment area, section 110(a)(2)(D) by its terms does not apply those constraints to sources in the upwind States. Rather, EPA's interpretation of the “contribute significantly to nonattainment” standard—which incorporates feasibility considerations in determining the implementation period for the upwind emissions controls—continues to apply.

Often, upwind emissions reductions affect at least several downwind areas with different attainment dates. The EPA does not read section 110(a)(2)(D) to require that the pace of upwind reductions be controlled by the earliest downwind attainment date. Rather, EPA views the pace of reductions as being determined by the time within which they may feasibly be achieved. In some cases, upwind sources are themselves in a nonattainment area that has a longer attainment date than the downwind area, and it may not be feasible for those upwind sources to implement reductions prior to the downwind attainment date. Therefore, the upwind emissions may be projected to continue to affect adversely nonattainment in the downwind area even after the downwind attainment date, in the manner described above. Further, emissions reductions after the attainment date may be important to prevent interference with maintenance of the standards.

The CAIR will achieve substantial reductions in time to help many nonattainment areas attain the standards by the applicable attainment dates. The design of the SO 2 program, including the declining caps in 2010 and 2015 and the banking provisions, will steadily reduce SO 2 emissions over time, achieving reductions in advance of the cap dates; and the 2009 and 2015 NO X reductions will be timely for many downwind nonattainment areas. Although many of today's nonattainment areas will attain before all the reductions required by CAIR will be achieved, it is clear that CAIR's reductions will still be needed through 2015 and beyond. The EPA has determined that each upwind State's 2010 and 2015 emissions reductions will be necessary because, for purposes of both PM 2.5 and 8-hour ozone, we reasonably predict that a downwind receptor linked to that upwind State will either: (i) Remain in nonattainment and continue to experience significant contribution to nonattainment from the upwind State's emissions; or (ii) attain the relevant NAAQS but later revert to nonattainment due, for example, to continued growth of the emissions inventory. This is discussed in detail in section III below.

iv. Factors To Consider in Future Rulemaking

In the January and June CAIR proposals, we discussed regional control requirements and budgets based on a showing of “significant contribution” by upwind States to nonattainment in downwind States (69 FR at 4611-13, 32720). The CAA section 110(a)(2)(D), which provides the authority for CAIR, states among other things that SIPs must contain adequate provisions prohibiting, consistent with the CAA, sources or other types of emissions activity within a State from emitting pollutants in amounts that will “contribute significantly to nonattainment in, or interfere with maintenance by, any other State with respect to” the NAAQS. In the CAIR, EPA has interpreted section 110(a)(2)(D) to require that certain States reduce emissions by specified amounts, and has determined those amounts based on the availability of highly cost effective controls for identified source categories. Following this interpretation, EPA has calculated CAIR's emissions reduction requirements based on the availability of highly cost-effective reductions of SO 2 and NO X from EGUs in States that meet EPA's proposed inclusion criteria.

One approach cited in the January 2004 CAIR proposal for ensuring that both the air quality component and the cost effectiveness component of the section 110 “contribute significantly” determination is met, is to consider a source category's contribution to ambient concentrations above the attainment level in all nonattainment areas in affected downwind states. Id. In the June supplemental proposal, we requested comment on a further refinement of this concept—i.e., whether a source category should be included in a broad regional rule promulgated pursuant to section 110(a)(2)(D) only if the proposed level of additional control of that category would meet a specified threshold. Under that approach, EPA said it might determine, for example, that in the context of a broad multi-state SIP call, emissions reductions from particular source category are “highly cost effective” only if emissions reductions from that source category would result in at least 0.5 percent of U.S. counties and/or parishes coming into attainment with a NAAQS. The EPA noted that, given the number of counties and parishes in the United States, this requirement would be met if at least 16 counties were brought into attainment with a NAAQS as a result of the proposed level of control on a particular source category.

The Agency received comments both supporting and opposing the adoption of this test as a part of the “highly cost effective” component of the “contribute significantly” requirement of CAA section 110(a)(2)(d). Commenters supporting this test asserted that it was consistent with the CAA's overall focus on State, rather than federal, control over which sources should be regulated, and also was consistent with ensuring that broad, regional SIP calls, such as the one at issue in this case, focus only on source categories the control of which will result in substantial overall improvements in air quality. Commenters opposing this screen with respect to the application of section 110(a)(2)(D) asserted, in general, that the test would be inconsistent with the analysis used by the Agency in the NO X SIP call and with the language of section 110(a)(2)(D).

We have determined that it is not appropriate to adopt a statutory interpretation embodying a “bright line” rule that 0.5 percent of the U.S. counties and/or parishes must be brought from nonattainment into attainment from controlling emissions from a particular source category, in order for reductions from that source category to be considered highly cost effective. We continue to believe, however, that broad multi-state rules under section 110(a)(2)(D), such as the one we are finalizing today, should play a limited role under the CAA and must be justified by a careful evaluation of the air quality improvement that will result from the controls under consideration. Therefore, we intend to undertake any future broad, multi-state rulemakings under section 110(a)(2)(D) regarding transported emissions only when, as here, they produce substantial air quality benefits across a broad area and have beneficial air quality impacts on a significant number of downwind nonattainment areas, including bringing many areas into attainment. We do not at this time anticipate the need for any such rulemakings in the future. We believe that today's action, coupled with current and upcoming national rules and local or subregional programs adopted by States, will be sufficient to address the remaining nonattainment problems.

In evaluating whether to undertake national or regional transport rulemakings in the future, we believe it is not only appropriate but necessary to consider the effectiveness of the proposed emissions reductions in improving downwind air quality. We believe it will be reasonable to initiate a broad multi-state rulemaking under section 110(a)(2)(D) based on a determination that particular emissions reductions are highly cost effective only when those reductions will bring a significant number of downwind areas into attainment. In adopting this approach for determining whether a future broad, multi-state SIP call is appropriate, we note that other CAA mechanisms, such as SIP disapproval authority and State petitions under section 126, are available to address more isolated instances of the interstate transport of pollutants.

The EPA projects that control of SO 2 and NO X through CAIR will bring 72 counties into attainment with the PM 2.5 and ozone NAAQS. The total number represents approximately 3 percent of the counties/parishes in the United States, and is clearly a significant number of areas. What will be considered a significant number of areas in any future cases will need to be determined on a case-by-case basis.

III. Why Does This Rule Focus on SO 2 and NO X, and How Were Significant Downwind Impacts Determined? Back to Top

This section discusses the basis for EPA's decision to require reductions in upwind emissions of SO 2 and NO X to address PM 2.5 transport and to require reductions in upwind emissions of NO X to address ozone-related transport. In addition, this section discusses how EPA determined which States are subject to today's rule because their sources' emissions will significantly contribute to nonattainment of the PM 2.5 or 8-hour ozone standards, or interfere with maintenance of those standards, in downwind States. The EPA assessed individual upwind States' ambient impacts on downwind States and established a threshold value to identify those States whose impact constitutes a significant contribution to air quality violations in the downwind States. The EPA used air quality modeling of emissions in each State to estimate the ambient impacts. The technical issues concerning the modeling platform and approach are discussed in section VI, Air Quality Modeling Approach and Results. Also, EPA considered the potential for upwind state emissions to interfere with maintenance of the PM 2.5 and 8-hour ozone NAAQS in downwind areas.

A. What Is the Basis for EPA's Decision To Require Reductions in Upwind Emissions of SO 2 and NO X To Address PM 2.5 Related Transport?

1. How Did EPA Determine Which Pollutants Were Necessary To Control To Address Interstate Transport for PM 2.5?

a. What Did EPA Propose Regarding This Issue in the NPR?

Section II of the January 2004 proposal summarized key scientific and technical aspects of the occurrence, formation, and origins of PM 2.5, as well as findings and observations relevant to formulating control approaches for reducing the contribution of transport to fine particle problems (69 FR 4575-87). Key concepts and provisional conclusions drawn from this discussion can be summarized as follows: [17]

(1) Fine particles (measured as PM 2.5 for the NAAQS) consist of a diverse mixture of substances that vary in size, chemical composition, and source. The PM 2.5 includes both “primary” particles that are emitted directly to the atmosphere as particles, and “secondary” particles that form in the atmosphere through chemical reactions from gaseous precursors. The major components of fine particles in the Eastern U.S. can be grouped into five categories: carbonaceous material (including both primary and secondary organic carbon and black carbon), sulfates, nitrates, ammonium, and crustal material, which includes suspended dust as well as some other directly emitted materials. The major gaseous precursors of PM 2.5 include SO 2, NO X, ammonia (NH 3), and certain volatile organic compounds.

(2) Examination of urban and rural monitors indicate that in the Eastern U.S., sulfates, carbonaceous material, nitrates, and ammonium associated with sulfates and nitrates are typically the largest components of transported PM 2.5, while crustal material tends to be only a small fraction.

(3) Atmospheric interactions among particulate ammonium sulfates and nitrates and gas phase nitric acid and ammonia vary with temperature, humidity, and location. Both ambient observations and modeling simulations suggest that regional SO 2 reductions are effective at reducing sulfate and associated ammonium, and, therefore, PM 2.5. Under certain conditions reductions in particulate ammonium sulfates can release ammonia as a gas, which then reacts with gaseous nitric acid to form nitrate particles, a phenomenon called “nitrate replacement.” In such conditions SO 2 reductions would be less effective in reducing PM 2.5, unless accompanied by reductions in NO X emissions to address the potential increase in nitrates.

(4) Reductions in ammonia can reduce the ammonium, but not the sulfate portion of sulfate particles. The relative efficacy of reducing nitrates through NO X or ammonia control varies with atmospheric conditions; the highest particulate nitrate concentrations in the East tend to occur in cooler months and regions. At present, our knowledge about sources, emissions, control approaches, and costs is greater for NO X than for ammonia. Existing programs to reduce NO X from stationary and mobile sources are well underway. From a chemical perspective, as NO X reductions accumulate relative to ammonia, the atmospheric chemical system would move towards an equilibrium in which ammonium nitrate reductions become more responsive to further NO X reductions relative to ammonia reductions.

(5) Much less is known about the sources of regional transport of carbonaceous material. Key uncertainties include how much of this material is due to biogenic as compared to anthropogenic sources, and how much is directly emitted as compared to formed in the atmosphere.

(6) Observational evidence suggests that the substantial reductions in SO 2 emissions in the eastern U.S. since 1990 have indeed caused observed reductions in PM 2.5 sulfate. The relatively small historical reductions in NO X emissions do not allow observations to be used similarly to test the effectiveness of NO X reductions.

Based on the understanding of current scientific and technical information, as well as EPA's air quality modeling, as summarized in the January 30 proposal, EPA concluded that it was both appropriate and necessary to focus on control of SO 2 and NO X emissions as the most effective approach to reducing the contribution of interstate transport to PM 2.5.

The EPA proposed not to control emissions that affect other components of PM 2.5, noting that “current information relating to sources and controls for other components identified in transported PM 2.5 (carbonaceous particles, ammonium, and crustal materials) does not, at this time, provide an adequate basis for regulating the regional transport of emissions responsible for these PM 2.5 components.” (69 FR 4582). For all of these components, the lack of knowledge of and ability to quantify accurately the interstate transport of these components limited EPA's ability to include these components in this rule.

b. How Does EPA Address Public Comments on Its Proposal To Address SO 2 and NO X Emissions and Not Other Pollutants?

i. Overview of Comments on This Issue

A large number of commenters including states, affected industries, environmental groups, academics, and other members of the public agreed with EPA's proposal to require cost-effective multipollutant reductions of SO 2 and NO X to address interstate transport contributions to PM 2.5 problems. Fewer commenters who supported controlling SO 2 and NO X commented on inclusion of additional pollutants, but several also agreed that it would be premature at this time to require control of emissions of other chemical components and precursors to address such transport. These commenters suggested that SO 2 and NO X emissions from EGUs and other sources indeed contribute significantly to downwind PM 2.5. They argued that control of other components is premature because of a lack of knowledge, either about the interstate contributions of other components or of control measures for these components. Generally, EPA accepts and agrees with these conclusions.

A number of commenters disagreed to varying degrees with part or all of EPA's proposed focus on SO 2 and NO X. The main points raised by these commenters can be grouped as follows:

(1) The focus on SO 2 and NO X is not appropriate because sulfates and nitrates may not be (or are not) the most important determinants of the health effects of PM 2.5.

(2) The EPA should mandate, or at least permit, states to control other precursors and particle emissions in addition to, or instead of, SO 2 and NO X. Commenters sometimes made specific recommendations with respect to additional pollutants, including carbonaceous (including organic) particles and precursors, ammonia, and other direct emissions, including crustal material.

(3) The focus on SO 2 may be appropriate, but the basis for requiring NO X control is not clear.

ii. Summary of EPA's Response to the Major Comments on This Issue

The following subsections summarize both key comments and EPA's responses organized by the major categories outlined above. As noted in Section I, EPA has developed and placed in the rulemaking docket a detailed response to these and other public comments.

(a) SO 2 and NO X May Be Less Important to Health Than Other Transport-Related Components

Comment: Several commenters argued that the proposed focus on SO 2 and NO X was premature, citing the potential for differential toxicity of various PM 2.5 components, and in some cases advancing evidence (e.g., the Electric Power Research Institute Aerosol Research and Inhalation Studies [ARIES]) [18] that other components such as organic particles appear to be more responsible for health effects of particles than sulfates and nitrates. Several argued that the relative contribution of components to health impacts is an important uncertainty that should be researched more carefully before proposing to control only SO 2 and NO X.

Response: Today's rulemaking establishes requirements for SIP submissions under section 110(a)(2)(D). Those SIP submissions must prohibit emissions that contribute significantly to nonattainment of a NAAQS in a downwind State. The EPA determined in the 1997 rulemaking promulgating the PM 2.5 NAAQS that specified levels of PM 2.5 adversely affect human health, and that sulfates and nitrates are components of PM 2.5 (62 FR 38652, July 18, 1997). SO 2 and NO X, in turn, are precursors to fine particulate sulfates and nitrates. Comments that sulfates and nitrates do not cause adverse health effects are more appropriately raised in the context of past or ongoing reviews of the PM NAAQS. Because today's action forms part of implementing and not establishing the PM NAAQS, comments relating to the evidence supporting or not supporting health effects of all or portions of pollutants regulated by the PM 2.5 NAAQS are not germane to this rulemaking.

Nevertheless, we discuss briefly EPA's current response regarding the contributions of different components of PM 2.5 to health effects. In establishing the current PM 2.5 NAAQS, EPA found that there was ample evidence to associate various health effects with the measured mass concentration of particles smaller than a nominal 2.5 micrometers (um), termed PM 2.5. The EPA recognizes that the toxicity of different chemical components of PM 2.5 may vary, and that the observed effects may be the result of the mixture of particles and gases. While research is underway to better identify whether some chemical components are more responsible for health effects than others, results now available from such research are limited and inconclusive. A number of studies included in the recent EPA PM criteria document [19] have found effects to be associated with one or more of the major components and sources of PM 2.5, including sulfates, nitrates, organic materials, PM 2.5 mass, coal combustion, and mobile sources. The criteria document concludes that these studies suggest that many different chemical components of fine particles and a variety of different types of source categories are all linked to premature mortality and other serious health effects, either independently or in combinations, but that it is not possible to reach clear conclusions about differential effects of PM components. Accordingly, individual studies or groups of studies such as ARIES cannot be used to single out any particular component of PM 2.5 as wholly responsible (or not at all responsible) for the array of health effects that have been found to be associated with various chemical and mass indicators of fine particles. Other Federal agencies and EPA continue to promote and support the epidemiological and toxicological studies needed to better understand the effects of different chemical components and different size particles on health effects.

In the meantime, EPA believes that, given the substantial evidence of significant health effects of fine particles, it is important to move forward expeditiously to address both transported and local sources of all the major components of fine particles in an effort to implement and attain the PM 2.5 standards. Today's rule is focused on the contribution of interstate transport of nitrate and sulfates to PM 2.5 in nonattainment areas. However, EPA has already adopted other rules that are reducing emissions and exposures to these and other major components of fine particles on a national, regional, and local basis. Recent national mobile rules and programs, in particular, have focused on carbonaceous materials emitted from gasoline and both highway and non-road diesel powered mobile sources (65 FR 6698; 66 FR 5002; 69 FR 38958). States with nonattainment areas will also be required to address local sources of PM 2.5 in order to meet progress and attainment requirements. Together, the collective effect of these programs ensures a balanced approach to reducing all of the major components of PM 2.5 from transported and local sources.

(b) Inclusion of Other PM 2.5 Precursors and Components

Comment: A number of commenters recommended that EPA either mandate or at least permit controls on the emissions that cause interstate transport of other components of PM 2.5, in addition to or as a substitute for, SO 2 and NO X controls. Several commenters recommended that EPA include emissions reductions related to the components of PM 2.5 other than sulfate and nitrate. While many commenters suggested addressing all of the important contributors to PM 2.5, including those not regulated under this Rule, others highlighted only one or two additional components as most important to include. Of the PM 2.5 components, direct emissions and precursors to carbonaceous PM 2.5 and ammonia emissions were the omitted contributors most frequently discussed.

Some of these commenters argued that, by limiting the rule to SO 2 and NO X and excluding other sources of ambient PM 2.5, EPA would be limiting the choices that states have to address their downwind interstate transport contributions. These commenters argued that this limitation is contrary to the CAA, which generally gives states the discretion to choose their own emission control strategies. Commenters further asserted that the roles of other components in PM 2.5 are sufficiently well understood that they should be included in state SIPs for PM 2.5 transport, and could partially satisfy the PM 2.5 reductions anticipated by this rule.

Response: The three main classes of PM 2.5 precursors that are not included in this rulemaking are carbonaceous material (including both primary emissions and VOC emissions that form secondary organic aerosol), ammonia, and crustal material. As noted in the proposal(69 FR 4576) and as mentioned in several comments, these components comprise a measurable faction of PM 2.5 throughout the Eastern U.S., and the contribution of carbonaceous material, in particular, is often substantial. In addition, emissions contributing to these components in one state likely do affect PM 2.5 concentrations in other states to some extent. However, the extent of those downwind contributions to nonattainment has not been quantified adequately and current scientific understanding makes such a determination more uncertain than is the case for SO 2 and NO X. Responses to recommendations for including each of these three classes in the transport rule are summarized below.

(i) Carbonaceous Material

For carbonaceous material, uncertainties in both the quantity and origins of emissions contributing to both primary and secondary carbonaceous material on regional scales (including emissions from fires and from biogenic sources) limit the quality of regional scale modeling of carbonaceous PM 2.5. This in turn causes substantial uncertainties in determining the amount of interstate transport from carbonaceous material and of the costs and effectiveness of emission controls. Modeling and monitoring the relative amount of organic particles that come from the formation of secondary organic particles, versus primary organic particles, is also highly uncertain.

In addition, comparison of urban and nearby rural PM composition monitors [20] in the eastern U.S. find a significantly larger amount of carbonaceous materials in urban areas as compared to rural areas, suggesting that a substantial fraction of carbonaceous particles in urban areas come from local sources. By contrast, urban and non-urban monitors in the East show greater homogeneity for regional sulfate concentrations as compared to carbonaceous materials, suggesting regional sources are most important for sulfates. Results for nitrates suggest both a mixture of regional and local sources. Furthermore, as noted above and in the proposal (69 FR 4577-78), while the relative contributions of different sources to regional sulfate and nitrates can be quantified with certainty, the contributions of different sources to carbonaceous materials on a regional scale are less clear. Moreover, as noted in the NPR preamble, some research into mechanisms of formation of organic particles suggests that both NO X and SO 2 reductions might be of some benefit in lowering the amount of secondary organic particles. [21] Current models are not, however, capable of quantifying such potential benefits.

While EPA does not believe that enough is known about the relative effectiveness or costs of reducing anthropogenic sources of carbonaceous particles on transported PM 2.5, EPA agrees that control of known source categories of these materials can have a significant benefit in reducing the significant local contribution. For this reason, EPA has already enacted other national rules that will reduce emissions of primary carbonaceous PM 2.5 from mobile sources, the largest contributor to such emissions. In addition to reducing PM 2.5 in nonattainment areas, these regulations will also have the benefit of reducing a large measure of whatever interstate transport of carbonaceous PM 2.5 occurs.

(ii) Ammonia

While current models are able to address the major chemical mechanisms involving particulate ammonium compounds, regional-scale ammonia emissions, particularly from agricultural sources, are highly uncertain. [22] Given the relative lack of experience in controlling such sources, the costs and effectiveness of actions to reduce regional ammonia emissions are not adequately quantified at present. As noted above, ammonium would not exist in PM 2.5 if not for the presence of sulfuric acid or nitric acid; hence, decreases in SO 2 and NO X can be expected ultimately to decrease the ammonium in PM 2.5 as well. The additional regional limits on SO 2 and NO X emissions outlined in today's notice added to those reductions provided under current programs would likewise be expected to reduce the PM 2.5 effectiveness of any ammonia control initiative. [23] Unlike ammonium, sulfuric acid has a very low vapor pressure and would exist in the particle with or without ammonia. Therefore, while SO 2 reductions would reduce particulate ammonium, changes in ammonia would be expected to have very little effect on the sulfate concentration.

In addition to the above considerations, because ammonium nitrates are highest in the winter, when ammonia emissions are lowest, reducing wintertime NO X emissions may represent a more certain path towards reducing this winter peak than ammonia reductions. Moreover, reductions in ammonia emissions alone would also tend to increase the acidity of PM 2.5 and of precipitation. As noted in the proposal, this might have untoward environmental or health consequences.

Some commenters highlighted ammonia as an important pollutant with multiple effects on the environment, including its contributions to PM 2.5. These commenters highlighted that ammonia emissions are not currently regulated extensively, and suggested that EPA strengthen its efforts to better understand the many effects of ammonia emissions and better research options for controlling ammonia, so that it can be regulated where appropriate in the future programs. Generally, EPA agrees with these commenters.

(iii) Crustal Material

The contributions of crustal materials to PM 2.5 nonattainment are usually small, and the interstate transport of crustal materials is even smaller. Emissions of crustal materials on regional scales are uncertain, highly variable in space and time, and may not be easily controlled in some cases, suggesting significant uncertainties in quantifying emissions and the costs and effectiveness of control actions. Emissions reductions of SO 2 and NO X will likely reduce some of the direct emissions of PM 2.5 from EGUs and other industries, which are responsible for a portion of the “crustal material” measured downwind at receptors.

(c) Summary of Response To Requiring or Allowing Reductions in Other Pollutants

After reviewing public comments in light of the current understanding of alternative pollutants as summarized above, EPA disagrees with those commenters who suggested that the final Clean Air Interstate Rule should require states to address the interstate transport of carbonaceous material (including VOCs), ammonia, and/or crustal material in the present rulemaking.

At present, the sources and emissions contributing to these components on regional scales are not sufficiently quantified. In addition, the representation of atmospheric physics and chemistry for these components in air quality models is in some cases poor in comparison with current understanding of SO 2 and NO X (most notably for sources and amounts of secondary organic aerosol production. [24] Consequently, quantification of the interstate transport of these components is significantly more uncertain than for SO 2 and NO X emissions. Given these uncertainties in regional emissions and interstate transport of these components, EPA has determined that it would be premature to quantify interstate impacts of these emissions through zero-out modeling, as was done for SO 2 and NO X emissions.

In addition, the costs of control measures, their effectiveness at reducing emissions, as well as their ultimate effectiveness at reducing PM 2.5 concentrations at downwind receptors are all uncertain. The EPA does not believe it could reasonably evaluate whether such State emissions contributed significantly to transport, or what level of control would address the significant contribution. Commenters have not provided us specific data and information to allow such assessments.

The EPA also disagrees with commenters who argue that EPA should, for the purposes of this rule, permit the States to substitute controls of sources of any of these other three components for the required limits on SO 2 and NO X. Given the greater uncertainties in estimating the contribution of alternative source emissions, States would have difficulty developing, and EPA would have difficulty in approving, SIPs that, by controlling these components, purport to reduce an upwind State's impact on downwind PM 2.5 nonattainment by an equivalent amount to that required in today's final rule.

As explained in the proposal, a decision not to regulate these components of PM 2.5 in the present rulemaking does not preclude state or local PM 2.5 implementation plans from reducing emissions of carbonaceous material, ammonia, or crustal material, in order to achieve attainment with PM 2.5 standards, in cases where there is evidence that such controls will be effective on a local basis. Although uncertainties exist in addressing long-range transport of these pollutants, state and local air quality management agencies will need to evaluate reasonable control measures for sources of these pollutants in developing SIPs due in 2008. We expect continuous improvements will be made in our understanding of source emissions and PM 2.5 components not addressed under CAIR. To assist future air quality management decisions, EPA is actively supporting research into better understanding the emissions, atmospheric processes, long range transport, and opportunities for control of these PM 2.5 components.

(d) Justification for Including NO X in Determining Significant Contributions and for Regulating NO X Emissions for PM 2.5 Transport

Some commenters questioned the EPA's basis for requiring emissions reductions of NO X, in addition to SO 2, for the purposes of controlling interstate transport of PM 2.5. These comments, and EPA's response, are discussed below. Other comments addressing EPA's basis for requiring NO X for ozone are addressed in a subsequent section.

Like SO 2, NO X emissions are understood to affect PM 2.5 on regional scales, due in part to the time needed to convert NO X emissions to nitrate. Like SO 2 but unlike precursors of other components of PM 2.5, emissions of NO X are well quantified for EGUs and with reasonable accuracy for other urban and regional sources, and the transport of NO X and PM 2.5 derived from NO X can also be quantified with a fair degree of certainty. In addition, SO 2 and NO X interact as part of the same chemical system in the atmosphere. Controlling SO 2 emissions without concurrently controlling NO X emissions can lead to nitrate replacement whereby SO 2 emissions reductions will be less effective than expected. Finally, SO 2 and NO X share common sources in fossil fuel combustion. As such, controlling emissions of both precursors in a coordinated way presents opportunities to reduce the overall cost of the control program. [25]

Commenters questioned EPA's methodology of evaluating whether an upwind State contributes significantly to PM 2.5 nonattainment by considering (through the “zero-out” air quality modeling technique) SO 2 and NO X emissions simultaneously. These commenters argued that zeroing out SO 2 and NO X emissions simultaneously precludes determining the contribution of each component to downwind nonattainment. Because sulfates generally comprise a greater fraction of PM 2.5 than nitrates in the Eastern U.S., these commenters argued that the basis for requiring NO X controls is weaker than for SO 2, and has not been determined directly by EPA.

The EPA's multi-pollutant approach of modeling SO 2 and NO X contributions at the same time is consistent both with sound science and with the requirements of CAA section 110(a)(2)(D), as EPA interpreted and applied them in the NO X SIP Call. This provision requires each State to submit a SIP to prohibit “any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will * * * contribute significantly to nonattainment” downwind. As discussed in section II above, in the NO X SIP Call, a rulemaking in which EPA regulated NO X emissions as precursors for ozone, EPA found that ozone resulted from the combined contributions of many emitters over a multistate region, a phenomenon that EPA termed “collective contribution” (63 FR 57356-86). As a result, EPA evaluated each State's contribution to nonattainment downwind by considering the impact of the entirety of that State's NO X emissions on downwind nonattainment. Once EPA determined the State's entire NO X emissions inventory to have at least a minimum downwind impact, then EPA required the State to eliminate the portion of those emissions that could be reduced through highly cost-effective controls. The EPA considered this approach to be consistent with the section 110(a)(2)(D) requirements.

In a companion rulemaking, the section 126 Rule, EPA found that certain, individual NO X emitters must be subject to Federal regulation due to their impact on downwind nonattainment (65 FR 2674). The EPA based this finding on the same notion of “collective contribution,” that is, NO X emissions from those individual sources were part of the upwind State's total NO X inventory, the total NO X inventory had a sufficiently high impact on downwind nonattainment, and therefore the individual NO X emitters should be subject to control without any separate determination as to their individual impacts on downwind nonattainment.

The DC Circuit accepted EPA's collective contribution approach upholding most of the NO X SIP Call regulation, in Michigan v. EPA, 213 F.3d 663 (DC Cir. 2000), cert. denied 532 U.S. 904 (2001). Similarly, the DC Circuit upheld most aspects of EPA's Section 126 Rule, including the collective contribution basis for finding that emissions from the individual sources should be subject to regulation. Appalachian Power Co. v. EPA, 249 F.3d 1032 (DC Cir. 2001) (per curium).

As discussed elsewhere, PM 2.5 is similar to ozone in that it is the result of emissions from many sources over a multi-state region. Accordingly, EPA considers that the phenomenon of “collective contribution” is associated with PM 2.5 as well.

In the CAIR NPR, EPA selected SO 2 and NO X as the appropriate precursors to be controlled for PM 2.5 transport, for several reasons presented above. As in the NO X SIP Call, today's rulemaking, under CAA section 110(a)(2)(D), requires EPA to evaluate whether a particular upwind State must submit a SIP that prohibits “any source or other type of emissions activity within the State from emitting any air pollutant in amounts which will * * * contribute significantly to nonattainment” downwind. In making this determination, EPA considers the effects of all of the appropriate precursors—here, both SO 2 and NO X—from all of the State's sources on downwind PM 2.5 nonattainment. If that collective contribution to downwind PM 2.5 nonattainment is sufficiently high, then EPA requires the upwind State to eliminate those precursors to the extent of the availability of highly cost-effective controls.

The EPA's approach to evaluating a State's impact on downwind nonattainment by considering the entirety of the State's SO 2 and NO X emissions is also consistent with the chemical interactions in the atmosphere of SO 2 and NO X in forming PM 2.5. The contributions of SO 2 and NO X emissions are generally not additive, but rather are interrelated due to the nitrate replacement phenomenon, as well as other complex chemical reactions that can include organic compounds as well. As commenters point out, the nature of these reactions can vary with location and time. The non-linear nature of some of these reactions can produce differing results depending on the relative amount of reductions and copollutants. Reductions in sulfates can increase nitrates and, in some conditions, modest reductions in nitrates can increase sulfates although through different mechanisms. Large regional reductions in both pollutants, however, are more likely to result in a significant reductions in fine particles. [26]

Based on its current understanding of regional air pollution and modeling results, EPA believes that adopting a broad new program of regional controls to continue the downward trajectory in both SO X and NO X begun in base programs such as the national mobile source rules and Title IV, as well as the NO X SIP call, will ultimately result in significant benefits not only in reducing PM 2.5 nonattainment, but improving public health, reducing regional haze, and addressing multimedia environmental concerns including acid deposition and nutrient loadings in sensitive coastal estuaries in the East. [27]

Some commenters argued that the benefits of combining NO X with SO 2 reductions, if any, would be small, and further argued that the effect of any nitrate reductions in the environment would be further diminished by measurement losses that can occur in the filter in the method used to measure PM 2.5. In so doing, they questioned the scientific basis for nitrate replacement, suggesting that this response to changes in SO 2 emissions may not happen in all places and at all times. The commenters referenced a study in the Southeastern U.S. by Blanchard and Hidy, [28] which they claim calls into question whether nitrate replacement actually occurs. In fact, the study finds evidence that nitrate replacement occurs: “the sulfate decreases were an input to the model calculations, but their effect on fine PM mass was modified by concomitant decreases in ammonium and increases in nitrate.” A second study by the same authors, using essentially the same dataset and methods, and referenced both by EPA in the NPR and by the commenters, gives very strong support for the existence of nitrate replacement, as well as for coordinating SO 2 and NO X reductions, as indicated by the following conclusions: “reductions in sulfate through SO 2 reduction at constant NO X levels would not result in proportional reduction in PM 2.5 mass because particulate nitrate concentrations would increase. However, if both NO X and SO 2 emissions are reduced, then it may be possible to achieve sulfate reductions without concomitant nitrate increases * * *” [29]

Nitrate replacement is well documented in the scientific literature as a possible response of PM 2.5 to changes in SO 2 emissions. [30] While these commenters are correct that nitrate replacement is not expected to occur at all places and at all times, even where average conditions are not favorable for nitrate replacement, hourly variability in those conditions can create conditions favorable for nitrate replacement at particular times. Nitrate replacement theory predicts no conditions under which SO 2 reductions would decrease nitrate, and suggests that nitrate may increase under fairly common conditions. [31] Consequently, the net effect of SO 2 reductions can be only to increase nitrate or not to have any effect. The variability of conditions occurring over a year means that SO 2 reductions would be expected to increase nitrate on balance.

Even if the studies referenced by these commenters showed that nitrate replacement does not occur in some circumstances, other studies suggest that the conditions for nitrate replacement are common in the Eastern U.S. [32] Suggesting that nitrate replacement does not occur under some conditions does not imply that NO X should not be controlled, when it is known that nitrate replacement occurs under other common conditions.

The EPA recognizes that the relative reductions in PM 2.5 from implementation of the CAIR will be greater for SO 2 than for NO X. Nevertheless, overall costs for reducing NO X in the CAIR region are much lower than SO 2 because a large portion of the region has already installed NO X controls for ozone in the summer months. Our revised modeling approaches took into account the differences commenters note between actual nitrate concentrations in the atmosphere and what is measured as PM 2.5. Nevertheless emissions of both pollutants clearly contribute to interstate transport of ambient fine particles, and EPA concludes that the best approach in this situation is to provide highly cost effective reductions for both pollutants. Moreover, in warmer conditions when apparent nitrate changes from NO X reductions as measured on PM 2.5 monitors are small, the actual reductions in particulate and gaseous nitrates in the ambient environment are larger; accordingly, NO X reductions combined with SO 2 reductions can be expected to reduce health risk, visibility impairment, and other environmental damages.

c. What Is EPA's Final Determination?

After considering the public comments, EPA concludes that it should adopt the approach it proposed for addressing interstate transport of pollutants that affect PM 2.5, for the reasons presented here and in the proposal. That is, in today's action, EPA is requiring states to take steps to control emissions of SO 2 and NO X on the basis of their contributions to nonattainment of PM 2.5 standards in downwind states. The EPA concludes that we do not now have a sufficient basis for including emissions of other components (carbonaceous material, ammonia, and crustal material) that contribute to PM 2.5 in determining significant contributions and in requiring emission reductions of these components.

2. What Is the Role for Local Emissions Reduction Strategies?

a. Summary of Analyses and Conclusions in the Proposal

In section IV.F of the proposed rule, we discussed two analyses that were completed to address the impact of local control measures relative to regional reductions of SO 2 and NO X (69 FR 4596-99). In the first analysis, we applied a list of readily identifiable control measures (NPR, Table IV-5) in the Philadelphia, Birmingham, and Chicago urban primary metropolitan statistical areas (PMSA) counties. In the second analysis, we applied a similar list of control measures to 290 counties representing the metropolitan areas we projected to contain any nonattainment county in 2010 in the baseline scenario. The three-city analysis estimated that these local measures would result in ambient PM 2.5 reductions of about 0.5 μg/m [3] to about 0.9 μg/m [3] , which is less than needed to bring any of the cities into attainment in 2010. The 290-county study, which included enough counties to produce regional as well as local reductions, found that while some of the 2010 nonattainment areas would be projected to attain, many would not. Moreover, much of the PM 2.5 reduction in the 290-county study resulted from assuming reduction in sulfates due to SO 2 reductions on utility boilers in the urban counties. Accordingly, we concluded that for a sizable number of PM 2.5 nonattainment areas it will be difficult if not impossible to reach attainment unless transport is reduced to a much greater degree than by the simultaneous adoption of controls within only the nonattainment areas.

b. Summary and Response to Public Comments

A number of commenters supported EPA's conclusion that regional reductions are necessary given the difficulty in achieving local emission reductions, and given that they are generally more cost-effective. Generally, EPA agrees with these commenters.

Other commenters were critical of the local measures analysis, and recommended that EPA should consider a more appropriate mix of regional and local controls before requiring substantial expenditures for controls on power plants or other regional sources potentially affected by this rule. These commenters believed that the proposed rule did not represent the optimal emissions reduction strategy. Other commenters believed that the local measures analysis underestimated the achievable local emissions reductions. Some commenters believed that EPA should include local control measures in the baseline scenario for the analysis. Finally, some commenters questioned the feasibility of doing a local measures analysis at all, given the uncertainties in the analysis, the uncertainties regarding nonattainment boundaries, and the work to be done by State and local areas to identify and evaluate strategies.

The EPA continues to conclude that it would be difficult if not impossible for many nonattainment areas to reach attainment through local measures alone, and EPA finds no information in the comments to alter this conclusion. While recognizing the uncertainties in conducting such an analysis (as noted in the preamble to the proposed rule), we continue to believe that the two local measures scenarios represent a highly ambitious set of measures and emissions reductions that may in fact be difficult to achieve in practice. This analysis was not intended to precisely identify local measures that may be available in a particular area. The EPA believes that a strategy based on adopting highly cost effective controls on transported pollutants as a first step would produce a more reasonable, equitable, and optimal strategy than one beginning with local controls. The local measures analyses we conducted were not, however, intended to develop a specific or “optimal” regional and local attainment strategy for any given area. Rather, the analysis was intended to evaluate whether, in light of available local measures, it is likely to be necessary to reduce significant regional transport from upwind states. We continue to believe that the two local measures analyses that were conducted for the proposal rule strongly support the need for regional reductions of SO 2 and NO X.

B. What Is the Basis for EPA's Decision To Require Reductions in Upwind Emissions of NO X To Address Ozone-Related Transport?

1. How Did EPA Determine Which Pollutants Were Necessary To Control To Address Interstate Transport for Ozone?

In the notice of proposed rulemaking, EPA provided the following characterization of the origin and distribution of 8-hour ozone air quality problems:

The ozone present at ground level as a principal component of photochemical smog is formed in sunlit conditions through atmospheric reactions of two main classes of precursor compound: VOCs and NO X (mainly NO and NO 2). The term “VOC” includes many classes of compounds that possess a wide range of chemical properties and atmospheric lifetimes, which helps determine their relative importance in forming ozone. Sources of VOCs include man-made sources such as motor vehicles, chemical plants, refineries, and many consumer products, but also natural emissions from vegetation. Nitrogen oxides are emitted by motor vehicles, power plants, and other combustion sources, with lesser amounts from natural processes including lightning and soils. Key aspects of current and projected inventories for NO X and VOC are summarized in section IV of the proposal notice and EPA websites (e.g., http://www.w.gov/ttn/chief.) The relative importance of NO X and VOC in ozone formation and control varies with local- and time-specific factors, including the relative amounts of VOC and NO X present. In rural areas with high concentrations of VOC from biogenic sources, ozone formation and control is governed by NO X. In some urban core situations, NO X concentrations can be high enough relative to VOC to suppress ozone formation locally, but still contribute to increased ozone downwind from the city. In such situations, VOC reductions are most effective at reducing ozone within the urban environment and immediately downwind.

The formation of ozone increases with temperature and sunlight, which is one reason ozone levels are higher during the summer. Increased temperature increases emissions of volatile man-made and biogenic organics and can indirectly increase NO X as well (e.g., increased electricity generation for air conditioning). Summertime conditions also bring increased episodes of large-scale stagnation, which promote the build-up of direct emissions and pollutants formed through atmospheric reactions over large regions. The most recent authoritative assessments of ozone control approaches 33, 34 have concluded that, for reducing regional scale ozone transport, a NO X control strategy would be most effective, whereas VOC reductions are most effective in more dense urbanized areas.

Studies conducted in the 1970s established that ozone occurs on a regional scale (i.e., 1000s of kilometers) over much of the Eastern U.S., with elevated concentrations occurring in rural as well as metropolitan areas. 35, 36 While progress has been made in reducing ozone in many urban areas, the Eastern U.S. continues to experience elevated regional scale ozone episodes in the extended summer ozone season.

Regional 8-hour ozone levels are highest in the Northeast and Mid-Atlantic areas with peak 2002 (3-year average of the 4th highest value for all sites in the region) ranging from 0.097 to 0.099 parts per million (ppm). [37] The Midwest and Southeast States have slightly lower peak values (but still above the 8-hour standard in many urban areas) with 2002 regional averages ranging from 0.083 to 0.090 ppm. Regional-scale ozone levels in other regions of the country are generally lower, with 2002 regional averages ranging from 0.059 to 0.082 ppm. Nevertheless, some of the highest urban 8-hour ozone levels in the nation occur in southern and central California and the Houston area.

In the notice of proposed rulemaking, EPA noted that we continue to rely on the assessment of ozone transport made in great depth by the OTAG in the mid-1990s. As indicated in the NO X SIP call proposal, the OTAG Regional and Urban Scale Modeling and Air Quality Analysis Work Groups reached the following conclusions:

A. Regional NO X emissions reductions are effective in producing ozone benefits; the more NO X reduced, the greater the benefit.

B. Controls for VOC are effective in reducing ozone locally and are most advantageous to urban nonattainment areas. (62 FR 60320, November 7, 1997).

The EPA proposed to reaffirm this conclusion in this rulemaking, and proposed to address only NO X emissions for the purpose of reducing interstate ozone transport.

Some commenters suggested that in this rulemaking EPA should require regional reductions in VOC emissions as well as NO X emissions in this rulemaking. [38] The EPA continues to believe based on the OTAG and NARSTO reports cited earlier, and the modeling completed as part of the analysis for this rule, that NO X emissions are chiefly responsible for regional ozone transport, and that NO X reductions will be most effective in reducing regional ozone transport. This understanding was considered an adequate basis for controlling NO X emissions for ozone transport in the NO X SIP call, and was upheld by the courts. As a result, EPA is requiring NO X reductions and not VOC reductions in this rulemaking.

However, EPA agrees, that VOCs from some upwind States do indeed have an impact in nearby downwind States, particularly over short transport distances. The EPA expects that States will need to examine the extent to which VOC emissions affect ozone pollution levels across State lines, and identify areas where multi-state VOC strategies might assist in meeting the 8-hour standard, in planning for attainment. This does not alter the basis for the CAIR ozone requirements in this rule; EPA's modeling supports the conclusion that NO X emissions from upwind states will significantly contribute to downwind nonattainment and interfere with maintenance of the 8-hour ozone standard.

2. How Did EPA Determine That Reductions in Interstate Transport, as Well as Reductions in Local Emissions, Are Warranted To Help Ozone Nonattainment Areas To Meet the 8-Hour Ozone Standard?

a. What Did EPA Say in Its Proposal Notice?

In the NPR, EPA noted that the Agency promulgated the NO X SIP call in 1998 to address interstate ozone transport problems in the Eastern U.S. The EPA noted that it made sense to re-evaluate whether the NO X SIP call was adequate at the same time that the Agency was assessing the need for emissions reductions to address interstate PM 2.5 problems because of overlap in the pollutants and relevant sources, and the timetables for States to submit local attainment plans. The EPA presented a new analysis of the extent of residual 8-hour ozone attainment projected to remain in 2010, and the extent and severity of interstate pollution transport contributing to downwind nonattainment in that year.

The proposal notice said that based on a multi-part assessment, EPA had concluded that:

  • “Without adoption of additional emissions controls, a substantial number of urban areas in the central and eastern regions of the U.S. will continue to have levels of 8-hour ozone that do not meet the national air quality standards.
  • * * * EPA has concluded that small contributions of pollution transport to downwind nonattainment areas should be considered significant from an air quality standpoint, because these contributions could prevent or delay downwind areas from achieving the standards.
  • * * * EPA has concluded that interstate transport is a major contributor to the projected (8-hour ozone) nonattainment problem in the eastern U.S. in 2010. * * * (T)he nonattainment areas analyzed receive a transport contribution of more than 20 percent of the ambient ozone concentrations, and 21 of 47 had a transport contribution of more than 50 percent.
  • Typically, two or more States contribute transported pollution to a single downwind area, so that the “collective contribution” is much larger than the contribution of any single State.

Also, EPA concluded that highly cost-effective reductions in NO X emissions were available within the eastern region where it determined interstate transport was occurring, and that requiring those highly cost effective reductions would reduce ozone in downwind nonattainment areas.

In addition, the proposal examined the effect of hypothetical across-the-board emissions reductions in nonattainment areas. The notice stated that EPA had conducted a preliminary scoping analysis in which hypothetical total NO X and VOC emissions reductions of 25 percent were applied in all projected nonattainment areas east of the continental divide in 2010, yet approximately 8 areas were projected to have ozone levels exceeding the 8-hour standard. Based on experience with state plans for meeting the one-hour ozone standard, EPA said this scenario was an indication that attaining the 8-hour standard will entail substantial cost in a number of nonattainment areas, and that further regional reductions are warranted.

b. What Did Commenters Say?

The Need for Reductions in Interstate Ozone Transport: Some commenters argued that EPA should not conduct another rulemaking to control interstate contributions to ozone because local contributions in nonattainment regions appear, according to the commenters, to have larger impacts than regional NO X emissions. The commenters cited EPA's sensitivity modeling of hypothetical 25 percent reductions as supporting this view.

The EPA disagrees that comparing the sensitivity modeling and the CAIR control modeling is a valid way to compare the effectiveness of local and regional controls. The two scenarios do not reduce emissions by equal tonnage amounts, equal percentages of the inventory, or equal cost. These scenarios therefore do not support an assessment of the relative effectiveness of local and regional controls. While EPA in general agrees that emissions reductions in a nonattainment area will have a greater effect on ozone levels in that area than similar reductions a long distance away, EPA does not agree that the modeling supports the conclusion that all additional controls to promote attainment with the 8-hour standard should be local. The level of reduction assumed was a hypothetical level, not a level determined to be reasonable cost nor a mandated level of reduction. The commenters provided no evidence that reasonable local controls alone would result in attainment throughout the East. However, EPA did receive comments that such a level would result in costly controls and might not be feasible in some areas that have previously imposed substantial controls.

The EPA believes it is clear that further reductions in emissions contributing to interstate ozone transport, beyond those required by the NO X SIP Call, are warranted to promote attainment of the 8-hour ozone standard in the eastern U.S. As explained elsewhere in this final rule, EPA analyzed interstate transport remaining after the NO X SIP Call, and determined—considering both the impact of interstate transport on downwind nonattainment, and the potential for highly cost effective reductions in upwind States—that 25 States significantly contribute to 8-hour ozone nonattainment downwind. The importance of transport is illustrated, as mentioned above, by EPA's findings for the final rule that (1) all the 2010 nonattainment counties analyzed were projected to receive a transport contribution of 24 percent or more of the ambient ozone concentrations, and (2) that 16 of 38 counties are projected to have a transport contribution of more than 50 percent.

In addition, EPA received multiple comments from State associations and individual States strongly agreeing that further reductions in interstate ozone transport are warranted to promote attainment with the 8-hour standard, to protect public health, and to address equity concerns of downwind states affected by transport. For example, comments from the Maryland Department of the Environment stated, “Our 15 year partnership with researchers from the University of Maryland has produced data that shows on many summer days the ozone levels floating into Maryland area are already at 80 to 90 percent of the 1-hour ozone standard and actually exceed the new 8-hour ozone standard before any Maryland emissions are added. * * * Serious help is needed from EPA and neighboring states to solve Maryland's air pollution problems. * * * Local reductions alone will not clean up Maryland's air.” The comments of the Ozone Transport Commission stated that even after levels of control envisioned by EPA in 2010 (under the Clear Skies Act), interstate transport from other states would continue to affect the Ozone Transport Region created by the CAA (Connecticut, Delaware, the District of Columbia, Maine, Maryland, Massachusetts, New Hampshire, New Jersey, New York, Pennsylvania, Rhode Island, Vermont, and Virginia). “Our modeling demonstrates that even in the extreme example of zero anthropogenic emissions within the OTR (Ozone Transport Region), 145 of 146 monitors show a significant (25%) increment of the 8-hour standard taken up by transport from outside the OTR.” Comments from the North Carolina Department of Environment and Natural Resources stated, “The reductions proposed in [EPA's rule] in the other states are needed to ensure that North Carolina can attain and maintain the health-based air quality standards for * * * 8-hour ozone.”

Magnitude of Ozone Reductions Achieved: Commenters stated that NO X reductions should not be pursued because the 8-hour ozone reductions in projected nonattainment counties resulting from the required NO X reductions are too small—1-2 ppb in only certain areas. According to commenters, these benefits are smaller than the threshold for determining significant contribution.

The EPA disagrees with the notion that if air quality improvements would be limited, then nothing further should be done to address interstate transport. Based on the difference between the base case and CAIR control case modeling results, EPA has concluded that interstate air quality impacts are significant from an air quality standpoint, and that highly cost effective reductions are available to reduce ozone transport. State comments have corroborated EPA's conclusion that a number of areas will face high local control costs, or even be unable to attain the 8-hour ozone standard, without further reductions in interstate transport. Therefore, EPA believes it is important for upwind states to modify their SIPs so that they contain adequate provisions to prohibit significant contributions to downwind nonattainment or interference with maintenance as the statute requires. The EPA has established an amount of required emissions reductions based on controls that are highly cost effective. The resulting improvements in downwind ozone levels are needed for attainment, public health and equity reasons.

The 2 ppb significance threshold that commenters cite is part of the test that EPA used to identify which States should be evaluated for inclusion in a rule requiring them to reduce emissions to reduce interstate transport. (See section VI.) This 2 ppb threshold is based on the impact on a downwind area of eliminating all emissions in an upwind State. The ozone reductions from CAIR will improve public health and will decrease the extent and cost of local controls needed for attainment in some areas. In addition, base case modeling for this rule shows that of the 40 counties projected in nonattainment in 2010, 16 counties are within 2 ppb of the standard, 6 counties are within 3 ppb, and 3 counties are within 4 ppb. In 2015, projected base case ozone concentrations in over 70 percent of nonattaining counties (i.e., 16 of 22 counties) are within 5 ppb of the standard.

Reducing NO X emissions has multiple health and environmental benefits. Controlling NO X reduces interstate transport of fine particle levels as well as ozone levels, as discussed elsewhere in this notice. Although EPA is not relying on other benefits for purposes for setting requirements in this rule, reducing NO X emissions also helps to reduce unhealthy ozone and PM levels within a State, as well as reduce acid deposition to soils and surface waters, eutrophication of surface and coastal waters, visibility degradation, and impacts on terrestrial and wetland systems such as changes in species composition and diversity.

EPA's Authority To Require Controls Beyond the NO X SIP Call: Commenters emphasized that in the NO X SIP Call, EPA determined the States whose emissions contribute significantly to nonattainment, EPA mandated NO X emissions reductions that would eliminate those significant contributions, and EPA indicated that it would reconsider the matter in 2007. This commenter argued that for the States included in the NO X SIP Call, EPA may not, as a legal matter, conduct further rulemaking at this time because the affected States are no longer contributing significantly to nonattainment downwind. In any event, the commenters said, EPA should abide by its statement that it would revisit the matter in 2007, and EPA should not do so earlier.

Sound policy considerations support re-examining interstate ozone transport at this time. At the time of the NO X SIP Call, EPA anticipated reassessing in 2007 the need for additional reductions in emissions that contribute to interstate transport, but EPA has accelerated that date in light of various circumstances, including the fact that we are undertaking similar action with the PM 2.5 NAAQS. In addition, in light of overlap in the pollutants, States, and sources likely to be affected, it is prudent to coordinate action under the 8-hour ozone standard. The EPA notes that evaluating PM 2.5 transport and ozone transport together at this time will enable States to consider the resulting rules in devising their PM 2.5 and 8-hour ozone attainment plans, and will enable States and sources to plan emissions reductions knowing their transport-related reduction requirements for both standards.

CAA section 110(a)(2)(D) requires that State SIPs contain “adequate provisions” prohibiting emissions that significantly contribute to nonattainment areas in, or interfere with maintenance by, other States. Over time, emissions of ozone precursors, the (projected) non-attainment status of receptors, the modeling tools that EPA and the states use to conduct their analyses, the data available to the states or EPA and other analytic tools or conditions may change. The EPA has conducted an updated analysis of upwind contribution to downwind nonattainment of 8-hour ozone nonattainment areas after the NO X SIP Call, including updated emissions projections, updated air quality modeling, and updated analysis of control costs. This has revealed a need for reductions beyond those required by the NO X SIP Call in order for upwind states to be in compliance with section 110(a)(2)(D). The EPA thus disagrees with commenters' assertions that the provisions of section 110(a)(2)(D) prevent EPA from conducting further evaluation of upwind contributions to downwind nonattainment at this time. The EPA also notes that the NO X SIP Call, a 1998 rulemaking, promulgated a set of requirements intended to eliminate significant contribution to downwind ozone nonattainment at the time of implementation, which EPA identified on the basis of modeling for the year 2007 (although implementation was required to occur several years earlier). In today's action, EPA is reviewing the transport component of 8-hour ozone nonattainment for the period beginning in 2010, consistent with the criteria in the NO X SIP Call as applied to present circumstances, concluding that even with implementation of the NO X SIP Call controls, upwind States will contribute significantly to downwind ozone nonattainment and interfere with maintenance at a point after 2007. No provision of the CAA prohibits this action.

Commenters added that the purpose of the CAIR rulemaking seemed to be to account for the fact that control costs have changed since the date of the NO X SIP Call. The commenters said that control costs will frequently fluctuate, but that such fluctuations should not merit revised rulemaking.

In response, we would note that EPA conducted an updated analysis for air quality impacts, not only costs, in determining that further reductions in interstate ozone transport are warranted. That air quality analysis showed a substantial, continuing interstate transport problem for areas after implementation of the NO X SIP Call. The EPA does have the legal authority to reconsider the scope of the area that significantly contributes and the level of control determined to be “highly cost-effective” based on new information. Updated information shows that lower NO X burners and SCR achieve better performance than previously estimated and as a result are more cost effective than previously anticipated. This rule follows the NO X SIP Call by six years; EPA does not believe that this represents a too-frequent re-evaluation, particularly given the stay of the 8-hour basis for the NO X SIP Call (See, e.g., CAA section 109(d)(1) requiring EPA to reevaluate the NAAQS themselves every five years.) So both updated air quality and cost information supports further NO X controls to reduce interstate transport.

Some commenters argued that EPA should delay imposing control obligations on upwind States for the 8-hour ozone NAAQS until after EPA has implemented local control requirements, and after all of the NO X SIP Call control requirements are implemented and evaluated. Others said EPA should not impose requirements on non-SIP-Call States until after all 8-hour controls—NO X SIP Call and local—are implemented.

We agree that the NO X SIP Call should be taken into account in evaluating the need for further interstate transport controls. We have taken the NO X SIP Call into account by including the effect of the NO X SIP Call in the base case used for the CAIR analysis, and by conducting analyses to confirm that CAIR will achieve greater ozone-season reductions than the SIP Call. The EPA disagrees that the Agency should wait for implementation of local controls before determining transport controls. There is no legal requirement that EPA wait to determine transport controls until after local controls are implemented. The EPA's basis for this legal interpretation is explained in section II.A. above. In addition, the Agency believes it is important to address interstate transport expeditiously for public health.

C. Comments on Excluding Future Case Measures From the Emissions Baselines Used To Estimate Downwind Ambient Contribution

The EPA received comments that the 2010 analytical baseline for evaluating whether upwind emissions meet the air quality portion of the “contribute significantly” standard should reflect local control measures that will be required in the downwind nonattainment areas, or broader statewide measures in downwind states, to attain the PM 2.5 or 8-hour ozone NAAQS by the relevant attainment dates, many of which are (or are anticipated to be) 2010 or earlier. This single target year was chosen both to address analytical tool constraints and to reasonably reflect future conditions in or near the initial attainment years for both ozone and PM nonattainment areas. The EPA did include in the baseline most of the specifically required measures that can be identified at this time, but did not include any further measures that would be needed for satisfying “rate of progress” requirements or for attainment of the PM 2.5 and 8-hour ozone standards. If EPA had included further local controls, the commenters contend, fewer upwind States would have exceeded our significant contribution thresholds.

We reject any notion that in determining the need for transport controls in upwind states, EPA should assume that the affected downwind areas must “go all the way first”—that is, assume that downwind areas put on local in-state controls sufficient to reach attainment, or assume that downwind states with nonattainment areas implement statewide control measures. The EPA does not believe these are appropriate assumptions. The former assumption would eviscerate the meaning of CAA section 110(a)(2)(D). The latter assumption would make the downwind state solely responsible for reductions in any case where a downwind state could attain through in-state controls alone, even if the upwind state contribution was significantly contributing to nonattainment problems in the downwind state. We do not believe that this approach would be consistent with the intent of section 110(a)(2)(D), which in part is to hold upwind states responsible for an appropriate share of downwind nonattainment and maintenance problems, and to prevent scenarios in which downwind states must impose costly extra controls to compensate for significant pollution contributions from uncontrolled or poorly controlled sources in upwind states. In addition, this approach could raise costs of meeting air quality standards because highly cost effective controls in upwind States would be foregone.

Rather, in the particular circumstances presented here, we think the adoption of regional controls at this time under section 110(a)(2)(D) is consistent with sound policy and section 110. Based on our analysis, the states covered by CAIR make a significant contribution to downwind nonattainment and the required reductions are highly cost effective. The reductions will reduce regional pollution problems affecting multiple downwind areas, will make it possible for States to determine the extent of local control needed knowing the reductions in interstate pollution that are required, will address interstate equity issues that can hamper control efforts in downwind States, and reflect considerations discussed in detail in section VII.

Although some commenters advocated specifically including statutorily mandated future nonattainment area controls in the analytical baseline, it would be difficult as a practical matter to predict the extent of local controls that will be required (beyond controls previously required) in each area in advance of final implementation rules interpreting the Act's requirements for PM 2.5 and 8-hour ozone, and before the state implementation plan process. Subpart 2 provisions that apply to certain ozone nonattainment areas are quite specific regarding some mandatory measures; we believe the CAIR baseline for the most part captures these measures. (See Response to Comments document in the docket.)As noted above, the choice of a single analytical year of 2010 was made to reflect baseline conditions at a date at or near the attainment dates for different pollutants and classes of areas. Because the attainment date for many ozone areas is 2009 or earlier, it should be noted that the analyses in 2010 may slightly overestimate the benefits of a number of national rules for mobile sources that grow with time. As noted elsewhere, these differences are unlikely to be significant.

D. What Criteria Should Be Used To Determine Which States Are Subject to This Rule Because They Contribute to PM 2.5 Nonattainment?

1. What Is the Appropriate Metric for Assessing Downwind PM 2.5 Contribution?

a. Notice of Proposed Rulemaking

In the NPR, we proposed as the metric for identifying a State as significantly contributing (depending upon further consideration of costs) to downwind nonattainment, the predicted change, due to the upwind State's emissions, in PM 2.5 concentration in the downwind nonattainment area that receives the largest ambient impact. The EPA proposed this metric in the form of a range of alternatives for a “bright line,” that is, ambient impacts at or greater than the chosen threshold level indicated that the upwind State's emissions do contribute significantly (depending on cost considerations), and that ambient impacts below the threshold mean that the upwind State's emissions do not contribute significantly to nonattainment. As detailed in section VI below, EPA conducted the analysis through air quality modeling that removed the upwind State's anthropogenic SO 2 and NO X emissions, and determined the difference in downwind ambient PM 2.5 levels before and after removal. The modeling results indicate a wide range of maximum downwind nonattainment impacts from the 37 States that we evaluated. The largest maximum contribution is 1.67 micrograms per cubic meter (μg/m [3] ), from Ohio to both Allegheny and Beaver counties in Pennsylvania.

b. Comments and EPA's Responses

The EPA proposed to use the maximum contribution on any downwind nonattainment area for assessing downwind PM 2.5 contributions. Many commenters expressed agreement with our proposed metric, however, many others disagreed. One group of these commenters indicated that EPA should distinguish the relative contribution from States using two parameters: (1) How many downwind nonattainment receptors they contribute to, and (2) how much they contribute to each such receptor. The commenters indicated that this approach would avoid inequities created by the disproportionate impact of some upwind contributors on their downwind neighbors. The EPA interprets these comments to suggest a metric that collectively includes both of these parameters, such as the sum of all downwind impacts on all affected receptors. This metric would result in higher values for States contributing to multiple receptors and at relatively high levels, and lower values for States contributing to fewer receptors and at relatively low levels.

The EPA's proposed metric does address how much each State contributes to a downwind neighbor; however, EPA does not believe that multiple downwind receptors need to be impacted in order for a particular state to be required to make emissions reductions under CAA section 110(a)(2)(D). Under this provision, an upwind State must include in the SIP adequate provisions that prohibit that State's emissions that “contribute significantly to nonattainment in * * *any other State* * *.” (Emphasis added.) Our interpretation of this provision is that the emphasized terms make clear that the upwind State's emissions must be controlled as long as they contribute significantly to a single nonattainment area.

One commenter agreed with EPA's use of maximum annual average downwind contribution, but suggested that EPA consider additional metrics such as: (a) Contributions to adverse health and welfare effects from short-term PM 2.5 concentrations; (b) contributions to worst 20 percent haze levels in Class 1 areas; and (c) contributions to adverse effects of sulfur and nitrogen deposition to acid sensitive surface waters and forest soils. The EPA appreciates that these metrics all have merit in their focus on the health and environmental consequences of emissions, however, in determining a metric for significant contributions, we must focus on implementation of CAA section 110(a)(2)(D) provisions regarding significant contribution to nonattainment of the PM 2.5 NAAQS.

Another commenter suggested EPA use the maximum annual average impact, as we proposed, but add the maximum daily PM 2.5 contribution. The commenter notes that this additional metric would indicate whether specific meteorological events drive the concentration change or whether there is a consistent pattern of transport from one area to another. It is not clear to EPA how the single data point of the maximum daily contribution indicates a consistent pattern of transport from one area to another since it is a measure from only a single day. Further, EPA does not agree that multiple days of impact is a relevant criterion for evaluating whether a State contributes significantly to nonattainment, since in theory, a single high-contribution event could be the cause or a substantial element of nonattainment of the annual average PM 2.5 standard. Because we currently do not observe nonattainment of the daily average PM 2.5 standard in Eastern areas, nonattainment of the annual average PM 2.5 standard is the relevant evaluative measure.

Some commenters suggested separately evaluating the NO X- and SO 2-related impacts (i.e., particulate nitrate and particulate sulfate) on nonattainment. As discussed in section II of this notice, EPA's approach to evaluating a State's impact on downwind nonattainment by considering the entirety of the State's SO 2 and NO X emissions is consistent with the chemical interactions in the atmosphere of SO 2 and NO X in forming PM 2.5. The contributions of SO 2 and NO X emissions are generally not additive, but rather are interrelated due to complex chemical reactions.

c. Today's Action

The EPA continues to believe that for each upwind State analyzed, the change in the annual PM 2.5 concentration level in the downwind nonattainment area that receives the largest impact is a reasonable metric for determining whether a State passes the “air quality” portion of the “contribute significantly” test, and therefore that State should be considered further for emissions reductions (depending upon the cost of achieving those reductions). This single concentration-based metric is adequate to capture the impact of SO 2 and NO X emissions on downwind annual PM 2.5 concentrations.

2. What Is the Level of the PM 2.5 Contribution Threshold?

a. Notice of Proposed Rulemaking

In the NPR, EPA proposed to establish a State-level annual average PM 2.5 contribution threshold from anthropogenic SO 2 and NO X emissions that was a small percentage of the annual air quality standard of 15.0 μg/m 3. The EPA based this proposal on the general concept that an upwind State's contribution of a relatively low level of ambient impact should be regarded as significant (depending on the further assessment of the control costs). We based our reasoning on several factors. The EPA's modeling indicates that at least some nonattainment areas will find it difficult or impossible to attain the standards without reductions in upwind emissions. In addition, our analysis of “base case” PM 2.5 transport shows that, in general, PM 2.5 nonattainment problems result from the combined impact of relatively small contributions from many upwind States, along with contributions from in-State sources and, in some cases, substantially larger contributions from a subset of particular upwind States. In the NO X SIP Call rulemaking, we termed this pattern of contribution—which is also present for ozone nonattainment—“collective contribution.”

In the case of PM 2.5, we have found collective contribution to be a pronounced feature of the PM 2.5 transport problem, in part because the annual nature of the PM 2.5 NAAQS means that throughout the entire year and across a range of wind patterns—rather than during just one season of the year or on only the few worst days during the year which may share a prevailing wind direction—emissions from many upwind States affect the downwind nonattainment area.

As a result, to address the transport affecting a given nonattainment area, many upwind States must reduce their emissions, even though their individual contributions may be relatively small. Moreover, as noted above, EPA's air quality modeling indicates that at least some nonattainment areas will find it difficult or impossible to attain the standards without reductions in upwind emissions. In combination, these factors suggest a relatively low value for the PM 2.5 transport contribution threshold is appropriate. For reasons specified in the NPR (69 FR 4584), EPA initially proposed a value of 0.15 μg/m 3 (1% of the annual standard) for the significance criterion, but also presented analyses based on an alternative of 0.10 μg/m 3 and called for comment on this alternative as well as on “the use of higher or lower thresholds for this purpose” (69 FR 4584).

The EPA adopted a conceptually similar approach to that outlined above for determining that the significance level for ozone transport in the NO X SIP Call rulemaking should be a small number relative to the NAAQS. The DC Circuit Court, in generally upholding the NO X SIP Call, viewed this approach as reasonable. Michigan v. EPA, 213 F.3d 663, 674-80 (DC Cir. 2000), cert. denied, 532 U.S. 904 (2001). After describing EPA's overall approach of establishing a significance level and requiring States with impacts above the threshold to implement highly cost-effective reductions, the Court explained: “EPA's design was to have a lot of States make what it considered modest NO X reductions * * *. ”Id. at 675. Indeed, the Court intimated that EPA could have established an even lower threshold for States to pass the air quality component:

The EPA has determined that ozone has some adverse health effects—however slight—at every level [citing National Ambient Air Quality Standards for Ozone, 62 FR 38856 (1997)]. Without consideration of cost it is hard to see why any ozone-creating emissions should not be regarded as fatally “significant” under section 110(a)(2)(D)(i)(I).”

213 F.3d at 678 (emphasis in original).

We believe the same approach applies in the case of PM 2.5 transport.

b. Comments and EPA's Responses

Many commenters indicated that EPA did not adequately justify the proposed annual average PM 2.5 contribution threshold level of 0.15 μg/m 3. Some commenters favor the alternative 0.10 μg/m 3 proposed by EPA, citing their agreement with EPA's rationale for 0.10 μg/m 3 while criticizing as arbitrary EPA's rationale for 0.15 μg/m 3.

Some commenters argued that the public health impact portion of EPA's rationale for establishing a relatively low-level threshold was not relevant. The commenters said that EPA previously determined, in establishing the PM 2.5 NAAQS, that ambient levels at or above 15.0 μg/m 3 were of concern for protecting public health, not the much lower levels that EPA proposed as the thresholds. In the NPR, we stated that we considered that there are significant public health impacts associated with ambient PM 2.5, even at relatively low levels. In generally upholding the NO X SIP Call, the DC Circuit noted a similar reason for establishing a relatively low threshold for ozone impacts. Michigan v. EPA, 213 F.3d 663, 678 (DC Cir. 2000), cert. denied, 532 U.S. 904 (2001). The EPA notes that by using a metric that focuses on the contribution of upwind areas to downwind areas that are above 15.0 μg/m 3, relatively low contributions to levels above the annual PM 2.5 standard are highly relevant to public health protection.

Many commenters offered alternative thresholds higher than 0.15 μg/m 3, citing previous EPA rules or policies as justification for the alternative level. Some suggested the PM 2.5 threshold should be equivalent in percentage terms to the threshold employed for assessing maximum downwind 8-hour ozone contributions. The threshold for maximum downwind 8-hour ozone concentration impact used in the NO X SIP Call, and proposed for use in the CAIR, is 2 parts per billion (ppb), or about 2.5 percent of the standard level of 80 ppb. Applying the 2.5 percent criterion to the 15.0 μg/m 3 annual PM 2.5 standard would yield a significance threshold of 0.35 μg/m 3.

The EPA disagrees with the comment that the thresholds for annual PM 2.5 and 8-hour ozone should be an equivalent percentage of their respective NAAQS. Both the forms and averaging times of the two standards are substantially different, with 8-hour ozone based on the average of the 4th highest daily 8-hour maximum values from each of 3 years, and PM 2.5 based on the average of annual means from 3 successive years. These fundamental differences in time scales, and thus in the patterns of transport that are relevant to contributing to nonattainment, do not suggest a transparent reason for presuming that the contribution thresholds should be equivalent. As discussed above, when more States make smaller individual contributions because of the annual nature of the PM 2.5 standard, it makes sense to have a threshold for PM 2.5 that is a smaller percentage of its NAAQS.

Other commenters suggested that in setting the maximum downwind PM 2.5 threshold, EPA should take into consideration the measurement precision of existing PM 2.5 monitors. The commenters assert that such measurement carries “noise” in the range of 0.5—0.6 μg/m 3. Because many daily average monitor readings are averaged to calculate the annual average, the precision of the annual average concentration is better than the figures cited by the commenters. Indeed, the annual standard is expressed as 15.0 μg/m 3, rounded to the nearest1/10μg, because such small differences are meaningful on an annual basis. While disagreeing with the specific amounts suggested by commenters, EPA recognizes that the PM 2.5 threshold specified in the proposal contains two digits beyond the decimal place, while the NAAQS specifies only one. The EPA agrees that specification of a threshold value of 0.15 μg/m 3 does suggest an overly precise test that might need to take into account modeled difference in PM 2.5 values as low as 0.001 μg/m 3.

Other commenters indicated that modeling “noise”—that is, imprecision—is a relevant consideration for establishing a threshold whose evaluation depends on air quality modeling analysis. These commenters indicated that a threshold of 5 percent of the NAAQS (i.e., 0.75 μg/m 3) is more reasonable considering modeling sensitivity. The commenters were not clear about what they mean by modeling “noise” and did not explain how it relates to the use of a threshold metric in the context of the CAIR.

In responding to the comment, we have considered some possible contributors to what the commenter describes as “noise.” There is the possibility that the air quality model has a systematic bias in predicting concentrations resulting from a given set of emissions sources. The EPA uses the model outputs in a relative, rather than an absolute, sense so that any modeling bias is constrained by real world results. As described further in section VI, EPA conducts a relative comparison of the results of a base case and a control case to estimate the percentage change in ambient PM 2.5 from the current year base case, holding meteorology, other source emissions, and other factors contributing to uncertainty constant. With this technique, any absolute modeling bias is cancelled out because the same model limitations and uncertainties are present in each set of runs.

Another possible source of noise is in the relative comparison of two model runs conducted on different computers. Since the computers used by EPA to run air quality models do not have any significant variability in their numerical processes, two model runs with identical inputs result in outputs that are identical to many significant digits. On the other hand, EPA believes it is not appropriate or necessary to carry such results to a level of precision that is beyond that required by the PM 2.5 NAAQS itself [39] .

Many commenters noted that EPA's proposed threshold of 0.15 μg/m 3, or one percent of the annual PM 2.5 NAAQS of 15.0 μg/m 3, is lower than the single-source contribution thresholds employed for PM 10 in certain other regulatory contexts. Commenters cited several different thresholds, including thresholds governing the applicability of the preconstruction review permit program and the emissions reduction requirement for certain major new or modified stationary sources located in attainment or unclassified areas; [40] and thresholds in the PSD rules that may relieve proposed sources from performing comprehensive ambient air quality analyses. [41]

Since the thresholds referred to by the commenters serve different purposes than the CAIR threshold for significant contribution, it does not follow that they should be made equivalent. The implication of the thresholds cited by the commenters is not that single-source contributions below these levels indicate the absence of a contribution. Rather, these thresholds address whether further more comprehensive, multi-source review or analysis of appropriate control technology and emissions offsets are required of the source. A source with estimated impacts below these levels is recognized as still affecting the airshed and is subject to meeting applicable control requirements, including best available control technology, designed to moderate the source's impact on air quality. The purpose of the CAIR threshold for PM 2.5 is to determine whether the annual average contribution from a collection of sources in a State is small enough not to warrant any additional control for the purpose of mitigating interstate transport, even if that control were highly cost effective.

One commenter suggested that EPA also establish and evaluate a threshold for a potential new tighter 24-hour PM 2.5 standard (e.g., 1 percent of 30 μg/m 3). The EPA must base its criteria on evaluation of the current PM 2.5 standards and not standards that may be considered in the future.

c. Today's Action

The EPA continues to believe that the threshold for evaluating the air quality component of determining whether an individual State's emissions “contribute significantly” to downwind nonattainment of the annual PM 2.5 standard, under CAA section 110(a)(2)(D) should be very small compared to the NAAQS. We are, however, persuaded by commenters arguments on monitoring and modeling that the precision of the threshold should not exceed that of the NAAQS. Rounding the proposal value of 0.15, the nearest single digit corresponding to about 1% of the PM 2.5 annual NAAQS is 0.2 μg/m 3. The final rule is based on this threshold. The EPA has decided to apply this threshold such that any model result that is below this value (0.19 or less)indicates a lack of significant contribution, while values of 0.20 or higher exceed the threshold. [42]

Using this metric for determining whether a State “contributes significantly” (before considering cost) to PM 2.5 nonattainment, our updated modeling shows that Kansas, Massachusetts, New Jersey, Delaware, and Arkansas (all included in the original proposal) no longer exceed the 0.2 μg/m 3 annual average PM 2.5 contribution threshold. Of these states, only Arkansas would exceed the threshold of 0.15 μg/m 3 that was included in the proposal.

E. What Criteria Should Be Used To Determine Which States Are Subject to This Rule Because They Contribute to Ozone Nonattainment?

1. Notice of Proposed Rulemaking

In assessing the contribution of upwind States to downwind 8-hour ozone nonattainment, EPA proposed to follow the approach used in the NO X SIP Call and to employ the same contribution metrics, but with an updated model and updated inputs that reflect current requirements (including the NO X SIP Call itself). [43]

The air quality modeling approach we proposed to quantify the impact of upwind emissions includes two different methodologies: Zero-out and source apportionment. As described in section VI, EPA applied each methodology to estimate the impact of all of the upwind State's NO X emissions on each downwind nonattainment areas.

The EPA's first step in evaluating the results of these methodologies was to remove from consideration those States whose upwind contributions were very low. Specifically, EPA considered an upwind State not to contribute significantly to a downwind nonattainment area if the State's maximum contribution to the area was either (1) less than 2 ppb, as indicated by either of the two modeling techniques; or (2) less than one percent of total nonattainment in the downwind area. [44]

If the upwind State's impact exceeded these thresholds, then EPA conducted a further evaluation to determine if the impact was high enough to meet the air quality portion of the “contribute significantly” standard. In doing so, EPA organized the outputs of the two modeling techniques into a set of “metrics.” The metrics reflect three key contribution factors:

  • The magnitude of the contribution (actual amount of ozone contributed by emissions in the upwind State to nonattainment in the downwind area);
  • The frequency of the contribution (how often contributions above certain thresholds occur); and
  • The relative amount of the contribution (the total ozone contributed by the upwind State compared to the total amount of nonattainment ozone in the downwind area).

The specific metrics on which EPA proposed to rely are the same as those used in the NO X SIP Call. Table III-1 lists them for each of the two modeling techniques, and identifies their relationship to the three key contribution factors.

Table III-1.—Ozone Contribution Factors and Metrics Back to Top
Factor Modeling technique
Zero-out Source apportionment
Magnitude of Contribution Maximum contribution Maximum contribution; and Highest daily average contribution (ppb andpercent).
Frequency of Contribution Number and percent of exceedances with contributions in various concentration ranges Number and percent of exceedances with contributions in variousconcentration ranges.
Relative Amount of Contribution Total contribution relative to the total exceedance ozone in the downwind area; and Population-weighted total contribution relative to the total population-weighted exceedance ozone in the downwind area Total average contribution to exceedance hours in the downwind area.

In the NPR, EPA proposed threshold values for the metrics. An upwind State whose contribution to a downwind area exceeded the threshold values for at least one metric in each of at least two of the three sets of metrics was considered to contribute significantly (before considering cost) to that downwind area. To reiterate, the three sets of metrics reflect the factors of magnitude of contribution, frequency of contribution, and relative percentage on nonattainment.

In fact, EPA noted in the NPR that for each upwind State, the modeling disclosed at least one linkage with a downwind nonattainment area in which all factors (magnitude, frequency, and relative amount) were found to indicate large and frequent contributions. In addition, EPA noted in the NPR that each upwind State contributed to nonattainment problems in at least two downwind States (except for Louisiana and Arkansas which contributed to nonattainment in only 1 downwind State).

In addition, EPA noted in the NPR that for most of the individual linkages, the factors yield a consistent result across all three sets of metrics (i.e., either (i) large and frequent contributions and high relative contributions or (ii) small and infrequent contributions and low relative contributions). In some linkages, however, not all of the factors are consistent. The EPA believes that each of the factors provides an independent, legitimate measure of contribution.

In the NPR, EPA applied the evaluation methodology described above to each upwind-downwind linkage to determine which States contribute significantly (before considering cost) to nonattainment in the 40 downwind counties in nonattainment for ozone in the East. The analysis of the metrics for each linkage was presented in the AQMTSD for the NPR. The modeling analysis supporting the final rule is an update to the NPR modeling, and is described in more detail in section VI below.

2. Comments and EPA Responses

Some commenters submitted comments specifically on the 8-hour ozone metrics. One commenter asserted that in calculating the “Relative Amount of Contribution” metric, EPA treats the modeled reductions from zeroing out a State's emissions as impacting only the portion of the downwind receptor's ambient ozone level that exceeds the 8-hour average 84 ppb level. The commenter asserted that this approach falsely treats the upwind state's emissions as contributing to the amount of ozone that exceeds the NAAQS, and thus inflates the ambient impact of those emissions. The commenter concluded that it would be more appropriate to treat the upwind emissions as impacting all of the downwind ozone level (not just the portion greater than 84 ppb). We interpret this comment to mean that in expressing an upwind State's contribution as a percentage, the denominator of the percentage should be the downwind area's total ozone contribution, rather than the downwind area's ozone excess above the NAAQS, but that the same threshold should be used to evaluate contribution. This would tend to result in fewer upwind States being found to be significant with respect to this metric.

We believe that it is important to examine the ozone contribution relative to the amount of ozone above the NAAQS as well as the amount relative to total nonattainment ozone. Both approaches have merit. The intent of the relative contribution metric, as calculated for the zero-out modeling, is to view the contribution of the upwind State relative to the amount that the downwind area is in nonattainment; that is, the amount of ozone above the NAAQS. However, our relative amount metric for the source apportionment modeling does treat the amount of contribution relative to the total amount of ozone when ozone concentrations are predicted to be above the NAAQS. To be found a significant contributor, an upwind State must be above the threshold for both the zero-out-based metric and the source-apportionment-based metric. Thus, our approach to considering the significance of interstate ozone transport captures both approaches for examining the relative amount of contribution and does not favor one approach over the other, as discussed above.

3. Today's Action

The EPA is finalizing the methodology proposed in the NPR, and discussed above, for evaluating the air quality portion of the “contribute significantly” standard for ozone.

F. Issues Related to Timing of the CAIR Controls

1. Overview

A number of commenters questioned the need for CAIR requirements considering that cap dates of 2010 and 2015 are later than the attainment dates that, in the absence of extensions, would apply to certain downwind PM 2.5 areas and ozone nonattainment areas. Other commenters, noting that states will be required to adopt controls in local attainment plans, questioned whether CAIR controls would still be needed to avoid significant contribution to downwind nonattainment, or whether the controls would still be needed to the extent required by the rule.

Of course, CAIR will achieve substantial reductions in time to help many nonattainment areas attain the standards by the applicable attainment dates. The design of the SO 2 program, including the declining caps in 2010 and 2015 and the banking provisions, will steadily reduce SO 2 emissions over time, achieving reductions in advance of the cap dates; and the 2009 and 2015 NO X reductions will be timely for many downwind nonattainment areas.

Although many of today's nonattainment areas will attain before all the reductions required by CAIR will be achieved, it is clear that CAIR's reductions will still be needed through 2015 and beyond. The EPA's air quality modeling has demonstrated that upwind States have a sufficiently large impact on downwind areas to require reductions in 2010 and 2015 under CAA section 110(a)(2)(D). Under this provision, SIPs must prohibit emissions from sources in amounts that “will contribute significantly to * * * nonattainment” or “will interfere with maintenance”. [45] The EPA has evaluated the attainment status of the downwind receptors in 2010 and 2015, and has determined that each upwind State's 2010 and 2015 emissions reductions are necessary to the extent required by the rule because a downwind receptor linked to that upwind State will either (i) remain in nonattainment and continue to experience significant contribution to nonattainment from the upwind State's emissions; or (ii) attain the relevant NAAQS but later revert to nonattainment due, for example, to continued growth of the emissions inventory.

The argument that the CAIR reductions are justified, in part, by the need to prevent interference with maintenance, is a limited one. The EPA does not believe that the “interfere with maintenance” language in section 110(a)(2)(D) requires an upwind state to eliminate all emissions that may have some impact on an area in a downwind state that is (or once was) in nonattainment and that, therefore, will need (or now needs) to maintain its attainment status. Instead, we believe that CAIR emission reductions are needed beyond 2010 and 2015, in part, to prevent upwind states from significantly interfering with maintenance in other states because our analysis shows it is likely that, in the absence of the CAIR, a current or projected attainment area will revert to nonattainment due to continued emissions growth or other relevant factors. We are not taking the position that CAIR controls are automatically justified to prevent interference with maintenance in every area initially modeled to be in nonattainment.

We also note that considering the emission controls needed for maintenance, along with the controls needed to reach attainment in the first place, is consistent with the goal of promoting a reasonable balance between upwind state controls and local (including all in-state) controls to attain and maintain the NAAQS. As discussed in section IV of this notice, in the ideal world, the states and EPA would have enough information (and powerful enough analytical tools) to allow us to identify a mix of control strategies that would bring every area of the country into attainment at the lowest overall cost to society. Under such an approach, we would evaluate the impact of every emissions source on air quality in all nonattainment areas, the cost of different options for controlling those sources, and the cost-effectiveness of those controls in terms of cost per increment of air quality improvement. Such an approach would obviously make it easier for a state to develop an appropriate set of control requirements for sources located in that state based on (1) the need to bring its own nonattainment areas into attainment and (2) its responsibility under section 110(a)(2)(D) to prevent significant contribution to nonattainment in downwind States and interference with maintenance in those States.

Such an approach would also make it much easier for the Agency to decide on efficiency grounds whether to take action under section 126 (or under section 110(a)(2)(D) if a State failed to meet its obligations under that section) for purposes of either attainment or maintenance of a NAAQS in another State. In the simplest example, we might need to consider a case in which a downwind State with a nonattainment area is seeking reductions from an upwind State based on the claim that emissions from the upwind state are contributing significantly to the nonattainment problem in the downwind State. In such a case, the first question is whether the upwind state should be required to take any action at all, and in the ideal world, it would be simple to answer this question. If emission reductions from sources in the upwind State are more cost-effective than emission reductions in the downwind State—in terms of cost per increment of improvement in air quality in the downwind nonattainment area—then the upwind State would need to take some action to control emissions from sources in that State. [46] On the other hand, if controls on sources in the upwind State are not more cost-effective in terms of cost per increment of improvement in air quality, then the Agency would not take action under sections 126 or 110(a)(2)(D); rather, the downwind State would need to meets its attainment and maintenance needs by controlling sources within its own jurisdiction. Of course, factors other than efficiency, such as equity or practicality, also might affect the decision.

Unfortunately, we do not have adequate information or analytical tools (ideally a detailed linear programming model that fully integrates both control costs and ambient impacts of sources in each State on each of the downwind receptors) to allow us to undertake the analysis described above at this time. However, the Agency believes that CAIR is consistent with this basic approach and will result in upwind States and downwind States sharing appropriate responsibility for attainment and maintenance of the relevant NAAQS, considering efficiency, equity and practical considerations. Under CAIR, the required reductions in upwind States (including those projected to occur after 2015) are highly cost effective, measured in cost-per-ton of emissions reduction, as documented in section IV. This suggests that, regardless of whether the CAIR reductions assist downwind areas in achieving attainment or in subsequently maintaining the relevant NAAQS, the upwind controls will be reasonable in cost relative to a further increment of local controls that, in most cases, will have a substantially higher cost per ton—particularly in areas that need greater local reductions and require reductions from a variety of source types. [47] Thus, we believe that CAIR is consistent with the goal of attaining and maintaining air quality standards in an efficient, as well as equitable, manner.

Another reason for considering both attainment and maintenance needs at this time is EPA's expectation that most nonattainment areas will be able to attain the PM 2.5 and 8-hour ozone standards within the time periods provided under the statute. Considering both types of downwind needs shows that there is a strong basis for CAIR's requirements despite the potential for most receptor areas to attain before all the emission reductions required by CAIR are achieved.

2. By Design, the CAIR Cap and Trade Program Will AchieveSignificant Emissions Reductions Prior to the Cap Deadlines

The EPA notes that Phase I of CAIR is the initial step on the slope of emissions reduction (i.e., the “glide path”) leading to the final control levels. Because of the incentive to make early emission reductions that the cap and trade program provides, reductions will begin early and will continue to increase through Phases I and II. Therefore, all the required Phase II emission reductions will not take place on January 1, 2015, the effective date of the second phase cap. Rather, these reductions will accrue throughout the implementation period, as the sources install controls and start to test and operate them. The resulting glide path of reductions with CAIR Phase II will provide important reductions to areas coming into attainment over the 2010 to 2014 period. [48]

3. Additional Justification for the SO 2 and NO X Annual Controls

Our modeling indicates that it is very plausible that a significant number of downwind PM 2.5 receptors are likely to remain in nonattainment in 2010 and beyond. As noted below (Preamble Table VI-10), the Agency has evaluated a wide range of emission control options and found that the average ambient reduction in PM 2.5 concentrations achievable through aggressive but feasible local controls is 1.26 μg/m [3] . In the 2010 base case (which does not consider potential local controls or 2010 CAIR controls, but does consider all other emission controls required to be in effect as of that date), nearly half the receptor counties would be in nonattainment by more than this amount. This indicates that nonattainment is of sufficient severity to make it likely that, in the absence of CAIR, many of these areas would need an attainment date extension of at least one year.

Our base case modeling further shows that every upwind state is linked to at least one receptor area projected to have nonattainment of this severity. Tables VI-10 and VI-11. Thus, there is a reasonable likelihood that CAIR controls will be needed from all of the upwind states to prevent significant contribution to these downwind receptors' nonattainment.

Nor is the amount of reduction in excess of what is needed for attainment. We project that even with CAIR controls, almost all of the upwind states in 2010 remain linked with at least one downwind receptor that would not attain by the same substantial margin exceeding the average of aggressive local controls. Tables VI-10 and VI-8. This not only indicates that the 2010 CAIR controls are not excessive, but that local controls will still be necessary for attainment.

In addition, there is potential for residual nonattainment in 2015 in view of the severity of PM 2.5 levels in some areas, uncertainties about the levels of reductions in PM 2.5 and precursors that will prove reasonable over the next decade, the potential for up to two 1-year extensions for areas that meet certain air quality levels in the year preceding their attainment date, and historical examples in which areas did not meet their statutory attainment dates for other NAAQS.

With respect to the argument that phase II emission reductions that will be achieved after 2015 are not needed because all receptors will have attained before 2015, we think it likely that some PM 2.5 nonattainment areas may qualify for 2014 attainment dates and eventually, one-year attainment date extensions, and that there may be residual nonattainment in 2015. We continue to project that nearly half the downwind receptors in the 2015 base case will be in nonattainment by amounts exceeding the average ambient reduction (again, 1.26 μg/m [3] ) attributable to local controls we believe would be aggressive but feasible for 2010. Table VI-11. The history of progress in development of emission reduction strategies and technologies indicates that greater local reductions could be achieved by 2015 than in 2010; nonetheless, this potential nonattainment is of sufficient severity to make it plausible that at least some of these areas will need an extension. In such cases, this would eliminate the issue of timing raised by commenters, since CAIR controls would no longer be following attainment dates.

Our modeling further shows that, in the 2015 base case (which does not include CAIR controls), all the upwind states in the CAIR region are linked to areas projected to exceed the standard by at least 2 μg/m [3] . Tables VI-11 and VI-8. Given the reasonable potential for continued nonattainment, it is reasonable to require 2015 CAIR controls from each upwind state to prevent significant contribution to nonattainment.

Moreover, even with 2015 CAIR controls (but not attainment SIP controls), almost all of the upwind states remain linked with at least one downwind receptor that would not attain by at least this same substantial margin (at least 1.26 μg/m [3] ). Id. This shows that the 2015 CAIR controls are not more than are necessary to attain the NAAQS (and also shows the necessity for local controls in order to attain). Thus, we conclude that the further PM 2.5 reductions achieved by the second phase cap will likely be needed to assure all relevant areas reach attainment by applicable deadlines.

Even if some of these areas make more progress than we predict, many downwind receptor areas would be likely in 2010 and 2015 to continue to have air quality only marginally better than the standard, and be at risk of returning to nonattainment. Air quality is unlikely to be appreciably cleaner than the standard because many areas will need steep reductions merely to attain, given that we project nonattainment by wide margins (as explained above).

Moreover, we project that without CAIR, PM 2.5 levels would worsen in 19 downwind receptor counties between 2010 and 2015, reflecting changes in local and upwind emissions. Air Quality Modeling Technical Support Document, November, 2004. This suggests a reasonable likelihood that, without CAIR, these areas would return to nonattainment. See 63 FR at 57379-80 (finding in NO X SIP Call that upwind emissions interfere with maintenance of 8-hour ozone standard under section 110(a)(2)(D)(i) where increases in emissions of ozone precursors are projected due to growth in emissions generating activity, resulting in receptors no longer attaining the standard). These downwind receptors link to all but two of the upwind states, and the remaining two upwind states are linked to receptors where projected PM 2.5 levels between 2010 and 2015 improve only slightly, leaving their air quality only marginally in attainment. Response to Comments, section III.C. In light of documented year-to-year variations in PM 2.5 levels, these receptors would have a reasonable probability of returning to nonattainment in the absence of CAIR.

Emissions trends after 2015 give rise to further maintenance concerns. Between 2015 and 2020, emissions of PM 2.5 and certain precursors are projected to rise. We do not have air quality modeling for 2020. However, for PM 2.5 and every precursor, the 2015-2020 emission trend is less favorable than the 2010-2015 emission trend. Given the PM 2.5 increases our air quality modeling found for 19 counties between 2010 and 2015, the emission trends suggest greater maintenance concerns in the 2015-2020 period than during the 2010-2015 period. See Response to Comments section III.C.

Accordingly, we believe that given these projected trends, and the likelihood of only borderline attainment, CAIR controls from every upwind state in the CAIR region are needed to prevent interference with maintenance of the PM 2.5 standard. The projected upwards pressure on PM 2.5 concentrations in most receptor areas indicates that the amount of upwind reductions is not more than necessary to prevent interference with maintenance of the standards, again given the likelihood of initial attainment by narrow margins.

4. Additional Justification for Ozone NO X Requirements

We believe that most 8-hour ozone areas will be able to attain by their attainment deadlines through existing measures, 2009 CAIR NO X reductions, and additional local measures. However, we also believe that a limited number of downwind receptor areas will remain in nonattainment with the ozone standard after 2010. This is due to the severity of projected ozone levels in certain areas, uncertainties about the levels of emissions reductions in that will prove reasonable over the next decade, and historical difficulties with attaining the 1-hour ozone standard.

For ozone, the historic difficulties that many areas, particularly large urban areas, have experienced in attaining the ozone NAAQS raises the possibility that some areas may not attain by their attainment dates, and may request a voluntary bump up to a higher classification pursuant to section 181(b)(2) to gain an extension, or may fail to attain by the attainment date and be bumped up under section 181(b)(2). These authorities were used in the course of implementing the 1-hour ozone NAAQS.

Our base case modeling (without CAIR, and without state controls implementing the 8-hour standard) projects geographically widespread nonattainment with the 8-hour ozone NAAQS in 2015. Tables VI-12 and VI-13. Five counties that link to 14 upwind states have projected ozone levels that exceed the 8-hour standard by 6 ppb or more, and 20 upwind states are linked to counties projected to exceed the 8-hour standard by more than 4 ppb. These two sets of linkages show that under a scenario in which several of the receptors with the highest ozone levels did not attain, CAIR reductions would be justified to prevent significant contributions from many of the upwind states in the CAIR ozone region.

The fact that receptors show significant nonattainment even after implementation of the phase II CAIR reductions, as shown in Table VI-13, indicates that these reductions would not be more than necessary to prevent significant contribution to nonattainment in residual areas. Even if all ozone nonattainment areas in the CAIR region could achieve reductions sufficient to meet the level of the 8-hour ozone standard in 2009 [49] based on local controls, 2009 CAIR NO X reductions, and existing programs, we believe that numerous downwind receptor areas would remain close enough to the standard to be at risk of falling back into nonattainment for the reasons discussed below. These receptor areas are linked to all states in the CAIR ozone region.

First, it is highly unlikely that the receptor areas will be able to attain by a wide margin. This is primarily because many of those areas will need substantial emissions reductions merely to attain. This is supported by modeling showing that in the 2010 base case, 30 percent of the receptors are projected to be in nonattainment by the wide margin of 6 ppb or more, indicating the steep emissions reductions necessary just to come into attainment. Table VI-12. We recognize that, unlike the trend in key PM receptor areas, our modeling projects that the ozone levels in ozone receptor areas will improve somewhat between 2010 and 2015 due chiefly to downward trends in NO X emissions projected under existing requirements. Nonetheless, as shown in detail in the Response to Comments, the projected improvements in ozone levels in the receptor areas are less (often considerably less) than historic variability in monitored 8-hour ozone design values from one three year period to the next. [50] We believe this variability is mostly attributable to changing weather conditions (which significantly affect the rate at which ozone is formed in the atmosphere and movement of ozone after it is formed), rather than variability in the emissions inventory. Thus, absent the second phase CAIR cap, these receptors remain vulnerable to falling back into nonattainment. The receptors for which this is the case link to each of the upwind States in the ozone CAIR region.

IV. What Amounts of SO 2 and NO X Emissions Did EPA Determine Should Be Reduced? Back to Top

In today's rule, EPA requires annual SO 2 and NO X emissions reductions and ozone-season NO X emissions reductions to eliminate the amount of emissions that contribute significantly to nonattainment of the NAAQS for PM 2.5 and ozone. The NO X reductions are phased in beginning in 2009, the SO 2 reductions beginning in 2010, and both caps are lowered in 2015. In this section of the preamble, EPA explains its analysis of the cost portion of the contribute-significantly test, which determines the amount of required emissions reductions. The cost portion requires analysis of whether the control program under review is highly cost effective, and other factors that are discussed below in section IV.A.

In section IV.A of today's preamble, EPA explains its methodology for determining the amounts of SO 2 and NO X emissions that must be eliminated for compliance with the CAIR. Section IV.A is divided into IV.A.1, IV.A.2, IV.A.3, and IV.A.4. In IV.A.1, EPA explains the methodology that the Agency used to model control costs for evaluation of cost effectiveness. In IV.A.2, EPA describes the methodology that was proposed in the NPR for determining the amounts of emissions that must be eliminated, including an overview of the proposed methodology, a description of the NO X SIP Call regulatory history in relation to the proposed methodology, and a description of EPA's proposed criteria for determining emission reduction requirements. Section IV.A.3 summarizes some comments received regarding the proposed methodology. Section IV.A.4 describes EPA's evaluation of highly cost-effective SO 2 and NO X emissions reductions based on controlling EGUs.

Section IV.A.4 is further divided into IV.A.4.a and IV.A.4.b, which address SO 2 and NO X emission reduction requirements, respectively. Section IV.A.4.a describes EPA's evaluation of highly cost-effective SO 2 reduction requirements, beginning with a summary of the proposal and then describing today's final determination. In IV.A.4.b., EPA describes its evaluation of highly cost-effective NO X reduction requirements, also beginning with a summary of the proposal and then describing today's final determination. Section IV.A.4.b first addresses annual NO X reductions, and then addresses ozone season NO X reductions. The final regionwide CAIR SO 2 and NO X control levels are provided within section IV.A, while a more detailed description of today's final emission reduction requirements is presented in section IV.D.

In section IV.B of today's preamble, EPA discusses other (non-EGU) sources that the Agency considered in developing today's rule.

Section IV.C of today's preamble explains the schedule for implementing today's SO 2 and NO X emissions reductions requirements. This section begins with an overview of the schedule (see section IV.C.1), then provides a detailed discussion of the engineering factors that affect timing for control retrofits (section IV.C.2). Within IV.C.2, EPA first describes the NPR discussion of engineering factors including the availability of boilermaker labor as a limitation (IV.C.2.a), then presents some comments received (IV.C.2.b) and EPA's responses (IV.C.2.c). In section IV.C.3, EPA discusses the financial stability of the power sector in relation to the schedule for the CAIR.

Section IV.D of today's preamble provides a detailed description of the final CAIR emission reduction requirements. Regionwide SO 2 and NO X control levels, projected base case emissions and emissions after the CAIR, and projected emissions reductions are presented. Section IV.D begins with a description of the criteria used to determine final control requirements and provides the details of the final requirements.

A. What Methodology Did EPA Use To Determine the Amounts of SO 2 and NO X Emissions That Must Be Eliminated?

1. The EPA's Cost Modeling Methodology

The EPA conducted analysis using the Integrated Planning Model (IPM) that indicates that its CAIR SO 2 and NO X reduction requirements are highly cost effective. Cost effectiveness is one portion of the contribute-significantly test. The EPA uses the IPM to examine costs and, more broadly, analyze the projected impact of environmental policies on the electric power sector in the 48 contiguous States and the District of Columbia. The IPM is a multi-regional, dynamic, deterministic linear programming model of the U.S. electric power sector. The EPA used the IPM to evaluate the cost and emissions impacts of the policies required by today's action to limit annual emissions of SO 2 and NO X and ozone season emissions of NO X from the electric power sector (on the assumption that all affected States choose to implement reductions by controlling EGUs using the model cap and trade rule).

The EPA conducted analyses for the final CAIR using the 2004 update of the IPM, version 2.1.9. Documentation describing the 2004 update is in the CAIR docket and on EPA's Web site. Some highlights of the 2004 update include: Updated inventory of electric generating units (EGUs) and installed pollution control equipment; updated State emission regulations; updated coal choices available to generating units; updated natural gas supply curves; updated SCR and SNCR cost assumptions; updated assumptions on performance of NO X combustion controls; updated title IV SO 2 bank assumptions; updated heat rates and SO 2 and NO X emission rates; and, updated repowering costs.

The National Electric Energy Data System (NEEDS) contains the generation unit records used to construct model plants that represent existing and planned/committed units in EPA modeling applications of the IPM. The NEEDS includes basic geographic, operating, air emissions, and other data on all the generation units that are represented by model plants in EPA's v.2.1.9 update of the IPM.

The IPM uses model run years to represent the full planning horizon being modeled. That is, several years in the planning horizon are mapped into a representative model run year, enabling the IPM to perform multiple-year analyses while keeping the model size manageable. Although the IPM reports results only for model run years, it takes into account the costs in all years in the planning horizon. In EPA's v.2.1.9 update of the IPM, the years 2008 through 2012 are mapped to run year 2010, and the years 2013 through 2017 are mapped to run year 2015. [51] Model outputs for 2009 and 2010 are from the 2010 run year. Model outputs for 2015 are from the 2015 run year.

The EPA used the IPM to conduct the cost-effectiveness analysis for the emissions control program required by today's action. The model was used to project the incremental electric generation production costs that result from the CAIR program. These estimates are used as the basis for EPA's estimate of average cost and marginal cost of emissions reductions on a per ton basis. The model was also used to project the marginal cost of several State programs that EPA considers as part of its base case.

In modeling the CAIR with the IPM, EPA assumes interstate emissions trading. While EPA is not requiring States to participate in an interstate trading program for EGUs, we believe it is reasonable to evaluate control costs assuming States choose to participate in such a program since that will result in less expensive reductions. The EPA's IPM analyses for the CAIR includes all fossil fuel-fired EGUs with generating capacity greater than 25 MW.

The EPA's IPM modeling accounts for the use of the existing title IV bank of SO 2 allowances. The projected EGU SO 2 emissions in 2010 and 2015 are above the cap levels, because of the use of the title IV bank. The annual SO 2 emissions reductions that are achieved in 2010 and 2015 are based on the caps that EPA determined to be highly cost effective, including the existence of the title IV bank.

The final CAIR requires annual SO 2 and NO X reductions in 23 States and the District of Columbia, and also requires ozone season NO X reductions in 25 States and the District of Columbia. Many of the CAIR States are affected by both the annual SO 2 and NO X reduction requirements and the ozone season NO X requirements.

The EPA initially conducted IPM modeling for today's final action using a control strategy that is similar but not identical to the final CAIR requirements. [52] Many of the analyses for the final CAIR are based on that initial modeling, as explained further below. The control strategy that EPA initially modeled included three additional States (Arkansas, Delaware and New Jersey) within the region required to make annual SO 2 and NO X reductions. However, these three States are not required to make annual reductions under the final CAIR. (In the “Proposed Rules” section of today's Federal Register, EPA is publishing a proposal to include Delaware and New Jersey in the CAIR region for annual SO 2 and NO X reductions.) The addition of these three States made a total of 26 States and the District of Columbia covered by annual SO 2 and NO X caps for the initial model run. The initial model run also included individual State ozone season NO X caps for Connecticut and Massachusetts, and did not include ozone season NO X caps for any other States.

The Agency conducted revised final IPM modeling that reflects the final CAIR control strategy. The final IPM modeling includes regionwide annual SO 2 and NO X caps on the 23 States and the District of Columbia that are required to make annual reductions, and includes a regionwide ozone season NO X cap on the 25 States and the District of Columbia that are required to make ozone season reductions. The EPA modeled the final CAIR NO X strategy as an annual NO X cap with a nested, separate ozone season NO X cap.

In this section of today's preamble, the projected CAIR costs and emissions are generally derived from the final IPM run reflecting the final CAIR. However, some of EPA's analyses are based on the initial IPM run, described above, which reflected a similar but not identical control strategy to the final CAIR. Analyses that are presented in this section of the preamble that are based on the initial IPM run include: IPM sensitivity runs that examine the effects of using the Energy Information Administration (EIA) natural gas price and electricity growth assumptions; marginal cost effectiveness curves developed using the Technology Retrofitting Updating Model; estimates of average annual SO 2 and NO X control costs and average non-ozone season NO X control costs, and projected control retrofits used in the feasibility analysis. The air quality analysis in section VI of today's preamble and the benefits analysis in section X, as well as the analyses presented in the Regulatory Impact Analysis (RIA), are based on emissions projections from the initial IPM run.

The EPA believes that the differences between the initial IPM run that the Agency used for many of the analyses for the CAIR, and the final IPM run reflecting the final CAIR requirements, have very little impact on projected control costs and emissions. For the two IPM runs, projected marginal costs of CAIR annual NO X reductions in 2009 and 2015 are identical. In addition, for the two IPM runs, projected marginal costs of CAIR annual SO 2 reductions in 2010 and 2015 are almost identical. Also, the 2009 and 2015 projected annual NO X emissions in the region encompassing the States that are affected by the final CAIR annual NO X requirements are virtually identical when compared between the two model runs (difference between projected NO X emissions is less than 1 percent for 2009 and less than 2 percent for 2015). In addition, the 2010 and 2015 projected annual SO 2 emissions in the region encompassing the States that are affected by the final CAIR annual SO 2 requirements are virtually the same when compared between the two runs (difference between projected SO 2 emissions is less than 1 percent for 2010 and less than 2 percent for 2015). These comparisons confirm EPA's belief that the initial IPM run very closely represents the final CAIR program.

The IPM output files for the model runs used in CAIR analyses are available in the CAIR docket. A Technical Support Document in the CAIR docket entitled “Modeling of Control Costs, Emissions, and Control Retrofits for Cost Effectiveness and Feasibility Analyses'' further explains the IPM runs used in the analyses for section IV of the preamble.

2. The EPA's Proposed Methodology To Determine Amounts of Emissions That Must be Eliminated

a. Overview of EPA Proposal for the Levels of Reductions and Resulting Caps, and Their Timing

In the NPR, the amounts of SO 2 and NO X emissions reductions that EPA proposed could be cost effectively eliminated in the CAIR region in 2010 and 2015, and the amount of the proposed EGU emissions caps for SO 2 and NO X that would exist if all affected States achieved those reductions by capping EGU emissions, appear in Tables IV-1 and IV-2, respectively.

Table IV-1.—Projected SO 2 and NO X Emission Reductions in the CAIR Region in 2010 and 2015 for the Proposed Rule Back to Top
Pollutant 2010 2015
[Million Tons]1
1CAIR Notice of Proposed Rulemaking (69 FR 4618, January 30, 2004). The proposed annual SO 2 and NO X caps covered a 27-State (AL, AR, DE, FL, GA, IL, IN, IA, KS, KY, LA, MD, MA, MI, MN, MO, NJ, NY, NC, OH, PA, SC, TN, TX, VA, WV, WI) plus DC region. In addition, we proposed an ozone-season only cap for Connecticut.
SO 2 3.6 3.7
NO X 1.5 1.8
Table IV-2.—Proposed Annual Electric Generating Unit SO 2 and NO X Emissions Caps in the CAIR Region Back to Top
Pollutant 2010-2014 2015 and later
[Million Tons]1
1CAIR Notice of Proposed Rulemaking (69 FR 4618, January 30, 2004). The proposed annual SO 2 and NO X caps covered a 27-State (AL, AR, DE, FL, GA, IL, IN, IA, KS, KY, LA, MD, MA, MI, MN, MO, NJ, NY, NC, OH, PA, SC, TN, TX, VA, WV, WI) plus DC region. In addition, we proposed an ozone-season only cap for Connecticut.
SO 2 3.9 2.7
NO X 1.6 1.3

In the NPR, EPA evaluated the amounts of SO 2 and NO X emissions in upwind States that contribute significantly to downwind PM 2.5 nonattainment and the amounts of NO X emissions in upwind States that contribute significantly to downwind ozone nonattainment. That is, EPA determined the amounts of emissions reductions that must be eliminated to help downwind States achieve attainment, by applying highly cost-effective control measures to EGUs and determining the emissions reductions that would result.

From past experience in examining multi-pollutant emissions trading programs for SO 2 and NO X, EPA recognized that the air pollution control retrofits that result from a program to achieve highly cost-effective reductions are quite significant and can not be immediately installed. Such retrofits require a large pool of specialized labor resources, in particular, boilermakers, the availability of which will be a major limiting factor in the amount and timing of reductions.

Also, EPA recognized that the regulated industry will need to secure large amounts of capital to meet the control requirements while managing an already large debt load, and is facing other large capital requirements to improve the transmission system. Furthermore, allowing pollution control retrofits to be installed over time enables the industry to take advantage of planned outages at power plants (unplanned outages can lead to lost revenue) and to enable project management to learn from early installations how to deal with some of the engineering challenges that will exist, especially for the smaller units that often present space limitations.

Based on these and other considerations, EPA determined in the NPR that the earliest reasonable deadline for compliance with the final highly cost-effective control levels for reducing emissions was 2015 (taking into consideration the existing bank of title IV SO 2 allowances). First, the Agency confirmed that the levels of SO 2 and NO X emissions it believed were reasonable to set as annual emissions caps for 2015 lead to highly cost-effective controls for the CAIR region.

Once EPA determined the 2015 emissions reductions levels, the Agency determined a proposed first (interim) phase control level that would commence January 1, 2010, the earliest the Agency believed initial pollution controls could be fully operational (in today's final action, the first NO X control phase commences in 2009 instead of in 2010, as explained in detail in section IV.C). The first phase would be the initial step on the slope of emissions reductions (the glide-path) leading to the final (second) control phase to commence in 2015. The EPA determined the first phase based on the feasibility of installing the necessary emission control retrofits, as described in section IV.C.

Although EPA's primary cost-effectiveness determination is for the 2015 emissions reductions levels, the Agency also evaluated the cost effectiveness of the first phase control levels to ensure that they were also highly cost effective. Throughout this preamble section, EPA reports both the 2015 and 2010 (and 2009 for NO X) cost-effectiveness results, although the first phase levels were determined based on feasibility rather than cost effectiveness. The 2015 emissions reductions include the 2010 (and 2009 for NO X) emissions reductions as a subset of the more stringent requirements that EPA is imposing in the second phase.

b. Regulatory History: NO X SIP Call

In the NPR, EPA generally followed the statutory interpretation and approach under CAA section 110(a)(2)(D) developed in the NO X SIP Call rulemaking. Under this interpretation, the emissions in each upwind State that contribute significantly to nonattainment are identified as being those emissions that can be eliminated through highly cost-effective controls.

In the NO X SIP Call, EPA relied primarily on the application of highly cost-effective controls in determining the amount of emissions that the affected States were required to eliminate. Specifically, EPA developed a reference list of the average cost effectiveness of recently promulgated or proposed controls, and compared the cost effectiveness of those controls to the cost effectiveness of the NO X SIP Call controls under consideration. In addition, EPA considered several other factors, including the fact that downwind nonattainment areas had already implemented ozone controls but upwind areas generally had not, the fact that some otherwise required local controls would be less cost-effective than the regional controls, and the overall ambient effects of the reductions required in the NO X SIP Call (63 FR 57399-57403; October 27, 1998).

i. Highly Cost-Effective Controls

In the NO X SIP Call, EPA presented control costs in 1990 dollars (1990$). For the electric power industry, these expenditures were the increase in annual electric generation production costs in the control region that result from the rule. In the CAIR NPR, SNPR, and today's final action, EPA presents the same type of electric generation as well as other costs in 1999$, and rounds all values related to the cost per ton of air emissions controls to the nearest 100 dollars.

In the NO X SIP Call, EPA's decision on the amount of required NO X emissions reductions was that this amount must be computed on the assumption of implementing highly cost-effective controls. The determination of what constituted highly cost effective controls was described as a two-part process: (1) The setting of a dollar-limit upper bound of highly cost-effective emissions reductions; and (2) a determination of what level of control below this upper-bound was appropriate based upon achievability and other factors.

With respect to setting the upper bound of potential highly cost-effective controls, EPA determined this level on the basis of average cost effectiveness (the average cost per ton of pollutant removed). The EPA explained that it relied on average cost effectiveness for two reasons:

Since EPA's determination for the core group of sources is based on the adoption of a broad-based trading program, average cost effectiveness serves as an adequate measure across sources because sources with high marginal costs will be able to take advantage of this program to lower their costs. In addition, average cost-effectiveness estimates are readily available for other recently adopted NO X control measures (63 FR 57399).

At that time, EPA acknowledged that average cost effectiveness did not directly address the fact that certain units might have higher costs relative to the average cost of reduction (e.g., units with lower capacity factors tend to have higher costs):

[I]ncremental cost effectiveness helps to identify whether a more stringent control option imposes much higher costs relative to the average cost per ton for further control. The use of an average cost effectiveness measure may not fully reveal costly incremental requirements where control options achieve large reductions in emissions (relative to the baseline) (63 FR 57399).

Examination of marginal cost effectiveness—which examines what the cost would be of the next ton of reduction after the defined control level—would fill this gap. However, for the NO X SIP Call rulemaking, adequate information concerning marginal cost effectiveness was not available.

For the NO X SIP Call, to determine the average cost effectiveness that should be considered to be highly cost effective, EPA developed a “reference list” of NO X emissions controls that are available and of comparable cost to other recently undertaken or planned NO X measures. The EPA explained that “the cost effectiveness of measures that EPA or States have adopted, or proposed to adopt, forms a good reference point for determining which of the available additional NO X control measures can most easily be implemented by upwind States whose emissions impact downwind nonattainment problems.” (63 FR 57400). The EPA explained that the measures on the reference list had already been implemented or were planned to be implemented, and therefore could be assumed to be less expensive than other measures to be implemented in the future. The EPA found that the costs of the measures on the reference list approached but were below $2,000 per ton (1990$). The EPA concluded that “controls with an average cost effectiveness [of] less than $2,000 [1990$, or $2,500 (1999$)] per ton of NO X removed [should be considered] to be highly cost-effective.” (63 FR 57400). Notably, the reference costs were taken from the supporting analyses used for the regulatory actions covering the NO X pollution controls—they are what regulatory decision makers and the public believed were the control costs.

Mindful of this $2,000 limit [1990$, or $2,500 (1999$)], EPA considered a control level that would have resulted in estimated average costs of approximately $1,800 (1990$) per ton. However, EPA concluded that because the corresponding level of controls—nominally a 0.12 lb/mmBtu control level—was not well enough established, EPA was “not as confident about the robustness” of the cost estimates. Moreover, EPA expressed concern that its “level of comfort” was not as high as it would have liked that the nominal 0.12 lb/mmBtu control level “will not lead to installation of SCR technology at a level and in a manner that will be difficult to implement or result in reliability problems for electric power generation” (63 FR 57401).

Accordingly, EPA selected the next control level that it had evaluated—a nominal 0.15 lb/mmBtu level—which would result in an average cost of approximately $1,500 [1990$, or $1,900 (1999$)] per ton. The EPA determined that this control level did not present the uncertainty concerns associated with the 0.12 level. The EPA added, in this 1998 rule: “With a strong need to implement a program by 2003 that is recognized by the States as practical, necessary, and broadly accepted as highly cost-effective, the Agency has decided to base the emissions budgets for EGUs on a 0.15 * * * level.” (63 FR 57401—57402). The EPA summarized its approach as determining “the required emission levels * * * based on the application of NO X controls that achieve the greatest feasible emissions reduction while still falling within a cost-per-ton reduced range that EPA considers to be highly cost-effective.* * *” (63 FR 57399).

The bulk of the cost for reducing NO X emissions for EGUs is in the capital investment in the control equipment, which would be the same whether controls are installed for ozone season only, or for annual controls. The increased costs to run the equipment annually instead of only in the ozone season is relatively small. Although the NO X SIP Call is an ozone season NO X reduction program, most of the NO X control costs on the reference list are for annual reductions. If the NO X SIP Call were an annual program instead of seasonal, its average control costs would be lower, relative to the annual control costs in the reference list.

ii. Other Factors

In the NO X SIP Call, although considering air quality and cost to be the primary factors for determining significant contribution, EPA identified several other factors that it generally considered. As one factor, EPA reviewed “overall considerations of fairness related to the control regimes required of the downwind and upwind areas,” particularly, the fact that the major urban nonattainment areas in the East had implemented controls on virtually all portions of their inventory of ozone precursors, but upwind sources had not implemented reductions intended to reduce their impacts downwind (63 FR 57404).

As another factor, EPA generally considered “the cost effectiveness of additional local reductions in the * * * ozone nonattainment areas.” The EPA included in the record information that nationally, on average, additional local measures would cost more than the cost of the upwind controls required under the NO X SIP Call. This consideration further indicated that the regional controls under the NO X SIP Call were highly cost effective (63 FR 57404).

In addition, EPA conducted air quality modeling to determine the impact of the controls, and found that they benefitted the downwind areas without being more than necessary for those areas to attain (63 FR 57403—57404).

c. Proposed Criteria for Emissions Reduction Requirements

i. General Criteria

In the CAIR NPR, EPA proposed criteria for determining the appropriate levels of annual emissions reductions for SO 2 and NO X and ozone-season emissions reductions for NO X. The EPA stated that it considers a variety of factors in evaluating the source categories from which highly cost-effective reductions may be available and the level of reduction assumed from that sector. These include:

  • The availability of information,
  • The identification of source categories emitting relatively large amounts of the relevant emissions,
  • The performance and applicability of control measures,
  • The cost effectiveness of control measures, and
  • Engineering and financial factors that affect the availability of control measures (69 FR 4611).

Further, EPA stated that overall, “We are striving * * * to set up a reasonable balance of regional and local controls to provide a cost-effective and equitable governmental approach to attainment with the NAAQS for fine particles and ozone.” (69 FR 4612)

The EPA has used these types of criteria in a number of efforts to develop regional and national strategies to reduce interstate transport of SO 2 and NO X. Starting in 1996, EPA performed analysis and engaged in dialogue with power companies, States, environmental groups and other interested groups in the Clean Air Power Initiative (CAPI). [53] In that study of national emission reduction strategies, EPA initially considered an emissions cap based on a 50 percent reduction in SO 2 emissions from title IV levels (i.e., 4.5 million tons nationwide) in 2010. For NO X, EPA initially looked at ozone season and non-ozone season caps. Commencing in 2000, the ozone season emissions cap would be based on an emission rate of 0.20 lb/mmBtu, and in 2005, the ozone season cap would be reduced to a level based on 0.15 lb/mmBtu (these cap levels would be similar to the phased caps adopted by the Ozone Transport Commission (OTC) States). The non-ozone season cap would be based on the proposed title IV phase II NO X rule. The EPA also considered other options in the CAPI study, including setting NO X caps based on emission rates of 0.20 lb/mmBtu and 0.25 lb/mmBtu; setting NO X caps based on rates of 0.15 lb/mmBtu and 0.20 lb/mmBtu but lowering the SO 2 allowance cap by 60 percent instead of 50 percent; and, keeping a NO X cap based on a rate of 0.15 lb/mmBtu but lowering the SO 2 allowance cap by 50 percent in 2005 instead of in 2010.

The EPA did a follow-up study in 1999 and discussed those results with various stakeholder groups, as well. [54] That study considered a variety of SO 2 emission caps ranging from a 40 percent reduction from title IV cap levels in 2010 to a 55 percent reduction from title IV cap levels in 2010. The 1999 study did not consider additional reductions in NO X emissions beyond those required under the NO X SIP Call.

In the last several years, EPA has performed significant additional analysis in support of the proposed Clear Skies Act. [55] That legislation, proposed in 2002 and 2003, would include nationwide SO 2 caps of 4.5 million tons in 2010 and 3.0 million tons in 2018 (i.e., 50 percent and 67 percent reductions from title IV cap levels). The Clear Skies Act also includes a two-phase, two-zone NO X emission cap program, with the first phase in 2008 and the second phase in 2018. In the 2003 legislation, the first phase NO X caps would result in effective NO X emissions rates of 0.16 lb/mmBtu in the east and 0.20 lb/mmBtu in the west, and the second phase would result in effective emission rates of 0.12 lb/mmBtu in the east and 0.20 lb/mmBtu in the west.

ii. Reliance on Average and Marginal Cost Effectiveness

In the CAIR NPR, EPA supported the conclusion that its emissions caps are highly cost effective based upon “(1) comparison to the average cost effectiveness of other regulatory actions and (2) comparison to the marginal cost effectiveness of other regulatory actions.” (69 FR 4585). We supplemented these comparisons of cost-effectiveness tables with an auxiliary evaluation of the marginal costs curves, which allowed us to show that the selected control levels would be “below the point at which there would be significant diminishing returns on the dollars spent for pollution control.” (69 FR 4614).

Although in the NO X SIP Call, EPA based the required controls on average cost alone, in today's rule, EPA uses both average and marginal costs, including an evaluation of the marginal cost curves. At the time of the NO X SIP Call, marginal cost information was not as readily available. Today, such information is available for both SO 2 and NO X controls, although marginal cost information remains more limited and EPA has had to specifically develop marginal cost estimates for use in this rulemaking.

Marginal costs are a useful measure of cost effectiveness because they indicate how much any additional level of control at the margin will cost relative to other actions that are available. Using both average and marginal control costs, provides a more complete picture of the costs of controls than using average costs alone. Average costs provide a means for a straightforward comparison between the CAIR and other emissions reductions programs for which average costs are generally the only type of costs available. Where marginal cost information is available, it enables EPA to compare the costs of the CAIR at the stringency level being considered to the costs of the last increment of control in other programs. Moreover, evaluation of marginal cost curves allows us to corroborate that the selected level of stringency of the selected program stops short of the point where the returns begin to diminish significantly.

Projected marginal cost information for controlling emissions from EGUs is now available for some State programs, because EPA includes the programs in its base case power sector modeling using the IPM to develop the incremental costs of electricity production for the CAIR. Marginal EGU control costs from State programs modeled using the IPM were compared to projected marginal EGU control costs under the CAIR, as discussed in more detail below.

3. What Are the Most Significant Comments That EPA Received About Its Proposed Methodology for Determining the Amounts of SO 2 and NO X Emissions That Must Be Eliminated, and What Are EPA's Responses?

Some commenters took issue with EPA's reliance on cost-per-ton-of-emissions-reductions as the metric for determining cost effectiveness. These commenters observed that this metric does not take into account that any given ton of pollutant reduction may have different impacts on ambient concentration and human exposure. Some of these commenters advocated use of a metric based on cost per unit of pollutant concentration reduced. Another stated that EPA should account for cost effectiveness based on geographical location relative to the area of nonattainment.

Still other commenters took a contrasting view. They argued that a metric based on cost-per-ambient-impact might be useful in justifying control cost effectiveness for source categories within an individual nonattainment area as part of an attainment SIP, but not for evaluating costs of controlling long-range transport. These commenters stated that it is impractical to calculate cost effectiveness of control on the basis of cost per unit reduction in ambient concentration. One queried: “Where would the ambient reduction be measured? 100 miles downwind? 1,500 miles downwind?”

The EPA agrees that optimally, the cost-per-ambient-impact of controls could play a major role in determining upwind control obligations (although equitable considerations and other factors identified in the NO X SIP Call rulemaking and today's action may also play a role). The EPA recognized the potential importance of this factor during the NO X SIP Call rulemaking and endeavored to develop technical information to support it. However, in that rulemaking, EPA was not able to develop an approach to quantify, with sufficient accuracy, cost-per-ambient impact because the NO X SIP Call region was large—covering approximately half of the continental U.S. and including approximately half the States—and many upwind States with different emissions inventories had widely varied impacts on many different nonattainment areas downwind.

This problem—the complexity of the task and the dearth of analytic tools—remains today for both PM 2.5 and 8-hour ozone regional transport. Not surprisingly, no commenter presented to EPA the analytic tools, which we would expect would consist of a complex, computerized program that could integrate, on a State-by-State basis, both control costs and ambient impacts by each State on each of its downwind receptors under the CAIR control scenario.

In the absence of a scientifically defensible, practicable method for implementing a program design approach based on the cost-per-ambient-impact of emissions reductions, EPA is not able to employ such an approach. However, EPA believes it appropriate to continue to examine ways to develop such an approach for future use.

A few commenters suggested that EPA should use a cost-benefit analysis for determining reduction levels. One noted that cost-benefit analysis can help find the reduction levels that maximize societal net benefit (benefits minus costs), and suggested the Agency should compare the marginal cost of each ton of pollutant reduced to the marginal benefit achieved, as well as compare the total costs to the total benefits. Another stated that an optimal allocation of resources is where the marginal cost equals the marginal benefit, and observed that comparing the average cost to the average benefit of the controls proposed in the CAIR NPR yields an average benefit significantly higher than the average cost. This commenter concluded that EPA should require controls beyond the controls described in the NPR as highly cost effective.

Although EPA strongly agrees that examination of costs and benefits is very useful, in today's rulemaking, EPA does not interpret CAA section 110(a)(2)(D) to base the amount of emissions reductions on benefits other than progress towards attainment of the PM 2.5 or the 8-hour ozone NAAQS. The EPA's interpretation does, however, use cost effectiveness per ton of pollutant reduced, and we are using that analytic tool for setting SO 2 and NO X emission reduction requirements. Additionally, EPA has prepared a cost-benefit analysis to inform the Agency and public of the many other important impacts of this rulemaking.

A few commenters suggested that the Agency should set its NO X and SO 2 reduction requirements based on Best Available Control Technology (BACT) emission rates for EGUs. Although not clearly stated, the commenters appear to suggest BACT level controls for both existing and new units.

The emission reduction requirements that EPA determined are based on the application of highly cost-effective controls that are a step that the Agency is taking at this time to eliminate emissions that contribute significantly to nonattainment of the ozone and fine particle NAAQS. As explained elsewhere, this step is reasonable in light of the current status of implementation for those NAAQS.

Basing emission reduction requirements on a presumption of BACT emission rates across the board would require scrubbers and SCRs on all coal-fired units and SCRs on all gas-fired and oil-fired units. The cost of these controls would vary considerably from source to source, be expensive for many sources, and may cause substantial fuel switching to natural gas and closure of smaller coal-fired units. Having considered this suggestion for deeper regional reductions that would not be as cost effective as the highly cost-effective reductions in today's rule, EPA believes that a more tailored approach, such as the CAIR level control as well as local controls under SIPs (where necessary), is a more reasonable approach to achieving the level of ambient improvement needed for attainment throughout the United States.

4. The EPA's Evaluation of Highly Cost-Effective SO 2 and NO X Emissions Reductions Based on Controlling EGUs

a. SO 2 Emissions Reductions Requirements

i. CAIR Proposal for SO 2

The NPR focused primarily on determining highly cost-effective amounts of emissions reductions based on, as in the NO X SIP Call, comparison to reference lists of the cost effectiveness of other regulatory controls. In the NPR, EPA developed reference lists for both the average cost effectiveness and the marginal cost effectiveness of those other controls. These reference lists indicated that the average annual costs per ton of SO 2 removed ranged from $500 to $2,100; and marginal costs of SO 2 removal ranged from $800 to $2,200.

Moreover, EPA further considered the cost effectiveness of alternative stringency levels for this regulatory proposal. That is, EPA examined changes in the marginal cost curve at varying levels of emissions reductions. The EPA determined in the NPR that the “knee” in the marginal cost-effectiveness curve—the point at which the marginal cost per ton of SO 2 removed begins to increase at a noticeably higher rate—appears to start above $1,200 per ton (69 FR 4613—4615).

In the NPR, EPA then provided further analysis of a two-phase SO 2 reduction program. The final (second) phase, in 2015, would reduce SO 2 emissions in the CAIR region by the amount that results from making a 65 percent reduction from the title IV Phase II allowance levels (taking into consideration the existing bank of title IV SO 2 allowances). The first phase, in 2010, would reduce SO 2 emissions in the CAIR region by a lesser amount, i.e., a 50 percent reduction from title IV Phase II allowance levels (again, taking into consideration the banked title IV SO 2 allowances). The EPA developed this target SO 2 control level for further evaluation because, based on all of the earlier work performed on multi-pollutant power plant reduction programs and general consideration, with technical support, of overall emissions reductions, costs to industry and the general public, ambient improvement, and consistency with the emerging PM 2.5 implementation program, we believed it would meet the criteria set forth above.

Then, EPA conducted cost analyses of this control level using the IPM as well as additional analysis of the implications of this control level to determine if it did indeed meet those criteria. The IPM analysis considered the increase in annual electric generation production costs in the CAIR region that result from the rule. The EPA evaluated the cost effectiveness of the final phase (2015) cap to determine if it is highly cost effective; and, we also evaluated the cost effectiveness of the 2010 cap. The EPA used the IPM to estimate cost effectiveness of the CAIR in the future. The IPM incorporates projections of future electricity demand, and thus heat input growth. The EPA's IPM analyses for the CAIR includes all fossil fuel-fired EGUs with capacity greater than 25 MW. A description of the IPM is included elsewhere in this preamble, and a detailed model documentation is in the docket.

The SO 2 annual control costs that were presented in the CAIR NPR were average costs of $700 per ton and $800 per ton for years 2010 and 2015, respectively, and marginal costs of $700 per ton and $1,000 per ton for years 2010 and 2015. In addition, the NPR included the results of sensitivity analyses that examined costs of the proposed SO 2 controls based on the Energy Information Administration's projections for electricity growth and natural gas prices. These sensitivity analyses showed marginal SO 2 control costs of $900 per ton and $1,100 per ton for years 2010 and 2015, respectively. The EPA proposed to consider the SO 2 emissions reductions proposed in the NPR as highly cost effective because they were consistent with the lower end of the reference list range of cost per ton of SO 2 reduction for controls on both an average and a marginal cost basis (69 FR 4613—4615).

ii. Analysis of SO 2 Emission Reduction Requirements for Today's Final Rule

(I) Reference Lists of Cost-Effective SO 2 Controls

For today's action, EPA updated the reference list of controls included in the NPR of the average and marginal costs per ton of recent SO 2 control actions. The EPA systematically developed a list of cost information from both recent actions and proposed actions. The EPA compiled cost information for actions taken by the Agency, and examined the public comments submitted after the NPR was published, to identify all available control cost information to provide the updated reference list for today's preamble. The updated reference list includes both average and marginal costs of control, to which EPA compares the CAIR control costs, and the list represents what regulatory decision makers and/or the public believes are the control costs. [56]

Table IV-3 provides average costs of SO 2 controls. This table includes average costs for recent BACT permitting decisions for SO 2. Under EPA's New Source Review (NSR) program, if a company is planning to build a new plant or modify an existing plant such that a significant net increase in emissions will occur, the company must obtain a NSR permit that addresses controls for air emissions. BACT is the type of control required by the NSR program for existing sources in attainment areas. The BACT decisions are determined on a case-by-case basis, usually by State or local permitting agencies, and reflect consideration of average and incremental cost effectiveness. These decisions are relevant for EPA's reference list of average costs of SO 2 controls, because they represent cost-effective controls that have been demonstrated.

Table IV-3.—Average Costs per Ton of Annual SO 2 Controls Back to Top
SO 2 control action Average cost per ton
1These numbers reflect a range of cost-effectiveness data entered into EPA's RACT/BACT/LAER Clearinghouse (RBLC) for add-on SO 2 controls (www.epa.gov/ttn/catc/). We identified actions in the data base for large, utility-scale, coal-fired boiler units for which cost effectiveness data were reported. The range of costs shown here is for boilers ranging from 30 MW to an estimated 790 MW (we used a conversion factor of 10 mmBtu/hr = 1 MW for units for which size was reported in mmBtu/hr). Emission limits for these actions ranged from 0.10 lb/mmBtu to 0.27 lb/mmBtu. Add-on controls reported for these units are dry or wet scrubbers (in one case with added alkali and in one case with a baghouse). Where the dollar-year was not reported we assumed 1999 dollars. The cost range presented in the NPR was $500-$2,100-today's range includes additional BACT costs that were entered into the clearinghouse after the NPR was published.
2Control of Emissions of Air Pollution From Nonroad Diesel Engines and Fuel; Final Rule (69 FR 39131; June 29, 2004). The value in this table represents the long-term cost per ton of emissions reduced from the total fuel and engine program (cost per ton of emissions reduced in the year 2030). 1999$ per ton.
3The EPA IPM modeling 2004, available in the docket. The EPA modeled the Regional Haze Requirements as source specific limits (90 percent SO 2 reduction or 0.1 lb/mmBtu rate; except the five state WRAP region for which we did not model SO 2 controls beyond what is done for the WRAP cap in the base case modeling). Estimated average costs based on this modeling are $2,600 per ton in 2015 and $3,400 per ton in 2020. 1999$ per ton.
Best Available Control Technology (BACT) Determinations 1$400-$2,100
Nonroad Diesel Engines and Fuel 2$800
Proposed Best Available Retrofit Technology (BART) for Electric Power Sector 3$2,600-$3,400

Table IV-4 provides the marginal cost per ton of recent State and regional decisions for annual SO 2 controls.

Table IV-4.—Marginal Costs per Ton of Annual SO 2 Controls Back to Top
SO 2 control action Marginal cost per ton
1The EPA IPM base case modeling August 2004, available in the docket. (1999$ per ton). We modeled New Hampshire's State Bill ENV-A2900, which caps SO 2 emissions at all existing fossil steam units.
2“An Assessment of Critical Mass for the Regional SO 2 Trading Program,” prepared for Western Regional Air Partnership Market Trading Forum by ICF Consulting Group, September 27, 2002, available in the docket. This analysis looked at the implications of one or more States choosing to opt-out of the WRAP regional SO 2 trading program. (1999$ per ton)
New Hampshire Rule 1$600
WRAP Regional SO 2 Trading Program 2$1,100-$2,200

(II) Cost Effectiveness of the CAIR Annual SO 2 Reductions

In the NPR, EPA evaluated an annual SO 2 control strategy based on a specified level of emissions reductions from EGUs. Available information indicated that emissions reductions from this industry would be the most cost effective. (As noted elsewhere, EPA considered control strategies for other source categories, but concluded that they would not qualify as highly cost-effective controls.) Of course, under today's rule, although EPA calculates the amount of emissions reductions States must achieve by evaluation of the EGU control strategy, States remain free to achieve those reductions by implementing controls on any sources they wish.

For today's action, EPA updated the predicted annual SO 2 control costs included in the NPR. The EPA analyzed the costs of the CAIR using an updated version of the IPM (documentation for the IPM update is in the docket). Further, EPA modified the modeling to match the final CAIR strategy (see section IV.A.1 for a description of EPA's CAIR IPM modeling).

The EPA also updated its analysis of the sensitivity of the marginal cost results to assumptions of higher electric growth and natural gas prices than we used in the base case. These sensitivity analyses were based on the Energy Information Administration's Annual Energy Outlook for 2004. [57]

In determining whether our control strategy is highly cost effective, EPA believes it is important to account for the variable levels of cost effectiveness that these sensitivity analyses indicate may occur if electricity demand or natural gas prices are appreciably higher than assumed in the IPM. Those two factors are key determinants of control costs and, over the relatively long implementation period provided under today's action, a meaningful degree of risk arises that these factors may well vary to the extent indicated by the sensitivity analyses. As a result, EPA wanted to examine the marginal costs that would occur under the scenarios modeled in the sensitivity analyses to see how they differed from the costs using EPA's assumptions.

Table IV-5 provides the average and marginal costs of annual SO 2 reductions under the CAIR for 2010 and 2015. (When presenting estimated CAIR control costs in section IV of this preamble, EPA uses “Main Case” to indicate the primary CAIR IPM analyses, as differentiated from other IPM analyses such as sensitivity runs used to examine the impacts of varying assumptions about natural gas price and electric growth.)

Table IV-5.—Estimated Costs Per Tons of SO 2 Controlled Under CAIR, Cap Levels Beginning in 2010 and 20151 Back to Top
Type of cost effectiveness 2010 2015
1The EPA IPM modeling 2004, available in the docket. $1999 per ton.
Average Cost—Main Case $500 $700
Marginal Cost—Main Case 700 1,000
Sensitivity Analysis: Marginal Cost Using EIA Electric Growth and Natural Gas Prices 800 1,200

These estimated SO 2 control costs under the CAIR reflect annual EGU SO 2 caps of 3.6 million tons in 2010 and 2.5 million tons in 2015 within the CAIR region. Based on IPM modeling, EPA projects that SO 2 emissions in the CAIR region will be about 5.1 million tons in 2010 and 4.0 million tons in 2015. The projected emissions are above the cap levels because of the use of the existing title IV bank of SO 2 allowances. Average costs shown for 2015 are an estimate of the average cost per ton to achieve the total difference in projected emissions between the base case conditions and the CAIR in the year 2015 (the 2015 average costs are not based on the increment in reductions between 2010 and 2015). (A more detailed description of the final CAIR SO 2 and NO X control requirements is provided below in today's preamble.)

(III) SO 2 Cost Comparison for CAIR Requirements

The EPA believes that if an SO 2 control strategy has a cost effectiveness that is at the low end of the updated reference tables, the approach should be considered to be highly cost effective. The costs in the reference range should be considered to be cost effective because they represent actions that have already been taken to reduce emissions. In deciding to require these actions, policymakers at the local, State and Federal levels have determined them to be cost-effective reductions to limit or reduce emissions. Thus, costs at the bottom of the range must necessarily be considered highly cost effective.

Today's action requires SO 2 emissions reductions (or an EGU emissions cap) in 2015. The EPA has determined that those emissions reductions are highly cost effective. In addition, today's action requires that some of those SO 2 emissions reductions (or a higher EGU emissions cap) be implemented by 2010. The EPA has examined the cost effectiveness of implementing those earlier emissions reductions (or cap) by 2010, and determined that they are also highly cost effective.

The cost of the SO 2 reductions required under today's action—if the States choose to implement those reductions through EGUs, for which the most cost-effective reductions are available—on average and at the margin, are at the lower end of the range of cost effectiveness of other, recent SO 2 control requirements. [58] This is true for our analysis of both the costs EPA generally expects as well as the somewhat higher costs that would result from higher than expected electricity demand and natural gas prices, as indicated in the sensitivity analyses that EPA has done.

Specifically, the average cost effectiveness of the SO 2 requirements is $700 per ton removed in 2015. This amount falls toward the low end of the reference range of average costs per ton removed of $400 to $3,400. Similarly, the marginal cost effectiveness of the SO 2 requirements ranges from $1,000 to $1,200 for 2015 (with the higher end of the range based on the sensitivity analyses). These amounts fall toward the lower end of the reference range of marginal cost per ton removed of $600 to $2,200.

The EPA believes that selecting as highly cost-effective amounts toward the lower end of our average and marginal cost ranges for SO 2 and NO X control is appropriate because today's rulemaking is an early step in the process of addressing PM 2.5 and 8-hour ozone nonattainment and maintenance requirements. The CAA requires States to submit section 110(a)(2)(D) plans to address interstate transport, and overall attainment plans to ensure the NAAQS are met in local areas. By taking the early step of finalizing the CAIR, we are requiring a very substantial air emission reduction that addresses interstate transport of PM 2.5 as well as a further reduction in interstate transport of ozone beyond that required by the NO X SIP Call Rule. Much of the air quality improvement resulting from reduced transport is likely to occur through broad and deep emissions reductions from the electric power sector, which has been a major part of the transport problem. Other air quality benefits will occur as the result of Federal mobile source regulations for new sources, which cover passenger vehicles and light trucks, heavy-duty trucks and buses, and non-road diesel equipment.

Against this backdrop of Federal actions that lower air emissions (as well as some substantial State control programs), States will develop plans designed to achieve the standards in their local nonattainment areas. The EPA has not yet promulgated rules interpreting the CAA's requirements for SIPs for PM 2.5 and ozone nonattainment areas, [59] nor have States developed plans to demonstrate attainment. As a result, there are significant uncertainties regarding potential reductions and control costs associated with State plans. We believe that some areas are likely to attain the standards in the near term through early CAIR reductions and local controls that have costs per ton similar to the levels we have determined to be highly cost effective. We expect that other areas with higher PM 2.5 or ozone levels will determine through the attainment planning process that they need greater emissions reductions, at higher costs per ton, to reach attainment within the CAA's timeframes. For those areas, States will need to assess targeted measures for achieving local attainment in a cost-effective (but not necessarily highly cost-effective) manner, in combination with the CAIR's significant reductions. Given the uncertainties that exist at this early stage of the implementation process, EPA believes this rule is a rational approach to determining the highly cost-effective reductions in PM 2.5 and ozone precursors that should be required for interstate transport purposes.

As discussed above, the Agency believes this approach is consistent with our action in the NO X SIP Call. While the cost level selected for the NO X SIP Call was not at the low end of the reference range of costs, if the NO X SIP Call costs were for annual rather than seasonal controls they would have been lower relative to the annual control costs on the list. This would make the relationship between the cost of the NO X SIP Call and the reference costs used in that rulemaking, more similar to relative costs of CAIR compared to its reference lists. Also, significant local controls for meeting the 1-hour ozone standard had already been adopted in many areas.

Although EPA's primary cost-effectiveness determination is for the 2015 emissions reductions levels, the Agency also evaluated the cost effectiveness of the interim phase control levels to ensure that they were also highly cost effective. For the SO 2 requirements for 2010, the average cost effectiveness is $500 per ton removed, and the marginal cost effectiveness ranges from $700 to $800. The 2010 costs indicate that the interim phase CAIR reductions are also highly cost-effective.

(IV) Cost Effectiveness: Marginal Cost Curves for SO 2 Control

As noted above, the Agency also considered another factor to corroborate its conclusion concerning the cost effectiveness of the selected levels of control:

The cost effectiveness of alternative stringency levels for today's action. Specifically, EPA examined changes in the marginal cost curve at varying levels of emissions reductions for EGUs. Figure IV-1 shows that the “knee” in the 2010 marginal cost-effectiveness curve—the point where the cost of controlling a ton of SO 2 from EGUs is increasing at a noticeably higher rate—appears to occur at about $2,000 per ton of SO 2. Figure IV-2 shows that the “knee” in the 2015 marginal cost-effectiveness curve also appears to occur at about $2,000 per ton of SO 2. (As discussed above, the projected marginal costs of SO 2 reductions for the CAIR are $700 per ton in 2010 and $1,000 per ton in 2015.) The EPA used the Technology Retrofitting Updating Model (TRUM), a spreadsheet model based on the IPM, for this analysis. (The EPA based these marginal SO 2 cost-effectiveness curves on the electric growth and natural gas price assumptions in the main CAIR IPM modeling run. Marginal cost effectiveness curves based on other electric growth and natural gas price assumptions would look different, therefore it would not be appropriate to compare the curves here to the marginal costs based on the IPM modeling sensitivity run that used EIA assumptions.) These results make clear that this rule is very cost effective because the control level is below the point at which the cost begins to increase at a significantly higher rate.

In this manner, these results corroborate EPA's findings above concerning the cost effectiveness of the emissions reductions. [60]

b. NO X Emissions Reductions Requirements

i. The CAIR Proposal for NO X and Subsequent Analyses for Regionwide Annual and Ozone Season NO X Control Levels

In this section, EPA describes its proposed method for determining regionwide NO X control levels and the method used for the final CAIR.

In the CAIR NPR, EPA updated the reference list included in the NO X SIP Call for the average annual cost effectiveness of recent or proposed NO X controls, and determined that these amounts ranged from approximately $200 to $2,800. In addition, in the NPR, EPA developed a reference list for marginal annual cost effectiveness for NO X controls, and determined that these amounts ranged from approximately $1,400 to $3,000 (69 FR 4614—4615).

In the NPR, EPA proposed a two-phased annual NO X control program, with a final phase in 2015 and a first phase in 2010. The regionwide emissions reduction requirements that EPA proposed—and the budget levels that would apply if all States chose to implement the reductions from EGUs—were based on using a combination of recent historical heat input and NO X emissions rates for fossil fuel-fired EGUs. For historical heat input, EPA proposed determining the highest heat input from units affected by the Acid Rain Program for each affected State for the years 1999-2002. The EPA then summed this heat input for all of the States affected for annual NO X reductions. For 2015, EPA calculated a proposed regionwide annual NO X budget by multiplying this heat input by an emission rate of 0.125 lb/mmBtu, and for 2010 by multiplying by 0.15 lb/mmBtu.

In developing the CAIR NPR, when EPA considered the appropriate amount of annual SO 2 emissions reductions, EPA relied on the existing title IV annual SO 2 cap as a starting point. However, in considering the appropriate amount of NO X reductions, the situation is different because title IV does not cap NO X emissions. Therefore, EPA and the States have focused on emissions caps based on a combination of heat input and NO X emission rates. Emission rates similar to the rates used to develop the CAIR NPR have been considered in the past. For example, the CAPI 1996 study, noted above, contemplated NO X caps based on an emission rate of 0.15 lb/mmBtu (and other options based on NO X rates of 0.20 lb/mmBtu and 0.25 lb/mmBtu). The NO X SIP Call is based on an emission rate of 0.15 lb/mmBtu.

The methodology described in the NPR is best understood as the means for developing the target 2015 annual NO X control level (or emissions budget) for further evaluation through IPM. The EPA developed this level mindful of its experience to date with the NO X SIP Call and the earlier work EPA has performed on multi-pollutant power plant reduction programs. The EPA also considered available technical information on pollution controls, costs to industry and the general public, ambient air improvement, and consistency with the emerging PM 2.5 implementation program, in developing its target control level.

Recent advances in combustion control technology for NO X reductions, as well as widespread use of selective catalytic reduction (SCR) on U.S. coal-fired EGU boilers achieving NO X emission rates of 0.06 lb/mmBtu and below, provide evidence that even lower average NO X emission rates are more highly cost-effective than rates considered in the past (based on analyzing EGUs), possibly on the order of 0.12 lb/mmBtu or less. The EPA developed the target annual NO X control level (or emissions budget) with the understanding that the evaluation of that level might indicate that average emission rates on the order of 0.12 lb/mmBtu or less might be highly cost effective for the final (2015) control phase, and an interim level resulting in an average emission rate of less than 0.15 lb/mmBtu might be feasible for the first phase.

The EPA did evaluate the target annual NO X control levels (or emissions budgets) using the IPM. The EPA confirmed that the 2015 level is highly cost effective. The Agency also evaluated the cost effectiveness of the proposed 2010 cap to assure that the interim phase reductions would also be highly cost effective. The EPA's IPM analyses for the CAIR includes all fossil fuel-fired EGUs with generating capacity greater than 25 MW.

The proposed cap for the first phase was developed taking into consideration how much pollution control for NO X and SO 2 could be installed without running into a shortage of skilled labor, in particular boilermakers (EPA's assumptions regarding boilermaker labor are described in section IV.C.2 of this preamble). The Agency focused on providing substantial reductions of both SO 2 and NO X emissions at the outset of the proposed program, leading to significant retrofits of Flue Gas Desulfurization units (FGD) for SO 2 control and SCR for NO X control.

In the NPR, EPA explained that using the highest Acid Rain Program heat input for each State to develop a regionwide heat input amount, rather than the average Acid Rain Program heat input, provided a cushion that represented a reasonable adjustment to reflect that there are some non-Acid Rain units that operate in these States that will be subject to the proposed CAIR emission reduction levels. The EPA explained that it did not use heat input data from non-Acid Rain units in the proposal because it did not have all the necessary data available at the time the NPR was developed. [61] Using the highest of recent years' Acid Rain Program heat input provided an approximation of the regionwide heat input, although it did not include heat input from non-Acid Rain sources. Multiplying the approximate recent heat input by 0.125 lb/mmBtu to develop a proposed regionwide annual 2015 NO X cap could reasonably be expected to yield an average effective NO X emission rate (considering all EGUs potentially affected by CAIR for annual reductions, not only the Acid Rain units, and considering growth in heat input) somewhat less than 0.125 lb/mmBtu. Likewise, multiplying the approximate recent heat input by 0.15 lb/mmBtu to develop a regionwide annual 2010 NO X cap could reasonably be expected to yield an average effective NO X emission rate for all CAIR units of about 0.15 lb/mmBtu or less.

Although EPA calculated—in essence, as a target level for further evaluation—the proposed regionwide annual NO X control levels (or emissions budgets) based on heat input from only Acid Rain Program units, the Agency evaluated the cost effectiveness of the control levels using heat input from all EGUs that potentially would be affected by the proposed CAIR. The EPA evaluated cost effectiveness using the IPM, which includes both Acid Rain units and non-Acid Rain units. Further, the IPM incorporates assumptions for electricity demand growth, and thus heat input growth.

Specifically, EPA evaluated these target annual NO X caps on EGUs for 2010 and 2015—and therefore the associated regionwide emissions reductions—using the IPM, which, in effect, demonstrated that these proposed NO X emissions cap levels can be met using highly cost-effective controls with the expected levels of electricity demand in 2010 and 2015, respectively. Those expected levels of electricity demand are higher than the electricity demand during the 1999 to 2002 years upon which EPA based heat input; and as a result, the amount of heat input necessary to meet the projected electricity demand is expected to be higher than the amount that EPA developed for evaluation purposes through the method described above. The projected average future emissions rates that would be associated with the 2010 and 2015 heat input levels needed to meet electricity demand (coupled with the NO X emissions budgets developed through the methodology described above) would be about 0.14 lb/mmBtu and 0.11 lb/mmBtu in 2010 and 2015, respectively. [62] These average rates would be for all units affected by annual NO X controls under CAIR, including non-Acid Rain units. Thus, the heat input is projected to be higher in 2010 and 2015 than the recent historic heat input used to develop the target emissions budgets, and the projected NO X emission rates in 2010 and 2015 are lower than the 0.15 lb/mmBtu and 0.125 lb/mmBtu rates that were used to develop the budgets. IPM determined the costs of meeting these average future NO X emission rates of 0.14 lb/mmBtu and 0.11 lb/mmBtu. The EPA considers these emission rates to be highly cost-effective and feasible.

In the NPR, EPA proposed an interim (Phase I) annual NO X phase in 2010 and a final (Phase II) annual NO X phase in 2015. However, in today's final rule, EPA is promulgating a Phase I for NO X in 2009 (with the Phase II for NO X in 2015, as proposed). The EPA determined the regionwide NO X control levels for 2009 and 2015 for today's final action using the same methodology as we used to determine proposed levels. The Agency evaluated the cost effectiveness of the final reduction requirements (and average NO X emission rates) using IPM and determined them to be highly cost-effective, assuming controls on EGUs. The EPA's evaluation of the cost effectiveness of the emission reduction strategy we assumed in establishing the final CAIR control levels is discussed further below.

The average NO X emission rates in the first and second phases of CAIR will be lower than the nominal emission rate on which the NO X SIP Call was based, which was 0.15 lb/mmBtu. In the NO X SIP Call, EPA also considered a control level based on a lower nominal emission rate, 0.12 lb/mmBtu. However, at that time the use of SCR was not sufficiently widespread to allow EPA to conclude that the controls necessary to meet a tighter cap could be installed in the required timeframe, without causing reliability problems for the electric power sector. Now, through the experience gained from the NO X SIP Call, EPA has confidence that with SCR technology average emissions rates lower than the NO X SIP Call nominal emission rate can be achieved on a regionwide basis.

In the CAIR NPR, after determining the regionwide control level and evaluating it to assure that it is highly cost-effective, the Agency then apportioned the regionwide budgets to the affected States. The EPA proposed to apportion regionwide NO X budgets to individual States on the basis of each State's share of recent average heat input. In the NPR, EPA used the average share of Acid Rain Program heat input. However, as discussed in the SNPR and the NODA, in order to distribute more equitably to States their share of the regionwide NO X budgets, EPA then considered each State's proportional share of recent average heat input using data from non-Acid Rain Program sources as well as Acid Rain Program sources. The EPA obtained EIA heat input data reported for non-Acid Rain sources and combined the EIA heat inputs with Acid Rain heat inputs to determine each State's share of combined average recent heat input.

The fact that EPA distributed the regionwide budget to individual States based on their proportional share of heat input from Acid Rain and non-Acid Rain units combined does not affect the determination of the regionwide budgets themselves. The regionwide budgets were determined to be highly cost-effective when tested for all units—both non-Acid Rain units as well as Acid Rain units—that would be affected by CAIR. (The EPA's method for apportioning regionwide NO X budgets to States is discussed in more detail elsewhere in today's preamble. That discussion includes an explanation of the differences between the State budgets that were presented in the NPR, the SNPR, and the NODA. In addition, see the TSD entitled “Regional and State SO 2 and NO X Emissions Budgets.”)

In the NPR, EPA proposed that Connecticut contributed significantly to downwind ozone nonattainment, but not to PM 2.5 nonattainment. Thus, the Agency proposed that Connecticut would not be subject to an annual NO X control requirement and was not included in the region proposed for annual controls. We proposed that Connecticut would be affected by an ozone season-only NO X control level, and proposed to calculate Connecticut's ozone season control level in a parallel way to how the regionwide annual NO X control levels were calculated. That is, EPA selected the highest of the same 4 years of (ozone season-only) heat input used for the regionwide budget calculation, and multiplied that heat input by the same NO X emission rates used to calculate the regionwide control levels. Connecticut is the only State for which an ozone season budget was proposed.

The EPA used the same methodology for developing regionwide budgets for today's final rule as was proposed in the NPR. For the final CAIR, EPA found that 23 States and the District of Columbia contribute significantly to downwind PM 2.5 nonattainment and found that 25 States and the District of Columbia contribute significantly to downwind ozone nonattainment (section III in today's preamble describes the significance determinations). CAIR requires annual NO X reductions in all States determined to contribute significantly to downwind PM 2.5 nonattainment, and requires ozone season NO X reductions in all States determined to contribute significantly to downwind ozone nonattainment (many of the CAIR States are affected by both annual and ozone season NO X reduction requirements). The final CAIR ozone season NO X reductions are required in two phases, with Phase I commencing in 2009 and Phase II in 2015, the same years as the annual NO X reduction requirements.

As described above, the Agency proposed ozone season NO X reduction requirements for Connecticut, and did not propose separate ozone season reduction requirements in any other State. For today's final rule, EPA requires ozone season reductions in all States contributing significantly to downwind ozone nonattainment. The EPA determined regionwide ozone season NO X control levels for the final CAIR using the same methodology as was used for the annual NO X reduction requirements (which is the same method that was proposed for Connecticut's ozone season budget). That is, EPA determined the highest (ozone season) heat input from Acid Rain Program units for the years 1999-2002 for each State, then summed this heat input for all of the States affected for ozone season NO X reductions. For the final 2015 control level, EPA calculated a regionwide ozone season NO X budget by multiplying this heat input by an emission rate of 0.125 lb/mmBtu, and for 2009 by multiplying by 0.15 lb/mmBtu. The Agency evaluated the cost effectiveness of these ozone season NO X control levels (and average NO X emission rates) using IPM and determined them to be highly cost-effective, assuming controls on EGUs. The EPA's evaluation of the cost effectiveness of the final CAIR control requirements is discussed further below.

Based on EPA's analysis of proposed annual NO X control levels, in the NPR the Agency presented average costs for annual NO X control of $800 per ton and $700 per ton for 2010 and 2015, and marginal costs of $1,300 per ton and $1,500 per ton for 2010 and 2015. In the NPR, EPA also presented marginal costs of annual NO X control from sensitivity analyses that used EIA assumptions for electricity growth and natural gas prices. Those marginal control costs were $1,300 per ton and $1,600 per ton for 2010 and 2015, respectively. The EPA also presented costs from a sensitivity model run that used EIA assumptions for electricity growth and natural gas price and higher SCR costs. These marginal control costs were $1,700 per ton and $2,200 per ton for 2010 and 2015, respectively. [63]

In the NPR, EPA also presented the average cost effectiveness for ozone season-only NO X control of $1,000 per ton and $1,500 per ton for 2010 and 2015, respectively, and a marginal cost for ozone season-only control of $2,200 per ton and $2,600 per ton for 2010 and 2015. The EPA also presented average costs for the non-ozone season (remaining seven months of the year) control of $700 per ton and $500 per ton in 2010 and 2015, respectively. (As noted above, the capital costs of installing NO X control equipment would be largely identical whether the equipment will be operated during the ozone season only or for the entire year. However, the amount of reductions would be less if the control equipment were operated only during the ozone season compared to annual operation.)

The EPA proposed the conclusion that these costs met the criteria for highly cost-effective emissions reductions for NO X (69 FR 4613-4615).

As with SO 2, EPA also considered the cost effectiveness of alternative stringency levels for this regulatory proposal (examining changes in the marginal cost curve at varying levels of emission reductions).

ii. What Are the Most Significant Comments That EPA Received About Proposed NO X Emission Reduction Requirements, and What Are EPA's Responses?

Some commenters expressed concern that EPA did not account for growth of heat input in calculating regionwide NO X emissions budgets, noting that growth was used in the calculation of the regional budget for the NO X SIP Call. Commenters suggest that, by not taking heat input growth into account, EPA developed regionwide budgets that are unduly stringent.

On the other hand, some commenters noted that they supported EPA's proposal to base regionwide budgets on historical heat input and did not want EPA to use growth projections for calculating regionwide NO X emissions budgets. Some stated that using actual, historic heat input numbers would be more straightforward than using growth projections, and some pointed to complications with the growth projection methodologies used in the NO X SIP Call.

The EPA recognizes that it employed a growth factor in the NO X SIP Call. There, EPA determined the amount of the regional emissions reductions and budgets by applying a growth factor to a historic heat input baseline. The DC Circuit, after first remanding that growth methodology for a better explanation, upheld it. West Virginia v. EPA, 362 F.3d 861 (DC Cir., 2004). See 67 FR 21 868 (May 1, 2002).

For CAIR, as described above, EPA developed a target level for the proposed NO X regionwide cap based on recent historic heat input and assumed emission rates of 0.125 lb/mmBtu and 0.15 lb/mmBtu for 2015 and 2010, respectively. The EPA evaluated these target NO X emissions levels using IPM, which indicated that those target caps—in conjunction with expected electricity demand for 2015 and 2010—would result from higher heat input levels and lower average emissions rates (about 0.11 lb/mmBtu and 0.14 lb/mmBtu for 2015 and 2010, respectively) than the amounts assumed in developing the target NO X caps. Most importantly, IPM indicated the cost levels associated with those projected 2015 and 2010 average NO X emission rates, and EPA has determined that those cost levels are highly cost-effective. For the final rule, EPA revised its analyses to reflect the 2009 initial NO X control phase, and determined that the final CAIR requirements are highly cost-effective. The EPA's methodology, in which the CAIR emissions reductions are predicted to be cost-effective under conditions of projected electricity growth that, in turn, projects heat input growth, in effect accounts for heat input growth. Moreover, the amount of heat input growth is the amount determined by IPM, a state-of-the-art model of the electricity sector (detailed documentation for IPM is in the docket).

Some commenters suggested that EPA adjust the NO X regionwide budget amounts to include heat input from non-Acid Rain units. For example, some suggested adding the non-Acid Rain unit heat input amounts that EPA used in apportioning regionwide NO X budgets to the States, to the total regionwide heat inputs that EPA used to calculate regionwide NO X budgets.

The regionwide budgets determined in the NPR were target levels developed as a starting point for further evaluation. The regionwide heat input amounts and NO X emission rates used to develop target budget levels were inherently imprecise. As discussed above, IPM modeling indicates that the projected future heat input amounts (based on electricity growth) are greater than the recent historic regionwide amount used to develop the target budget levels, and the future average emission rates for all units affected by CAIR annual NO X controls (including non-Acid Rain units) are less than the rates used to develop the target budget levels. IPM indicates that the target regionwide NO X budget levels (and corresponding future average NO X emission rates and heat input levels) are highly cost-effective for all CAIR units, including non-Acid Rain units. The EPA does not believe it is necessary to adjust the target regionwide budget levels to include the relatively small additional amount of heat input from non-Acid Rain units. The method the Agency used to develop target levels was not intended to be a precise methodology for determining the NO X caps; rather, it was a reasonable method for selecting a target level to be evaluated further. Upon evaluation of the target level, EPA determined that it can be achieved using highly cost-effective controls for all affected EGUs, including non-Acid Rain units.

iii. Analysis of NO X Emission Reduction Requirements for Today's Final Rule

(I) Reference Lists of Cost-Effective Controls

For today's action, EPA updated the reference list of controls included in the NPR of the average and marginal costs per ton of recent NO X control actions. The EPA systematically developed a list of cost information from recent actions and proposed actions. The Agency sought cost information for actions taken by EPA, and examined the comments submitted after the NPR was published, to identify all available control cost information to provide the updated reference list for today's preamble. The updated reference list includes both average and marginal costs of control to which EPA compares the CAIR control costs, although the Agency has limited information on marginal costs of other programs.

The EPA's updated summary of average costs of annual NO X controls are shown in Table IV-6. The results of this reexamination show that costs of recent actions are generally very similar to those identified in the NO X SIP Call. The cost figures are presented in 1999 dollars. [64]

Table IV-6.—Average Costs per Ton of Annual NO X Controls Back to Top
NO X control action Average cost per ton
1Control of Emissions of Air Pollution From Nonroad Diesel Engines and Fuel; Final Rule (69 FR 39131; June 29, 2004). The value in this table represents the long-term cost per ton of emissions reduced from the total fuel and engine program (cost per ton of emissions reduced in the year 2030). This value includes the cost for NO X plus NMHC reductions. 1999$ per ton.
2Control of Air Pollution from New Motor Vehicles: Heavy-Duty Engine and Vehicle Standards and Highway Diesel Fuel Sulfur Control Requirements; Final Rule (66 FR 5102; January 18, 2001). The values shown for 2007 Highway HD Diesel Stds are discounted costs. Costs shown in this table include a VOC component. 1999$ per ton.
3Proposed Revision of Standards of Performance for Nitrogen Oxide Emissions From New Fossil-Fuel Fired Steam Generating Units; Proposed Revision to Reporting Requirements for Standards of Performance for New Fossil-Fuel Fired Steam Generating Units; Proposed Rule (62 FR 36953; July 9, 1997), Table 4 (the Agency's estimate of average control costs was unchanged for the NSPS revisions final rule, published September 5, 1998). In the CAIR NPR, we included a value from the range of NO X controls for coal-fired EGUs from Table 2 in the proposed NSPS proposed rule (62 FR 36951). 1999$ per ton.
4Costs shown in this table are the range of project costs reported for projects that were FY 2002-2003 recipients of the TERP Emission Reductions Incentive Grants Program. These costs may not be in 1999 dollars. (www.tnrcc.state.tx.us/oprd/sips/grants.html)
5The EPA IPM modeling 2004 of the proposed BART for the electric power sector (69 FR 25184, May 5, 2004), available in the docket. The EPA modeled the Regional Haze Requirements as a source specific 0.2 lb/mmBtu NO X emission rate limit. Estimated average costs based on this modeling are $800 per ton in 2015 and 2020. 1999$ per ton.
Marine Compression Ignition Engines Up to $2002
Off-highway Diesel Engine $400-$7002
Nonroad Diesel Engines and Fuel $6001
Marine Spark Ignition Engines $1,200-$1,8002
Tier 2 Vehicle Gasoline Sulfur $1,300-$2,3002
Revision of New Source Performance Standards for NO X Emissions-EGUs $1,7003
2007 Highway Heavy Duty Diesel Standards $1,600-$2,1002
National Low Emission Vehicle $1,9002
Tier 1 Vehicle Standards $2,100-$2,8002
Revision of New Source Performance Standards for NO X Emissions-Industrial Units $2,2003
On-board Diagnostics $2,3002
Texas NO X Emission Reduction Grants FY 2002-2003 $300-$12,7004
Best Available Retrofit Technology (BART) for Electric Power Sector $8005

Table IV-7 presents modeled marginal costs for recent State annual NO X rules.

Table IV-7.—Marginal Costs per Ton of Reduction, Recent Annual NO X Rules Back to Top
NOX control action Marginal cost per ton
1The EPA IPM base case modeling August 2004, available in the docket. 1999$ per ton. We modeled Senate Bill 7 and Ch. 117, which impose varying NO X control requirements in different areas of the State; the range of marginal costs shown here reflects the range of requirements.
Texas Rules $2,000-$19,6001

The EPA does not believe that it has sufficient information, for today's rulemaking, to treat controls on source categories other than certain EGUs as providing highly cost-effective emissions reductions. The CAA Section 110 permits States to choose the sources and source categories that will be controlled in order to meet applicable emission and air quality requirements. This means that some States may choose to meet their CAIR obligations by imposing control requirements on sources other than EGUs.

As examples of cost-effective actions that States can take in efforts to provide for attainment with the air quality standards, Table IV-8 presents estimated average costs for potential local mobile source NO X control actions. The EPA received these cost data during the public comments on the NPR.

Table IV-8.—Average Costs of Potential Local Mobile Source Control Actions To Reduce NO X Emissions Back to Top
Source category Average cost per ton
[$ per Ton]1
1Washington DC Metro Area MWCOG Analysis of Potential Reasonably Available Control Measures (RACM). Projects determined to be “Possible” by MWCOG but not RACM because benefits from the possible control measures do not meet the 8.8 tpd NO X or 34.0 tpd VOC threshold necessary for RACM. These costs may not be in 1999 dollars. (www.mwcog.org/uploads/committee-documents/z1ZZXg20040217144350.pdf)Comments submitted to the EPA CAIR docket from the Clean Air Task Force et al., dated March 30, 2004, included costs from the MWCOG analysis.
MWCOG Analysis: Mobile Source, Bicycle racks in DC $9,000
MWCOG Analysis: Mobile Source, Telecommuting Centers 7,300
MWCOG Analysis: Mobile Source, Government Action Days (ozone action days) 5,000
MWCOG Analysis: Mobile Source, Permit Right Turn on Red 1,200
MWCOG Analysis: Mobile Source, Employer Outreach 3,500
MWCOG Analysis: Mobile Source, Mass Marketing Campaign 2,900
MWCOG Analysis: Mobile Source, Transit Prioritization 8,500

(II) Cost Effectiveness of CAIR Annual NO X Reductions

Table IV-9 provides the average and marginal costs of annual NO X reductions under CAIR for 2009 and 2015. These costs are updated from the NPR figures—the EPA analyzed the costs of the CAIR using an updated version of IPM (documentation for the IPM update is in the docket). Further, EPA modified the modeling to match the final CAIR strategy (see section IV.A.1 for a description of EPA's CAIR IPM modeling).

CAIR provides for a Compliance Supplement Pool (CSP) of NO X allowances that can be used for compliance with the annual NO X reduction requirements. The CSP is discussed in detail later in this preamble. The EPA used IPM to model marginal costs of CAIR with the CSP. The magnitude of the NO X CSP is relatively small compared to the annual NO X budget, [65] thus the CSP does not significantly impact the marginal costs (see Table IV-9).

As with SO 2 marginal costs, EPA considered the sensitivity of the NO X marginal cost results to assumptions of higher electric growth and future natural gas prices than the Agency used in the base case, as shown in Table IV-9.

Table IV-9.—Estimated Costs per Ton of Annual NO X Controlled Under CAIR1 Back to Top
Type of cost effectiveness 2009 2015
1The EPA IPM modeling 2004, available in the docket. 1999$ per ton.
Average Cost—Main Case $500 $700
Marginal Cost—Main Case 1,300 1,600
Marginal Cost—With Compliance Supplement Pool (CSP) 1,300 1,600
Sensitivity Analysis: Marginal Cost Using Alternate Electricity Growth and Natural Gas Price Assumptions 1,400 1,700

These estimated NO X control costs under CAIR reflect annual EGU NO X caps of 1.5 million tons in 2009 and 1.3 million tons in 2015 within the CAIR annual NO X control region (the 23 States and DC that must make annual reductions). In both the main IPM modeling case and the modeling case that includes the CSP, projected annual NO X emissions in the CAIR region will be about 1.5 million tons in 2009 and 1.3 million tons in 2015. The projected emissions are very similar in both modeling cases because the CSP is relatively small compared to the annual NO X budget.

Average costs shown for 2015 are based on the amount of reductions that would achieve the total difference in projected emissions between the base case conditions and CAIR in the year 2015. These costs are not based on the increment in reductions between 2009 and 2015. (A more detailed description of the final CAIR SO 2 and NO X control requirements is provided later in today's preamble.)

Most of the States subject to today's PM 2.5 control requirements have been subject to the NO X SIP Call requirements. Some sources in these States have installed SCRs, and run them during the ozone season. These sources might comply with the PM 2.5 annual NO X requirements by, at least in part, running the SCR controls for the remaining months of the year. Under these circumstances, the compliance costs for the PM 2.5 SIP requirements are lower.

Table IV-10 provides estimated costs per ton of NO X for non-ozone season reductions under CAIR. These figures are updated from the NPR calculations—the EPA analyzed the costs of the CAIR using an updated version of IPM (documentation for the IPM update is in the docket) and modeled controls on a region that more closely matches the region affected by CAIR.

Table IV-10.—Predicted Costs per Ton of Non-Ozone Season NO X Controlled Under CAIR1 Back to Top
Type of cost effectiveness 2009 2015
1The EPA IPM modeling 2004, available in the docket. 1999$ per ton.
Average Cost $500 $500

The estimated non-ozone season NO X costs, like the annual NO X costs, are on the low end of the cost effectiveness range described in Table IV-6. The EPA considers the 2015 and also the 2009 costs to represent highly cost-effective controls.

Environmental Defense reached similar conclusions regarding the cost effectiveness of non-ozone season NO X reductions, as described in their report “A Plan for All Seasons: Costs and Benefits of Year-Round NO X Reductions in Eastern States (2002).” As stated in that report, “[As Figure 4 shows,] extending NO X reductions throughout the year results in dramatic decreases in the per-ton costs of NO X emission reductions for the 19 NO X SIP Call States. This is because the bulk of the cost for reducing NO X emissions from power plants lies in the capital investment in the control equipment. Once the primary investment has been made, it costs relatively little to continue running the control equipment beyond the summer months required by EPA's NO X SIP Call.” Environmental Defense based these conclusions on analysis conducted by Resources for the Future (RFF). In an RFF paper, “Cost-Effective Reduction of NO X Emissions from Electricity Generation (July 2001),” RFF draws similar conclusions.

(III) NO X Cost Comparison for CAIR Requirements

The EPA believes that selecting as highly cost-effective amounts at the lower end of these average and marginal cost ranges is appropriate for reasons explained above in this section of the preamble.

As discussed above, although in the NO X SIP Call the cost level selected was not at the low end of the reference range of costs, if the NO X SIP Call costs were for annual rather than seasonal controls they would have been lower relative to the other control costs on the reference list which were mostly for annual programs.

For annual NO X, the range of average cost effectiveness extends broadly, from under $200 to thousands of dollars (Table IV-6). The 2015 estimated average costs for CAIR annual NO X control of $700 are consistent with the lower end of this range.

Less information is available for the marginal costs of controls than for average costs. Looking at the available marginal costs (Table IV-7), the 2015 CAIR marginal costs for annual NO X controls are at the lower end of the range. The EPA also evaluated the cost effectiveness of the 2009 cap, and concluded that the 2009 requirements are highly cost-effective.

(IV) Cost Effectiveness: Marginal Cost Curves for Annual NO X Control

As with SO 2 controls, EPA also considered the cost effectiveness of alternative stringency levels for NO X control for today's action by examining changes in the marginal cost curve at varying levels of emissions reductions. Figure IV-3 shows that the “knee” in the 2010 marginal cost effectiveness curve for EGUs—the point where the cost of controlling a ton of NO X begins to increase at a noticeably higher rate—appears to occur at over $1,700 per ton of NO X. Although EPA conducted this marginal cost curve analysis based on an initial NO X control phase in 2010, the results would be very similar for 2009, which is the initial NO X phase in the final CAIR. Figure IV-4 shows that the “knee” in the 2015 marginal cost effectiveness curve for EGUs appears to occur at over $1,700 per ton of NO X. (The EPA based these marginal NO X cost effectiveness curves on the electricity growth and natural gas price assumptions in the main CAIR IPM modeling run. Marginal cost effectiveness curves based on other electric growth and natural gas price assumptions would look different, therefore it would not be appropriate to compare the curves here to the marginal costs based on the IPM modeling sensitivity run that used EIA assumptions.) The EPA used the Technology Retrofitting Updating Model (TRUM), a spreadsheet model based on IPM, for this analysis. These results make clear that this rule is very cost-effective because the control level is below the point at which the cost begins to increase at a significantly higher rate.

In this manner, these results corroborate EPA's findings above concerning the cost effectiveness of the emissions reductions. [66]

BILLING CODE 6560-50-P

(V) Cost Effectiveness of Ozone Season NO X Reductions

The CAIR requires ozone season NO X emissions reduction for all States determined to contribute significantly to ozone nonattainment downwind (25 States and the District of Columbia). The EPA used IPM to model average and marginal costs of the ozone season reductions assuming EGU controls. In this modeling case, EPA modeled an ozone season NO X cap for the region affected by CAIR for downwind ozone nonattainment, but did not include the CAIR annual SO 2 or NO X caps. Based on that modeling, Table IV-11 provides estimated average and marginal costs of regionwide ozone season NO X reductions for 2009 and 2015. Table IV-11 shows the estimated cost effectiveness of today's ozone season NO X control requirements for 8-hour transport SIPs.

Table IV-11.—Estimated Costs per Ton of Ozone Season NO X Controlled Under CAIR1 Back to Top
Type of cost effectiveness 2009 2015
1The EPA IPM modeling 2004, available in the docket. 1999$ per ton.
Average Cost $900 $1,800
Marginal Cost 2,400 3,000

These estimated NO X control costs are based on ozone season EGU NO X caps of 0.6 million tons in 2009 and 0.5 million tons in 2015 within the CAIR ozone season NO X control region. Average costs shown for 2015 are based on the amount of reductions that would achieve the total difference in projected emissions between the base case conditions and CAIR in the year 2015. These costs are not based on the increment in reductions between 2009 and 2015. (A more detailed description of the final CAIR SO 2 and NO X control requirements is provided later in today's preamble.)

The EPA believes that selecting as highly cost-effective amounts at the lower end of the average and marginal cost ranges is appropriate for reasons explained above in section IV in this preamble.

In the NO X SIP Call, EPA identified average costs of $2,500 (1999$) (or $2,000 (1990$)) as highly cost-effective. [67] The estimated average costs of regionwide ozone season NO X control under CAIR are $1,800 per ton in 2015 and $900 per ton in 2009. Thus, with respect to average costs the controls for the final phase (2015) cap, which are below the $2,500 identified in the NO X SIP Call, are also highly cost-effective, as are those for the 2009 cap. In addition, the estimated average costs of CAIR ozone season NO X control are at the lower end of the reference range of average annual NO X control costs (the reference list of average annual NO X control costs is presented above).

Similarly, the estimated marginal costs [68] of ozone season CAIR NO X controls are within EPA's reference range of marginal costs, at the lower end of the range (the reference list of marginal annual NO X control costs is presented above). We note that the marginal costs in the reference range are for annual NO X reductions, and would likely be higher for ozone season only programs. Considering both average and marginal costs, the CAIR ozone season control level is highly cost-effective.

For purposes of estimating costs of ozone season control under CAIR, EPA set up this modeling case with CAIR ozone season NO X requirements but without the annual NO X requirements. The Agency believes that the cost of the ozone season CAIR requirements will actually be lower than the costs presented here because interactions will occur between the CAIR annual and ozone season NO X control requirements. [69] In addition, for States in both programs, the same controls achieving annual reductions for PM purposes will achieve ozone season reductions for ozone purposes; this is not reflected in our cost-per-ton estimates.

As with SO 2 controls, and annual NO X controls, EPA also considered the cost effectiveness of alternative stringency levels for CAIR NO X reductions for ozone purposes by examining changes in the marginal cost curve at varying levels of emissions reductions. Figure IV-5 shows that the “knee” in the 2010 marginal cost effectiveness curve for ozone season NO X reductions from EGUs—the point where the cost of controlling an ozone season ton of NO X begins to increase at a noticeably higher rate—appears to occur somewhere between $3,000 and $4,000 per ton of NO X. Although EPA conducted this marginal cost curve analysis based on an initial NO X control phase in 2010 the results would be very similar for 2009, which is the initial NO X phase in the final CAIR. Figure IV-6 shows that the “knee” in the 2015 marginal cost effectiveness curve for ozone season NO X reductions from EGUs appears to occur somewhere between $3,000 and $4,000 per ton of NO X. The EPA used the Technology Retrofitting Updating Model (TRUM), a spreadsheet model based on the IPM, for this analysis. These results make clear that CAIR NO X reductions for ozone purposes are very cost-effective because the control level is below the point at which the cost begins to increase at a significantly higher rate.

In this manner, these results corroborate EPA's findings above concerning the cost effectiveness of the emissions reductions. [70]

B. What Other Sources Did EPA Consider When Determining Emission Reduction Requirements?

1. Potential Sources of Highly Cost-Effective Emissions Reductions

In today's rulemaking, EPA determines the amount of regionwide emissions reductions required by determining the amount of emissions reductions that could be achieved through the application of highly cost-effective controls on certain EGUs. The EPA has reviewed other source categories, but concludes that for purposes of today's rulemaking, there is insufficient information to conclude that highly cost-effective controls are available for other source categories.

a. Mobile and Area Sources

In the NPR (69 FR 4610), EPA explained that “it did not identify highly cost-effective controls on mobile or area sources.” No comments were received suggesting that mobile or area sources should be controlled. Therefore, in developing emission reduction requirements, EPA is not assuming any emissions reductions from mobile or area sources.

b. Non-EGU Boilers and Turbines

The largest single category of stationary source non-EGUs are large non-EGU boilers and turbines. This source category emits both SO 2 and NO X. In the CAIR NPR, EPA proposed not to include any potential SO 2 or NO X emissions reductions from non-EGU boilers and turbines as constituting “highly cost-effective” reductions and thus to be taken into account in establishing emissions requirements because EPA believed it had insufficient information on their control costs, particularly costs associated with the integration of NO X and SO 2 controls. In addition, based on information EPA does have, projected base case (without the CAIR) emissions of SO 2 and NO X from these sources are significantly lower than projected EGU emissions. The EPA projects that in 2010 under base case conditions, EGUs would contribute 70 percent of SO 2 in the CAIR region compared to 15 percent from non-EGU boilers and turbines in the CAIR region. The Agency also predicts that in 2010 under the base case, EGUs would contribute 25 percent of NO X emissions in the CAIR region compared to 16 percent from non-EGU boilers and turbines in the CAIR region. Thus, simply on an absolute basis, non-EGU emissions are relatively less significant than emissions from EGUs. The EPA is finalizing its proposed approach to these sources and has not based today's requirements on any presumed availability of highly cost-effective emissions reductions from non-EGU boilers and turbines.

A number of commenters believe EPA should determine that emissions reductions from non-EGUs should be taken into account in establishing emission requirements because, they believe, highly cost-effective controls are available for these sources. These commenters argued that highly cost-effective controls are available for these sources and that EPA should have sufficient emissions and control cost information because the same sources were included in the NO X SIP Call.

In addition, while it is true that these sources were included in the NO X SIP Call, EPA only addressed NO X reductions from these sources. Neither SO 2 reductions nor monitoring of SO 2 emissions is required by the NO X SIP Call. As a result, for these sources, EPA has less reliable SO 2 emissions data and very little information on the integration of NO X and SO 2 controls. Although EPA has more information on NO X emissions from these sources because of the NO X SIP Call (and other programs in the northeastern U.S.), the geographic coverage of the CAIR includes some States that were not included in the NO X SIP Call, some of which States contain significant amounts of industry. The EPA has even less emissions data from non-EGUs in these non-SIP call States affected by the CAIR. While EPA has incorporated State-submitted emissions inventory data for 1999 into its analysis for the CAIR, even this data is generally lacking information on fuel, sulfur content, and existing controls. Without this data, it is very difficult to assess the emission reduction opportunities available for non-EGU boilers and turbines. Furthermore, with regards to NO X, many non-EGU boilers and turbines are making reductions using low NO X burners (the control technology EPA assumed in making the cost-effectiveness determinations in the NO X SIP Call). Since these controls are operated year-round, annual emissions reductions are already being obtained from many of these units. Additional reductions would likely be less cost effective.

Another commenter stated that non-EGU “major sources” are subject to the requirements of title V of the CAA and, therefore, EPA should have adequate emissions data provided as part of the sources' permitting obligations. However, title V simply requires that a source's permit include the substantive requirements (such as emission monitoring requirements) imposed by other sections of the CAA and does not itself impose any substantive requirements. Thus, the mere fact that a source is a major source required to have a title V permit does not mean that the source is monitoring and submitting emissions, fuel, and control device data. Many such sources do not, in fact, provide such data.

One commenter submitted cost information for FGD technology applications on industrial boilers. However, the information submitted by the commenter was based on the use of a limited number of technologies and for a limited number of boiler sizes. The EPA does not believe that the limited information demonstrates that SO 2 emissions from these sources could be controlled in a highly cost-effective manner across the entire sector in question, or to what level the emissions could be controlled.

Some commenters recommended including non-EGU boilers and turbines because in the future, after reductions from EGUs are made, the relative contribution of non-EGU boilers and turbines to the total NO X and SO 2 emissions will increase. The EPA agrees that the relative contribution of non-EGUs to total NO X and SO 2 emissions will increase in the future if States choose to meet their CAIR emissions reduction obligations solely by way of emission reductions made by EGUs. However, EPA does not believe that this, by itself, provides any basis for determining that in the context of this rule emissions reductions from non-EGUs should be determined to be highly cost-effective. As discussed above, EPA believes it is necessary to have more reliable emissions data and better control cost information for these sources before assuming reductions from them in the CAIR. The EPA is working to improve its inventory of emissions and control cost information for non-EGU boilers and turbines. Specifically, we are assessing the emission inventory submittals for 2002 made by States in response to the relatively new requirements of 40 CFR part 51 (the Consolidated Emission Reporting Rule), and we will work with States whose submissions appear to have gaps in required data. We also note that EPA provides financial and technical support for the efforts of the five Regional Planning Organizations to coordinate among and assist States in improving emission inventories.

Another commenter expressed concern that if the decision whether to control large industrial boilers is left to the States, the result may be inequitable treatment of EGUs on a State-by-State basis, particularly with respect to allowances, and therefore it would make sense to require NO X and SO 2 reductions from large industrial boilers. Section 110 of the CAA leaves the ultimate choice of what sources to control to the States, and EPA cannot require States to control non-EGUs. Even if EPA had included reductions from non-EGUs in determining the total amount of reductions required under the CAIR, EPA could not have required any State to achieve those reductions through emission limitations on non-EGUs.

The recent economic circumstances faced by the manufacturing sector accentuates EPA's concerns about the lack of reliable emissions data and control information regarding non-EGUs. We note that the U.S. manufacturing sector was adversely affected by the latest business cycle slowdown. As noted in the 2004 Economic Report of the President, the manufacturing sector was hit earlier, longer, and harder than other sectors of the economy. The 2004 Report also points out that, although manufacturing output has dropped much more than the real gross domestic product (GDP) during past business cycles, the latest recovery has been unusual because it has been weaker for the manufacturing sector than the recovery in the real GDP. The disparity across sectors (and even within individual sectors) in the economic condition of firms reinforces EPA's concerns about moving forward to consider emission controls on non-EGUs at this time.

As explained elsewhere in this preamble, although the CAIR does not require that States achieve the required emissions reductions by controlling particular source categories, we expect that States will meet their CAIR obligations by requiring emissions reductions from EGUs because such reductions are highly cost effective. We believe the States are in the best position to make decisions regarding any additional control requirements for non-EGU sources. In making such decisions, States may take into consideration all relevant factors and information, such as differences across States in the need for control, differences in relative contribution of various sources, and differences in the operating and economic conditions across sources.

c. Other Non-EGU Stationary Sources

In the NPR and in the technical support document entitled “Identification and Discussion of Sources of Regional Point Source NO X and SO 2 Emissions Other Than EGUs (January 2004),” EPA applied a similar rationale for non-EGU stationary sources other than boilers and turbines. For SO 2, EPA noted that the emissions from such sources were a relatively small part of the emissions inventory, and we also noted the lack of information on costs. For NO X, we explained that more information was available than for SO 2. This is because the NO X SIP Call included consideration of emissions control measures for internal combustion (IC) engines and cement kilns, and developed cost estimates for other NO X-emitting categories such as process heaters and glass manufacturing. However, we believed—as for boilers and turbines, discussed above—that insufficient information on emission control options and costs, was available to apply these measures to the entire geographic area covered by the proposed rule.

No adverse comments were received suggesting inclusion of SO 2 emissions reductions from non-EGU stationary sources other than boilers and turbines. Accordingly, EPA has determined not to consider SO 2 reductions from these other non-EGU stationary sources.

Several commenters suggested that EPA should have been able to consider NO X emissions reductions from non-EGU categories other than boilers and turbines, such as internal combustion (IC) engines and refinery fluid catalytic cracking units. These commenters believed such reductions were demonstrated to be cost effective, and questioned EPA's assertion that insufficient information is available. Finally, some commenters believe EPA should have, at a minimum, required that controls for NO X SIP Call sources—including large IC engines and cement kilns—should be extended from the ozone season to the entire year.

We believe it likely that inclusion in today's requirements of reductions from any highly cost-effective controls—if available—for these categories would have very small effects. First, most of the States included in the CAIR rule were also included in the NO X SIP Call, so that many of the emissions reductions that would be available from these sources have already occurred due to implementation of the NO X SIP Call. Second, in the States included in the CAIR rule, but which were not covered by the NO X SIP Call, only a small portion of NO X emissions come from cement kilns and IC engines compared to EGUs. Moreover, in some parts of this geographic area, in particular for Texas, many sources in these source categories are already regulated under ozone nonattainment plans (including SIPs for the Texas cities of Houston, Galveston, and Dallas).

Regarding the commenters' recommendation that extending NO X SIP Call control requirements to a year-round basis for large IC engines and cement kilns should be considered to be highly cost effective, EPA believes that few emissions reductions would be achieved from doing so. The types of controls that were applied in the NO X SIP Call States, while required to be in place only during the ozone season, will, as a practical matter, be applied on a year-round basis, whether or not so required by today's rule. Most, if not all, of the NO X SIP Call States have developed regulations to control NO X emissions from IC engines and cement kilns during the ozone season. The control of choice to meet these reductions from large lean burn IC engines is low emission combustion (LEC), which for retrofit applications is a substantial equipment modification of the engine's combustion system. The engine will operate with LEC year round because this modification is a permanent change to the engine. Most, if not all, new large lean-burn IC engines have LEC. In addition, year-round emissions controls are already required for rich-burn engines greater than 500 hp which will likely install nonselective catalyst reduction to comply with the recently adopted hazardous air pollutant standards (see final rule for reciprocating IC engines, 69 FR 33474, June 15, 2004). For cement kilns, the controls of choice are low NO X burners and mid-kiln firing. Low NO X burners (LNB) are a permanent part of the kiln, so that the kiln will operate year-round with LNB. Mid-kiln firing is a kiln modification for which a solid and slow burning fuel (typically tires) is injected in the mid-kiln area. Due to tipping fees and fuel credits, mid-kiln firing results in an operating cost savings. After this system is installed, year-round operation is expected.

C. Schedule for Implementing SO 2 and NO X Emissions Reduction Requirements for PM 2.5 and Ozone

1. Overview

In the NPR, EPA proposed a two-phased schedule for implementing the CAIR annual emission reduction requirements: implementation of the first phase would be required by January 1, 2010 (covering 2010-2014), and that for the second phase by January 1, 2015 (covering after 2014). The EPA based its proposal on its analysis of engineering, financial, and other factors that affect the timing for installing the emission controls that would be most cost-effective—and are therefore the most likely to be adopted—for States to meet the CAIR requirements. Those air pollution controls are primarily retrofitted FGD systems (i.e., scrubbers) for SO 2 and SCR systems for NO X on coal-fired power plants.

The EPA's projections showed a significant number of affected sources installing these controls. The proposed two-phased schedule allowed the implementation of as much of the controls as feasible by an early date, with a later time for the remaining controls.

The EPA received detailed, technical comments from commenters who argued that the controls could not be implemented until later than proposed, and from other commenters who argued that the controls could be implemented sooner than proposed. The EPA has reviewed the comments and has conducted additional research and analyses to verify availability of adequate industrial resources, including boilermakers, for constructing the emission control retrofits required by CAIR. These analyses are based on conservative assumptions, including those suggested by the commenters, to ensure that the requirements imposed by CAIR do not result in shortages of the required resources that could substantially increase construction costs for pollution controls and reduce the cost effectiveness of this program.

Today, EPA is taking final action to require the annual emissions reductions on the same two-phase schedule as proposed. However, the requirements for the first phase include two separate compliance deadlines: Implementation of NO X reductions are required by January 1, 2009 (covering 2009-2014) and for SO 2 reductions by January 1, 2010 (covering 2010-2014). The compliance deadline requirements for the second phase are the same as proposed. The EPA believes that its action is consistent with the Agency's obligations under the CAA to require emission reductions for obtaining NAAQS to be achieved as soon as practicable. The EPA applied the same criterion in implementing the NO X SIP Call, which was based on a single-phased schedule. [71]

2. Engineering Factors Affecting Timing for Control Retrofits

a. NPR

In the NPR, EPA identified the availability of boilermakers as an important constraint for the installation of significant amounts of SCR and FGD retrofits. Boilermakers are skilled laborers that perform various specialized construction activities, including welding and rigging, for boilers and high pressure vessels. The air pollution control devices, such as scrubber and SCR vessels, require boilermakers for their construction. Apprentices with no prior work-related experience complete a four-year training program, to become full boilermakers. For apprentices with relevant experience, this training period could be shorter. For example, union members representing the shipbuilding trade could be expedited into the boilermaker division within a year.

The boilermaker constraint was considered more important for the initiation of the first phase of CAIR, since the NO X SIP Call experience had shown that many sources would be adverse to committing significant funds to install controls until after SIPs were finalized. With the States required to finalize SIPs in 18 months after the signing of the final rule, the sources would have three years in which to complete purchasing, construction, and startup activities associated with these controls, to meet the proposed CAIR deadline.

The EPA's projections showed power plants installing 51.4 gigawatts (GW) of FGD and 28.2 GW of SCR retrofits during the first CAIR phase. These projections include retrofits for CAIR as well as retrofits for base case policies (i.e., retrofits for existing regulatory requirements). We estimated the total boilermaker-years required for installing these controls at 12,700, which was based on the boilermakers being utilized over a period of 18 months during the installation process. Also, based on the projected boilermaker population in the timeframe relevant to the installation of these controls, we estimated that 14,700 boilermaker-years were available over the same 18-month period. The availability of approximately 15 percent more boilermaker-years than required, as shown by these estimates, confirms the adequacy of this critical resource for CAIR and EPA assumed this to be a reasonable contingency factor.

The EPA also determined that installation of the projected amounts of FGD and SCR retrofits could be completed within the three-year period available for CAIR. This determination was based on a previous report prepared by EPA for the proposed Clear Skies Act, “Engineering and Economic Factors Affecting the Installation of Control Technologies for Multi-Pollutant Strategies,” (docket no. OAR-2003-0053-0106). According to this report, an average of 21 months are required to install SCR on one unit, and 27 months to install a scrubber on one unit. For multiple units within the same plant, installation of controls would normally be staggered to avoid operational disruptions. The EPA projected that the maximum number of multiple-unit controls required for each affected facility could all be installed within three years.The NPR proposal included a second phase, with a compliance deadline of January 1, 2015. The EPA's projections showed power plants installing 19.1 GW of FGD and 31.7 GW of SCR retrofits by 2015, which included retrofits for CAIR as well as retrofits for base case policies (i.e., retrofits for existing regulatory requirements). Availability of boilermaker labor was not an important constraint for this phase.

b. Comments

The EPA received several comments relating to the requirements for the two-phased implementation program, the emission caps and compliance deadline for each phase, and resources required to install necessary controls. The commenters offered opposing viewpoints, which can be broadly categorized as follows.

Several commenters indicated that the compliance deadline of 2010 for the first phase was not attainable and argued that EPA should either extend the deadline, or set higher emission caps for this phase. The commenters raised the following specific points in support of their concerns:

  • The time allowed for completing various activities from planning to startup of the required controls was not sufficient. Other related activities, including project financing and obtaining a landfill permit for the scrubber waste, could also require more time than what the rule allowed. In addition, the short implementation period would require simultaneous outages of too many units to tie the new equipment into the existing systems, which would affect the reliability of the electrical grid.
  • Implementation of controls to the required large number of units would cause shortages in the supply of critical industrial resources, especially boilermakers. An analysis performed by a commenter showed a shortfall in the supply of boilermaker labor during the construction period relevant to CAIR retrofits. This commenter anticipated that certain key variables would be greater in value than those used by EPA and based their analysis on higher SCR prices, EIA-projected higher natural gas prices and electricity demand factors, and more stringent boilermaker duty rates (boilermaker-year/MW) and availability factors.

Commenters who favored more stringent compliance deadlines argued that the required controls could be installed in less time and more controls could be built in early years. These commenters raised the following specific points in support of their concerns.

  • The compliance deadlines for the two phases did not support the ozone and fine particulate (PM 2.5) attainment dates mandated by the CAA. The Phase I deadline should be accelerated to meet these attainment dates. Sufficient industrial resources, including boilermakers, would be available to support such an acceleration. While some commenters supported an earlier Phase I deadline of January 1, 2008, the others supported a deadline of January 1, 2009. Some of these commenters also suggested that the Phase I deadline be accelerated only for NO X.
  • The EPA's estimates for the boilermaker availability were too conservative. A boilermaker labor analysis performed by one commenter showed an adequate supply of this resource to support installation of all Phase I and II controls by the start of the first phase (by 2010), thereby eliminating the need for two phases.
  • The time allowed for installing controls for Phase II was excessive. The initiation of this phase could be moved forward.

Several commenters supported EPA's assumptions used in support of the adequacy of the implementation period and resources to build the required CAIR controls. These assumptions included the overall construction schedule durations for SCR and FGD systems and boilermaker unit rates.

c. Responses

The EPA reviewed the above comments and performed additional research and analyses, including new IPM runs that incorporated higher SCR and natural gas costs and greater electric demand. We also found that more units had installed SCR under the NO X SIP Call and other regulatory actions than what our records previously showed. This increase in the number of existing SCR installations was also incorporated into these IPM runs. In addition, the number of existing FGD installations was also revised slightly downward, for the same reason.

The revised IPM analyses for today's final action show that the amounts of controls that need to be put on for Phase I are 39.6 GW of FGD and 23.9 GW of SCR. These amounts represent a reduction from the estimates for the NPR. For Phase II, the amount of the required controls are 32.4 GW of FGD and 26.6 GW of SCR. These amounts represent an increase from the estimates for the NPR. The amounts shown for both phases reflect all retrofits required for the CAIR and base case (non-CAIR) policies. The retrofit projections for the base case policies are included, since some of the available boilermaker labor would be consumed in building these retrofits during the CAIR time-frame.

The EPA also contacted the International Brotherhood of Boilermakers (IBB), U.S. Bureau of Labor Statistics (BLS), and National Association of Construction Boilermaker Employers (NACBE) to verify its assumptions on boilermakers population, percentage of boilermakers available to work on the control retrofit projects, and average annual hours of boilermaker employment. Except for the boilermaker population, the information received as a result of these investigations validated EPA's assumptions. IBB also confirmed that the boilermaker population would at least be maintained at the current level of 26,000 members, during the period relevant to construction of CAIR retrofits. It did not want to forecast growth and historically has not done so. Therefore, instead of the 28,000 boilermaker forecasted population used in the NPR, we have conservatively used a boilermaker population of 26,000 for the final CAIR. A detailed discussion on these assumptions and the information received from these sources is available in the docket to this rulemaking as a technical support document (TSD), entitled “Boilermaker Labor and Installation Timing Analysis, (docket no. OAR-2003-0053-2092).”

The responses to the most significant comments on these issues are summarized in the following sections.

i. Issues Related to Compliance Deadline Extension

(I) Adequacy of Phase I Implementation Period

Today's action initiates State activities in conjunction with EPA to set up the administrative details of CAIR. With the first phase compliance deadline of January 1, 2009, for NO X and January 1, 2010, for SO 2, the affected sources would have approximately 33/4and 43/4years for the implementation of the overall requirements for this phase, respectively. The final SIPs would be submitted at the end of the first 18 months of these implementation periods. The remaining 21/4and 31/4years would be available for the sources to complete activities required for the procurement and installation of NO X and SO 2 controls, respectively. For the reasons outlined below, EPA believes that these deadlines provide enough time to install the required Phase I controls.

(A) Engineering/Construction Schedule Issues

The EPA notes that, for CAIR, the States would finalize the SIPs in 18 months after the rule is signed, and that until then, the majority of sources required to install controls may not initiate activities that require commitment of major funds. However, some activities, such as planning, preparation of conceptual designs, selection of technologies, and contacts with equipment suppliers can be started or completed prior to the finalization of SIPs, at least for major sources expected to require longer implementation periods. In addition, other activities, such as permitting and financing can be started after the rule is finalized. This is based on the NO X SIP Call experience.

After the SIPs are finalized, the sources would have approximately 21/4and 31/4years in which to complete purchasing, detailed design, fabrication, construction, and startup of the required NO X and SO 2 controls, respectively. This assumes that activities, such as planning and selection of technologies, have already been started or completed, prior to the start of these 21/4- and 31/4-year periods. As discussed in the NPR proposal, EPA projects an average single-unit installation time of 21 months for SCR and 27 months for a scrubber. Our revised IPM analysis for the final rule shows that many facilities would install controls on multiple units (a maximum of six for SCR and five for FGD) at the same plant. We expect these facilities to stagger these installations to minimize operational disruptions.

The EPA also projects that SCRs and scrubbers could be installed on the multiple units in the available time periods of 21/4and 31/4years, respectively. The issues related to the availability of boilermakers and the ability of the plants requiring multiple-unit controls to stagger their installations during these periods are discussed later in this preamble.

As compared to projections in the NPR proposal, earlier signing of the final rule adds approximately three additional months to the overall implementation periods for SO 2 controls. Furthermore, EPA's projections for the final rule show fewer Phase I NO X and SO 2 controls being added than the projections in the NPR proposal. Since the compliance deadline for NO X has been moved up a year from the proposal, a three-month earlier rule promulgation provides more time for implementing SO 2 controls only. However, since it does allow use of critical resources, such as boilermakers, for SO 2 controls to be spread over a longer period of time, the net effect would be to make more of these resources available for both SO 2 and NO X controls (as compared to a scenario where promulgation was not three months earlier). This is especially true since the implementation periods for both NO X and SO 2 controls would start at the same time and the plants installing these controls would be competing for the same resources until January 1, 2009, the compliance deadline for NO X. The EPA, therefore, believes that 21/4- and 31/4-year time periods provide reasonable amounts of time from the approval of State programs by September 2006, until the commencement of compliance deadlines for meeting the NO X and SO 2 emission requirements.

Certain commenters have provided their own estimates of schedule requirements for installing the required controls. In some cases, these estimates are longer than those determined by EPA. For scrubbers, including spray dryer and wet limestone or lime type systems, the control implementation requirements provided by the commenters range from 30 to 54 months for the overall project and 18 to 36 months for the phase following equipment awards. In this case, the lowest 18-month schedule requirement cited applies to spray dryers, whereas the shortest schedule cited for wet scrubbers for the activities following the equipment awards is 24 months. For SCR, the control implementation requirements cited by the commenters range from 24 to 36 months for the overall project and 17 to 25 months for the phase following the equipment awards.

One commenter has pointed out that the construction schedule requirements for the FGD and SCR retrofit projects have shortened, because of the lessons learned from a significant number of such projects completed during the last few years. The EPA notes that a recent announcement for a new 485 MW limestone scrubber facility indicates a construction schedule duration (from equipment award to startup) of only 18 months. [72] This is well below the schedule requirement cited by the commenters for a wet limestone scrubber.

The EPA also notes that most of the commenters' schedule estimates are consistent with the time periods available for completing the CAIR-related NO X and SO 2 projects. Some of the longer schedules submitted by commenters would exceed the CAIR Phase I dates. However, EPA considers these longer schedules to be speculative, as these commenters did not justify them. The major factors that influence schedule requirements include size of the installation, degree of retrofit difficulty, and plant location. The EPA does not expect these factors to make a difference of more than a few months between the schedule requirements of various installations. The commenters who have cited long schedule requirements that fall at the higher end of the above ranges have not provided any data to support the wide differences between their schedules and those proposed by others, including EPA. It should also be noted that EPA's schedules are based on information from several actual SCR and scrubber installations. Therefore, EPA cannot accept the excessive schedule requirements proposed by these commenters.

(B) Landfill Permit Issue

The EPA contacted several key States requiring FGD retrofits, to investigate the amount of time required to obtain a landfill permit for scrubber waste. We note that not all scrubber installations would require landfills, as some scrubber designs produce saleable waste products, such as gypsum.

Specifically, EPA contacted Georgia, Ohio, Indiana, Alabama, Pennsylvania, West Virginia, Tennessee, and Kentucky. [73] Except for Kentucky, all States indicated that their permit approval periods ranged from 12 to 27 months. Some of these States indicated that permit approval may require more time than 27 months, but only for the cases in which major landfill design issues persist or the permit applicant has not provided complete and proper information with the permit application.

The Kentucky Department of Environmental Protection indicated that, based on their historical records, the average permit approval period was 31/2years. They also stated that the State was sensitive to an applicant's time restrictions and the permit approval times had varied depending on the level of urgency surrounding a permit application. They further confirmed that they would work with the industry to meet compliance deadlines, such as those required by CAIR, as efficiently as possible.

Based on the above investigations, EPA notes that the landfill permitting requirements quoted by all States fall well within the 43/4-year implementation period for Phase I. Also, landfill permitting activities as well as its design and construction can be accomplished, independent of the design and construction of the FGD system. The EPA, therefore, believes that landfill permitting is not a constraint for compliance with the rule.

(C) Project Financing Issue

Commenters representing small units or units owned by the co-operatives raised concerns that arrangement of financing for control retrofits could take long periods of time. However, EPA's projections show a larger portion of the smaller units installing controls only during the second phase. These projections also show that only a few co-operative units would require installation of controls. Therefore, EPA believes that the Phase I implementation periods of approximately 33/4and 43/4years for NO X and SO 2 controls, respectively, provide enough time for completing the financing activity for all controls. Of course, if individual sources face difficulties in meeting deadlines to implement controls, they may use the allowance-trading provisions of CAIR to defer implementation of controls.

(D) Electrical Grid Reliability Issue

Based on available data for the NO X SIP Call, approximately 68 GW of SCR retrofits were started up during the years from 2001 to 2003. This included approximately 42 GW of SCRs in 2003 alone, which exceeds the combined capacity of SCR and FGD retrofits for CAIR that we expect to be started up in any one year. The EPA projects that startup of the 23.9 GW of SCR and 39.6 GW of FGD capacity required for Phase I would be spread over a period of two years (2008 and 2009). The total capacity of units starting up in each year is therefore expected to be approximately 32 GW (half of the combined SCR and FGD capacity of 63.5 GW).

The NO X SIP Call experience shows that outages required to complete installation of the large SCR capacity, especially during 2003, did not have an adverse impact on the electrical grid reliability. The EPA notes that the outage requirement for SCR usually exceeds that for scrubbers, since SCR is located closer to the boiler and it may be more intrusive to the existing equipment. As shown above, the CAIR retrofits are projected to include more scrubbers than SCRs and the capacity of these retrofits starting up in any one year is below the capacity of the NO X SIP Call units that started up in 2003. Therefore, the overall outage requirement for CAIR would be less than that experienced for the NO X SIP Call.

Based on published industry data, the planned outage times for coal-fired units from 2001-2002 (SCR buildup years) decreased by over two percent compared to the previous two years from 1998-1999. [74] The reduction in the overall outage time in the 2001-2002 period also shows that the SCR retrofits did not adversely affect the grid reliability. Therefore, EPA believes that the concern regarding electrical grid reliability is unwarranted for CAIR retrofits.

(II) Availability of Boilermaker Labor in Phase I

The EPA has performed several analyses to verify the adequacy of the available boilermaker labor for the installation of CAIR's Phase I controls. These analyses were not just based on using EPA's assumptions for the key factors affecting the boilermaker availability, but also the assumptions suggested by commenters for these factors to determine how sure we could be on our key conclusions. If there was insufficient labor for the amount of air pollution controls that will need to be installed, the program would be in jeopardy. For instance, shortages in manpower could lead to high wage rates that could substantially increase construction costs for pollution controls and reduce the cost effectiveness of this program. During the peak of the NO X SIP Call SCR construction period, the power industry did experience an increase in the SCR construction costs. One of the reasons cited for these higher costs was an increased demand for boilermaker labor. The EPA strongly wanted to avoid this possibility for CAIR. The EPA also wanted to be very sure that the levels of controls and timing of the program's start were appropriate. Therefore, EPA tended to make conservative assumptions and to test the sensitivity of key assumptions that were uncertain.

Boilermakers population, percentage of boilermakers available to work on the control retrofit projects, and average annual hours of boilermaker employment are some of the key factors that affect boilermaker availability. As discussed previously, EPA's assumptions on these factors were validated or revised through our discussions with IBB, BLS, and NACBE.

Two other key factors that also have an impact on boilermaker availability include the number of required SCR and FGD retrofits and boilermaker duty rates (boilermaker-year/MW, i.e., the number of boilermaker years needed to install SCR or FGD on one MW of electric generation capacity). The EPA's projections for the required SCR and FGD retrofits are based on the IPM analyses performed for the final rule. The basis for the boilermaker duty rates used by EPA is a report prepared by EPA for the proposed Clear Skies Act, “Engineering and Economic Factors Affecting the Installation of Control Technologies for Multi-Pollutant Strategies.”

Some commenters have suggested use of EIA's projections of natural gas prices and electricity demand rates that are higher than EPA's projections used in the IPM analyses. Use of higher values for these parameters would increase the number of required control retrofits. While not agreeing with these commenters that EIA's projections should replace the data that EPA uses, we acknowledge that there is reasonable uncertainty concerning these assumptions and that addressing the uncertainty explicitly by considering EIA's alternative assumptions is prudent, given the importance of having sufficient labor resources to meet the program's requirements in 2010. Therefore, EPA has performed a sensitivity analysis to determine the required control retrofits resulting from the use of these EIA projections, and then used the increased amounts of the required control retrofits to determine their impacts on the boilermaker availability.

The EPA also received comments suggesting that the SCR costs used in our IPM analyses were below the levels experienced in recent SCR installations. We note that the SCR costs were revised in the IPM analyses performed for the final rule, to reflect recent industry experience. One commenter reported SCR capital costs that exceeded our revised costs. The EPA does not agree with these reported costs, as they are not supported by the overall cost data submitted by the commenter. However, to address the concern with the SCR costs in general, we have performed a sensitivity analysis to determine the impact of increasing the SCR capital and fixed OM costs by 30 percent.

An increase in the SCR costs would affect the amounts of the required control retrofits. Table IV-12 shows the projected Phase I SCR and FGD retrofits for the above two alternate cases, based on using EIA's projections for natural gas prices and electricity demand rates and higher SCR costs.

Table IV-12.—IPM Projections for Total Capacities of FGD and SCR Retrofit Projects for Coal-Fired Electric Generation Units for CAIR Phase I Using EPA and Commenter Assumptions Back to Top
Retrofit type EPA base case assumptions EIA projections1 EIA projections and higher SCR costs2
1The required control retrofits shown are based on using EIA projections for natural gas prices and electricity demand rates.
2The required control retrofits shown are based on using EIA projections for natural gas prices and electricity demand rates as well as 30 percent higher SCR capital and fixed OM costs.
CAIR FGD, GW 37 45.4 47.9
Non-CAIR FGD, GW 2.6 3.7 Included Above
CAIR SCR, GW 18.2 20.6 25.2
Non-CAIR SCR, GW 5.7 4.6 Included Above

As shown in Table IV-12 above, the alternate case using just the EIA's projections for natural gas prices and electricity demand rates requires the largest amounts of control retrofits. Therefore, a boilermaker availability analysis was performed for just this case.

One commenter has suggested use of higher boilermaker duty rates for both SCR and FGD retrofits, based on an industry survey they had conducted. Use of higher duty rates would result in more boilermakers being needed to install the controls. Table IV-13 shows the boilermaker duty rates used by EPA as well as those suggested by this commenter.

Table IV-13.—Boilermaker Duty Rates for SCR and FGD Systems for Coal-Fired Electric Generation Units Back to Top
Source FGD SCR
1The duty rate values shown are average values calculated by using the FGD and SCR correlations provided by the commenter along with the MW size of individual units projected by the IPM to require FGD or SCR controls for Phase I of CAIR.
EPA's estimate, boilermaker-year/MW 0.152 0.175
Commenter-suggested, boilermaker-year/MW1 0.269 0.343

Our review of the limited supporting information submitted by the commenter about their survey for these duty rates shows that they are based on data from a small number of installations and represent scope of work at each power plant that is well above the average installation conditions used in determining the duty rates used by EPA. Therefore, EPA considers these commenter-suggested duty rates to represent the upper end of the range of values that would be expected for the SCR and FGD controls under consideration. This is also supported by the average duty rate (0.199) submitted by one other commenter for installing FGDs, which is well below the average duty rate (0.269) suggested by the first commenter. However, EPA also notes that the duty rate suggested by the second commenter is higher than that (0.152) used by EPA.

The EPA conducted the boilermaker analysis for the final rule using alternative assumptions for boilermaker duty rates. These alternative assumptions yield a range of estimates of the amount of control that could feasibly be installed. In keeping with EPA's desire to be very sure that there is sufficient boilermaker labor available during the CAIR's Phase I construction period, the Agency has considered the most stringent duty rates suggested by the first commenter, as well as other duty rates (see Table IV-13), in analyzing the impact on the boilermaker availability. The EPA considers this to be a bounding analysis in which the estimates based on the most stringent duty rates reflect conditions with the highest retrofit difficulty level that EPA could realistically expect to occur. We expect that the average boilermaker duty rates applicable to the overall boiler population required to retrofit controls under this rule would not fall outside of the values used by EPA and those suggested by the first commenter.

In the NPR, only the union boilermakers belonging to the IBB were considered in the EPA's availability analysis. Some commenters have pointed out that additional sources of boilermakers will be available for CAIR. Two such sources include non-union and Canadian boilermakers. IBB has confirmed that 1,325 Canadian boilermakers were brought in to support the NO X SIP Call SCR work in 2003. The EPA also projects that approximately 15 percent of FGDs and 43 percent of SCRs will be installed for Phase I in the traditionally non-union States and believes there will be nonunion labor available in these States. One source has confirmed that a substantial amount of SCR retrofit work during the 2000-2002 period was executed by non-union labor. [75] Based on these data, we have conservatively assumed that 1,000 boilermakers from Canada will be available and 10 percent of the retrofits would be installed by non-union boilermakers for Phase I.

Based on EPA data, an average 32 GW of new gas-fired, combined cycle generating capacity was being added annually, during the NO X SIP Call SCR construction years of 2002 and 2003. A substantial number of boilermakers were involved in the construction of these gas-fired projects. Since projections for the timeframe relevant to CAIR retrofits show only a small amount of new electric generating capacity being added, the number of boilermakers involved in the building of new plants would be smaller and more of the boilermaker population would be available to work on the Phase I retrofits. As pointed out by one commenter, the boilermakers available due to this projected drop in the building of new generation capacity represents a third additional source of boilermakers for CAIR.

The EPA projects only an insignificant amount of new coal-fired generating capacity being added during Phase I. The most recent EIA's projections also do not show any new coal fired capacity being added between 2007 and 2010, the timeframe relevant to boilermaker-related construction activities for CAIR. [76] However, EPA's projections do show approximately 15 GW of new or repowered gas-fired capacity being added, during 2007-2010. The EIA's projections for new gas-fired capacity addition during Phase I are well below those of EPA's. We used the more conservative EPA projections for new generating capacity additions and the gas-fired capacity additions during the NO X SIP Call period to estimate the additional boilermaker labor that would become available for the Phase I retrofits. This estimate shows that approximately 28 percent more boilermakers would be available to work on the CAIR retrofits, because of a slowdown in the construction of new power plants. [77]

In the boilermaker availability analyses performed by EPA, the required boilermaker-years were determined for each case, based on the amounts of SCR and FGD retrofits being installed and the pertinent boilermaker availability factors and duty rates. The required boilermaker-years were then compared to the available boilermaker years to verify adequacy of the boilermaker labor. All sources of boilermakers were considered in these analyses, including the union boilermakers and the boilermakers from the three additional sources discussed previously.

The EPA's boilermaker availability analyses firmly support CAIR's Phase I requirements. Using EPA's projections of FGD and SCR retrofits installed for Phase I and EPA's assumptions for boilermaker duty rates, there are ample boilermakers available with a large contingency factor to support the predicted levels of CAIR retrofits. For the most conservative analysis using the boilermaker duty rates suggested by one commenter and the EIA's projections for natural gas prices and electricity demand rates, there are sufficient boilermakers available with a contingency factor of approximately 14 percent.

In the NPR proposal, EPA estimated that a contingency factor of 15 percent was available to offset any increases in boilermaker requirements due to unforeseen events, such as sick leave, time lost due to inclement weather, time lost due to travel between job-sites, inefficiencies created due to project scheduling issues, etc. The EPA had considered this 15 percent contingency factor to be adequate for these unforeseen events. We also note that EPA did not receive any comments suggesting a need for a higher contingency factor.

The EPA also notes that the above boilermaker labor estimates have not considered the benefits of the experiences gained by the U.S. construction industry from the recent buildup of large amounts of air pollution controls, including the NO X SIP Call SCRs. As pointed out by one commenter, such experiences include use of modular construction, which can result in a significant reduction in the required boilermaker labor for CAIR retrofits. Also, as a result of this controls buildup, an increased number of experienced designers and construction personnel have become available to the industry. Some of these benefits may be offset by factors, such as the increased level of retrofit difficulty expected for the CAIR retrofits, especially for the small size units. However, we believe that the net effect of this experience is a more efficient use of the boilermaker labor in the construction of the air pollution control retrofits projects. Unfortunately, EPA cannot quantify the value of this experience in determining its overall impact on boilermaker requirements.

Therefore, EPA considers the 14 percent contingency in the available boilermaker-years for the above bounding analysis using commenter-suggested assumptions to be adequate.

ii. Issues Related to Compliance Deadline Acceleration

(I) Acceleration of Phase I Compliance Deadline

As a result of EPA's review of the comments received and further investigations conducted by the Agency for the final rule, the compliance deadline for implementing Phase I NO X controls has been moved up by one year. We believe that the affected plants would have sufficient time with this change to meet the CAIR requirements associated with NO X emissions, as long as the compliance deadline for implementing SO 2 controls is not changed. The EPA does not agree that accelerating the originally proposed Phase I compliance deadline of January 1, 2010, for implementing both NO X and SO 2 controls is possible. These issues are discussed below.

(A) Two-Year Phase I Acceleration for NO X and SO 2 Controls

With today's final action and allowing 18 months for the SIPs, sources installing controls would have approximately 31/4years for implementing the rule's requirements. Some commenters suggested moving Phase I forward by 2 years, with a new compliance deadline of January 1, 2008, which would reduce the implementation period to 11/4years. It is recognized that sources generally would not initiate any implementation activities that require major funding, before the final SIPs are available.

The EPA's projections show that, for SCR installation on one unit, an average 21-month schedule is required to complete purchasing, construction, and startup activities. For the same activities for FGD, an average 27-month schedule is required. As can be seen, the total time required for just one SCR or FGD installation exceeds the 11/4-year implementation period available for Phase I, if the compliance deadline is moved to January 1, 2008.

(B) One-Year Phase I Acceleration for NO X and SO 2 Controls

If the Phase I compliance deadline for both NO X and SO 2 controls is moved up by 1 year, the affected facilities would have 21/4years or 27 months to complete installation of these controls. As discussed in the preceding section, FGD installation on one unit requires an average 27-month schedule to complete purchasing, construction, and startup activities.

The sources installing controls on more than one unit at the same facility would likely stagger the outage-related activities, such as final hookup of the new equipment into the existing plant settings and startup, to minimize operational disruptions and avoid losing too much generating capacity at one time. The EPA projects that an average 2-month period is required to complete the outage construction activities and a 1-month period to complete the startup activities for FGD. Therefore, if back-to-back outages are assumed for a plant installing FGD on just two units, the 27 months needed to install FGD on the first unit and an additional 3 months needed for outage activities on the second unit would result in an overall schedule requirement of 30 months. This 30-month schedule exceeds the available 27-month implementation period, if the compliance deadline is moved up by 1 year. For plants installing FGD controls on more than two units and performing hookup construction and startup activities in back-to-back outages, an additional 3 months would be added to the 30-month schedule requirement for each additional unit.

The EPA notes that certain plants installing multiple-unit controls may be able to meet the compliance deadline requirement by using alternative approaches, such as simultaneous unit outages and purchase of allowances to defer installation of controls on some units. However, our projections for the final rule show that some facilities would be installing FGD controls on five multiple units at a single site. Moreover, these projections show 26 plants requiring FGD retrofit on more than one unit, which represents a major portion of the total number of plants required to install such controls under CAIR. We believe it would not be appropriate to expect this number of plants to resort to alternative means to accommodate such installations, such as simultaneous unit outages or purchasing of allowances.

For FGD retrofits, some plants would be required to obtain solid waste landfill permits. As discussed previously, the time required to obtain these permits could range from one to 31/2years. With the compliance deadline moved up by one year, the overall implementation period would be reduced from 43/4to 33/4years. For those plants subjected to a 31/2-year permit approval period, only 3 months would be available to prepare the permit applications at the beginning of the compliance period and to prepare the landfill area for accepting the waste after permit approval. The EPA does not believe that 3 months is adequate for such activities. These plants would, therefore, need the 43/4-year implementation period to complete activities related to landfills associated with the FGD systems.

The EPA also performed an analysis to verify if the available boilermaker labor is adequate to support the January 1, 2009, compliance deadline for both NO X and SO 2. This analysis was performed, using commenter-suggested boilermaker duty rates and EIA's assumptions for the natural gas prices and electricity demand rates. The results show that given these assumptions sufficient number of boilermakers will not be available and that there will be a shortfall of approximately 32 percent in the boilermakers available to support Phase I activities for this case.

Considering the constraints identified in the above analyses for the FGD installation schedule requirements and boilermaker labor availability, EPA believes that it is not reasonable to move the Phase I compliance deadline for both NO X and SO 2 caps to January 1, 2009.

(C) One-Year Phase I Acceleration for NO X Controls Only

A 1 year acceleration would result in a compliance deadline of January 1, 2009, for installing Phase I NO X controls. With this change, the affected sources installing these controls would have approximately 21/4years for implementing the rule's requirements, following the approval of State programs. However the implementation period for installing FGD controls would still be at 31/4years.

As shown previously, 21 months would be required to complete purchasing, construction, and startup of SCR on one unit. For multiple-unit installations with back-to-back unit outages for the tie-in construction and startup, the available 21/4-year implementation period would permit staggering of SCR installations on a maximum of three units (see the above referenced TSD). For a plant requiring SCR retrofit on more than three units, simultaneous outages of two units would become necessary. However, EPA notes that there are only six plants projected to require SCR installation on more than three units and, therefore, it is expected that simultaneous outages of two units at each of these plants would not have an adverse impact on the reliability of the electrical grid.

In addition, the plants installing SCR on more than three units at the same site would have two other options to meet the rule's requirements, without having to resort to simultaneous two-unit outages. First, these plants would be able to defer installation of SCRs on some of the units by receiving allocated allowances or purchasing allowances from the 200,000-ton Compliance Supplement Pool being made available as part of CAIR. [78] Second, the outage activities for some of the units at these plants could be extended into the first quarter of 2009, which is beyond the compliance deadline of January 1, 2009, since these units would not generate NO X emissions during an outage and therefore not require any allowances to compensate for them. The EPA's projections show that, of the above six plants installing SCR on more than three units, four of them require SCR retrofits on four units each. If it is assumed that these four plants would perform outage activities on the fourth unit during the first quarter of 2009, there would only be two plants left that would be required to either purchase allowances or perform work during simultaneous outages.

The EPA also notes that the total schedule requirements for multiple-unit plants can be reduced further by performing some of the activities, especially those related to planning and engineering, prior to the 21/4-year period. Also, with the total installation time requirement for FGD being more than that for SCR, EPA expects the outages associated with most Phase I FGDs to take place after January 1, 2009. The overall impact of the outages taken for these SCR and FGD retrofits would, therefore, be minimized.

The EPA also performed an analysis to determine the impact of an 1-year acceleration in the NO X compliance deadline on Phase I boilermaker labor requirements. Since the amounts of the required Phase I NO X and FGD retrofits are not affected by this change, the overall boilermaker requirements for this phase will remain the same as previously reported for the case with the same compliance deadline for both NO X and SO 2. However, with the new NO X compliance deadline, installation of all NO X retrofits would have to be completed by January 1, 2009, and some of the FGD construction work requiring boilermakers would also be done during this period. The EPA assumed that, along with completing installation of all SCRs, 35 percent of the boilermaker labor required to install all FGDs would be used in the period prior to January 1, 2009. This is a conservative assumption, since the amount of boilermaker labor used for this period would be greater than 50 percent of the total Phase I boilermaker labor requirement. The analysis performed by EPA shows that sufficient boilermakers would be available with a contingency factor of approximately 14 percent to install all SCR controls and 35 percent of the FGD retrofit work by January 1, 2009. This analysis is based on the most conservative assumptions, using the boilermaker duty rates suggested by one commenter and the EIA's projections for natural gas prices and electricity demand rates. Based on the above analyses, EPA believes that moving the compliance deadline for Phase I for both NO X and SO 2 is not practical. However, a 1-year acceleration in the compliance deadline for NO X only is feasible. Since EPA is obligated under the CAA to require emission reductions for obtaining NAAQS to be achieved as soon as practicable, we have based the final rule on two separate Phase I compliance deadlines of January 1, 2009, and January 1, 2010, for NO X and SO 2, respectively.

(II) Implementing All Controls in Phase I

The EPA proposed a phased program with the consideration that for engineering and financial reasons, it would take a substantial amount of time to install the projected controls. This program would require one of the most extensive capital investment and engineering retrofit programs ever undertaken in the U.S. for pollution control. The capital investment for pollution control for CAIR that would be installed by 2015 is estimated to be approximately 15 billion dollars. By 2015, close to 340 control unit retrofits will occur. This is occurring at a time when the industry also faces another major infrastructure challenge—upgrading transmission capacity to make the grid more reliable and economic to operate. This also will cost tens of billions of dollars.

The proposed program's objective was to eliminate upwind states' significant contribution to downwind nonattainment, providing air quality benefits as soon as practicable. A phased approach was also considered necessary because more of the difficult-to-retrofit and finance, smaller size units would be included in the second phase, which would allow them to complete activities necessary for implementing the required controls as well as provide them an opportunity to benefit from the lessons learned during the first phase.

In general, environmental controls resulting from legislative or regulatory actions are applied to those units first that offer superior choices from constructability and cost-effectiveness standpoints. Experience gained by the industry from these installations can then be used to develop innovative solutions for any constructability issues and to improve cost effectiveness, as these technologies are applied to harder-to-control units. The EPA believes that this phenomenon applies to the application of the SCR and FGD technologies at coal-fired power plants.

In the last few years, SCR and FGD systems have been added to several existing coal-fired units, under the NO X SIP Call and Acid Rain Program. These were mainly large units that had features, such as spacious layouts, amenable to the retrofit of the new air pollution control equipment. The units installing controls during Phase I of CAIR would, in general, be smaller in size and would offer relatively more difficult settings to accommodate the new equipment. These units would certainly benefit from the experience the industry has gained from the installations completed in recent years.

A large portion of the units (47 percent) projected to implement controls during the second phase consists of even smaller units, less than 200 MW in size. Compared to larger units, the retrofits for these smaller units would be more difficult to plan, design, and build. Historically, smaller units have been built with less equipment redundancy, smaller capacity margins, and more congested layouts. It is likely, therefore, to be more difficult and require additional design efforts to accommodate the new equipment into the existing settings for the smaller units. Use of lessons learned by firms constructing these units from the previous installations, including those to be built during the first phase, would help streamline this process and maintain the cost effectiveness of these installations. Moving a large portion of the retrofits required for these smaller units to the second phase also provides more time to complete the required retrofit activities.

Because EPA's projections for the second phase include a large proportion of smaller units, the total number of units requiring NO X and SO 2 controls exceeds that in the first phase (186 vs. 153). Requiring an acceleration of the second phase controls to be completed in the first phase would, therefore, more than double the number of retrofits required for the first phase from 153 to 339. Based on data available from EPA and other sources, the industry completed 95 SCR installations for the NO X SIP Call in 2002 and 2003. If the 2004 projections for the NO X SIP Call are added to this number, the total number of SCR retrofits over the 2002-2004 period would be 140. This is less than half the number that would be required for CAIR during a similar period, if the Phase II requirements are implemented along with the Phase I requirements. Also, the combined capacity for FGD and SCR retrofits required for Phase I would be 122.5 GW, which is approximately 57 percent greater than the installed SIP-Call SCR capacity for the 2002-2004 period. Such a change in the rule would therefore amount to imposing a requirement over the power industry that is significantly more demanding and burdensome than what the industry was required to do under the NO X SIP Call rule.

The EPA notes that critical resources other than the boilermakers are needed for the installation of SCR and FGD controls, such as construction equipment, engineering and construction staffs belonging to different trades, construction materials, and equipment manufacturers. Some commenters, based on their experience with NO X SIP Call, also pointed out that the requirement for some of these resources, especially construction equipment (e.g., large cranes used to mount SCR and scrubber vessels above ground), construction materials, equipment manufacturing shop capacities, and engineering and construction management teams overseeing these projects, is affected directly by the number of installations. The greater the requirement is to install a large number of retrofits by 2010, the greater would be the need for all these resources, which would be limited in the short term, as demands from equipment vendors, project teams, and material suppliers ramp up. In the NO X SIP Call, this led to shortages and bottlenecks in projects in certain areas, causing increased project times and costs. The EPA wants to avoid creating a similar situation by requiring too much at once.

The EPA has also acknowledged the increase in SCR costs during the NO X SIP Call implementation period, most likely due to an increase in construction costs (resulting from increased demand for boilermaker labor) and steel prices. The EPA has revised its estimates of SCR capital costs in the IPM runs for the final rule and believes the conservatism in its FGD capital costs also accounts for this factor.

The EPA believes that moving the Phase II requirements to the Phase I period could cause near-term shortages in some of the critical resources. This would further increase compliance costs and could remove the highly cost-effective nature of these controls and lead to a greater demand for natural gas.

In addition to the above, financing a large amount of controls for Phase I may prove challenging, especially for the coal plants owned by deregulated generators. As discussed later in this section, such generators are continuing to face serious financial challenges, and many have below investment grade credit ratings. This significantly complicates the financing of costly retrofit controls. Such plants would also not have the certainty of regulatory recovery of investments in pollution control, and would have to rely on the market to recover their costs. Having a second phase cap would allow these companies additional time to strengthen their finances and improve their cash flow.

In the interest of being prudent in evaluating the need to phase in the program, EPA also performed an analysis to determine if the available boilermaker labor would be adequate to support installation of all Phase I and II controls in 2010. This analysis was conservatively based on using commenter-suggested boilermaker duty rates and EIA's projections for gas prices and electricity demand rates. The results show that a sufficient number of boilermakers will not be available and that there will be a shortfall of approximately 25 percent in the boilermakers available to support Phase I activities for this case.

Based on the above analyses, EPA believes that implementation of controls for both phases in Phase I is impractical. We also believe that it is prudent and reasonable in requiring the industry to undertake this massive retrofit program on a two-phase schedule, to be largely completed in less than a decade.

(III) Acceleration of Phase II Compliance Deadline

The EPA does not believe that acceleration of the compliance deadline for the second phase is reasonable. As pointed out earlier, a large portion of the units projected to install controls during the second phase consists of small units, less than 200 MW in size. Due to the issues related to financing of the retrofit projects for some of these units and considering that planning and designing of controls for these units is likely to take longer, EPA does not consider the schedule acceleration to be appropriate.

The EPA notes that Phase I of CAIR is the initial step on the slope of emissions reduction (the glide-path) leading to the final control levels. Because of the incentive to make early emission reductions that the cap-and-trade program provides, reductions will begin early and will continue to increase through Phases I and II. The EPA, therefore, does not believe that all of the required Phase II emission reductions would take place on January 1, 2015, the compliance deadline. These reductions are expected to accrue throughout the implementation period, as the sources install controls and start to test and operate them.

The EPA also notes that the 5-year implementation period for Phase II is consistent with other regulations and statutory requirements, such as title IV for SO 2 and NO X controls. In addition, some commenters have cited a need for a 6-year period for obtaining financing for plants owned by the co-operatives. These facilities are likely to commit funds for major activities, only after financing has been obtained. Therefore, for such facilities, a period of approximately four years would be available for procuring, installing, and startup activities, assuming that the financing activities were started right after the rule is finalized. Since the plants owned by co-operatives are usually small in size, they are likely to require and be benefitted by the extra time allowed to them by this four-year implementation period.

The EPA also performed an analysis to verify adequacy of the available boilermaker labor for pollution control retrofits the power industry will install to comply with the Phase II CAIR requirements. A 36-month construction period requiring boilermakers was conservatively selected for this analysis. Based on the IPM analysis for the final rule, conservatively, the power industry will build 27.5 GW of FGD and 26.6 GW of SCR retrofits for compliance with lower emission caps that go into effect for NO X and SO 2 in 2015. The analysis was based on using EIA's projections for the natural gas prices and electricity demand rates and the commenter-suggested boilermaker duty rates. The results show availability of ample boilermakers with a contingency factor of 46 percent to support Phase II activities.

The EPA notes that the retrofits that will occur in Phase II will be smaller, more numerous, and more challenging, since the easiest controls will likely be installed in Phase I. Therefore, having a greater contingency factor (as we do) is warranted. This is further supported when the uncertainty in predicting the construction activities in the areas outside of air pollution controls is considered. Notably after 2010, the excess generation capacity that we have today is no longer expected to be present and there may be a shift towards a requirement for increasing generation capacity. Increased construction of new power plants will have a direct impact on the availability of boilermakers for the Phase II controls. The EPA believes that a higher contingency factor for Phase II is desirable to ensure that the industry will succeed in getting the required reductions at the required time.

Any acceleration of the Phase II compliance deadline will also cause an appreciable reduction in the above estimated contingency factor for boilermaker labor. For example, based on EPA analysis, an acceleration of one year is projected to reduce this contingency factor to only about one percent. Therefore, EPA believes that acceleration of the Phase II compliance deadline cannot be justified.

3. Assure Financial Stability

The EPA recognizes that the power sector will need to devote large amounts of capital to meet the control requirements of the first phase. Furthermore, over the next 10 years, the power sector is facing additional financial challenges unrelated to environmental issues, including economic restructuring impacts, investments related to domestic security and investments related to electrical infrastructure. Among the consideration of other factors, EPA believes it is important to take into account the ability of the power sector to finance the controls required under CAIR. A detailed assessment of the status of the financial health of the U.S. Utility Industry, particularly of the unregulated sector is offered in the TSD, “U.S. Utility Industry Financial Status and Potential Recovery.”

Commenters have noted that they appreciate EPA's growing realization that many companies may have difficulty securing financing, and the agency's establishment of a two-phase reduction program on both technical and financial grounds.

Utilities and non-utility generating companies have felt significant financial pressure over the past 5 years. The years 2000 and 2001 saw the escalation and fallout from the California energy crisis, the bankruptcy of Enron, and a massive building program, largely on the side of the merchant generating sector. Subsequent low power margins and large debt obligations have led to a significant number of credit downgrades of utilities and power generators and the bankruptcy of coal-generating merchant companies. According to Standard and Poor's, a leading provider of investment ratings, there were almost ten times more downgrades of utility credit in 2002 and 2003 than there were upgrades. While more recently the sector has stabilized, a significant number of owners of coal-fired capacity in the CAIR region, particularly those with deregulated capacity, are still at below investment-grade credit ratings.

In general, EPA believes that regulated plants, given appropriate regulatory requirements, should not face significant financial problems meeting their obligations under CAIR. While EPA recognizes that issues such as the expiration of rate caps and the time lags associated with regulatory approval and recovery may provide cash flow challenges, regulated electricity rates are generally seen as a positive factor in credit ratings, as entities are allowed a recovery on prudent investment through rate cases (and, in some jurisdictions, the recovery of allowance expenditures through fuel adjustment clauses).

Deregulated coal capacity (operating in an environment of market prices rather than electricity rates set by regulators) has no such guarantees, and would need to recover investments in pollution control from market prices (which in many cases are not set by coal units). Additionally, deregulated entities, because of their more aggressive building and borrowing strategies and reliance on market prices (which now reflect the current capacity overbuild), have faced more significant financial difficulties (including a number of bankruptcies) and are currently in a weaker position financially. [79] A number of firms that have avoided financial distress in the near term have done so by renegotiating their pending debt, postponing payment. A good portion of this debt is of a shorter-term nature, and will be coming due in the next five years.

Such financial difficulties increase the cost of capital necessary for capital expenditures and affect the availability of such capital, making required controls more expensive. Recent financial troubles have been cited as the reason for the deferment or cancellation of pollution control expenditures. Should interest rates rise in the future, it will become more difficult and costly for utilities seeking financing.

These problems impact a significant segment of coal generators, as deregulated coal capacity makes up about a third of all U.S. coal capacity and almost 90 percent of this deregulated capacity would be affected by CAIR requirements.

Given the lead times needed to plan and construct such equipment, as well as the financial uncertainty many of the plant owners are confronting, companies may find it difficult to install controls at their plants too quickly. The EPA believes that the choice of timing of the emission caps in CAIR would allow firms time to improve their current and near-term financial difficulties (through reorganization, mergers, sales, etc.). Phasing in the more stringent emission caps by 2015 would also spread investment requirements and resulting cash flow demands, rather than forcing firms to finance a large spike in investments in a very short time period, while they are still trying to recover financially.

The timing of controls expected to be installed as a result of CAIR are similar to that noted in EPA's analysis of the Clear Skies proposal. The EPA looked in detail at the potential financial impact of the Clear Skies program (particularly focusing on the deregulated coal sector). The EPA found that some individual deregulated coal plants might be adversely affected, but on average such plants would actually experience a small financial improvement under Clear Skies. Baseload deregulated coal plants would benefit from even slight increases in the price of natural gas ( units burning natural gas generally set the wholesale price of electricity on the margin in the regions where deregulated coal is located). These units would also be recipients of allocated allowances. Overall, the phased in nature of CAIR, the fact that most coal plants continue to be regulated and the fact that sources would also receive allowances, would all mitigate the financial impact of this rule.

The EPA believes that the timing requirements finalized today reflect a prudent and cautious approach designed to assure that the industry will succeed in implementing this program. The EPA believes that deferring the second phase to 2015 will provide enough time for companies to raise additional capital needed to install controls. Also, we believe that the implementation period should account (at least broadly) for the possibility that electricity demand or natural gas prices may increase more than assumed, and therefore that additional control equipment would be needed. Allowing until 2015 for implementation of the more stringent control levels in today's rule will provide more flexibility in the event of greater electricity demand and will ensure that power plants in the CAIR region will have the ability, both technical and financial, to make the pollution control retrofits required.

Currently, EPA is cooperating with the National Association of Regulatory Utility Commissioners (NARUC) in developing a menu of policy options and financial incentives for encouraging improved environmental performance for generation. A survey of a number of States was conducted as part of this effort, and policies such as pre-approval statutes for compliance plans, state income tax credits, accelerated depreciation, and special treatment of allowance transactions were cited as examples of such policies [80] . Such policies will ease some of the financial pressures of CAIR by providing greater regulatory certainty and lowering the effective costs of controls.

D. Control Requirements in Today's Final Rule

1. Criteria Used To Determine Final Control Requirements

The EPA's general approach to developing emission reduction requirements—basing the requirements on the application of highly cost-effective controls—was adopted in the NO X SIP Call and has been sustained in court. In the NPR, the Agency proposed this approach for developing SO 2 and NO X emission reduction requirements. The majority of commenters accepted this basic approach for determining reduction requirements. Some commenters did suggest other approaches, however, as discussed above.

Many commenters suggested that the CAIR regionwide SO 2 and NO X control levels should be more or less stringent than the levels proposed in the NPR. The EPA has determined that the control levels that we are finalizing today are highly cost-effective and feasible, and constitute substantial reductions that address interstate transport, at the outset of State and EPA efforts to bring about attainment of the PM 2.5 NAAQS (EPA believes that most if not all States will obtain CAIR reductions by capping emissions from the power sector). Today, EPA finalizes the use of both average and marginal cost effectiveness of controls as the basis for determining the highly cost-effective amounts.

In the CAIR NPR, EPA proposed criteria for determining the appropriate levels of SO 2 and NO X emissions reductions, and stated that EPA considered a variety of factors in evaluating the source categories from which highly cost-effective reductions may be available and the level of reduction assumed from that sector (69 FR 4611). The EPA has reviewed comments on its NPR, SNPR and NODA and conducted further analyses with respect to the proposed criteria, and is finalizing its control requirements in today's action. Following is a brief summary of EPA's conclusions based on the criteria.

The availability of information, and the identification of source categories emitting relatively large amounts of the relevant emissions, are two criteria used in EPA's evaluation of the CAIR program. In the NPR, EPA stated that EGUs are the most significant source of SO 2 emissions and a very substantial source of NO X in the affected region, and further stated that highly cost-effective control technologies are available for achieving significant SO 2 and NO X emissions reductions from EGUs. We requested comment on sources of information for emissions and costs from other sectors (69 FR 4610). A detailed discussion regarding non-EGU sources is provided above. The EPA has not received additional information that would change its proposed control strategy.

Another criterion is the performance and applicability of control measures. The NPR included a detailed discussion of the performance and applicability of SO 2 and NO X control technologies for EGUs. In particular, EPA discussed FGD for SO 2 removal and SCR for NO X removal, both of which are fully demonstrated and available pollution control technologies on coal-fired EGU boilers (69 FR 4612). None of the commenters provided information that differed from EPA's assessment of the performance of these control measures. In addition, the commenters generally supported EPA's assumptions on the applicability of these controls.

The cost effectiveness of control measures is another criterion used in EPA's analysis. As discussed in detail above, EPA determined that the proposed control levels are highly cost-effective, and is finalizing the levels in today's action. The EPA used IPM to analyze the cost effectiveness of the proposed and final CAIR control requirements. IPM incorporates assumptions about the capital costs and fixed and variable operations and maintenance costs of control measures for EGUs. Several commenters suggested that the SCR control cost assumptions that we used in IPM analysis for the NPR were too low. Consequently, we increased the SCR control cost assumptions in IPM and conducted cost effectiveness modeling for the final control requirements using these updated costs. [81] Commenters generally supported our FGD control costs assumptions, which are largely unchanged from the NPR modeling to the modeling for today's final rule.

And finally, EPA considered engineering and financial factors that affect the availability of control measures. The EPA conducted a detailed analysis of engineering factors that affect timing of control retrofits, including an evaluation of the comments received. The EPA's analysis supports its compliance schedule, a two-phase emissions control program with the final phase commencing in 2015, and with a first phase commencing in 2010 for SO 2 reductions and in 2009 for NO X reductions. Further, EPA's analysis demonstrates that it would not be realistically possible to start the program sooner, or to impose more stringent emissions caps in the first phase.

Based on EPA's review of comments and analysis, EPA determined that the proposed control requirements are reasonable with respect to engineering factors. As discussed above, EPA also considered how to avoid creating financial instability for the affected sector, and how to ensure the capital needed for the required controls would be readily available. Assuming States choose to control EGUs, the power sector will need to devote large amounts of capital to meet the CAIR control requirements.

The EPA explained that implementing CAIR as a two-phase program, with the more stringent control levels commencing in the second phase, will allow time for the power sector to address any financial challenges. The EPA's evaluation of engineering and financial factors supports the decision to implement CAIR as a two-phase program, with the final (second) compliance level commencing in 2015 and a first phased-in level starting in 2010 for SO 2 reductions and in 2009 for NO X reductions. A description of the final CAIR control requirements follows.

2. Final Control Requirements

Today's final rule implements new annual SO 2 and NO X emissions control requirements to reduce emissions that significantly contribute to PM 2.5 nonattainment. The final rule also requires new ozone season NO X emissions control requirements to reduce emissions that significantly contribute to ozone nonattainment.

The final rule requires annual SO 2 and NO X reductions in the District of Columbia and the following 23 States: Alabama, Florida, Georgia, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Texas, Virginia, West Virginia, and Wisconsin. (In the “Proposed Rules” section of today's action, EPA is publishing a proposal to include Delaware and New Jersey in the CAIR region for annual SO 2 and NO X reductions.)

In addition, the final rule requires ozone season NO X reductions in the District of Columbia and the following 25 States: Alabama, Arkansas, Connecticut, Delaware, Florida, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Massachusetts, Michigan, Mississippi, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia, and Wisconsin.

The CAIR requires many of the affected States to reduce annual SO 2 and NO X emissions as well as ozone season NO X emissions. However, there are three States for which only annual emission reductions are required (Georgia, Minnesota and Texas). Likewise, there are five States for which only ozone season reductions are required (Arkansas, Connecticut, Delaware, Massachusetts, and New Jersey). The following 20 States and the District of Columbia are required to make both annual and ozone season reductions: Alabama, Florida, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Mississippi, Missouri, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia and Wisconsin.

Table IV-14 shows the amounts of regionwide annual SO 2 and NO X emissions reductions under CAIR that EPA projects, if States choose to meet their CAIR obligations by controlling EGUs. Table IV-15 shows the amounts of regionwide ozone season NO X emissions reductions under CAIR that EPA projects, if States choose to meet their CAIR obligations by controlling EGUs. If all affected States choose to implement these reductions through controls on EGUs, the regionwide annual SO 2 and NO X emissions caps that would apply for EGUs are also shown in the Table IV-14, and ozone season NO X caps for EGUs are in Table IV-15. Base case emissions levels for affected EGUs as well as emissions with CAIR are also shown in Table IV-14 and Table IV-15, based on IPM modeling.

The EPA is finalizing the regionwide EGU SO 2 emissions caps—if States choose to comply by controlling EGUs—as shown in Table IV-14 [82] . As indicated above, EPA identified SO 2 budget amounts, as target levels for further evaluation, by adding together the title IV Phase-II allowances for all of the States in the CAIR region, and making a 50 percent reduction for the 2010 cap and a 65 percent reduction for the 2015 cap. The EPA determined, through IPM analysis, that the resulting regionwide emissions caps (if all States choose to obtain reductions from EGUs) are highly cost-effective levels.

Also, EPA is finalizing the regionwide EGU annual and ozone season NO X emission caps—if States choose to comply by controlling EGUs—as shown in Table IV-14 and Table IV-15. [83] As indicated above, EPA identified NO X budget amounts, as target levels for further evaluation, through the methodology of determining the highest recent Acid Rain Program heat input from years 1999-2002 for each affected State, summing the highest State heat inputs into a regionwide heat input, and multiplying the regionwide heat input by 0.15 lb/mmBtu and 0.125 lb/mmBtu for 2009 and 2015, respectively. The EPA determined, through IPM analysis, that the resulting regionwide emissions caps (if all States choose to obtain reductions from EGUs) are highly cost-effective levels.

The emission reductions, EGU emissions caps, and emissions shown in Table IV-14 are for the 23 States and the District of Columbia that are required to make annual SO 2 and NO X reductions for CAIR. (Table IV-14 does not include information for the five States that are required to make ozone season reductions only.)

The emission reductions, EGU emissions caps, and emissions shown in Table IV-15 are for the 25 States and the District of Columbia that are required to make ozone season NO X reductions for CAIR. (Table IV-15 does not include information for the three States that are required to make annual reductions only.)

The EPA is requiring the CAIR SO 2 and NO X emissions reductions in two phases. For States affected by annual SO 2 and NO X emission reductions requirements, the final (second) phase commences January 1, 2015, and the first phase begins January 1, 2010 for SO 2 reductions and January 1, 2009 for NO X reductions. For States affected by ozone season NO X emission reductions requirements, the final (second) phase commences May 1, 2015 and the first phase starts May 1, 2009. Notably, the first phase control requirements are effective in years 2010 through 2014 for SO 2 and in years 2009 through 2014 for NO X, and the 2015 requirements are for that year and thereafter.

Table IV-14.—Final Rule SO 2 and NO X Annual Base Case Emissions, Emission Caps, Emissions After CAIR and Emission Reductions in the Region Required To Make AnnualSO 2 and NO X Reductions (23 State and DC) for the Interim Phase (2010 for SO 2 and 2009 for NO X) and Final Phase (2015 for SO 2 and NO X) for EGUs Back to Top
Base case emissions CAIR emissions caps Emissions after CAIR Emissions reduced
(Million Tons)84
Notes: Numbers may not add due to rounding.
1. The emission caps that EPA used to make its determination of highly cost-effective controls and the emission reductions associated with those caps are shown in Table IV-14. For a discussion of the emission reduction requirements if States control source categories other than EGUs, see section VII in this preamble. Emissions shown here are for EGUs with capacity greater than 25 MW.
2. The District of Columbia and the following 23 States are affected by CAIR for annual SO 2 and NO X controls: AL, FL, GA, IA, IL, IN, KY, LA, MD, MI, MN, MO, MS, NY, NC, OH, PA, SC, TN, TX, VA, WV, WI.
3. The 2010 SO 2 emissions cap applies to years 2010 through 2014. The 2009 NO X emissions cap applies to years 2009 through 2014. The 2015 caps apply to 2015 and beyond.
4. Due to the use of the existing bank of SO 2 allowances, the estimated SO 2 emissions in the CAIR region in 2010 and 2015 are higher than the emissions caps.
5. Over time the banked SO 2 emissions allowances will be consumed and the 2015 cap level will be reached. SO 2 emissions levels can be thought of as on a flexible “glide path” to meet the 2015 CAIR cap with increasing reductions over time. The annual SO 2 emissions levels in 2020 with CAIR are forecasted to be 3.3 million tons within the region encompassing States required to make annual reductions, an annual reduction of 4.4 million tons from base case levels.
First phase (2010 for SO 2 and 2009 for NO X)        
SO 2 8.7 3.6 5.1 3.5
NO X 2.7 1.5 1.5 1.2
Sum 11.4 NA 6.6 4.8
Second Phase (2015 for SO 2 and NO X)        
SO 2 7.9 2.5 4.0 3.8
NO X 2.8 1.3 1.3 1.5
Sum 10.6 NA 5.3 5.3

Table IV-15.—Final Rule NO X Ozone Season Base Case Emissions, Emissions Caps, Emissions after CAIR and Emission Reductions in the Region Required to Make Ozone Season NO X Reductions (25 States and DC) for the Interim Phase (2009) and Final Phase (2015) for Electric Generation Units Back to Top
Ozone Season NO X
Phase Base case emissions CAIR emissions caps Emissions after CAIR Emissions reduced
(Million Tons)85
Notes:
1. The emission caps that EPA used to make its determination of highly cost-effective controls and the emission reductions associated with those caps are shown in Table IV-15. For a discussion of the emission reduction requirements if States control source categories other than EGUs, see section VII in this preamble. Emissions shown here are for EGUs with capacity greater than 25 MW.
2. The District of Columbia and the following 25 States are affected by CAIR for ozone season NO X controls: AL, AR, CT, DE, FL, IA, IL, IN, KY, LA, MA, MD, MI, MO, MS, NJ, NY, NC, OH, PA, SC, TN, VA, WV, WI.
3. The 2009 NO X emissions cap applies to years 2009 through 2014. The 2015 cap applies to 2015 and beyond.
2009 0.7 0.6 0.6 0.1
2015 0.7 0.5 0.5 0.2

Table IV-16 shows the estimated amounts of regionwide annual SO 2 and NO X emissions reductions that would occur if EPA finalizes its proposal to find that Delaware and New Jersey contribute significantly to downwind PM 2.5 nonattainment, and if all affected States choose to control EGUs (the proposal is published in the “Proposed Rules” section of today's action). In that case, the estimated regionwide annual SO 2 and NO X emissions caps that would apply for EGUs are as shown in Table IV-16. Annual base case emissions levels for EGUs in the CAIR region (including Delaware and New Jersey) as well as emissions with CAIR are also shown in the Table, based on IPM modeling. If EPA finalizes its proposal to include Delaware and New Jersey for PM 2.5 requirements, then the ozone season requirements would not change for States required to make ozone season reductions for CAIR.

Based on EPA modeling with Delaware and New Jersey included in the PM 2.5 region (and if all affected States choose to control EGUs), the EGU emissions caps and the ozone season NO X emissions and emission reductions associated with those caps, for the 25 States and the District of Columbia that are required to make ozone season NO X reductions, would be as shown in Table IV-15, above. [86]

Table IV-16.—SO 2 and NO X Annual Base Case Emissions, Emissions Caps, Emissions After CAIR and Emission Reductions in the Region Required to Make Annual SO 2 and NO X Reductions (25 States and DC) for the Initial Phase (2010 for SO 2 and 2009 for NO X) and Final Phase (2015 for SO 2 and NO X) for Electric Generation Units if EPA Finalizes Its Proposal to Include Delaware and New Jersey for PM 2.5 Requirements Back to Top
First phase (2010 for SO 2 and 2009 forNO X)
Base case emissions CAIR emissions caps Emissions after CAIR Emissions reduced
[Million tons]87
Note: Numbers may not add due to rounding.
1The emission caps that EPA used to make its determination of highly cost-effective controls and the emission reductions associated with those caps are shown in Table IV-16. For a discussion of the emission reduction requirements if States control source categories other than EGUs, see section VII in this preamble. Emissions shown here are for EGUs with capacity greater than 25 MW.
2The District of Columbia and the following 25 States would be affected by CAIR for annual SO 2 and NO X controls if EPA finalizes its proposal to include DE and NJ: AL, DE, FL, GA, IA, IL, IN, KY, LA, MD, MI, MN, MO, MS, NJ, NY, NC, OH, PA, SC, TN, TX, VA, WV, WI.
3The 2010 SO 2 emissions cap would apply to years 2010 through 2014. The 2009 NO X emissions cap would apply to years 2009 through 2014. The 2015 caps would apply to 2015 and beyond.
4Due to the use of the existing bank of SO 2 allowances, the estimated SO 2 emissions in the CAIR region in 2010 and 2015 would be higher than the emissions caps.
5Over time the banked SO 2 emissions allowances would be consumed and the 2015 cap level would be reached. SO 2 emissions levels can be thought of as on a flexible “glide path” to meet the 2015 CAIR cap with increasing reductions over time. The annual SO 2 emissions levels in 2020 with CAIR, within the region of States required to make annual reductions (including Delaware and New Jersey), are forecasted to be 3.3 million tons, an annual reduction of 4.4 million tons from base case levels.
SO 2 8.8 3.7 5.2 3.6
NO X 2.8 1.5 1.5 1.2
Sum 11.5 NA 6.7 4.8
Second phase
(2015 for SO 2 andNO X)
Base case emissions CAIR emissions caps Emissions after CAIR Emissions reduced
SO 2 7.9 2.6 4.1 3.9
NO X 2.8 1.3 1.3 1.5
Sum 10.7 NA 5.3 5.4

TheEPA apportioned the EGU caps—and associated required regionwide emission reductions—on a State-by-State basis. The affected States may determine the necessary controls on SO 2 and NO X emissions to achieve the required reductions. The EPA's apportionment method and the resulting State EGU emissions budgets are described in Section V in today's preamble.

To achieve the required SO 2 and NO X reductions in the most cost-effective manner, EPA suggests that States implement these reductions by controlling EGUs under a cap and trade program that EPA would implement.

However, the States have flexibility in choosing the sources that must reduce emissions. If the States choose to require EGUs to reduce their emissions, then States must impose a cap on EGU emissions, which would in effect be an annual emissions budget. Provisions for allocating SO 2 and NO X allowances to individual EGUs—which apply if a State chooses to control EGUs and elects to allow them to participate in the interstate cap and trade program—are presented elsewhere in today's preamble. If a State wants to control EGUs, but does not want to allow EGUs to participate in the interstate cap and trade program, the State has flexibility in allocating allowances, but it must cap EGUs. Sources that are subject to the emission reduction requirements under title IV continue to be subject to those requirements.

If the States choose to control other sources, then they must employ methods to assure that those other sources implement controls that will yield the appropriate amount of annual emissions reduction. See section VII (SIP Criteria and Emissions Reporting Requirements) in today's preamble.

Implementation of the cap and trade program is discussed in section VIII in today's preamble.

For convenience, we use specific terminology to refer to certain concepts. “State budget” refers to the statewide emissions that may be used as an accounting technique to determine the amount of annual or ozone season emissions reductions that controls may yield. It does not imply that there is a legally enforceable statewide cap on emissions from all SO 2 or NO X sources. “Regionwide budget” refers to the amount of emissions, computed on a regionwide basis, which may be used to determine State-by-State requirements. It does not imply that there is a legally enforceable regionwide cap on emissions from all SO 2 or NO X sources. “State EGU budget” refers to the legally enforceable annual or ozone season emissions cap on EGUs a State would apply should it decide to control EGUs.

V. Determination of State Emissions Budgets Back to Top

The EPA outlined in the NPR and SNPR its proposals regarding a methodology for setting both regional and State-level SO 2 and NO X budgets. Section IV explains how the regionwide budgets were developed. This section V describes how EPA apportions the regionwide emissions reductions—and the associated EGU caps—on a State-by-State basis, so that the affected States may determine the necessary controls of SO 2 and NO X emissions.

In the NPR and SNPR, EPA proposed annual SO 2 and NO X caps for States contributing to fine particle nonattainment and separate ozone-season only caps for States contributing to ozone—but not fine particle—nonattainment. The EPA is finalizing an annual cap for both SO 2 and NO X for States that contribute to fine particle nonattainment. In addition, EPA is finalizing an ozone-season only cap for NO X for all States that contribute to ozone nonattainment.

States have several options for reducing emissions that significantly contribute to downwind nonattainment. They can adopt EPA's approach of reducing the emissions in a cost-effective manner through an interstate cap and trade program. This approach would, by definition, achieve the required cost-effective reductions. Alternately, States could achieve all of the necessary emissions reductions from EGUs, but choose not to use EPA's interstate emissions trading program. In this case, a State would need to demonstrate that it is meeting the EGU budgets outlined in this section. Finally, States could obtain at least some of their required emissions reductions from sources other than EGUs. Additional detail on these options is provided in section VII.

A. What Is the Approach for Setting State-by-State Annual Emissions Reductions Requirements and EGU Budgets?

This section presents the final methodologies used for apportioning regionwide emission reduction requirements or budgets to the individual States.

In the CAIR NPR, EPA proposed methods for determining the SO 2 and NO X emission reduction requirements or budgets for each affected State. In the June 2004 SNPR, EPA proposed corrections and improvements to the proposals in the CAIR NPR. In the August 2004 NODA, EPA presented the corrected NO X budgets resulting from the improvements proposed in the SNPR.

1. SO 2 Emissions Budgets

a. State Annual SO 2 Emission Budget Methodology

As noted elsewhere in today's preamble, the regionwide annual budget for 2015 and beyond is based on a 65 percent reduction of title IV allowances allocated to units in the CAIR States for SO 2 control. The regionwide annual SO 2 budget for the years 2010-2014 is based on a 50 percent reduction from title IV allocations for all units in affected States.

In the NPR and SNPR, EPA also proposed calculating annual State SO 2 budgets based on each State's allowances under title IV of the 1990 CAA Amendments. We are finalizing this proposed approach for determining State annual SO 2 budgets.

State annual budgets for the years 2010-2014 (Phase I) are based on a 50 percent reduction from title IV allocations for all units in the affected State. The State annual budget for 2015 and beyond (Phase II) is based on a 65 percent reduction of title IV allowances allocated to units in the affected State for SO 2 control.

Some commenters criticized EPA's basing State budgets on title IV allocations since these were based largely on 1985-1987 historic heat input data. Commenters argue that the initial allocation was not equitable and that in any event, the electric power sector has changed significantly. They conclude that State budgets should reflect those differences. Commenters have also commented that tying SO 2 allocations to title IV also does not let States account for units that are exempt from title IV or for new units that have come online since 1990.

While acknowledging these concerns, EPA believes, for a number of reasons, that setting State budgets according to title IV allowances represents a reasonable approach.

The EPA believes that basing budgets on title IV allowances is necessary in order to ensure the preservation of a viable title IV program, which is important for reasons discussed in section IX of this preamble. Such reasons include the desire to maintain the trust and confidence that has developed in the functioning market for title IV allowances. The EPA believes it is important not to undermine such confidence (which is an essential underpinning to a viable market-based system) recognizing that it is a key to the success of a trading program under the CAIR.

The title IV program represents a logical starting point for assessing emissions reductions for SO 2, since it is the current effective cap on SO 2 emissions for Acid Rain units, which make up the large majority of affected EGU CAIR units. It is from this starting emissions cap, that further CAIR reductions are required. Consequently, EPA proposes State-level reductions based on reductions from the initial allocations of title IV allowances to individual units at sources (power plants) in States covered by the CAIR.

The setting of SO 2 budgets differs from the setting of NO X budgets for the CAIR, in part, because of this difference in starting points—since there is no existing NO X regional annual cap, and no currency for emissions, on which sources rely. Furthermore, Congress, as part of title IV of the CAA, decided upon the allocations of title IV allowances specifically for the control of SO 2, and not for NO X.

Moreover, Congress decided to allocate title IV allowances in perpetuity, realizing that the electricity sector would not remain static over this time period. Congress clearly did not choose a policy to regularly revisit and revise these allocations, believing that its allocations methodology for title IV allowances would be appropriate for future time periods.

The EPA realizes, putting aside concerns of linkage to title IV, that there are numerous potential methodologies of dividing up the regional budgets among the States. Also, EPA believes, that while initial allocations of State budgets are important for distributional reasons, under a cap and trade system, they would not impact the attainment of the environmental objectives or the overall cost of this rule.

Each of the alternate methods also has certain shortcomings, many of which have been identified by commenters. Basing allowances on historic emissions, for instance, would penalize States that have already gone through significant efforts to clean up their sources. Basing allowances on heat input has advantages, but cannot accommodate States that have worked to improve their energy efficiency. Basing allowances on output would provide gas-fired units with many more allowances than they need, rather than giving them to the coal-fired units that will be incurring the greatest costs from the tighter caps.

The EPA did look at a number of allowance outcomes using alternate potential methods for allocating SO 2 allowances. These methods included allocating on the basis of historic emissions, heat input (with alternatives based on heat input from all fossil generation, and heat input from coal- and oil-fired generation only) and output (with alternatives based on all generation and all fossil-fired generation). Allocating allowances based on title IV yields results that fall within a reasonable range of results obtained from using these alternate methodologies. In fact, calculating State budgets using title IV allowances yields budgets generally at or within the ranges of budgets calculated using the other methods in more than two-thirds of the States, which account for over 85 percent of the total heat input in the region from 1999-2002. This analysis is discussed further in the response to comments document.

b. Final SO 2 State Emission Budget Methodology

The EPA is finalizing the budgets as noted in the SNPR, adjusting for the proper inclusion of States covered under the final CAIR. The final State budgets are included in Table V-1 below. Details of the data and methodology used to calculate these budgets are included in the accompanying “Regional and State SO 2 and NO X Emissions Budgets” Technical Support Document.

Table V-1.—Final Annual Electric Generating Units SO 2 Budgets Back to Top
State State SO 2 budget 2010* State SO 2 budget 2015**
[Tons]
*Annual budget for SO 2 tons covered by allowances for 2010-2014.
**Annual budget for SO 2 tons covered by allowances for 2015 and thereafter.
Alabama 157,582 110,307
District of Columbia 708 495
Florida 253,450 177,415
Georgia 213,057 149,140
Illinois 192,671 134,869
Indiana 254,599 178,219
Iowa 64,095 44,866
Kentucky 188,773 132,141
Louisiana 59,948 41,963
Maryland 70,697 49,488
Michigan 178,605 125,024
Minnesota 49,987 34,991
Mississippi 33,763 23,634
Missouri 137,214 96,050
New York 135,139 94,597
North Carolina 137,342 96,139
Ohio 333,520 233,464
Pennsylvania 275,990 193,193
South Carolina 57,271 40,089
Tennessee 137,216 96,051
Texas 320,946 224,662
Virginia 63,478 44,435
West Virginia 215,881 151,117
Wisconsin 87,264 61,085
Total 3,619,196 2,533,434

c. Use of SO 2 Budgets

These specific levels of the proposed State budgets would actually provide binding statewide caps on EGU emissions for States that choose to control only EGUs but do not want to participate in the trading program. For States choosing to participate in the trading program, these State budgets would not be binding, instead, the States' SO 2 reductions would be achieved solely through the application of required retirement ratios as discussed in section VII of this preamble. For States controlling both EGUs and non-EGUs (or controlling only non-EGUs), these State budgets would be used to calculate the emissions reductions requirements for non-EGUs and the remaining reduction requirement for EGUs. This is described in more detail in the section VII discussion on SIP approvability.

2. NO X Annual Emissions Budgets

a. Overview

In this section, EPA discusses the apportioning of regionwide NO X annual emission reduction requirements or budgets to the individual States. In the January 2004 proposal, we proposed State EGU annual NO X budgets based on each State's average share of recent historic heat input. In the SNPR, we proposed the same input-based methodology, but revised the budgets based on more complete heat input data. Also, EPA took comment on an alternative methodology that determines State budgets by multiplying heat input data by adjustment factors for different fuels. In the August NODA, EPA presented the corrected annual NO X budgets resulting from the improved methodology proposed in the SNPR.

b. State Annual NO X Emissions Budget Methodology

Proposed and Discussed NO X Emission Budget Methodology

As noted elsewhere in today's preamble, EPA determined historical annual heat input data for Acid Rain Program units in the applicable States and multiplied by 0.15 lb/mmBtu (for 2009) and 0.125 lb/mmBtu (for 2015) to determine total annual NO X regionwide budgets for the CAIR region. The EPA applied these rates to each individual State's total highest annual heat input for any year from 1999 through 2002. Thus, EPA used the heat input total for the year in which a State's total heat input was the highest.

In the January 2004 proposal, we proposed annual NO X State budgets for a 28-State (and D.C.) region based on each jurisdiction's average heat input—using heat input data from Acid Rain Program units—over the years 1999 through 2002. We summed the average heat input from each of the applicable jurisdictions to obtain a regional total average annual heat input. Then, each State received a pro rata share of the regional NO X emissions budget based on the ratio of its average annual heat input to the regional total average annual heat input.

In the SNPR, EPA proposed to revise its determination of State NO X budgets by supplementing Acid Rain Program unit data with annual heat input data from the U.S. Energy Information Administration (EIA), for the non-Acid Rain unit data. A number of commenters had suggested that this would better reflect the heat input of the units that will be controlled under the CAIR, and EPA agrees.

In the SNPR, EPA asked for, and subsequently received, comments on determining State budgets by multiplying heat input data by adjustment factors for different fuels. The factors would reflect the inherently higher emissions rate of coal-fired units, and consequently the greater burden on coal units to control emissions.

Today's Rule

As noted earlier in the case of SO 2, EPA recognizes that the choice of method in setting State budgets, with a given regionwide total annual budget, makes little difference in terms of the levels of resulting regionwide annual SO 2 and NO X emissions reductions. If States choose to control EGUs and participate in the cap and trade program, allowances could be freely traded, encouraging least-cost compliance over the entire region. In such a case, the least-cost outcome would not depend on the relative levels of individual State budgets.

A number of commenters have stated, without supporting analysis or evidence, that budgets based on heat input, (and particularly those that would use different fuel factors) do not encourage efficiency. Economic theory indicates that neither a heat input, nor an output-based approach, if allocated once and based on a historical baseline, would provide any incentives for more or less efficient generation (changes in future behavior would have no impact on allocations). The cap and trade system itself, regardless of how the allowances are distributed, provides the primary incentive for more efficient, cleaner generation of electricity.

The EPA is finalizing an approach of calculating State budgets through a fuel-adjusted heat-input basis. State budgets would be determined by multiplying historic heat input data (summed by fuel) by different adjustment factors for the different fuels. These factors reflect for each fuel (coal, gas and oil), the 1999-2002 average emissions by State, summed for the CAIR region, divided by average heat input by fuel by State, summed for the CAIR region. The resulting adjustment factors from this calculation are 1.0 for coal, 0.4 for gas and 0.6 for oil. The factors would reflect the inherently higher emissions rate of coal-fired plants, and consequently the greater burden on coal plants to control emissions.

Such an approach provides States with allowances more in proportion with their historical emissions. It provides for a more equitable budget distribution by recognizing that different States are facing the reduction requirements with different starting stocks of generation, with different starting emission profiles. [88] The fuel burned is a key factor in differentiating the generation.

However, this approach is not equivalent to an approach based strictly on historical emissions (which would give fewer allowances to States which have already cleaned up their coal plants). Under the approach we are finalizing today, heat input from all coal, whether clean or uncontrolled, would be counted equally in determining State budgets. Likewise, all heat input from gas, whether clean or uncontrolled, from a steam-gas unit or from a combined-cycle plant, would be counted equally in determining State budgets.

It is not expected that this decision would disadvantage States with significant gas-fired generation. One reason is that the calculation of the adjusted heat input for natural gas generation generally includes significant historic heat input and emissions from older, less efficient and dirtier steam gas units. These units' capacity factors are declining and are expected to decline further over time as new, cleaner and more efficient combined-cycle gas units increase their generation.

It is important to note that the methodology by which the NO X State budgets are determined need not be used by individual States in determining allocations to specific sources. As discussed in section VIII of this document (Model Trading Rule), EPA is offering States the flexibility to allocate allowances from their budgets as they see fit.

Finally, EPA discussed in the January 2004 proposal, a methodology used in the NO X SIP Call (67 FR 21868) that applied State-specific growth rates for heat input in setting State budgets. [89] The EPA, in the SNPR, noted that it is not proposing to use this method for the CAIR because we believe that other methods are reasonable, and that methods involving State-specific growth rates present certain challenges due to the inherent difficulties in predicting State-specific growth in heat input over a lengthy period, especially for jurisdictions that are only a part of a larger regional electric power dispatch region. Several commenters stated their support for incorporating growth, believing that not taking growth into account would penalize States with higher growth. However, a significant number of commenters stated their opposition to using growth in setting State budgets, noting the problems that arose in the NO X SIP Call. The EPA believes that setting budgets using a heat input approach, without a growth adjustment, is fair, would be simpler and would involve less risk of resulting litigation.

c. Final Annual State NO X Emission Budgets

The final annual State NO X emission budgets following this method are included in Table V-2 below. Details of the numbers and methodology used to calculate these budgets are included in the “Regional and State SO 2 and NO X Emissions Budgets” Technical Support Document.

Table V-2.—Final Annual Electric Generating Units NO X Budgets Back to Top
State State NO X budget 2009* State NO X budget 2015**
[Tons]
*Annual budget for NO X tons covered by allowances for 2009-2014.
**Annual budget for NO X tons covered by allowances for 2015 and thereafter.
Alabama 69,020 57,517
District of Columbia 144 120
Florida 99,445 82,871
Georgia 66,321 55,268
Illinois 76,230 63,525
Indiana 108,935 90,779
Iowa 32,692 27,243
Kentucky 83,205 69,337
Louisiana 35,512 29,593
Maryland 27,724 23,104
Michigan 65,304 54,420
Minnesota 31,443 26,203
Mississippi 17,807 14,839
Missouri 59,871 49,892
New York 45,617 38,014
North Carolina 62,183 51,819
Ohio 108,667 90,556
Pennsylvania 99,049 82,541
South Carolina 32,662 27,219
Tennessee 50,973 42,478
Texas 181,014 150,845
Virginia 36,074 30,062
West Virginia 74,220 61,850
Wisconsin 40,759 33,966
Total 1,504,871 1,254,061

d. Use of Annual NO X Budgets

These proposed State budgets would serve as effective binding caps on State emissions, if States chose to control only EGUs, but did not want to participate in the trading program. For States controlling both EGUs and non-EGUs (or controlling only non-EGUs), these budgets would be compared to a baseline level of emissions to calculate the emissions reductions requirements for non-EGUs and the required caps for EGUs. This process is described in more detail in the section VII discussion on SIP approvability.

e. NO X Compliance Supplement Pool

As is discussed in section I, EPA is establishing a NO X compliance supplement pool of 198,494 tons, which would result in a total compliance supplement pool of approximately 200,000 tons of NO X when combined with EPA's proposed rulemaking to include Delaware and New Jersey. The EPA is apportioning the compliance supplement pool to States based on the assumption that a State's need for allowances from the pool is proportional to the magnitude of the State's required emissions reductions (as calculated using the State's base case emissions and annual NO X budget). The EPA is apportioning the 200,000 tons of NO X on a pro-rata basis, based on each State's share of the total emissions reductions requirement for the region in 2009. This is consistent with the methodology used in the NO X SIP Call. Table V-3 presents each State's compliance supplement pool.

Table V-3.—State NO X Compliance Supplement Pools Back to Top
State Base case 2009 emissions 2009 State annual NO X budget Reduction requirement Compliance supplement pool*
[Tons]
*Rounding to the nearest whole allowance results in a total compliance supplement pool of 199,997 tons.
Alabama 132,019 69,020 62,999 10,166
District of Columbia 0 144 0 0
Florida 151,094 99,445 51,649 8,335
Georgia 143,140 66,321 76,819 12,397
Illinois 146,248 76,230 70,018 11,299
Indiana 233,833 108,935 124,898 20,155
Iowa 75,934 32,692 43,242 6,978
Kentucky 175,754 83,205 92,549 14,935
Louisiana 49,460 35,512 13,948 2,251
Maryland 56,662 27,724 28,938 4,670
Michigan 117,031 65,304 51,727 8,347
Minnesota 71,896 31,443 40,453 6,528
Mississippi 36,807 17,807 19,000 3,066
Missouri 115,916 59,871 56,045 9,044
New York 45,145 45,617 0 0
North Carolina 59,751 62,183 0 0
Ohio 263,814 108,667 155,147 25,037
Pennsylvania 198,255 99,049 99,206 16,009
South Carolina 48,776 32,662 16,114 2,600
Tennessee 106,398 50,973 55,425 8,944
Texas 185,798 181,014 4,784 772
Virginia 67,890 36,074 31,816 5,134
West Virginia 179,125 74,220 104,905 16,929
Wisconsin 71,112 40,759 30,353 4,898
CAIR region subtotal 198,494
Delaware 9,389 4,166 5,223 843
New Jersey 16,760 12,670 4,090 660
Total 199,997

B. What Is the Approach for Setting State-by-State Emissions Reductions Requirements and EGU Budgets for States With NO X Ozone Season Reduction Requirements?

1. States Subject to Ozone-Season Requirements

In the NPR, EPA proposed that Connecticut contributes significantly to ozone nonattainment in another State, but not to fine particle nonattainment. As a result of subsequent air quality modeling, EPA has also found that Massachusetts, New Jersey, Delaware and Arkansas contribute significantly to ozone nonattainment in another State, but not to fine particle nonattainment. In this final rule, EPA is establishing a regionwide ozone-season budget for all States that contribute significantly to ozone nonattainment in another State, regardless of their contribution to fine particle nonattainment. The following 25 States, plus the District of Columbia, are found to contribute significantly to ozone nonattainment: Alabama, Arkansas, Connecticut, Delaware, Florida, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Massachusetts, Michigan, Mississippi, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia, and Wisconsin.

These States are subject to an ozone season NO X cap, which covers the 5 months of May through September. The EPA is calculating the ozone season cap level for the 25 States plus the District of Columbia region by multiplying the region's ozone season heat input by 0.15 lb/mmBtu for 2009 and 0.125 lb/mmBtu for 2015. Heat input for the region was estimated by looking at reported ozone season Acid Rain heat inputs for each State for the years 1999 through 2002, and selecting the single year highest heat input for each State as a whole.

As is the case for the annual NO X State Budgets, EPA is finalizing an approach of calculating ozone season NO X State budgets through a fuel-adjusted heat input basis. State budgets would be determined by multiplying State-level average historic ozone-season heat input data (summed by fuel) by different adjustment factors for the different fuels (1.0 for coal, 0.4 for gas, and 0.6 for oil). The total ozone season State budgets are then determined by calculating each State's share of total fuel-adjusted heat input, and multiplying this share by the regionwide budget.

The budgets for these States in 2009 and 2015 are included in Table V-4 below.

Table V-4.—Final Seasonal Electricity Generating Unit NO X Budgets Back to Top
State State NO X budget 2009 * State NO X budget 2015 **
[Tons]
* Seasonal budget for NO X tons covered by allowances for 2009-2014. For States that have lower EGU budgets under the NO X SIP Call than their 2009 CAIR budget, table V-4 includes their SIP Call budget. For Connecticut, the NO X SIP Call budget is also used for 2015 and beyond.
** Seasonal budget for NO X tons covered by allowances for 2015 and thereafter.
Alabama 32,182 26,818
Arkansas 11,515 9,596
Connecticut 2,559 2,559
Delaware 2,226 1,855
District of Columbia 112 94
Florida 47,912 39,926
Illinois 30,701 28,981
Indiana 45,952 39,273
Iowa 14,263 11,886
Kentucky 36,045 30,587
Louisiana 17,085 14,238
Maryland 12,834 10,695
Massachusetts 7,551 6,293
Michigan 28,971 24,142
Mississippi 8,714 7,262
Missouri 26,678 22,231
New Jersey 6,654 5,545
New York 20,632 17,193
North Carolina 28,392 23,660
Ohio 45,664 39,945
Pennsylvania 42,171 35,143
South Carolina 15,249 12,707
Tennessee 22,842 19,035
Virginia 15,994 13,328
West Virginia 26,859 26,525
Wisconsin 17,987 14,989
Total 567,744 484,506

VI. Air Quality Modeling Approach and Results Back to Top

Overview

In this section we summarize the air quality modeling approach used for the proposed rule, we address major comments on the fundamental aspects of EPA's proposed approach, and we describe the updated and improved approach, based on those comments, that we are finalizing today. This section also contains the results of EPA's final air quality modeling, including: (1) Identifying the future baseline PM 2.5 and 8-hour ozone nonattainment counties in the East; (2) quantifying the contribution from emissions in upwind States to nonattainment in these counties; (3) quantifying the air quality impacts of the CAIR reductions on PM 2.5 and 8-hour ozone; and (4) describing the impacts on visibility in Class I areas of implementing CAIR compared to implementing the regional haze requirement for best available retrofit technology (BART).

We present the air quality models, model configuration, and evaluation; and then the emissions inventories and meteorological data used as inputs to the air quality models. Next, we provide the updated interstate contributions for PM 2.5 and 8-hour ozone and those States that make a significant contribution to downwind nonattainment, before considering cost. Finally, we present the estimated impacts of the CAIR emissions reductions on air quality and visibility. As described below, our air quality modeling for today's rule utilizes the Community Multiscale Air Quality (CMAQ) model in conjunction with 2001 meteorological data for simulating PM 2.5 concentrations and associated visibility effects and the Comprehensive Air Quality Model with Extensions (CAMx) with meteorological data for three episodes in 1995 for simulating 8-hour ozone concentrations. Our approach to modeling both PM 2.5 and 8-hour ozone involves applying these tools (i.e., CMAQ for PM 2.5 and CAMx for 8-hour ozone) using updated emissions inventory data for 2001, 2010, and 2015 to project future baseline concentrations, interstate transport, and the impacts of CAIR on projected nonattainment of PM 2.5 and 8-hour ozone. We provide additional information on the development of our updated CAIR air quality modeling platform, the modeling analysis techniques, model evaluation, and results for PM 2.5 and 8-hour ozone modeling in the CAIR Notice of Final Rulemaking Emissions Inventory Technical Support Document (NFR EITSD) and the Air Quality Modeling Technical Support Document (NFR AQMTSD).

A. What Air Quality Modeling Platform Did EPA Use?

1. Air Quality Models

a. The PM 2.5 Air Quality Model and Evaluation

Overview

In the NPR, we used the Regional Model for Simulating Aerosols and Deposition (REMSAD) as the tool for simulating base year and future concentrations of PM 2.5. Like most photochemical grid models, the predictions of REMSAD are based on a set of atmospheric specie mass continuity equations. This set of equations represents a mass balance in which all of the relevant emissions, transport, diffusion, chemical reactions, and removal processes are expressed in mathematical terms. The modeling domain used for this analysis covers the entire continental United States and adjacent portions of Canada and Mexico.

The EPA applied REMSAD for an annual simulation using meteorology and emissions for 1996. We used the results of this 1996 Base Year model run to evaluate how well the modeling system (i.e., the air quality model and input data sets) replicated measured data over the time period and domain simulated. We performed a model evaluation for PM 2.5 and speciated components (e.g., sulfate, nitrate, elemental carbon, organic carbon, etc.) as well as nitrate, sulfate and ammonium wet deposition, and visibility. The evaluation used available 1996 ambient measurements paired with REMSAD predictions corresponding to the location and time periods of the measured data. We quantified model performance using various statistical and graphical techniques. Additional information on the model evaluation procedures and results are included in the Notice of Proposed Rulemaking Air Quality Modeling Technical Support Document (NPR AQMTSD).

The EPA received numerous comments on various elements of the proposed PM 2.5 air quality modeling approach. The major comments are responded to below. Other comments are addressed the Response to Comment (RTC) document. Regarding REMSAD, commenters argued that: (1) The REMSAD model is an inappropriate tool for modeling PM 2.5; (2) the scientific formulation of the model is simplistic and outdated and that other models with better science are available and should be used; and (3) results from REMSAD are directionally correct but better tools should be used as the basis for the final determinations on transport and projected nonattainment.

We agree that models with more refined science are available for PM 2.5 modeling and we have selected one of these models, the CMAQ as the tool for PM 2.5 modeling for the final CAIR. The CMAQ model is a publicly available, peer-reviewed, state-of-the-science model with a number of science attributes that are critical for accurately simulating the oxidant precursors and non-linear organic and inorganic chemical relationships associated with the formation of sulfate, nitrate, and organic aerosols. Several of the important science aspects of CMAQ that are superior to REMSAD include: (1) Updated gaseous/heterogeneous chemistry that provides the basis for the formation of nitrates and includes a current inorganic nitrate partitioning module; (2) in-cloud sulfate chemistry, which accounts for the non-linear sensitivity of sulfate formation to varying pH; (3) a state-of-the-science secondary organic aerosol module that includes a more comprehensive gas-particle partitioning algorithm from both anthropogenic and biogenic secondary organic aerosol; and (4) the full CB-IV chemistry mechanism, which provides a complete simulation of aerosol precursor oxidants.

However, even though REMSAD does not have all the scientific refinements of CMAQ, we believe that REMSAD treats the key physical and chemical processes associated with secondary aerosol formation and transport. Thus, we believe that the conclusions based on the proposal modeling using REMSAD are valid and therefore support today's findings based only on CMAQ that: (1) There will be widespread PM 2.5 nonattainment in the eastern U.S. in 2010 and 2015 absent the reductions from CAIR; (2) upwind States in the eastern part of the United States contribute to the PM 2.5 nonattainment problems in other downwind States; (3) States with high emissions tend to contribute more than States with low emissions; (4) States close to nonattainment areas tend to contribute more than other States farther upwind; and (5) the CAIR controls will produce major benefits in terms of bringing areas into or closer to attainment.

Comments and Responses

(i) REMSAD Science and Evaluation

Comment: Some commenters stated that REMSAD is an inappropriate model for use in simulating PM 2.5. Other commenters said, more specifically, that the chemical mechanism in REMSAD (i.e., micro CB-IV) is simplified and not validated, and that the model has not been scientifically peer-reviewed.

Response: The EPA disagrees with comments claiming that REMSAD is an inappropriate tool for modeling PM 2.5. The EPA believes that REMSAD is appropriate for regional and national modeling applications because the model does include the key physical and chemical processes associated with secondary aerosol formation and transport. [90]

Specifically, REMSAD simulates both gas phase and aerosol chemistry. The gas phase chemistry uses a reduced-form version of Carbon Bond chemical mechanism (micro-CB-IV). Formation of inorganic secondary particulate species, such as sulfate and nitrate, are simulated through chemical reactions within the model. Aerosol sulfate is formed in both the gas phase and the aqueous phase. The REMSAD model also accounts for the production of secondary organic aerosols through chemistry processes involving volatile organic compounds (VOC) and directly emitted organic particles. Emissions of non-reactive particles (e.g., elemental carbon) are treated as inert species which are advected and deposited during the simulation.

With regard to comments on the micro CB-IV chemical mechanism, although this mechanism treats fewer organic carbon species compared to the full CB-IV, the inorganic portion of the reduced mechanism is identical to the full chemical mechanism. The intent of the CB-IV mechanism is to: (a) Provide a faithful representation of the linkages between emissions of ozone precursor species and secondary aerosol precursor species; (b) treat the oxidizing capacity of the troposphere, represented primarily by the concentrations of radicals and hydrogen peroxide; and (c) simulate the rate of oxidation of the nitrogen oxide (NO X) and sulfur dioxide (SO 2), which are precursors to secondary aerosols. The EPA agrees that micro CB-IV is simplified compared to the full CB-IV mechanism. However, performance testing of micro CB-IV indicates that this simplified mechanism is similar to the full CB-IV chemical mechanism in simulating ozone formation and approximates other species reasonably well (e.g., hydroxyl radical, hydroperoxy radical, the operator radical, hydrogen peroxide, nitric acid, and peroxyacetyl nitrate). [91]

The REMSAD model was subjected to a scientific peer-review (Seigneur et al., 1999) and EPA has incorporated the major science improvements that were recommended by the peer-review panel. These improvements were included in the version of REMSAD used for the NPR modeling. Specifically, the following updates have been implemented into REMSAD Version 7.06, which was used for the proposed CAIR control strategy simulations: (1) The nighttime chemistry treatment was updated to improve the treatment of the gas phase species NO 3 and N 2 O 5; (2) the effects of temperature and pressure dependence on chemical rates were added; (3) the MARS-A aerosol partitioning module was added for calculating particle and gas phase fractions of nitrate; (4) aqueous phase formation of sulfate was updated by including reactions for oxidation of SO 2 by ozone and oxygen, (5) peroxynitric acid (PNA) chemistry was added; and (6) a module for calculating biogenic and anthropogenic secondary organic aerosols was developed and integrated into REMSAD. We believe that these changes adequately respond to the peer review comments and have bolstered the scientific credibility of this model.

(ii) Use of CMAQ Instead of REMSAD for PM 2.5 Modeling

Comment: Some commenters claimed that REMSAD is outdated and that other models with more sophisticated science are available. Commenters said that EPA should utilize the best available science through use of the most comprehensive photochemical model for simulating aerosols. Commenters specifically stated that EPA should use more recently developed models such as the CMAQ model or the aerosol version of the Comprehensive Air Quality Model with Extensions (CAM X-PM).

Response: The EPA agrees that photochemical models are now available that are more scientifically sophisticated than REMSAD. In this regard, and in response to commenters' recommendations on specific models, EPA has selected CMAQ as the modeling tool for the final CAIR modeling analysis. As stated above, the CMAQ model is a publicaly available, peer-reviewed, state-of-the-science model with a number of science attributes that are critical for accurately simulating the oxidant precursors and non-linear organic and inorganic chemical relationships associated with the formation of sulfate, nitrate, and organic aerosols. As listed above, the important science aspects of CMAQ that are superior to REMSAD include: (1) Updated gaseous/heterogeneous chemistry that provides the basis for the formation of nitrates and includes a current inorganic nitrate partitioning module; (2) in-cloud sulfate chemistry, which accounts for the non-linear sensitivity of sulfate formation to varying pH; (3) a state-of-the-science secondary organic aerosol module that includes a more comprehensive gas-particle partitioning algorithm from both anthropogenic and biogenic secondary organic aerosol; and (4) the full CB-IV chemistry mechanism, which provides a complete simulation of aerosol precursor oxidants.

(iii) Model Evaluation

Comment: A number of commenters claimed that EPA's air quality model evaluation for 1996 was deficient because it lacked sufficient ambient measurements, especially in urban areas, to judge model performance. Commenters said that EPA should: (1) Update the evaluation to a more recent time period in order to take advantage of greatly expanded ambient PM 2.5 species measurements, especially in urban areas; and (2) calculate model performance statistics over monthly and/or seasonal time periods using daily/weekly observed/model-predicted data pairs.

Some commenters said that the 1996 data were so limited that it is not possible to determine whether REMSAD could be used with confidence to assess the effects of emissions changes. Still, other commenters said that the performance of REMSAD for the 1996 modeling platform was poor.

Commenters acknowledged that there are no universally accepted or EPA-recommended quantitative criteria for judging the acceptability of PM 2.5 model performance. In the absence of such model performance acceptance criteria, some commenters said that performance should be judged by comparing EPA's model performance results to the range of results obtained by other groups in the air quality modeling community who conducted other recent regional PM 2.5 model applications. A few commenters also identified specific model performance ranges and criteria that they said should be achievable for sulfate and PM 2.5, given the current state-of-science for aerosol modeling and measurement uncertainty. The specific values cited by these commenters are ±30 percent to ±50 percent for fractional bias, 50 percent to 75 percent for fractional error, and 50 percent for normalized error.

Response: The EPA agrees that the limited amount of ambient PM 2.5 species data available in 1996 affected our ability to evaluate model performance, especially in urban areas, and there were deficiencies in the performance of REMSAD using the 1996 model inputs. Also, EPA agrees that a model evaluation should be performed for a more recent time period in order to address these concerns. Thus, we conclude that the 1996 modeling platform which includes 1996 emissions, 1996 meteorology, and 1996 ambient data should be updated and improved, as recommended by commenters.

The EPA has developed a new modeling platform which includes emissions, meteorological data, and other model inputs for 2001. This platform was used to confirm the ability of our modeling system to replicate ambient PM 2.5 and component species in both urban and rural areas and, thus, establish the credibility of this platform for PM 2.5 modeling as part of CAIR. [92] In 2001, there was an extensive set of ambient PM 2.5 measurements including 133 urban Speciation Trends Network (STN) monitoring sites across the nation, with 105 of these in the East. This network did not exist in 1996. Also, the number of mainly suburban and rural monitoring sites in the Clean Air Status and Trends Network (CASTNET) and Interagency Monitoring of Protected Visual Environments (IMPROVE) network has increased to over 200 in 2001, compared to approximately 120 operating in 1996.

The EPA evaluated CMAQ for the 2001 modeling platform using the extensive set of 2001 monitoring data for PM 2.5 species. The evaluation included a statistical analysis in which the model predictions and measurements were paired in space and in time (i.e., daily or weekly to be consistent with the sampling protocol of the monitoring network). Model performance statistics were calculated for each network with separate statistics for sites in the West and the East. [93] In response to comments that performance statistics should be calculated over monthly and/or seasonal time periods, we elected to use seasonal time periods in order to be consistent with our use of quarterly average PM 2.5 species as part of the procedure for projecting future concentrations, as described below in section VI.B.1. In addition, the sampling frequency at the CASTNET, IMPROVE, and STN sites may not provide sufficient samples in a 1-month period to provide a robust calculation of model performance statistics. Details of EPA's model evaluation for CMAQ using the 2001 modeling platform are in the report “Updated CMAQ Model Performance Evaluation for 2001” which can be found in the docket for today's rule.

The EPA agrees that there are no universally accepted performance criteria for PM 2.5 modeling and that performance should be judged by comparison to the performance found by other groups in the air quality modeling community. In this respect, we have compared our CMAQ 2001 model performance results to the range of performance found in other recent regional PM 2.5 model applications by other groups. [94] Details of this comparison can be found in the CMAQ evaluation report. Below is a summary of performance results from other, non-EPA modeling studies, for summer sulfate and winter nitrate. It CAIR. Overall, the general range of fractional bias (FB) and fractional error (FE) statistics for the better performing model applications are as follows:

—Summer sulfate is in the range of −10 percent to +30 percent for FB and 35 percent to 50 percent for FE; and

—Winter nitrate is in the range of +50 percent to +70 percent for FB and 85 percent to 105 percent for FE.

The corresponding performance statistics for EPA's 2001 CMAQ application as well as the 1996 REMSAD application used for the proposal modeling are provided in Table VI-1.

Table VI-1.—Selected Performance Evaluation Statistics From the CMAQ 2001 Simulation and the REMSAD 1996 Simulation Back to Top
Eastern U.S. CMAQ 2001 REMSAD 1996
FB(%) FE(%) FB(%) FE(%)
Sulfate (Summer):        
STN 14 44
Improve 10 42 −20 51
CASTNet 3 22 −21 59
Nitrate (Winter)        
STN 15 73
Improve 21 92 67 103

The results indicate that the performance for CMAQ in 2001 is within the range or better than that found by other groups in recent applications. The performance also meets the benchmark goals suggested by several commenters. In addition, the CMAQ performance is considerably improved over that of the REMSAD 1996 performance for summer sulfate and winter nitrate, which were near the bounds or outside the range of other recent applications.

The CMAQ model performance results give us confidence that our applications of CMAQ using the new modeling platform provide a scientifically credible approach for assessing PM 2.5 concentrations for the purposes of CAIR.

b. Ozone Air Quality Modeling Platform and Model Evaluation

Overview

The EPA used the CAM X, version 3.10 in the NPR to assess 8-hour ozone concentrations and the impacts of ozone and ozone precursor transport on elevated levels of ozone across the eastern U.S. The CAM X is a publicly available Eulerian model that accounts for the processes that are involved in the production, transport, and destruction of ozone over a specified three-dimensional domain and time period. The CAM X model was run with 1995/96 base year emissions to evaluate the performance of the modeling platform to replicate observed concentrations during the three 1995 episodes. This evaluation was comprised principally of statistical assessments of hourly, 1-hour daily maximum, and 8-hour daily maximum ozone predictions. As described in the NPR AQMTSD, model performance of CAM X for ozone was judged against the results from previous regional ozone model applications. This analysis indicates that model performance was comparable to or better than that found in previous applications and is, therefore, acceptable for the purposes of CAIR ozone modeling.

The EPA did not receive comments on the CAM X model or the model performance for ozone. The EPA did receive comments on the choice of episodes for ozone modeling, the meteorological data for these episodes, the spatial resolution of our modeling, and consistency between ozone and PM 2.5 modeling in terms of methods for projecting future air quality concentrations. As described below and in the RTC document and NFR AQMTSD, we continue to believe that: (1) The three 1995 episodes are representative episodes for regional modeling of 8-hour ozone; and (2) the meteorological data for these episodes and spatial resolution are adequate for use in our modeling for CAIR. Thus, the ozone air quality assessments in today's rule rely on CAM X modeling of meteorological data for the three 1995 episodes for the domain and spatial resolution used for the NPR. As discussed below, we ran CAM X for the updated 2001 emissions inventory and the updated 2010 and 2015 base case inventories as part of the process to project 8-hour ozone for these future year scenarios. We revised our method of projecting future ozone concentrations to be consistent with the method we are using for PM 2.5.

c. Model Grid Cell Configuration

As described in the NPR AQMTSD, the PM 2.5 modeling for the proposal was performed for a domain (i.e., area) covering the 48 States and adjacent portions of Canada and Mexico. Within this domain, the model predictions were calculated for a grid network with a spatial resolution of approximately 36 km. Our 8-hour ozone modeling for proposal was performed using a nested grid network. The outer portion of this grid has a spatial resolution of approximately 36 km. The inner “nested” area, which covers a large portion of the eastern U.S., has a resolution of approximately 12 km.

Comment: Some commenters said that the 36 km grid cell size used by EPA in modeling PM 2.5 and the 36 km/12 km grid resolution used for ozone modeling are too coarse and are inconsistent with EPA's draft modeling guidance.

Response: We disagree with these comments and continue to believe that the grid dimensions for our PM 2.5 modeling and our 8-hour ozone modeling are not too coarse nor are they inconsistent with our draft guidance documents for PM 2.5 modeling [95] and ozone modeling. [96] The draft guidance for PM 2.5 modeling states that 36 km resolution is acceptable for regional scale applications in portions of the domain outside of nonattainment areas. For portions of the domain which cover nonattainment areas, 12 km resolution or less is recommended by the guidance. However, as stated in the guidance document, these recommendations were based on guidance for 8-hour ozone modeling because there was a lack of PM 2.5 modeling at different grid resolutions at the time the guidance was drafted. In addition, the PM 2.5 guidance states that exceptions to these recommendations can be made on a case-by-case basis.

For several reasons, we believe that 36 km resolution is sufficient for PM 2.5 modeling for the purposes of CAIR. First, recent analyses that compare 36 km to 12 km modeling of PM 2.5 [97] indicate that spatial mean concentrations of gas phase and aerosol species at 36 km and 12 km are quite similar. A comparison of model predictions versus observations indicates that the model performance is similar at 12 km and 36 km in both rural and urban areas. Thus, using 12 km resolution does not necessarily provide any additional confidence in the results. Second, ambient measurements of sulfate and to a significant extent nitrate, which are the pollutants of most importance for CAIR, do not exhibit large spatial differences between rural and urban areas, as described elsewhere in today's rule. This implies that it is not necessary to use fine resolution modeling in order to properly capture the regional concentration patterns of these pollutants.

Our draft 8-hour ozone modeling guidance recommends using 36 km resolution for regional modeling with nested grid cells not exceeding 12 km over urban portions of the modeling domain. The guidance states that 4 to 5 km resolution for urban areas is preferred, if feasible. In addition, if 12 km modeling is used then plume-in-grid treatment for large point sources of NO X should be considered.

Our modeling for CAIR is consistent with this guidance in that we use 36 km resolution for the outer portions of the region; 12 km resolution covering nearly all urban areas in the domain; and a plume-in-grid algorithm for major NO X point sources in the region. In addition, analyses that compare model 12 km resolution to 4 km resolution for portions of our 1995 episodes indicate that the spatial fields predicted at both 12 km and 4 km have many common features in terms of the areas of high and low ozone. [98] In a comparison of model predictions to observation, the 12 km modeling was found to be somewhat more accurate than the finer 4 km modeling.

2. Emissions Inventory Data

For the proposed rule, emissions inventories were created for the 48 contiguous States and the District of Columbia. These inventories were estimated for a 2001 base year to reflect current emissions and for 2010 and 2015 future baseline scenarios. The inventories were prepared for electric generating units (EGUs), industrial and commercial sources (non-EGUs), stationary area sources, on-road vehicles, and non-road engines. The inventories contained both annual and typical summer season day emissions for the following pollutants: oxides of nitrogen (NO X); volatile organic compounds (VOC); carbon monoxide (CO); sulfur dioxide (SO 2); direct particulate matter with an aerodynamic diameter less than 10 micrometers (PM 10) and less than 2.5 micrometers (PM 2.5); and ammonia (NH 3). A summary of the development of these inventories is provided below. Additional information on the emissions inventory used for proposal can be found in the NPR AQMTSD.

Because the complete 2001 National Emission Inventory (NEI) and future-year projections consistent with that NEI were not available in a form suitable for air quality modeling when needed for the proposal, we developed a reasonably representative “proxy” inventory for 2001. For the EGU, mobile, and non-road emissions sectors, 1996-to-2001 adjustment ratios were created by dividing State-level total emissions for each pollutant for 2001 by the corresponding consistent 1996 emissions. These adjustment ratios were then multiplied by the REMSAD-ready 1996 emissions for these two sectors to produce REMSAD-ready files for the 2001 proxy. For non-EGUs and stationary area sources, linear interpolations were performed between the REMSAD-ready 1996 emissions and the REMSAD-ready 2010 base case emissions to produce 2001 proxy emissions for these two sectors. Details on the creation of the 2001 proxy inventory used for proposal are provided in the NPR AQMTSD.

The NPR future 2010 and 2015 base case emissions reflect projected economic growth and control programs that are to be implemented by 2010 and 2015, respectively. Control programs included in these future base cases include those State, local, and Federal measures already promulgated and other significant measures expected to be promulgated before the final rule is implemented. Future year 2010 and 2015 base case EGU emissions were obtained from versions 2.1 and 2.1.6 of the Integrated Planning Model (IPM).

Comment: Several commenters stated that the emission inventory used for the “proxy” 2001 base year was not sufficient for the rulemaking, primarily because it was developed from a 1996 modeling inventory by applying various adjustment factors. Commenters suggested that: (1) More up-to-date inventories were now available and should be used; (2) the most recent Continuous Emissions Monitoring (CEM) data or throughput information should be used to derive a 2001 EGU inventory; and (3) EPA should use the 2001 MOBILE6 and NONROAD2002 models for estimating on-road mobile and non-road engine emissions, respectively.

Response: The EPA believes that the base year for modeling should be as recent as possible, given the availability of nationally complete emissions estimates and ambient monitoring data. For the analyses of the final rule, EPA has used a base year inventory developed specifically for 2001. The base year inventory for the electric utility sector now uses measured CEM emissions data for 2001. The non-EGU point source and stationary-area source sectors are based on the final 1999 NEI data submittals from State, local, and Tribal air agencies. This inventory is the latest available quality-assured and reviewed national emission data set for these sectors. The 1999 data for non-EGU point and stationary-area sources were projected to represent a 2001 inventory using State/county-specific and sector-specific growth rates. The on-road mobile inventory uses MOBILE version 6.2 and the non-road engines inventory uses the NONROAD2004 model, both with updated input parameters to calculate emissions for 2001. More detailed information on the development of the emissions inventories can be found in the NFR EITSD.

Comment: Commenters stated that EPA failed to develop an accurate and comprehensive ammonia emission inventory from soil, fertilizer, and animal husbandry sources.

Response: The 2001 inventory used for the analyses for the final rule includes a new national county-level ammonia inventory developed by EPA using the latest emission rates selected based on a comprehensive literature review, and activity levels as provided by the U.S. Census of Agriculture for animal husbandry. The 2001 inventory from fertilizer application sources was compiled from State and local submissions to EPA for 1999, augmented as necessary with EPA estimates, and grown to 2001 using State/county-specific and category-specific growth rates. With regard to background soil emissions of NH 3, EPA believes that the current state of understanding of background soil ammonia releases and sinks is insufficient to warrant including these emission sources in modeling inventories at this time.

Comment: Two commenters indicated that EPA should revise 2010 and 2015 base case emissions by improving the methods for estimating economic growth and not rely on the Bureau of Economic Analysis (BEA) data used for proposal.

Response: In response to these comments, EPA has refined its economic growth projections. In addition to updated versions of the MOBILE6, NONROAD, and IPM models, EPA developed new economic growth rates for stationary, area, and non-EGU point sources. For these two sectors, the final approach uses a combination of: (1) Regional or national fuel-use forecast data from the U.S. Department of Energy for source types that map to fuel use sectors (e.g., commercial coal, industrial natural gas); (2) State-specific growth rates from the Regional Economic Model, Inc. (REMI) Policy Insight® model, version 5.5; and (3) forecasts by specific industry organizations and Federal agencies. For more detail on the growth methodologies, please refer to the NFR EITSD.

3. Meteorological Data

In order to solve for the change in pollutant concentrations over time and space, the air quality model requires certain meteorological inputs that, in part, govern the formation, transport, and destruction of pollutant material. Two separate sets of meteorological inputs were used in the air quality modeling completed as part of the NPR. The meteorological input files for the proposal PM 2.5 modeling were developed from a Fifth-Generation NCAR/Pennsylvania State Mesoscale Model (MM5) model simulation for the entire year of 1996. The gridded meteorological data for the three 1995 ozone episodes were developed using the Regional Atmospheric Modeling System (RAMS). Both of these models are publicly-available, widely-used, prognostic meteorological models that solve the full set of physical and thermodynamic equations which govern atmospheric motions. Further, each of these specific meteorological data sets has been utilized in past EPA rulemaking modeling analyses (e.g., the Nonroad Land-based Diesel Engines Standards).

Comment: Several commenters claimed that the 1996 meteorological modeling data used to support the fine particulate modeling were outdated and non-representative. We also received recommendations from commenters on benchmarks to be used as goals for judging the adequacy of meteorological modeling.

Response: The EPA draft PM 2.5 modeling guidance which provides general recommendations on meteorological periods to model for PM 2.5 purposes lists three primary general criteria for consideration: (a) Variety of meteorological conditions; (b) existence of an extensive air quality/meteorological data bases; and (c) sufficient number of days. The approach recommended in the guidance for modeling annual PM 2.5 is to use a single, representative year. Based on the comments received and the criteria outlined in the guidance, EPA developed meteorological data for the entire calendar year of 2001. This year was chosen for the PM 2.5 modeling platform based on several factors, specifically: (a) It corresponds to the most recent set of emissions data; (b) there are considerable ambient PM 2.5 species data for use in model evaluation (as described in section VI.A.1., above); and (c) Federal Reference Method (FRM) PM 2.5 data for this year are included in the calculation of the most recent PM 2.5 design values used for designating PM 2.5 nonattainment areas. In view of these factors, EPA believes that 2001 meteorology are representative for PM 2.5 modeling for the purposes of this rule.

The new 2001 meteorological data used for PM 2.5 modeling were derived from an updated version of the MM5 model used for the 1996 meteorology used for proposal. The version of MM5 used for the 2001 simulation contains more sophisticated physics options with respect to features like cloud microphysics and land-surface interactions, and more refined vertical resolution of the atmosphere compared to the version used for modeling 1996 meteorology. While there are currently no universally accepted criteria for judging the adequacy of meteorological model performance, EPA compared the 2001 MM5 model performance against the benchmark goals [99] recommended by some commenters. The benchmark goals suggest that temperature bias should be within the range of approximately ± 0.5 degrees C and errors less than or equal to 2.0 degrees C are typical.

In general, the model performance statistics for our 2001 meteorological modeling are in line with the above benchmark goals. Specfically, the mean temperature bias of our 2001 meteorological modeling was approximately 0.6 degrees C and the mean error was approximately 2.0 degrees C. The evaluation of the 2001 MM5 for humidity (water vapor mixing ratio) shows biases of less than 0.5 g/kg and errors of approximately 1 g/kg, which compare favorably to the goals of ± 1 g/kg for bias and 2 g/kg or less error. Model performance for winds in our 2001 simulation was also improved compared to what has historically been found in MM5 modeling studies. The index of agreement for surface winds in the 2001 case equaled 0.86, which is far better than the benchmark goal of 0.60. The precipitation evaluation results show that the model generally replicates the observed data, but is overestimating precipitation in the summer months. More information about the model performance evaluation and the MM5 configuration is provided in the NFR AQMTSD.

Comment: Several groups criticized the lack of quantitative meteorological model evaluation data for the 1995 RAMS meteorological modeling used for episodic ozone modeling.

Response: A peer-reviewed, quantitative evaluation of the RAMS model performance for this meteorological period is provided by Hogrefe, et al. [100] This analysis was performed using RAMS predictions for June through August of 1995. The results show that the RAMS biases and errors are generally in line with past meteorological model simulations by other groups outside EPA. The EPA remains satisfied that the 1995 RAMS meteorological inputs for the three CAM X ozone modeling episodes are of sufficient quality and we have continued to use these inputs for the ozone analyses for the final rule.

Comment: The EPA received several comments on the episodes selected for ozone modeling. There was general criticism that the ozone modeling did not follow EPA's own guidance for the selection of episodes. Additionally, there was specific criticism that the episodes did not provide for a reasonable test of the 8-hour ozone NAAQS in some areas.

Response: The draft 8-hour ozone guidance recommends, at a minimum, that four criteria be used to select episodes which are appropriate to model. This guidance is generally intended for local attainment demonstrations, as opposed to regional transport analyses, but it does recommend that in applying a regional model one should choose episodes meeting as many of the criteria as possible, though it acknowledges there may be tradeoffs. Given the large number of nonattainment areas within the ozone domain, it would be extremely difficult to assess the criteria on a area-by-area basis. However, from a general perspective, the 1995 episodes address all of the primary criteria, which include: (1) A variety of meteorological conditions; (2) measured ozone values that are close to current air quality; (3) extensive meteorological and air quality data; and (4) a sufficient number of days. More detail is provided in the NFR AQMTSD, but here is a brief description of how each of the four primary criteria are met by the 1995 cases.

With regard to the criteria of meteorological variations, we have completed inert tracer simulations for each of the three 1995 episodes that show different transport patterns in all three cases. For example the June case involves east-to-west transport; the July case involves west-to-east transport; and the August case involves south-to-north transport. In a separate analysis to determine whether the 1995 modeling days correspond to commonly occurring and ozone-conducive meteorology, EPA has applied a multi-variate statistical approach for characterizing daily meteorological patterns and investigating their relationship to 8-hour ozone concentrations in the eastern U.S. Across the 16 sites for which the analysis was completed, there were five to six distinct sets of meteorological conditions, called regimes, that occurred during the ozone seasons studied. An analysis of the 8-hour daily maximum ozone concentrations for each of the meteorological regimes was undertaken to determine the distribution of ozone concentrations and the frequency of occurrence of each regimes. The EPA determined that between 60 and 70 percent of the episode days we modeled are associated with the most frequently occurring, high ozone potential, meteorological regimes. These results also provide support that the episodes being modeled are representative of conditions present when high ozone concentrations are measured throughout the modeling domain. For the second criteria, EPA has completed an analysis which shows that the 1995 episodes contain observed 8-hour daily maximum ozone values that approximate recent ambient concentrations over the eastern U.S. Additional analyses performed by EPA and others have concluded that each of the three episodes involves widespread areas of elevated ozone concentrations. The synoptic meteorological pattern of the July 1995 episode has been identified by one of the commenters as representing a classic set of conditions necessary for high ozone over the eastern U.S. While the ozone was not quite as widespread in the June and August 1995 episodes, these periods also contained exceedances of the 8-hour ozone NAAQS in most portions of the region.

We believe that there is ample meteorological and air quality data available to support an evaluation of the modeling for these episodes. Specifically, there were over 700 ozone monitors reporting across the domain for use in model evaluation. As noted above, the model performance for these episodes compares favorably to the recommendations in EPA's urban modeling guidance. In addition, the modeling period is comprised of 30 days, not including model ramp-up periods which is considerably more than is typically used in an attainment demonstration modeling submitted to EPA by a State. Finally, EPA's draft ozone guidance also indicates as one of four secondary criteria that extra weight can be assigned to modeling episodes for which there is prior experience in modeling. The 1995 CAIR ozone episodes have been successfully used to drive the air quality modeling completed for several recent notice-and-comment rulemakings (Tier-2, Heavy Duty Engine, and NonRoad). Based on the analyses discussed above and the adherence to the modeling guidance, EPA is satisfied that the 1995 CAM X episodes are appropriate for continued use.

B. How Did EPA Project Future Nonattainment for PM 2.5 and 8-Hour Ozone?

1. Projection of Future PM 2.5 Nonattainment

a. Methodology for Projecting Future PM 2.5 Nonattainment

In the NPR, we assessed the prospects for future attainment and nonattainment in 2010 and 2015 of the PM 2.5 annual NAAQS. The approach for identifying areas expected to be nonattainment for PM 2.5 in the future involved using the model predictions in a relative way to forecast current PM 2.5 design values to 2010 and 2015. The modeling portion of this approach included annual simulations for 2001 proxy emissions and for 2010 and 2015 base case emissions scenarios. As described below, the predictions from these runs were used to calculate relative reduction factors (RRFs) which were then applied to current PM 2.5 design values from FRM sites in the East. This approach is consistent with the procedures in the draft of EPA's PM 2.5 modeling guidance.

To determine the current PM 2.5 air quality for use in projecting design values to the future, we selected the higher of the 1999-2001 or 2000-2002 design value (the most recent ambient data at the time of the proposal) for each monitor that measured nonattainment in 2000-2002. For those sites that were attaining the PM 2.5 standard based on their 2000-2002 design value, we used the value from this period as the starting point for projecting 2010 and 2015 air quality at these sites.

The procedure for calculating future year PM 2.5 design values is called the Speciated Modeled Attainment Test (SMAT). The test uses model predictions in a relative sense to estimate changes expected to occur in each major PM 2.5 species. These species are sulfate, nitrate, organic carbon, elemental carbon, crustal, and un-attributed mass. The relative change in model-predicted species concentrations were applied to ambient species measurements in order to project each species for the future year scenarios. We applied a spatial interpolation to the IMPROVE and STN speciation data as a means for estimating species composition fractions for the FRM monitoring sites. Future year PM 2.5 was calculated by summing the projected concentrations of each species. The SMAT technical procedures, as applied for the NPR, are contained in the NPR and NPR AQMTSD.

As noted above, the procedures for determining future year PM 2.5 concentrations were applied for each FRM site. For counties with only one FRM site, the forecast design value for that site was used to determine whether or not the county was predicted to be nonattainment in the future. For counties with multiple monitoring sites, the site with the highest future concentration was selected for that county. Those counties with future year concentrations of 15.1 μg/m 3 (as rounded up from 15.05 μg/m 3) or more were predicted to be nonattainment. Based on the modeling performed for the NPR, 61 counties in the East were forecast to be nonattainment for the 2010 base case. Of these, 41 were forecast to remain nonattainment for the 2015 base case.

Comment: Some commenters said that EPA has not established the credibility of using models in a relative sense to estimate future PM 2.5 concentrations and that poor performance of REMSAD for 1996 calls into question the use of models to adequately determine the effects of changes in emissions. One commenter said that a mechanistic model evaluation, in which model predictions of PM 2.5 precursor photochemical oxidants are compared to corresponding measurements, is an approach for gaining confidence in the ability of a model to provide a credible response to emission changes.

Response: The EPA believes the future year nonattainment projections should be based on using model predictions in a relative sense. By applying the model in a relative way, each measured component of PM 2.5 is adjusted upward or downward based on the percent change in that component, as determined by the ratio of future year to base year model predictions. The EPA feels that by using this approach, we are able to reduce the risk that overprediction or underprediction of PM 2.5 component species may unduly affect our projection of future year nonattainment.

The EPA agrees with commenters that one way to establish confidence in the credibility of this approach is to determine whether model predictions of PM 2.5 precursors are generally comparable to corresponding measured data. In this regard, we compared the CMAQ predictions to observations for several precursor gases for which measurements were available in 2001. These gases include sulfur dioxide, nitric acid, and ozone.

The results for the East are summarized in Table VI-2. Additional details on this analysis can be found in the CMAQ evaluation report. The results indicate that for both summer and winter ozone, the fractional bias and error is within the recommended range for urban scale ozone modeling included in EPA's draft guidance for 8-hour ozone modeling. For the other species examined, there are limited ambient data and few other studies against which to compare our findings. Still, our performance results for these species are within the range suggested as acceptable by commenters for sulfate (i.e., ±30 percent to ±60 percent for fractional bias and 50 percent to 75 percent for fractional error). Thus, CMAQ is considered appropriate and credible for use in projecting changes in future year PM 2.5 concentrations and the resultant health/economic benefits due to the emissions reductions.

Table VI-2.—CMAQ Model Performance Statistics for Ozone, Total Nitrate, and Nitric Acid in the East Back to Top
Eastern U.S. CMAQ 2001
FB (%) FE (%)
Ozone:    
AIRS (Summer) 13 21
AIRS (Winter) −9 31
Sulfur Dioxide:    
CASTNet (Summer) 31 48
CASTNet (Winter) 39 43
Nitric Acid:    
CASTNet (Summer) 29 39
CASTNet (Winter) −21 55

Comment: Several commenters said that EPA's SMAT approach is flawed and suggested alternative methods for attributing individual species mass to the FRM measured PM 2.5 mass. One commenter detailed several different methods to apportion the FRM mass to individual PM 2.5 species. They refer to two different estimation methods as the “FRM equivalent” approach and the “best estimate” approach.

Response: The EPA agrees that alternative methodologies can be used to apportion PM 2.5 species fractions to the FRM data. We believe that revising SMAT to use a methodology similar to an “FRM equivalent” methodology, as described in the Notice of Data Availability (69 FR 47828; August 6, 2004), is warranted. Since nonattainment designation determinations and future year nonattainment projections are based on measured FRM data, we believe that the PM 2.5 species data should be adjusted to best conform to what is measured on the FRM filters. Based on comments, EPA has revised our technique for projecting current PM 2.5 data to incorporate some aspects of the commenter's “FRM equivalent” methodology. As described in more detail in the NFR AQMTSD, we believe our revised methodology to be the most technically appropriate way of estimating what is measured on the FRM filters.

Full documentation of the revised EPA SMAT methodology is contained in the updated SMAT report [101] . In brief, we revised the SMAT methodology to take into account several known differences between what is measured by speciation monitors and what is measured on FRM filters. Among the revisions were calculations to account for nitrate, ammonium, and organic carbon volatilization, blank PM 2.5 mass, particle bound water, the degree of neutralization of sulfate, and the uncertainty in estimating organic carbon mass.

Comment: Several commenters noted that the future year design values were based on projections of the 1999-2001 and/or 2000-2002 FRM monitoring data and that there are more recent design value data available for the 2001-2003 design value period. Commenters also noted that the 2001-2003 data shows lower PM 2.5 concentrations at the majority of sites and therefore, by projecting the highest design value, we are overestimating the future year PM 2.5 values.

Response: As stated above, the PM 2.5 projection methodology in the NPR used the higher of the 1999-2001 or 2000-2002 PM 2.5 design value data. The draft modeling guidance for PM 2.5 specifies the use of the higher of the three design value periods which straddle the emissions year. The emissions year is 2001 and therefore the three periods would be 1999-2001, 2000-2002, and 2001-2003. Since the 2001-2003 data is now available, we are using it as part of the current year PM 2.5 calculations for the final rule.

The observation by a commenter that the 2001-2003 data are generally lower than in the previous two design value periods (i.e., 1999-2001 and 2000-2002) leads to the issue of how to reduce the influence of year-to-year variability in meteorology and emissions on our estimate of current air quality. As a consequence of this year-to-year variability in concentrations, relying on design values from any single period, as in the approach used for proposal, may not provide a robust representation of current air quality for use in forecasting the future. Specifically, the lower PM 2.5 values in 2001-2003 may not be representative of the current modeling period. To address the issue of year-to-year variability in the ambient data we have modified our methodology to use an average of the three design value periods that straddle the base year emissions year (i.e., 2001). In this case it is the average of the 1999-2001, 2000-2002, and 2001-2003 design values. The average of the three design values is not a straight 5-year average. Rather, it is a weighted average of the 1999-2003 period. That is, by averaging 1999-2001, 2000-2002, and 2001-2003, the value from 2001 is weighted three times; 2000 and 2002 are each weighted twice and 1999 and 2003 are each weighted once. This approach has the desired benefits of: (1) weighting the PM 2.5 values towards the middle year of the 5-year period, which is the 2001 base year for our emissions projections; and (2) smoothing out the effects of year-to-year variability in emissions and meteorology that occurs over the full 5-year period. We have adopted this method for use in projecting future PM 2.5 nonattainment for the final rule analysis. We plan to incorporate this new methodology into the next draft version of our PM 2.5 modeling guidance.

b. Projected 2010 and 2015 Base Case PM 2.5 Nonattainment Counties

For the final rule, we have revised the projected PM 2.5 nonattainment counties for 2010 and 2015 by applying CMAQ for the entire year (i.e., January through December) of 2001 using 2001 Base Year and 2010 and 2015 future base case emissions from the new modeling platform, as described in section VI.A.2. The 2010 and 2015 base case PM 2.5 nonattainment counties were determined applying the updated SMAT method using current 1999-2003 PM 2.5 air quality coupled with the PM 2.5 species from the 2001 Base Year and 2010 and 2015 base case CMAQ model runs. For counties with multiple monitoring sites, the site with the highest future concentration was selected for that county. Those counties with future year design values of 15.05 μg/m [3] or higher were predicted to be nonattainment. The result is that, without controls beyond those included in the base case, 79 counties in the East are projected to be nonattainment for the 2010 base case. For the 2015 base case, 74 counties in the East are projected to be nonattainment for PM 2.5.

In light of the uncertainties inherent in regionwide modeling many years into the future, of the 79 nonattainment counties projected for the 2010 base case, we have the most confidence in our projection of nonattainment for those counties that are not only forecast to be nonattainment in 2010, based on the SMAT method, but that also measure nonattainment for the most recent period of available ambient data (i.e., 2001-2003). In our analysis for the 2010 base case, there are 62 such counties in the East that are both “modeled” nonattainment and currently have “monitored” nonattainment. We refer to these counties as having “modeled plus monitored” nonattainment. Out of an abundance of caution, we are using only these 62 “modeled plus monitored” counties as the downwind receptors in determining which upwind States make a significant contribution to PM 2.5 in downwind States.

The 79 counties in the East that we project will be nonattainment for PM 2.5 in 2010 and the subset of 62 counties that are also “monitored” nonattainment in 2001-2003, are identified in Table VI-3. The 2015 base case PM 2.5 nonattainment counties are provided in Table VI-4.

Table VI-3.—Projected PM 2.5 Concentrations (μg/m3) for Nonattainment Counties in the 2010 Base Case Back to Top
State County 2010 Base “Modeled + Monitored”
Alabama DeKalb Co 15.23 No.
Alabama Jefferson Co 18.57 Yes.
Alabama Montgomery Co 15.12 No.
Alabama Morgan Co 15.29 No.
Alabama Russell Co 16.17 Yes.
Alabama Talladega Co 15.34 No.
Delaware New Castle Co 16.56 Yes.
District of Columbia 15.84 Yes.
Georgia Bibb Co 16.27 Yes.
Georgia Clarke Co 16.39 Yes.
Georgia Clayton Co 17.39 Yes.
Georgia Cobb Co 16.57 Yes.
Georgia DeKalb Co 16.75 Yes.
Georgia Floyd Co 16.87 Yes.
Georgia Fulton Co 18.02 Yes.
Georgia Hall Co 15.60 No.
Georgia Muscogee Co 15.65 No.
Georgia Richmond Co 15.68 No.
Georgia Walker Co 15.43 Yes.
Georgia Washington Co 15.31 No.
Georgia Wilkinson Co 16.27 No.
Illinois Cook Co 17.52 Yes.
Illinois Madison Co 16.66 Yes.
Illinois St. Clair Co 16.24 Yes.
Indiana Clark Co 16.51 Yes.
Indiana Dubois Co 15.73 Yes.
Indiana Lake Co 17.26 Yes.
Indiana Marion Co 16.83 Yes.
Indiana Vanderburgh Co 15.54 Yes.
Kentucky Boyd Co 15.23 No.
Kentucky Bullitt Co 15.10 No.
Kentucky Fayette Co 15.95 Yes.
Kentucky Jefferson Co 16.71 Yes.
Kentucky Kenton Co 15.30 No.
Maryland Anne Arundel Co 15.26 Yes.
Maryland Baltimore City 16.96 Yes.
Michigan Wayne Co 19.41 Yes.
Missouri St. Louis City 15.10 No.
New Jersey Union Co 15.05 Yes.
New York New York Co 16.19 Yes.
North Carolina Catawba Co 15.48 Yes.
North Carolina Davidson Co 15.76 Yes.
North Carolina Mecklenburg Co 15.22 No.
Ohio Butler Co 16.45 Yes.
Ohio Cuyahoga Co 18.84 Yes.
Ohio Franklin Co 16.98 Yes.
Ohio Hamilton Co 18.23 Yes.
Ohio Jefferson Co 17.94 Yes.
Ohio Lawrence Co 16.10 Yes.
Ohio Mahoning Co 15.39 Yes.
Ohio Montgomery Co 15.41 Yes.
Ohio Scioto Co 18.13 Yes.
Ohio Stark Co 17.14 Yes.
Ohio Summit Co 16.47 Yes.
Ohio Trumbull Co 15.28 No.
Pennsylvania Allegheny Co 20.55 Yes.
Pennsylvania Beaver Co 15.78 Yes.
Pennsylvania Berks Co 15.89 Yes.
Pennsylvania Cambria Co 15.14 Yes.
Pennsylvania Dauphin Co 15.17 Yes.
Pennsylvania Delaware Co 15.61 Yes.
Pennsylvania Lancaster Co 16.55 Yes.
Pennsylvania Philadelphia Co 16.65 Yes.
Pennsylvania Washington Co 15.23 Yes.
Pennsylvania Westmoreland Co 15.16 Yes.
Pennsylvania York Co 16.49 Yes.
Tennessee Davidson Co 15.36 No.
Tennessee Hamilton Co 16.89 Yes.
Tennessee Knox Co 17.44 Yes.
Tennessee Sullivan Co 15.32 No.
West Virginia Berkeley Co 15.69 Yes.
West Virginia Brooke Co 16.63 Yes.
West Virginia Cabell Co 17.03 Yes.
West Virginia Hancock Co 17.06 Yes.
West Virginia Kanawha Co 17.56 Yes.
West Virginia Marion Co 15.32 Yes.
West Virginia Marshall Co 15.81 Yes.
West Virginia Ohio Co 15.14 Yes.
West Virginia Wood Co 16.66 Yes.
Table VI-4.—Projected PM 2.5 Concentrations (μg/m3) for Nonattainment Counties in the 2015 Base Case Back to Top
State County 2015 Base
Alabama DeKalb Co 15.24
Alabama Jefferson Co 18.85
Alabama Montgomery Co 15.24
Alabama Morgan Co 15.26
Alabama Russell Co 16.10
Alabama Talladega Co 15.22
Delaware New Castle Co 16.47
District of Columbia 15.57
Georgia Bibb Co 16.41
Georgia Chatham Co 15.06
Georgia Clarke Co 16.15
Georgia Clayton Co 17.46
Georgia Cobb Co 16.51
Georgia DeKalb Co 16.82
Georgia Floyd Co 17.33
Georgia Fulton Co 18.00
Georgia Hall Co 15.36
Georgia Muscogee Co 15.58
Georgia Richmond Co 15.76
Georgia Walker Co 15.37
Georgia Washington Co 15.34
Georgia Wilkinson Co 16.54
Illinois Cook Co 17.71
Illinois Madison Co 16.90
Illinois St. Clair Co 16.49
Illinois Will Co 15.12
Indiana Clark Co 16.37
Indiana Dubois Co 15.66
Indiana Lake Co 17.27
Indiana Marion Co 16.77
Indiana Vanderburgh Co 15.56
Kentucky Boyd Co 15.06
Kentucky Fayette Co 15.62
Kentucky Jefferson Co 16.61
Kentucky Kenton Co 15.09
Maryland Baltimore City 17.04
Maryland Baltimore Co 15.08
Michigan Wayne Co 19.28
Mississippi Jones Co 15.18
Missouri St. Louis City 15.34
New York New York Co 15.76
North Carolina Catawba Co 15.19
North Carolina Davidson Co 15.34
Ohio Butler Co 16.32
Ohio Cuyahoga Co 18.60
Ohio Franklin Co 16.64
Ohio Hamilton Co 18.03
Ohio Jefferson Co 17.83
Ohio Lawrence Co 15.92
Ohio Mahoning Co 15.13
Ohio Montgomery Co 15.16
Ohio Scioto Co 17.92
Ohio Stark Co 16.86
Ohio Summit Co 16.14
Ohio Trumbull Co 15.05
Pennsylvania Allegheny Co 20.33
Pennsylvania Beaver Co 15.54
Pennsylvania Berks Co 15.66
Pennsylvania Delaware Co 15.52
Pennsylvania Lancaster Co 16.28
Pennsylvania Philadelphia Co 16.53
Pennsylvania York Co 16.22
Tennessee Davidson Co 15.36
Tennessee Hamilton Co 16.82
Tennessee Knox Co 17.34
Tennessee Shelby Co 15.17
Tennessee Sullivan Co 15.37
West Virginia Berkeley Co 15.32
West Virginia Brooke Co 16.51
West Virginia Cabell Co 16.86
West Virginia Hancock Co 16.97
West Virginia Kanawha Co 17.17
West Virginia Marshall Co 15.52
West Virginia Wood Co 16.69

2. Projection of Future 8-Hour Ozone Nonattainment

a. Methodology for Projecting Future 8-Hour Ozone Nonattainment

The approach for projecting future 8-hour ozone concentrations used by EPA in the NPR was based on applying the model in a relative sense to estimate the change in ozone between the base year (2001) and each future scenario. Projected 8-hour ozone design values in 2010 and 2015 were estimated by combining the relative change in model predicted ozone from 2001 to the future scenario with an estimate of the base year ambient 8-hour ozone design value. These procedures for calculating future case ozone design values are consistent with EPA's draft modeling guidance for 8-hour ozone attainment demonstrations. The draft guidance specifies the use of the higher of the design values from (a) the period that straddles the emissions inventory base year or (b) the design value period which was used to designate the area under the ozone NAAQS. At the time of the proposal, 2000-2002 was the design value period which both straddled the 2001 base year inventory and was also the latest period available.

Comment: Commenters noted that the procedures used by EPA for projecting future 8-hour ozone concentrations differ from the procedures used for projecting PM 2.5. These commenters said that EPA should harmonize the two approaches.

Response: In response to comments, we have made several changes in the approach to projecting future 8-hour ozone nonattainment in order to follow an approach that is consistent with the manner in which PM 2.5 projections are determined. The approach we are using to project PM 2.5 for the final rule analysis is described in section VI.B.1, above. In order to harmonize the ozone approach with the approach used for PM 2.5, we are using the weighted average of the design values for the periods that straddle the emission base year (i.e., 2001). These periods are 1999-2001, 2000-2002, and 2001-2003. In this approach, the fourth-high ozone value from 2001 is weighted three times, 2000 and 2002 are weighted twice, and 1999 and 2003 are weighted once. This has the desired effect of weighting the projected ozone values towards the middle year of the 5-year period, which is the emissions year (2001), while accounting for the emissions and meteorological variability that occurs over the full 5-year period. The average weighted concentration is expected to be more representative as a starting point for future year projections than choosing (a) the single design value period that straddles the base year or (b) the design value used for designations. We plan to incorporate this new methodology into the next draft version of our ozone modeling guidance.

Comment: One commenter claimed that the 2010 and 2015 ozone projections in the proposal base cases were too optimistic, that is, that the modeling was underestimating the number of areas that may be in nonattainment in the future. The commenter urged a more conservative approach to assessing the future attainment status of areas.

Response: The technical basis for the comment stemmed from the assertion that the regional ozone modeling that EPA performed for the proposal was not of “SIP-quality.” The EPA response to the specific technical issues with regard to episode selection and grid resolution can be found in section VI.A as well as in the response to comments document. The EPA remains confident that the CAIR 8-hour ozone modeling platform is appropriate for assessing potential levels of future nonattainment.

b. Projected 2010 and 2015 Base Case 8-Hour Ozone Nonattainment Counties

For the final rule, we have revised our projections of ozone nonattainment for the 2010 and 2015 base cases by applying CAMx for the three 1995 ozone episodes using 2001 Base Year and 2010 and 2015 future base case emissions from the new modeling platform, as described in section VI.A.2. The revised 2010 and 2015 base case 8-hour ozone nonattainment counties were determined by applying the relative change in 8-hour ozone predicted by these CAMx model runs to the weighted average 1999-2003 8-hour ozone concentrations as described above and, in more detail, in the NFR AQMTSD. For counties with multiple monitoring sites, the site with the highest future concentration was selected for that county. Those counties with future year design values of 85 parts per billion (ppb) or higher were predicted to be nonattainment.

As a result of our updated modeling we project that, without controls beyond those in the base case, there will be 40 8-hour ozone nonattainnment counties in 2010 and 22 nonattainment counties in 2015. All of the 40 counties that we are projecting to be nonattainment for the 2010 base case are also measuring nonattainment based on the most recent design value period (i.e., 2001-2003). We refer to these counties as “modeled plus monitored” nonattainment, as described above in section IV.B.1 for PM 2.5. We are using these 40 counties as the downwind receptors to determine which States make a significant contribution to 8-hour ozone nonattainment in downwind States.

The counties we are projecting to be nonattainment for 8-hour ozone in the 2010 base case and 2015 base case are listed in Table VI-5 and Table VI-6, respectively.

Table VI-5.—Projected 2010 Base Case 8-hour Ozone Nonattainment Counties and Concentrations (ppb) Back to Top
State County 2010 Base
Connecticut Fairfield Co 92.6
Connecticut Middlesex Co 90.9
Connecticut New Haven Co 91.6
Delaware New Castle Co 85.0
District of Columbia 85.2
Georgia Fulton Co 86.5
Maryland Anne Arundel Co 88.8
Maryland Cecil Co 89.7
Maryland Harford Co 93.0
Maryland Kent Co 86.2
Michigan Macomb Co 85.5
New Jersey Bergen Co 86.9
New Jersey Camden Co 91.9
New Jersey Gloucester Co 91.8
New Jersey Hunterdon Co 89.0
New Jersey Mercer Co 95.6
New Jersey Middlesex Co 92.4
New Jersey Monmouth Co 86.6
New Jersey Morris Co 86.5
New Jersey Ocean Co 100.5
New York Erie Co 87.3
New York Richmond Co 87.3
New York Suffolk Co 91.1
New York Westchester Co 85.3
Ohio Geauga Co 87.1
Pennsylvania Bucks Co 94.7
Pennsylvania Chester Co 85.7
Pennsylvania Montgomery Co 88.0
Pennsylvania Philadelphia Co 90.3
Rhode Island Kent Co 86.4
Texas Denton Co 87.4
Texas Galveston Co 85.1
Texas Harris Co 97.9
Texas Jefferson Co 85.6
Texas Tarrant Co 87.8
Virginia Arlington Co 86.2
Virginia Fairfax Co 85.7
Wisconsin Kenosha Co 91.3
Wisconsin Ozaukee Co 86.2
Wisconsin Sheboygan Co 88.3
Table VI-6.—Projected 2015 Base Case 8-hour Ozone Nonattainment Counties and Concentrations (ppb) Back to Top
State County 2015 Base
Connecticut Fairfield Co 91.4
Connecticut Middlesex Co 89.1
Connecticut New Haven Co 89.8
Maryland Anne Arundel Co 86.0
Maryland Cecil Co 86.9
Maryland Harford Co 90.6
Michigan Macomb Co 85.1
New Jersey Bergen Co 85.7
New Jersey Camden Co 89.5
New Jersey Gloucester Co 89.6
New Jersey Hunterdon Co 86.5
New Jersey Mercer Co 93.5
New Jersey Middlesex Co 89.8
New Jersey Ocean Co 98.0
New York Erie Co 85.2
New York Suffolk Co 89.9
Pennsylvania Bucks Co 93.0
Pennsylvania Montgomery Co 86.5
Pennsylvania Philadelphia Co 88.9
Texas Harris Co 97.3
Texas Jefferson Co 85.0
Wisconsin Kenosha Co 89.4

C. How Did EPA Assess Interstate Contributions to Nonattainment?

1. PM 2.5 Contribution Modeling Approach

For the proposed rule, EPA performed State-by-State zero-out modeling to quantify the contribution from emissions in each State to future PM 2.5 nonattainment in other States and to determine whether that contribution meets the air quality prong (i.e., before considering cost) of the “contribute significantly” test. The zero-out modeling technique provides an estimate of downwind impacts by comparing the model predictions from the 2010 base case to the predictions from a run in which all anthropogenic SO 2 and NO X emissions are removed from specific States. Counties forecast to be nonattainment for PM 2.5 in the proposal 2010 base case were used as receptors for quantifying interstate contributions of PM 2.5. For each State-by-State zero-out run we projected the annual average PM 2.5 concentration at each receptor using the proposed SMAT technique, as described in the NPR AQMTSD. The contribution from an upwind State to nonattainment at a given downwind receptor was determined by calculating the difference in PM 2.5 concentration between the 2010 base case and the zero-out run at that receptor. We followed this process for each State-by-State zero-out run and each receptor. For each upwind State, we identified the largest contribution from that State to a downwind nonattainment receptor in order to determine the magnitude of the maximum downwind contribution from each State. The maximum downwind contribution was proposed as the metric for determining whether or not the contribution was significant. As described in section III, EPA proposed, in the alternative, a criterion of 0.10 μg/m 3 and 0.15 μg/m 3 for determining whether emissions in a State make a significant contribution (before considering cost) to PM 2.5 nonattainment in another State. Details on these procedures can be found in the NPR AQMTSD.

Comments: Commenters questioned the use of zero-out modeling and said that EPA should support the development of a source apportionment model for PM 2.5 contributions. The commenter recommended that EPA delay the final rule until such a technique can be used. Another commenter provided results of a sulfate source apportionment technique currently under development along with modeling results which showed that the zero-out technique and source apportionment for sulfate provide similar results in terms of the magnitude and extent of downwind impacts. The commenter noted that the results suggest that zero-out modeling may somewhat underestimate the transport of sulfate.

Response: The EPA continues to believe that the zero-out technique is a credible method for quantifying interstate PM 2.5 contributions. This is supported by a commenter's results showing that the zero-out technique and source apportionment appear to give similar results. We accept the commenter's modeling for sulfate source apportionment results which indicate that the zero-out technique does not overestimate interstate transport. Moreover, EPA rejects the notion that we should delay needed reductions while we await alternative assessment techniques.

2. 8-Hour Ozone Contribution Modeling Approach

In the proposal, EPA quantified the impact of emissions from specific upwind States on 8-hour ozone concentrations in projected downwind nonattainment areas. The procedures we followed to assess interstate ozone contribution for the proposal analysis are summarized below. We are using these same procedures along with the updated CAM X modeling platform, as described in section VI.A., to assess ozone contributions for today's rule. Details on these procedures can be found in the NFR AQMTSD.

We applied two different modeling techniques, zero-out and source apportionment, to assess the contributions of emissions in upwind States on 8-hour ozone nonattainment in downwind States. The outputs of the two modeling techniques were evaluated in terms of three key contribution factors to determine which States make a significant contribution to downwind ozone nonattainment as described in section VI.B.2. The zero-out and source apportionment modeling techniques provide different, but equally valid, technical approaches to quantifying the downwind impact of emissions from upwind States. The zero-out modeling analysis provides an estimate of downwind impacts by comparing the model predictions from the 2010 base case and the predictions from a model run in which all anthropogenic NO X and VOC emissions are removed from specific States. The source apportionment modeling quantifies downwind impacts by tracking and allocating the amounts of ozone formed from man-made NO X and VOC emissions in upwind States. Because large portions of the six States along the western border of the modeling domain [102] are outside the area covered by our modeling, EPA did not analyze the contributions to downwind ozone nonattainment for these States.

In the analysis done at proposal, EPA considered three fundamental factors for evaluating whether emissions in an upwind State make large and/or frequent contributions to downwind nonattainment: (1) The magnitude of the contribution; (2) the frequency of the contribution; and (3) the relative amount of the contribution when compared against contributions from other areas. The factors are the basis for several metrics that can be used to assess a particular impact. The metrics used in this analysis were the same as those used in the NO X SIP Call.

Within these three factors, eight specific metrics were calculated to assess the contribution of each of the 31 States to the residual nonattainment counties. For the zero-out modeling, EPA considered: (1) The maximum contribution (magnitude); (2) the number and percentage of exceedances with contributions in certain concentration ranges (frequency); (3) the total contribution relative to the total exceedance level ozone in the receptor area (relative amount); and (4) the population-weighted total contribution relative to the total population-weighted exceedance level ozone in the receptor area (relative amount). For the source apportionment modeling EPA considered: (5) The maximum contribution (magnitude); (6) the highest daily average contribution (magnitude); (7) the number and percentages of exceedances with contributions in certain concentration ranges (frequency); and (8) the total average contribution to exceedance ozone in the downwind area (relative amount). The values for these metrics were calculated using only those periods during which the model predicted 8-hour average ozone concentrations greater than or equal to 85 ppb in at least one of the model grid cells associated with the receptor county in the 2010 base case. Grid cells were linked to a specific nonattainment county if any part of the grid cell covered any portion of the projected 2010 nonattainment county.

The first step in evaluating the contribution factors was to screen out linkages for which the contributions were clearly small. This initial screening was based on two criteria: (1) The maximum contribution had to be greater than or equal to 2 ppb from either of the two modeling techniques; and (2) the total average contribution to exceedance of ozone in the downwind area had to be greater than 1 percent. If either screening test was not met, then the linkage was not considered significant. Those linkages that had contributions which exceeded the screening criteria were evaluated further in steps 2 through 4.

In step 2, we evaluated the contributions in each linkage based on the zero-out modeling and in step 3 we evaluated the contributions in each linkage based on the source apportionment modeling. In step 4, we considered the results of both step 2 and step 3 to determine which of the linkages were significant. For both techniques, EPA determined whether the linkage is significant by evaluating the magnitude, frequency, and relative amount of the contributions. Each upwind State that made relatively large and/or frequent contributions to nonattainment in the downwind area, based on these factors, was considered to contribute significantly to nonattainment in the downwind area.

The EPA believes that each of the factors provides an independent measure of contribution, however, there had to be at least two different factors that indicated large and/or frequent contributions in order for the linkage to be found significant. In this regard, the finding of a significant contribution for an individual linkage was not based on any single factor. Further, each of the modeling approaches had to show at least one indicator of a large and/or frequent contribution in order for the linkage to be found significant. The EPA received several general comments on the procedures for assessing interstate contributions of ozone to projected residual nonattainment areas, as discussed below.

Comment: A commenter opposed the use of population-weighted metrics to determine whether an upwind State's impact on a location in another State is significant.

Response: The commenter's concern was that transport contributions to rural areas with low populations were not being weighted appropriately. This is not a valid concern because the relative contribution factor from the zero-out modeling is presumed to be met if either of the two criteria (population-weighted, or non-population-weighted) show large contributions.

Comment: Also, EPA received a specific comment on a certain linkage that was deemed to be significant in the analysis done to support the NPR. The commenter objected to the conclusion that Mississippi significantly contributes to residual ozone exceedances near Memphis. The objection resulted from issues with grid resolution, episode selection, and the fact that the zero-out and source apportionment modeling for Mississippi included some emissions from Tennessee and Arkansas due to the irregular State boundaries.

Response: As noted in section VI.B.2, Crittenden County, AR is no longer projected to be a nonattainment area in the 2010 base case. As a result, the issue of Mississippi's contribution to ozone in the Memphis area is moot.

D. What Are the Estimated Interstate Contributions to PM 2.5 and 8-Hour Ozone Nonattainment?

1. Results of PM 2.5 Contribution Modeling

In this section, we present the interstate contributions from emissions in upwind States to PM 2.5 nonattainment in downwind nonattainment counties. States which contribute 0.2 μg/m [3] or more to PM 2.5 nonattainment in another State are determined to contribute significantly (before considering cost). We calculated the interstate PM 2.5 contributions using the State-by-State zero-out modeling technique, as indicated above in section VI.C.1. This technique is described in the NFR AQMTSD. We performed zero-out modeling using CMAQ for each of 37 States individually (i.e., Alabama, Arkansas, Connecticut, Delaware, Florida, Georgia, Illinois, Indiana, Iowa, Kansas, Kentucky, Louisiana, Maine, Maryland combined with the District of Columbia, Massachusetts, Michigan, Minnesota, Mississippi, Missouri, Nebraska, New Hampshire, New Jersey, New York, North Carolina, North Dakota, Ohio, Oklahoma, Pennsylvania, Rhode Island, South Carolina, South Dakota, Tennessee, Texas, Vermont, Virginia, West Virginia, and Wisconsin).

We calculated each State's contribution to PM 2.5 in each of the 62 counties that are projected to be nonattainment in the 2010 base case (i.e., “modeled” nonattainment) and are also “monitored” nonattainment in 2001-2003, as described in section VI.B.1.b. The maximum contribution from each upwind State to downwind PM 2.5 nonattainment is provided in Table VI-7. The contributions from each State to nonattainment in each nonattainment county are provided in the NFR AQMTSD. Based on the State-by-State modeling, there are 23 States and the District of Columbia [103] which contribute 0.2 μg/m [3] or more to downwind PM 2.5 nonattainment (Alabama, the District of Columbia, Florida, Georgia, Illinois, Indiana, Iowa, Kentucky, Louisiana, Maryland, Michigan, Minnesota, Mississippi, Missouri, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Texas, Virginia, West Virginia, and Wisconsin). In Table VI-8, we provide a list of the downwind nonattainment counties to which each upwind State contributes 0.2 μg/m [3] or more (i.e., the upwind State-to-downwind nonattainment “linkages”).

Table VI-7.—Maximum Downwind PM 2.5 Contribution (μg/m3) for each of 37 States Back to Top
Upwind State Maximum downwind contribution
Alabama 0.98
Arkansas 0.19
Connecticut 0.05
Delaware 0.14
Florida 0.45
Georgia 1.27
Illinois 1.02
Indiana 0.91
Iowa 0.28
Kansas 0.11
Kentucky 0.90
Louisiana 0.25
Maine 0.05
Maryland/DC 0.69
Massachusetts 0.07
Michigan 0.62
Minnesota 0.21
Mississippi 0.23
Missouri 1.07
Nebraska 0.07
New Hampshire 0.05
New Jersey 0.13
New York 0.34
North Carolina 0.31
North Dakota 0.11
Ohio 1.67
Oklahoma 0.12
Pennsylvania 0.89
Rhode Island 0.05
South Carolina 0.40
South Dakota 0.05
Tennessee 0.65
Texas 0.29
Vermont 0.05
Virginia 0.44
West Virginia 0.84
Wisconsin 0.56
Table VI-8.—Upwind State-to-Downwind Nonattainment County Significant “Linkages” for PM 2.5. Back to Top
Upwind states Total linkages Downwind counties
AL 21 Bibb GA Cabell WV Catawba NC Clark IN.
Clarke GA Clayton GA Cobb GA Davidson NC.
DeKalb GA Dubois IN Fayette KY Floyd GA.
Fulton GA Hamilton OH Hamilton TN Jefferson KY.
Knox TN Lawrence OH Scioto OH Vanderburgh IN.
Walker GA      
FL 7 Bibb GA Clarke GA Clayton GA Cobb GA.
DeKalb GA Jefferson AL Russell AL  
GA 17 Butler OH Cabell WV Catawba NC Clark IN.
Davidson NC Fayette KY Hamilton OH Hamilton TN.
Jefferson AL Jefferson KY Kanawha WV Knox TN.
Lawrence OH Montgomery OH Russell AL Scioto OH.
Vanderburgh IN      
IL 23 Allegheny PA Butler OH Cabell WV Clark IN.
Cuyahoga OH Dubois IN Fayette KY Franklin OH.
Hamilton OH Hamilton TN Jefferson AL Jefferson KY.
Kanawha WV Lake IN Lawrence OH Mahoning OH.
Marion IN Montgomery OH Scioto OH Stark OH.
Summit OH Vanderburgh IN Wayne MI
IN 46 Allegheny PA Beaver PA Berkeley WV Bibb GA.
Brooke WV Butler OH Cabell WV Cambria PA.
Catawba NC Clarke GA Clayton GA Cobb GA.
Cook IL Cuyahoga OH Davidson NC DeKalb GA.
Fayette KY Floyd GA Franklin OH Fulton GA.
Hamilton OH Hamilton TN Hancock WV Jefferson AL.
Jefferson KY Jefferson OH Kanawha WV Knox TN.
Lancaster PA Lawrence OH Madison IL Mahoning OH.
Marion WV Marshall WV Montgomery OH Ohio WV.
Russell AL St. Clair IL Scioto OH Stark OH.
Summit OH Walker GA Wayne MI Washington PA.
Westmoreland PA Wood WV    
IA 5 Cook IL Lake IN Madison IL Marion IN.
St. Clair IL      
KY 35 Allegheny PA Butler OH Cabell WV Catawba NC.
Clark IN Clarke GA Cobb GA Cuyahoga OH.
Davidson NC Dubois IN Floyd GA Franklin OH.
Hamilton OH Hamilton TN Jefferson AL Jefferson OH.
Kanawha WV Knox TN Lawrence OH Madison IL.
Mahoning OH Marion IN Marion WV Marshall WV.
Montgomery OH Ohio WV St. Clair IL Scioto OH.
Stark OH Summit OH Vanderburgh IN Walker GA.
Washington PA Westmoreland PA Wood WV.  
LA 2 Jefferson AL Russell AL    
MD/DC 13 Berkeley WV Berks PA Cambria PA Dauphin PA.
Delaware PA District of Columbia Lancaster PA New Castle DE.
New York NY Philadelphia PA Union NJ Westmoreland PA.
York PA      
MI 36 Allegheny PA Beaver PA Berks PA Brooke WV.
Butler OH Cabell WV Cambria PA Clark IN.
Cook IL Cuyahoga OH Dauphin PA Delaware PA.
Fayette KY Franklin OH Hamilton OH Hancock WV.
Jefferson OH Lake IN Lancaster PA Lawrence OH.
Mahoning OH Marion IN Marion WV Marshall WV.
Montgomery OH New Castle DE Ohio WV Philadelphia PA.
Scioto OH Stark OH Summit OH Union NJ.
Washington PA Westmoreland PA Wood WV York PA.
MN 2 Cook IL Lake IN    
MO 9 Clark IN Cook IL Dubois IN Jefferson KY.
Lake IN Madison IL Marion IN St. Clair IL.
Vanderburgh IN.      
MS 1 Jefferson AL      
NY 5 Berks PA Lancaster PA New Castle DE New Haven CT.
Union NJ      
NC 7 Anne Arundel MD Baltimore City Bibb GA Clarke GA.
District of Columbia Kanawha WV Knox TN.  
OH 51 Anne Arundel MD Allegheny PA Baltimore City MD Beaver PA.
Berkeley WV Berks PA Bibb GA Brooke WV.
Cabell WV Cambria PA Catawba NC Clark IN.
Clarke GA Clayton GA Cobb GA Cook IL.
Dauphin PA Davidson NC DeKalb GA Delaware PA.
District of Columbia Dubois IN Fayette KY Floyd GA.
Fulton GA Hamilton TN Hancock WV Jefferson AL.
Jefferson KY Kanawha WV Knox TN Lake IN.
Lancaster PA Madison IL Marion IN Marion WV.
Marshall WV New Castle DE New York NY Ohio WV.
Philadelphia PA Russell AL St. Clair IL Union NJ.
Vanderburgh IN Walker GA Washington PA Wayne MI.
Westmoreland PA Wood WV York PA  
PA 25 Anne Arundel MD Baltimore City Berkeley WV Brooke WV.
Cabell WV Catawba NC Clarke GA Cuyahoga OH.
Davidson NC District of Columbia Hancock WV Jefferson OH.
Kanawha WV Lawrence OH Mahoning OH Marion WV.
Marshall WV New Castle DE New York NY Ohio WV.
Stark OH Summit OH Union NJ Wayne MI.
Wood WV      
SC 9 Bibb GA Catawba NC Clarke GA Clayton GA.
Cobb GA Davidson NC DeKalb GA Fulton GA.
Russell AL      
TN 23 Bibb GA Butler OH Cabell WV Catawba NC.
Clark IN Clarke GA Clayton GA Cobb GA.
Davidson NC DeKalb GA Dubois IN Fayette KY.
Floyd GA Fulton GA Hamilton OH Jefferson AL.
Jefferson KY Kanawha WV Lawrence OH Russell AL.
Scioto OH Vanderburgh TN Walker GA.
TX 2 Madison IL St Clair IL    
VA 13 Anne Arundel MD Baltimore City MD Berkeley WV Berks PA.
Catawba NC Dauphin PA Davidson NC Delaware PA.
District of Columbia Lancaster PA New Castle DE Philadelphia PA.
York PA      
WV 33 Anne Arundel MD Allegheny PA Baltimore City MD Beaver PA.
Berks PA Butler OH Cambria PA Catawba NC.
Clarke GA Cuyahoga OH Dauphin PA Davidson NC.
Delaware PA District of Columbia Fayette KY Franklin OH.
Hamilton OH Jefferson OH Knox TN Lancaster PA.
Lawrence OH Mahoning OH Montgomery OH New Castle DE.
New York NY Philadelphia PA Scioto OH Stark OH.
Summit OH Union NJ Washington PA Westmoreland PA.
York PA      
WI 4 Cook IL Lake IN Marion IN Wayne MI.

2. Results of 8-Hour Ozone Contribution Modeling

In this section, we present the results of air quality modeling to determine which upwind States contribute significantly (before considering cost) to 8-hour ozone nonattainment in downwind States. The analytical procedures to determine which States make a significant contribution are based on the zero-out and source apportionment modeling techniques using CAM X, as described in section VI.C.2 and in the NFR AQMTSD. We performed ozone contribution modeling using both of these techniques for 31 States in the East and the District of Columbia (i.e., Alabama, Arkansas, Connecticut, Delaware, Georgia, Florida, Iowa, Illinois, Indiana, Kentucky, Louisiana, Massachusetts, Maine, Maryland combined with the District of Columbia, Michigan, Minnesota, Mississippi, Missouri, New Hampshire, New Jersey, New York, North Carolina, Ohio, Pennsylvania, Rhode Island, South Carolina, Tennessee, Vermont, Virginia, West Virginia, and Wisconsin).

We evaluated the interstate ozone contributions from each of the 31 upwind States and the District of Columbia to each of the 40 counties that are projected to be nonattainment in the 2010 base case (i.e., “modeled” nonattainment) and are also “monitored” nonattainment in 2001-2003, as described in section VI.B.2.b. We analyzed the contributions from upwind States to these counties in terms of various metrics, described above and in more detail in the NFR AQMTSD.

Based on the State-by-State modeling, there are 25 States and the District of Columbia [104] which make a significant contribution (before considering cost) to 8-hour ozone nonattainment in downwind States (i.e., Alabama, Arkansas, Connecticut, Delaware, the District of Columbia, Florida, Iowa, Illinois, Indiana, Kentucky, Louisiana, Massachusetts, Maryland, Michigan, Mississippi, Missouri, New Jersey, New York, North Carolina, Ohio, Pennsylvania, South Carolina, Tennessee, Virginia, West Virginia, and Wisconsin). In Table VI-9, we provide a list of the downwind nonattainment counties to which each upwind State makes a significant contribution (i.e., the upwind State-to-downwind nonattainment “linkages”).

Table VI-9.—Upwind State-to-Downwind Nonattainment County Significant “Linkages” for 8-hour Ozone. Back to Top
Upwind states Total linkages Downwind counties
AL 3 Fulton GA Harris TX Jefferson TX.
AR 3 Galveston TX Harris TX Jefferson TX.
CT 2 Kent RI Suffolk NY.    
DE 13 Bucks PA Camden NJ Chester PA Gloucester NJ.
Hunterdon NJ Mercer NJ Middlesex NJ Monmouth NJ.
Montgomery PA Morris NJ Ocean NJ Philadelphia PA.
Suffolk NY      
FL 1 Fulton GA      
IA 3 Kenosha WI Macomb MI Sheboygan WI.
IL 5 Geauga OH Kenosha WI Macomb MI Ozaukee WI.
Sheboygan WI.      
IN 5 Geauga OH Kenosha WI Macomb MI Ozaukee WI.
Sheboygan WI.      
KY 3 Fulton GA Geauga OH Macomb MI.
LA 3 Galveston TX Harris TX Jefferson TX.
MA 2 Kent RI Middlesex NJ.    
MD/DC 23 Arlington VA Bergen NJ Bucks PA Camden NJ.
Chester PA District of Columbia Erie NY Fairfax VA.
Fairfield CT Gloucester NJ Hunterton NJ Mercer NJ.
Middlesex NJ Monmouth NJ Montgomery PA Morris NJ.
New Castle DE New Haven CT Ocean NJ Philadelphia PA.
Richmond NY Suffolk NY Westchester NY
MI 19 Anne Arundel MD Bergen NJ Bucks PA Camden NJ.
Cecil MD Chester PA Erie NY Geauga OH.
Gloucester NJ Kent MD Mercer NJ Middlesex NJ.
Monmouth NJ Morris NJ New Castle DE Ocean NJ.
Philadelphia PA Richmond NY Suffolk NY
MO 4 Geauga OH Kenosha WI Ozaukee WI Sheboygan WI.
MS 2 Harris TX Jefferson TX.    
NC 8 Anne Arundel MD Fulton GA Harford MD Kent MD.
Newcastle DE Suffolk NY Bucks PA Chester PA.
NJ 10 Erie NY Fairfield CT Kent RI Middlesex CT.
Montgomery PA New Haven CT Philadelphia PA Richmond NY.
Suffolk NY Westchester NY.    
NY 9 Fairfield CT Kent RI Mercer NJ Middlesex CT.
Middlesex NJ Monmouth NJ Morris NJ New Haven CT.
Ocean NJ.      
Anne Arundel MD Arlington VA Bergen NJ Bucks PA.
OH 28 Camden NJ Cecil MD Chester PA District of Columbia.
Fairfax VA Fairfield CT Gloucester NJ Harford MD.
Hunterton NJ Kent MD Kent RI Macomb MI.
Mercer NJ Middlesex CT Middlesex NJ Monmouth NJ.
Montgomery PA Morris NJ New Castle DE New Haven CT.
Ocean NJ Philadelphia PA Suffolk NY Westchester NY.
PA 25 Anne Arundel MD Arlington VA Bergen NJ Camden NJ.
Cecil MD District of Columbia Erie NY Fairfax VA.
Fairfield CT Gloucester NJ Harford MD Hunterton NJ.
Kent MD Kent RI Mercer NJ Middlesex CT.
Middlesex NJ Monmouth NJ Morris NJ New Castle DE.
New Haven CT Ocean NJ Richmond NY Suffolk NY.
Westchester NY.      
SC 1 Fulton GA.      
TN 1 Fulton GA.      
VA 26 Anne Arundel MD Bergen NJ Bucks PA Camden NJ.
Cecil MD Chester PA District of Columbia Erie NY.
Fairfield CT Gloucester NJ Harford MD Hunterton NJ.
Kent MD Kent RI Mercer NJ Middlesex CT.
Middlesex NJ Monmouth NJ Morris NJ New Castle DE.
New Haven CT Ocean NJ Philadelphia PA Richmond NY.
Suffolk NY Westchester NY.    
WI 2 Erie NY Macomb MI.    
WV 25 Anne Arundel MD Bergen NJ Bucks PA Camden NJ.
Cecil MD Chester PA Fairfax VA Fairfield CT.
Fulton GA Gloucester NJ Harford MD Hunterton NJ.
Kent MD Mercer NJ Middlesex NJ Monmouth NJ.
Montgomery PA Morris NJ New Castle DE New Haven CT.
Ocean NJ Philadelphia PA Richmond NY Suffolk NY.
Westchester NY      

E. What are the Estimated Air Quality Impacts of the Final Rule?

In this section, we describe the air quality modeling performed to determine the projected impacts on PM 2.5 and 8-hour ozone of the SO 2 and NO X emissions reductions in the control region modeled. The modeling used to estimate the air quality impact of these reductions assumes annual SO 2 and NO X controls for Arkansas, Delaware, and New Jersey in addition to the 23-States plus the District of Columbia. Since Arkansas, Delaware, and New Jersey are not included in the final CAIR region for PM 2.5, the modeled estimated impacts on PM 2.5 are overstated for today's final rule. However, EPA plans to include Delaware and New Jersey in the CAIR region for PM 2.5 through a separate regulatory process. Thus, the estimates are reflective of the total impacts expected for CAIR assuming Delaware and New Jersey will become part of the annual SO 2 and NO X trading programs.

As discussed in section IV, EPA analyzed the impacts of the regional emissions reductions in both 2010 and 2015. These impacts are quantified by comparing air quality modeling results for the regional control scenario to the modeling results for the corresponding 2010 and 2015 base case scenarios. The 2010 and 2015 emissions reductions from the power generation sector include a two-phase cap and trade program covering the control region modeled (i.e., the 23 States plus the District of Columbia included in today's rule and Arkansas, Delaware, and New Jersey). [105] Phase 1 of the regional strategy (the 2010 reductions) is forecast to reduce total EGU SO 2 emissions [106] in the control region modeled by 40 percent in 2010. Phase 2 (the 2015 reductions) is forecast to provide a 48 percent reduction in EGU SO 2 emissions compared to the base case in 2015. When fully implemented post-2015, we expect this rule to result in more than a 70 percent reduction in EGU SO 2 emissions compared to current emissions levels. The reductions at full implementation occur post-2015 due to the existing title IV bank of SO 2 allowances, which can be used under the CAIR program. The net effect of the strategy on total SO 2 emissions in the control region modeled considering all sources of emissions, is a 28 percent reduction in 2010 and a 32 percent reduction in 2015.

For NO X, Phase 1 of the strategy is forecast to reduce total EGU emissions by 44 percent in 2009. Total NO X emissions across the control region (i.e., includes all sources) are 11 percent lower in the 2010 CAIR scenario compared to the emissions in the 2010 base case. In Phase 2, EGU NO X emissions are projected to decline by 54 percent in 2015 in this region. Total NO X emissions from all anthropogenic sources are projected to be reduced by 14 percent in 2015. The percent change in emissions by State for SO 2 and NO X in 2010 and 2015 for the regional control strategy modeled are provided in the NFR EITSD.

1. Estimated Impacts on PM 2.5 Concentrations and Attainment

We determined the impacts on PM 2.5 of the CAIR regional strategy by running the CMAQ model for this strategy and comparing the results to the PM 2.5 concentrations predicted for the 2010 and 2015 base cases. In brief, we ran the CMAQ model for the regional strategy in both 2010 and 2015. The model predictions were used to project future PM 2.5 concentrations for CAIR in 2010 and 2015 using the SMAT technique, as described in section VI.B.1. We compared the results of the 2010 and 2015 regional strategy modeling to the corresponding results from the 2010 and 2015 base cases to quantify the expected impacts of CAIR.

The impacts of the SO 2 and NO X emissions reductions expected from CAIR on PM 2.5 in 2010 and 2015 are provided in Table VI-10 and Table VI-11, respectively. In these tables, counties shown in bold/italics are projected to come into attainment with CAIR.

Table VI-10.—Projected PM 2.5 Concentrations (μg/m3) for the 2010 Base Case and CAIR and the Impact of CAIR Regional Controls in 2010 Back to Top
State County 2010 Base case 2010 CAIR Impact of CAIR
Alabama DeKalb Co 15.23 13.97 −1.26
Alabama Jefferson Co 18.57 17.46 −1.11
Alabama Montgomery Co 15.12 14.10 −1.02
Alabama Morgan Co 15.29 14.11 −1.18
Alabama Russell Co 16.17 15.15 −1.02
Alabama Talladega Co 15.34 14.00 −1.34
Delaware New Castle Co 16.56 14.84 −1.72
District of Columbia 15.84 13.68 −2.16
Georgia Bibb Co 16.27 15.17 −1.10
Georgia Clarke Co 16.39 14.96 −1.43
Georgia Clayton Co 17.39 16.29 −1.10
Georgia Cobb Co 16.57 15.35 −1.22
Georgia DeKalb Co 16.75 15.70 −1.05
Georgia Floyd Co 16.87 15.87 −1.00
Georgia Fulton Co 18.02 16.98 −1.04
Georgia Hall Co 15.60 14.28 −1.32
Georgia Muscogee Co 15.65 14.57 −1.08
Georgia Richmond Co 15.68 14.64 −1.04
Georgia Walker Co 15.43 14.22 −1.21
Georgia Washington Co 15.31 14.22 −1.09
Georgia Wilkinson Co 16.27 15.22 −1.05
Illinois Cook Co 17.52 16.88 −0.64
Illinois Madison Co 16.66 15.96 −0.70
Illinois St. Clair Co 16.24 15.54 −0.70
Indiana Clark Co 16.51 15.15 −1.36
Indiana Dubois Co 15.73 14.37 −1.36
Indiana Lake Co 17.26 16.48 −0.78
Indiana Marion Co 16.83 15.54 −1.29
Indiana Vanderburgh Co 15.54 14.26 −1.28
Kentucky Boyd Co 15.23 13.38 −1.85
Kentucky Bullitt Co 15.10 13.67 −1.43
Kentucky Fayette Co 15.95 14.17 −1.78
Kentucky Jefferson Co 16.71 15.44 −1.27
Kentucky Kenton Co 15.30 13.72 −1.58
Maryland Anne Arundel Co 15.26 12.98 −2.28
Maryland Baltimore city 16.96 14.88 −2.08
Michigan Wayne Co 19.41 18.23 −1.18
Missouri St. Louis City 15.10 14.40 −0.70
New Jersey Union Co 15.05 13.60 −1.45
New York New York Co 16.19 14.95 −1.24
North Carolina Catawba Co 15.48 14.07 −1.41
North Carolina Davidson Co 15.76 14.36 −1.40
North Carolina Mecklenburg Co 15.22 13.92 −1.30
Ohio Butler Co 16.45 15.03 −1.42
Ohio Cuyahoga Co 18.84 17.11 −1.73
Ohio Franklin Co 16.98 15.13 −1.85
Ohio Hamilton Co 18.23 16.61 −1.62
Ohio Jefferson Co 17.94 15.64 −2.30
Ohio Lawrence Co 16.10 14.11 −1.99
Ohio Mahoning Co 15.39 13.40 −1.99
Ohio Montgomery Co 15.41 13.83 −1.58
Ohio Scioto Co 18.13 15.98 −2.15
Ohio Stark Co 17.14 15.08 −2.06
Ohio Summit Co 16.47 14.69 −1.78
Ohio Trumbull Co 15.28 13.50 −1.78
Pennsylvania Allegheny Co 20.55 18.01 −2.54
Pennsylvania Beaver Co 15.78 13.61 −2.17
Pennsylvania Berks Co 15.89 13.56 −2.33
Pennsylvania Cambria Co 15.14 12.72 −2.42
Pennsylvania Dauphin Co 15.17 12.88 −2.29
Pennsylvania Delaware Co 15.61 13.94 −1.67
Pennsylvania Lancaster Co 16.55 14.09 −2.46
Pennsylvania Philadelphia Co 16.65 14.98 −1.67
Pennsylvania Washington Co 15.23 12.99 −2.24
Pennsylvania Westmoreland Co 15.16 12.60 −2.56
Pennsylvania York Co 16.49 14.20 −2.29
Tennessee Davidson Co 15.36 14.26 −1.10
Tennessee Hamilton Co 16.89 15.57 −1.32
Tennessee Knox Co 17.44 16.16 −1.28
Tennessee Sullivan Co 15.32 14.01 −1.31
West Virginia Berkeley Co 15.69 13.43 −2.26
West Virginia Brooke Co 16.63 14.42 −2.21
West Virginia Cabell Co 17.03 15.08 −1.95
West Virginia Hancock Co 17.06 14.89 −2.17
West Virginia Kanawha Co 17.56 15.27 −2.29
West Virginia Marion Co 15.32 12.90 −2.42
West Virginia Marshall Co 15.81 13.46 −2.35
West Virginia Ohio Co 15.14 12.81 −2.33
West Virginia Wood Co 16.66 14.14 −2.52
Table VI-11.—Projected PM 2.5 Concentrations (μg/m 3) for the 2015 Base Case and CAIR and the Impact of CAIR Regional Controls in 2015 Back to Top
State County 2015 Base case 2015 CAIR Impact of CAIR
Alabama DeKalb Co 15.24 13.46 −1.78
Alabama Jefferson Co 18.85 17.36 −1.49
Alabama Montgomery Co 15.24 13.87 −1.37
Alabama Morgan Co 15.26 13.85 −1.41
Alabama Russell Co 16.10 14.66 −1.44
Alabama Talladega Co 15.22 13.35 −1.87
Delaware New Castle Co 16.47 14.41 −2.06
District of Columbia 15.57 13.11 −2.46
Georgia Bibb Co 16.41 14.83 −1.58
Georgia Chatham Co 15.06 13.86 −1.20
Georgia Clarke Co 16.15 14.10 −2.05
Georgia Clayton Co 17.46 15.85 −1.61
Georgia Cobb Co 16.51 14.67 −1.84
Georgia DeKalb Co 16.82 15.29 −1.53
Georgia Floyd Co 17.33 15.79 −1.54
Georgia Fulton Co 18.00 16.47 −1.53
Georgia Hall Co 15.36 13.48 −1.88
Georgia Muscogee Co 15.58 14.06 −1.52
Georgia Richmond Co 15.76 14.23 −1.53
Georgia Walker Co 15.37 13.65 −1.72
Georgia Washington Co 15.34 13.67 −1.67
Georgia Wilkinson Co 16.54 15.01 −1.53
Illinois Cook Co 17.71 16.95 −0.76
Illinois Madison Co 16.90 16.07 −0.83
Illinois St. Clair Co 16.49 15.64 −0.85
Illinois Will Co 15.12 14.27 −0.85
Indiana Clark Co 16.37 14.79 −1.58
Indiana Dubois Co 15.66 14.16 −1.50
Indiana Lake Co 17.27 16.36 −0.91
Indiana Marion Co 16.77 15.38 −1.39
Indiana Vanderburgh Co 15.56 14.17 −1.39
Kentucky Boyd Co 15.06 12.95 −2.11
Kentucky Fayette Co 15.62 13.54 −2.08
Kentucky Jefferson Co 16.61 15.13 −1.48
Kentucky Kenton Co 15.09 13.26 −1.83
Maryland Baltimore city 17.04 14.50 −2.54
Maryland Baltimore Co 15.08 12.75 −2.33
Michigan Wayne Co 19.28 17.95 −1.33
Mississippi Jones Co 15.18 14.06 −1.12
Missouri St. Louis city 15.34 14.50 −0.84
New York New York Co 15.76 14.33 −1.43
North Carolina Catawba Co 15.19 13.45 −1.74
North Carolina Davidson Co 15.34 13.61 −1.73
Ohio Butler Co 16.32 14.67 −1.65
Ohio Cuyahoga Co 18.60 16.67 −1.93
Ohio Franklin Co 16.64 14.57 −2.07
Ohio Hamilton Co 18.03 16.10 −1.93
Ohio Jefferson Co 17.83 15.26 −2.57
Ohio Lawrence Co 15.92 13.71 −2.21
Ohio Mahoning Co 15.13 12.94 −2.19
Ohio Montgomery Co 15.16 13.33 −1.83
Ohio Scioto Co 17.92 15.55 −2.37
Ohio Stark Co 16.86 14.58 −2.28
Ohio Summit Co 16.14 14.18 −1.96
Ohio Trumbull Co 15.05 13.08 −1.97
Pennsylvania Allegheny Co 20.33 17.47 −2.86
Pennsylvania Beaver Co 15.54 13.09 −2.45
Pennsylvania Berks Co 15.66 12.99 −2.67
Pennsylvania Delaware Co 15.52 13.52 −2.00
Pennsylvania Lancaster Co 16.28 13.33 −2.95
Pennsylvania Philadelphia Co 16.53 14.53 −2.00
Pennsylvania York Co 16.22 13.46 −2.76
Tennessee Davidson Co 15.36 14.02 −1.34
Tennessee Hamilton Co 16.82 14.94 −1.88
Tennessee Knox Co 17.34 15.61 −1.73
Tennessee Shelby Co 15.17 14.19 −0.98
Tennessee Sullivan Co 15.37 13.77 −1.60
West Virginia Berkeley Co 15.32 12.73 −2.59
West Virginia Brooke Co 16.51 14.05 −2.46
West Virginia Cabell Co 16.86 14.64 −2.22
West Virginia Hancock Co 16.97 14.54 −2.43
West Virginia Kanawha Co 17.17 14.66 −2.51
West Virginia Marshall Co 15.52 12.87 −2.65
West Virginia Wood Co 16.69 13.88 −2.81

As described in section VI.B.1, we project that 79 counties in the East will be nonattainment for PM 2.5 in the 2010 base case. We estimate that, on average, the regional strategy will reduce PM 2.5 in these 79 counties by 1.6 μg/m 3. In over 90 percent of the nonattainment counties (i.e., 74 out of 79 counties), we project that PM 2.5 will be reduced by at least 1.0 μg/m 3. In over 25 percent of the 79 nonattainment counties (i.e., 23 of the 79 counties), we project PM 2.5 concentrations will decline by of more than 2.0 μg/m 3. Of the 79 counties that are nonattainment in the 2010 Base, we project that 51 counties will come into attainment as a result of the SO 2 and NO X emissions reductions expected from the regional controls. Even those 28 counties that remain nonattainment in 2010 after implementation of the regional strategy will be closer to attainment as a result of these emissions reductions. Specifically, the average reduction of PM 2.5 in the 28 residual nonattainment counties is projected to be 1.3 μg/m 3. After implementation of the regional controls, we project that 18 of the 28 residual nonattainment counties in 2010 will be within 1.0 μg/m 3 of the NAAQS and 12 counties will be within 0.5 μg/m 3 of attainment.

In 2015 we are projecting that PM 2.5 in the 74 base case nonattainment counties will be reduced by 1.8 μg/m 3, on average, as a result of the SO 2 and NO X reductions in the regional strategy. In over 90 percent of the nonattainment counties (i.e., 67 of the 74 counties) concentrations of PM 2.5 are predicted to be reduced by at least 1.0 μg/m 3. In over 35 percent of the counties (i.e., 27 of the 74 counties), we project the regional strategy to reduce PM 2.5 by more than 2.0 μg/m 3. As a result of the reductions in PM 2.5, 56 nonattainment counties are projected to come into attainment in 2015. The remaining 18 nonattainment counties are projected to be closer to attainment with the regional strategy. Our modeling results indicate that PM 2.5 will be reduced in the range of 0.7 μg/m 3 to 2.9 μg/m 3 in these 18 counties. The average reduction across these 18 residual nonattainment counties is 1.5 μg/m 3.

Thus, the SO 2 and NO X emissions reductions which will result from the regional strategy will greatly reduce the extent of PM 2.5 nonattainment by 2010 and beyond. These emissions reductions are expected to substantially reduce the number of PM 2.5 nonattainment counties in the East and make attainment easier for those counties that remain nonattainment by substantially lowering PM 2.5 concentrations in these residual nonattainment counties.

2. Estimated Impacts on 8-Hour Ozone Concentrations and Attainment

We determined the impacts on 8-hour ozone of the regional strategy by running the CAM X model for this strategy and comparing the results to the ozone concentrations predicted for the 2010 and 2015 base cases. In brief, we ran the CAM X model for the regional strategy in both 2010 and 2015. The model predictions were used to project future 8-hour ozone concentrations for the regional strategy in 2010 and 2015 using the Relative Reduction Factor technique, as described in section VI.B.1. We compared the results of the 2010 and 2015 regional strategy modeling to the corresponding results from the 2010 and 2015 base cases to quantify the expected impacts of the regional controls.

The results of the regional strategy ozone modeling are expressed in terms of the expected reductions in projected 8-hour concentrations and the implications for future nonattainment. The impacts of the regional NO X emissions reductions on 8-hour ozone in 2010 and 2015 are provided in Table VI-12 and Table VI-13, respectively. In these tables, counties shown in bold/italics are projected to come into attainment with the regional controls.

Table VI-12.—Projected 8-Hour Concentrations (ppb) for the 2010 Base Case and CAIR and the Impact of CAIR Regional Controls in 2010 Back to Top
State County 2010 Base case 2010 CAIR Impact of CAIR
Connecticut Fairfield Co 92.6 92.2 −0.4
Connecticut Middlesex Co 90.9 90.6 −0.3
Connecticut New Haven Co 91.6 91.3 −0.3
District of Columbia District of Columbia 85.2 85.0 −0.2
Delaware New Castle Co 85.0 84.7 −0.3
Georgia Fulton Co 86.5 85.1 −1.4
Maryland Anne Arundel Co 88.8 88.6 −0.2
Maryland Cecil Co 89.7 89.5 −0.2
Maryland Harford Co 93.0 92.8 −0.2
Maryland Kent Co 86.2 85.8 −0.4
Michigan Macomb Co 85.5 85.4 −0.1
New Jersey Bergen Co 86.9 86.0 −0.9
New Jersey Camden Co 91.9 91.6 −0.3
New Jersey Gloucester Co 91.8 91.3 −0.5
New Jersey Hunterdon Co 89.0 88.6 −0.4
New Jersey Mercer Co 95.6 95.2 −0.4
New Jersey Middlesex Co 92.4 92.1 −0.3
New Jersey Monmouth Co 86.6 86.4 −0.2
New Jersey Morris Co 86.5 85.5 −1.0
New Jersey Ocean Co 100.5 100.3 −0.2
New York Erie Co 87.3 86.9 −0.4
New York Richmond Co 87.3 87.1 −0.2
New York Suffolk Co 91.1 90.8 −0.3
New York Westchester Co 85.3 84.7 −0.6
Ohio Geauga Co 87.1 86.6 −0.5
Pennsylvania Bucks Co 94.7 94.3 −0.4
Pennsylvania Chester Co 85.7 85.4 −0.3
Pennsylvania Montgomery Co 88.0 87.6 −0.4
Pennsylvania Philadelphia Co 90.3 89.9 −0.4
Rhode Island Kent Co 86.4 86.2 −0.2
Texas Denton Co 87.4 86.8 −0.6
Texas Galveston Co 85.1 84.6 −0.5
Texas Harris Co 97.9 97.4 −0.5
Texas Jefferson Co 85.6 85.0 −0.6
Texas Tarrant Co 87.8 87.2 −0.6
Virginia Arlington Co 86.2 86.0 −0.2
Virginia Fairfax Co 85.7 85.4 −0.3
Wisconsin Kenosha Co 91.3 91.0 −0.3
Wisconsin Ozaukee Co 86.2 85.8 −0.4
Wisconsin Sheboygan Co 88.3 87.7 −0.6
Table VI-13.—Projected 8-Hour Concentrations (ppb) for the 2015 Base Case and CAIR and the Impact of CAIR Regional Controls in 2015 Back to Top
State County 2015 Base case 2015 CAIR Impact of CAIR
Connecticut Fairfield Co 91.4 90.6 −0.8
Connecticut Middlesex Co 89.1 88.4 −0.7
Connecticut New Haven Co 89.8 89.1 −0.7
Maryland Anne Arundel Co 86.0 84.9 −1.1
Maryland Cecil Co 86.9 85.4 −1.5
Maryland Harford Co 90.6 89.6 −1.0
Michigan Macomb Co 85.1 84.2 −0.9
New Jersey Bergen Co 85.7 84.5 −1.2
New Jersey Camden Co 89.5 88.3 −1.2
New Jersey Gloucester Co 89.6 88.2 −1.4
New Jersey Hunterdon Co 86.5 85.4 −1.1
New Jersey Mercer Co 93.5 92.4 −1.1
New Jersey Middlesex Co 89.8 88.8 −1.0
New Jersey Ocean Co 98.0 96.9 −1.1
New York Erie Co 85.2 84.2 −1.0
New York Suffolk Co 89.9 89.0 −0.9
Pennsylvania Bucks Co 93.0 91.8 −1.2
Pennsylvania Montgomery Co 86.5 84.9 −1.6
Pennsylvania Philadelphia Co 88.9 87.5 −1.4
Texas Harris Co 97.3 96.4 −0.9
Texas Jefferson Co 85.0 84.1 −0.9
Wisconsin Kenosha Co 89.4 88.8 −0.6

As described in section VI.B.1, we project that 40 counties in the East would be nonattainment for 8-hour ozone under the assumptions in the 2010 base case. Our modeling of the regional controls in 2010 indicates that 3 of these counties will come into attainment of the 8-hour ozone NAAQS and that ozone in 16 of the 40 nonattainment counties will be reduced by 1 ppb or more. In addition, our modeling predicts that 8-hour ozone exceedances (i.e., 8-hour ozone of 85 ppb or higher) within nonattainment areas are expected to decline by 5 percent in 2010 with CAIR. Of the 37 counties that are projected to remain nonattainment in 2010 after the regional strategy, nearly half (i.e., 16 of the 37 counties) are within 2 ppb of attainment.

In 2015, we project that 6 of the 22 counties which are nonattainment for 8-hour ozone in the base case will come into attainment with the regional strategy. Ozone concentrations in over 70 percent (i.e., 16 of 22 counties) of the 2015 base case nonattainment counties are projected to be reduced by 1 ppb or more as a result of the regional strategy. Exceedances of the 8-hour ozone NAAQS are predicted to decline in nonattainment areas by 14 percent with regional controls in place in 2015. Thus, the NO X emissions reductions which will result from the regional strategy will help to bring 8-hour ozone nonattainment areas in the East closer to attainment by 2010 and beyond.

F. What are the Estimated Visibility Impacts of the Final Rule?

1. Methods for Calculating Projected Visibility in Class I Areas

The NPR contained example future year visibility projections for the 20 percent worst days and 20 percent best days at Class I areas that had complete IMPROVE monitoring data in 1996. Changes in future visibility were predicted by using the REMSAD model to generate relative visibility changes, then applying those changes to measured current visibility data. Details of the visibility modeling and calculations can be found in the NPR AQMTSD. An example visibility calculation was given in Appendix M of the NPR AQMTSD along with the predicted improvement in visibility (in deciviews) on the 20 percent best and worst days at 44 Class I areas. The data contained in Appendix M was for informational purposes only and was not used in the significant contribution determination or control strategy development decisions.

The SNPR contained visibility calculations in support of the “better-than-BART” analysis. The better-than-BART analysis employed a two-pronged test to determine if the modeled visibility improvements from the CAIR cap and trade program for EGU's were “better” than the visibility improvements from a nationwide BART program. The analysis used the visibility calculation methodology detailed in the NPR TSD. Detailed results of the SNPR better-than-BART analysis are contained in the SNPR AQMTSD. The better-than-BART analysis for the final rule is addressed in section IX.C.2 of the preamble. Additional information on the visibility calculation methodology is contained in the NFR AQMTSD.

2. Visibility Improvements in Class I Areas

For the NFR we have modeled several new CAIR [107] and CAIR + BART cases to re-examine the better-than-BART two-pronged test. We have modeled an updated nationwide BART scenario as well as a CAIR in the East/BART in the West scenario. The results were analyzed at 116 Class I areas that have complete IMPROVE data for 2001 or are represented by IMPROVE monitors with complete data. Twenty-nine of the Class I areas are in the East and 87 are in the West. The results of the visibility analysis are summarized in section IX.C.2. Detailed results for all 116 Class I areas are presented in the NFR AQMTSD.

VII. SIP Criteria and Emissions Reporting Requirements Back to Top

This section describes: (1) The criteria we will use in determining approvability of SIPs submitted to meet the requirements of today's rulemaking; (2) the dates for submittal of the SIPs that are required under the CAIR; (3) the consequences of either failing to submit such a SIP or submitting a SIP which is disapproved; and (4) the emissions inventory reporting requirements for States.

A. What Criteria Will EPA Use To Evaluate the Approvability of a Transport SIP?

1. Introduction

The approvability criteria for CAIR SIP submissions are finalized today in 40 CFR 51.123 (NO X emissions reductions) and in 40 CFR 51.124 (SO 2 emissions reductions). Most of the criteria are substantially similar to those that currently apply to SIP submissions under CAA section 110 or part D (nonattainment). For example, each submission must describe the control measures that the State intends to employ, identify the enforcement methods for monitoring compliance and managing violations, and demonstrate that the State has legal authority to carry out its plan.

This part of the preamble explains additional approvability criteria specific to the CAIR that were proposed and discussed in the CAIR NPR or in the CAIR SNPR, and are being promulgated today. As explained in both the CAIR NPR and the CAIR SNPR, EPA proposed that each affected State must submit SIP revisions containing control measures that assure that a specified amount of NO X and SO 2 emissions reductions are achieved by specified dates.

Although EPA determined the amount of emissions reductions required by identifying specific, highly cost-effective control levels for EGUs, EPA explained in the CAIR NPR and the CAIR SNPR that States have flexibility in choosing which sources to control to achieve the required emissions reductions. As long as a State's emissions reductions requirements are met, a State may impose controls on EGUs only, on non-EGUs only, or on a combination of EGUs and non-EGUs. The SIP approvability criteria are intended to provide as much certainty as possible that, whichever sources a State chooses to control, the controls will result in the required amount of emissions reductions.

In the CAIR NPR, EPA proposed a “hybrid” approach for the mechanisms used to ensure emissions reductions are achieved. This approach incorporates elements of an emissions “budget” approach (requiring an emissions cap on affected sources) and an “emissions reduction” approach (not requiring an emissions cap). In this hybrid approach, if States impose control measures on EGUs, they would be required to impose an emissions cap on all EGUs, which would effectively be an emissions budget. And, as stated in the CAIR NPR, if States impose control measures on non-EGUs, they would be encouraged but not required to impose an emissions cap on non-EGUs. In the CAIR NPR, we requested comment on the issue of requiring States to impose caps on any source categories that the State chooses to regulate.

In the CAIR SNPR, we proposed to modify the hybrid approach and require States that choose to control large industrial boilers or turbines (greater than 250 MMBTU/hr) to impose an emissions cap on all such sources within their State. This is similar to EPA's approach in the NO X SIP Call which required States to include an emissions cap on such sources as well as on EGUs if the SIP submittals included controls on such sources. (See 40 CFR 51.121(f)(2)(ii).)

A few commenters supported the use of emissions caps on any source category subject to CAIR controls, including non-EGUs, because it would be the most effective way to demonstrate compliance with the budget. A few other commenters opposed the use of an emissions cap on non-EGUs, saying either that States should have the flexibility to determine whether to impose a cap, or that such a requirement would result in increased costs for non-EGUs including cogeneration units that are non-EGUs. No commenter opposing such a requirement provided any information indicating that such a requirement would be ineffective or impracticable. Today EPA is adopting the modified approach, as described in the CAIR SNPR, that States choosing to control EGUs or large industrial boilers or turbines must do so by imposing an emissions cap on such sources, similar to what was required in the NO X SIP Call.

Extensive comments were received regarding the need for an ozone season NO X cap in States identified to be contributing significantly to the region's ozone nonattainment problems. In proposal, EPA stated that the annual NO X cap under CAIR reduced NO X emissions sufficiently enough to not warrant a regional ozone season NO X cap. Commenters remained very concerned that the annual NO X cap would not aid ozone attainment. While EPA feels that the annual NO X limit will most likely be protective in the ozone season, a seasonal cap will provide certainty, which EPA agrees is very important in the effort to help areas achieve ozone attainment. Today, EPA is finalizing an ozone season NO X cap for States shown to contribute significantly for ozone. As is further explained in section VIII, EPA is also finalizing an ozone season trading program that States may use to achieve the required emissions reductions. This program will subsume the existing NO X SIP Call trading program. Therefore, any State that wishes to continue including its sources in an interstate trading program run by EPA to achieve the emissions reductions required by EPA must modify its SIP to conform with this new trading program.

The EPA will automatically find that a State is continuing to meet its NO X SIP Call obligation if it achieves all of its required CAIR emissions reductions by capping EGUs, it modifies its existing NO X SIP Call to require its non-EGUs currently participating in the NO X SIP Call budget trading program to conform to the requirements of the CAIR ozone season NO X trading program with a trading budget that is the same or tighter than the budget in the currently approved SIP, and it does not modify any of its other existing NO X SIP Call rules. If a State chooses to achieve the ozone season NO X emissions reduction requirements of CAIR in another way, it will also be required to demonstrate that it continues to meet the requirements of the NO X SIP Call.

Specific criteria for approval of CAIR SIP submissions as promulgated by today's action are described below. The criteria are dependent on the types of sources a State chooses to control.

2. Requirements for States Choosing To Control EGUs

a. Emissions Caps and Monitoring

As explained in the CAIR NPR (69 FR 4626), and in the CAIR SNPR (69 FR 32691), EPA proposed requiring States to apply the “budget” approach if they choose to control EGUs; that is, each State must cap total EGU emissions at the level that assures the appropriate amount of reductions for that State. The requirement to cap all EGUs is important because it prevents shifting of utilization (and resulting emissions) to uncapped EGUs. The EGUs are part of a highly interconnected electricity grid that makes utilization shifting likely and even common. The units are large and offer the same market product (i.e., electricity), and therefore the units that are least expensive to operate are likely to be operated as much as possible. If capped and uncapped units are interconnected, the uncapped units' costs would tend to decrease relative to the capped units, which must either reduce emissions or use or buy allowances, and the uncapped units' utilization would likely increase. The cap ensures that emissions reductions from these interconnected sources are actually achieved rather than emissions simply shifting among sources. The caps constitute the State EGU Budgets for SO 2 and NO X. Additionally, EPA proposed that, if States choose to control EGUs, they must require EGUs to follow part 75 monitoring, recordkeeping, and reporting requirements. Part 75 monitoring and reporting requirements have been used effectively for determining NO X and SO 2 emissions from EGUs under the title IV Acid Rain program and the NO X SIP Call program and in combination with emissions caps are an integral part of those programs. (Additional explanation for the need for Part 75 monitoring is given in the NPR and SNPR and is incorporated here.) Therefore, today, EPA adopts the requirements for emission caps and Part 75 monitoring for EGUs in these States.

b. Using the Model Trading Rules

As proposed, if a State chooses to allow its EGUs to participate in EPA-administered interstate NO X and SO 2 emissions trading programs, the State must adopt EPA's model trading rules, as described elsewhere in today's preamble and in §§ 96.101-96.176 (for NO X) and §§ 96.201-96.276 (for SO 2), set forth below. Additionally, EPA proposed that for the States for which EPA made a finding of significant contribution for both ozone and PM 2.5, participation in both the NO X and SO 2 trading programs would be required in order to be included in the EPA-administered program. States for which the finding was for ozone only could choose to participate in only the EPA-administered NO X trading program through adoption of the NO X model trading rule. The EPA stated that States adopting EPA's model trading rules, modified only as specifically allowed by EPA, will meet the requirement for applying an emissions cap and requirement to use part 75 monitoring, recordkeeping, and reporting for EGUs.

Some commenters opposed EPA's proposal to require participation in both the NO X and SO 2 trading programs because some States may want to participate in the EPA-administered trading programs for only NO X or only SO 2. A few commenters claimed that the requirement to participate in both programs would limit State flexibility or is an “all or nothing” approach; other commenters objected that there was no environmental basis for such a requirement; and one commenter suggested that States not affected by CAIR but that volunteer to control emissions should be permitted to join the program for one or both pollutants. Additionally, commenters cited a need for an ozone season NO X program.

The EPA has taken the comments into account and in today's action agrees to allow a State identified to contribute significantly for PM 2.5 (and therefore required to make annual SO 2 and NO X reductions) to participate in the EPA-administered CAIR trading program for either SO 2 or NO X, not necessarily both, so long as the State adopts the model rule for the applicable trading program.

In response to extensive comments relating to EPA's proposal to forego a seasonal NO X cap because EPA demonstrated that the annual NO X cap was sufficiently stringent, EPA is finalizing an ozone season NO X trading program for States identified as contributing significantly for ozone. These States will be subject to an ozone season NO X cap and an annual NO X cap if the State is also identified as contributing significantly for PM 2.5. Therefore, today's action includes an additional model rule for an ozone season NO X trading program (40 CFR 96, subparts AAAA through IIII). The States that may use the ozone season NO X trading program but not the annual NO X trading program are those States in the CAIR region identified as contributing significantly for ozone only (Arkansas, Connecticut, Delaware, Massachusetts, and New Jersey).

As discussed in the proposal, EPA is finalizing the option for New Hampshire and Rhode Island to participate in the regional trading program through use of the CAIR ozone season NO X model rule because sources in these States have made investments in NO X controls in the past based on the existence of a regional ozone season NO X trading program. Additionally, the States' combined projected 2010 and 2015 NO X emissions are less than one-half of one percent of the total CAIR regional NO X cap and therefore would not create a significant increase in the CAIR cap. All comments received were supportive of this approach and EPA is finalizing it today.

None of these States (Arkansas, Connecticut, Delaware, Massachusetts, New Hampshire, New Jersey, or Rhode Island) has the option to participate in the EPA-administered CAIR SO 2 trading program nor the annual CAIR NO X trading program because there are no PM 2.5-related emissions reductions required under today's action in those States. (Of course, sources in these States will still be subject to the Acid Rain SO 2 cap and trade program.) Likewise, Texas, Minnesota and Georgia may not participate in the ozone season NO X program, because they have not been shown to contribute significantly to the regional ozone problem. They are, however, required to make annual NO X and SO 2 reductions and may choose to participate in the annual NO X and annual SO 2 trading program to meet their CAIR obligations.

Except for the special cases of Rhode Island and New Hampshire, other States outside of the CAIR region may not participate in the CAIR trading programs for either pollutant, because they were not shown to contribute significantly to PM 2.5 or ozone nonattainment in the CAIR region. Allowing States outside of the CAIR region to participate would generally create an opportunity—through net sales of allowances from the non-CAIR States to CAIR States—for emission increases in States that have been shown to contribute significantly to nonattainment in the CAIR region. [108]

A State may not participate in the EPA-administered trading programs if they choose to get a portion of CAIR reductions from non-EGUs. (This is also discussed in Section VIII.) The EPA maintains that requiring certain consistencies among States in the regionwide trading programs that EPA has offered to run does not unfairly limit States' flexibility to choose an approach for achieving CAIR mandated reductions that is best suited for a particular State's unique circumstances. States are free to achieve the reductions through whatever alternative mechanisms the States wish to design; for example, a group of States could cooperatively implement their own multi-State trading programs that EPA would not administer.

c. Using a Mechanism Other Than the Model Trading Rules

If States choose to control EGUs through a mechanism other than the EPA-administered NO X and SO 2 emissions trading programs, then the States (i) must still impose an emissions cap on total EGU emissions and require part 75 monitoring, recordkeeping, and reporting requirements on all EGUs, and (ii) must use the same definition of EGU as EPA uses in its model trading rules, i.e., the sources described as “CAIR units” in § 96.102, § 96.202, and § 96.302. A few commenters expressed concern that these requirements limit States' discretion in designing control measures to meet the CAIR requirements, but failed to offer any reason why the requirements would be impracticable or ineffective. The EPA believes that the requirements are necessary for a number of reasons. The requirements to cap all EGUs and to use the same definition of EGU are important because they prevent shifting of utilization (and resulting emissions) from capped to uncapped sources. In this case, not requiring a cap on total EGU emissions in these States is likely to result in increased utilization and consequently increased emissions in these States. The requirement to use part 75 monitoring ensures the accuracy of monitored data and consistency of reporting among sources (and thus the certainty that emissions reductions actually occurred) across all States. Furthermore, most EGUs are currently monitoring and reporting using part 75 so it does not impose an additional requirement. Therefore, EPA is finalizing the proposed approach.

If a State chooses to design its own intrastate or interstate NO X or SO 2 emissions trading programs, the State must, in addition to meeting the requirements of the rules finalized in today's action, consider EPA's guidance, “Improving Air Quality with Economic Incentive Programs,” January, 2001 (EPA-452/R-01-001) (available on EPA's Web site at: http://www.epa.gov/ttn/ecas/incentiv.html). The State's programs are subject to EPA approval. The EPA will not administer a State-designed trading program. Additionally, it should be noted that allowances from any alternate trading program may not be used in the EPA-administered trading programs.

d. Retirement of Excess Title IV Allowances

The CAIR NPR proposed requirements on SIPs relating to the effects of title IV SO 2 allowance allocations for 2010 and beyond that are in excess of the State's CAIR EGU SO 2 emissions budget. The requirements were intended to ensure that the excess is not used in a manner that would lead to a significant increase in supply of title IV allowances, the collapse of the price of title IV allowances, the disruption of operation of the title IV allowance market and the title IV SO 2 cap and trade system, and the potential for increased emissions in all States prior to 2010 and in non-CAIR States in 2010 and later. These negative impacts on the title IV allowance market and on air quality, which are discussed in detail in section IX.B. below, would undermine the efficacy of the title IV program and could erode confidence in cap and trade programs in general. To avoid these impacts, EPA proposed to require retirement of the excess title IV allowances through a retirement ratio mechanism.

The EPA proposed, as a mechanism for removing these additional allowances and meeting the 50 percent reduction required under phase I (2010-2014), that each affected EGU had to hold, and EPA would retire, two vintage 2010-2014 allowances for every ton of SO 2 that the unit emits. Further, EPA proposed that, for phase II (which begins in 2015) when a 65 percent reduction is required, each affected EGU had to hold, and EPA would retire, three vintage 2015 and beyond allowances for every ton of SO 2 that the unit emits. This 3-to-1 ratio would result in slightly more reductions than EPA has determined were necessary to eliminate the significant contribution by an upwind State.

In the CAIR SNPR, EPA proposed two alternatives for addressing the issue of the additional allowances. Under the first alternative, affected EGUs had to hold, and EPA would retire, vintage 2015 and beyond allowances at a rate of 2.86-to-1 rather than 3-to-1, which would result in exactly the amount of reductions EPA has determined are necessary to eliminate a State's significant contribution.

Alternatively, also in the CAIR SNPR, EPA proposed requiring the retirement of 2015 and beyond vintage allowances at a 3-to-1 ratio and permitting States to convert the additional reductions into allowances in their rules. The EPA also suggested that some States may want to use these reserved allowances to create an incentive for additional local emissions reductions that will be needed to bring all areas into attainment with the PM 2.5 NAAQS.

As part of today's final CAIR rulemaking, EPA is finalizing a ratio of 2.86-to-one. The ratio ultimately represents a reduction of 65 percent from the final title IV cap level, which has been found to be highly cost-effective. For a detailed discussion regarding EPA's determination of highly cost-effective, please refer to Section IV of the final CAIR preamble. As discussed earlier, EPA must employ a uniform ratio across sources to ensure consistency and the same cost-effectiveness level across sources. Therefore, EPA will use a Phase II ratio of 2.86-to-1 for all States affected by CAIR who choose to participate in the trading program.

Today, EPA is finalizing the general requirement that all SIPs must include a mechanism to ensure that excess SO 2 allowances are retired. Furthermore, for States that participate in the EPA-administered cap and trade program, EPA is finalizing a specific mechanism that States must use.

i. States Participating in the EPA-Administered SO 2 Trading Program

If a State chooses to participate in the EPA-administered trading program, the State's excess title IV allowance retirement mechanism must follow the provisions of the SO 2 model trading rule that requires that vintage 2010 through 2014 title IV allowances be retired at a ratio of two allowances for every ton of emissions and that vintage 2015 and beyond title IV allowances be retired at a ratio of 2.86 allowances for every ton of emissions. Pre-2010 vintage allowances would be retired at a ratio of one allowance for every ton of emissions. (See discussion of the model SO 2 cap and trade rule in section VIII of today's preamble.) States using the model SO 2 cap and trade rule satisfy the requirement for retirement of excess title IV allowances.

ii. States Not Participating in the EPA-Administered SO 2 Trading Program

In the CAIR NPR, EPA stated that if a State does not choose to participate in the EPA-administered trading programs but controls only EGUs, the State may choose the specific method to retire allowances in excess of its budget. The EPA considered alternative ways for retiring these excess allowances and, as stated in the CAIR SNPR, believed that the use by different States of different means to address this concern could undermine the regionwide emissions reduction goals of the CAIR rulemaking. The EPA further described its concerns in section II of the preamble to the CAIR SNPR. (See 69 FR 32686-32688.) Because of these concerns, in the CAIR SNPR, EPA withdrew the CAIR NPR proposal on this point and re-proposed that all States use a 2-for-1 retirement ratio for vintage 2010 through 2014 allowances and a 2.86-for-1 or a 3-for-1 retirement ratio for vintage 2015 and beyond allowances to address concerns about title IV allowances that exceed State budgets. The EGUs would have a total emissions cap enforced by the State.

The SNPR described that for sources affected by both title IV and CAIR, allowance deductions and associated compliance determinations would be sequential. That is, title IV compliance would be determined and then CAIR compliance would be determined. So, in 2010-2014, after surrendering one vintage 2010 through 2014 allowance for each ton of emissions for title IV compliance, the source would then surrender one additional allowance (for a total of two allowances for each ton which meets the CAIR requirement). Similarly, in 2015 and beyond, after surrendering one vintage 2015 and beyond allowance for each ton of emissions for title IV compliance, the source would surrender 1.86 or 2 additional allowances and therefore meet the CAIR requirement. Commenters argued that in States where EGUs are not trading under CAIR that the excess title IV allowances could be removed in a variety of ways and that EPA did not need to require each State do this the same way, only that each State ensure that they are removed.

Today, EPA adopts the following requirement: If a State does not choose to participate in the EPA-administered trading programs but controls only EGUs, the State must include in its SIP a mechanism for retiring the excess title IV allowances (i.e., the difference between total allowance allocations in the State and the State EGU SO 2 budget). To meet this requirement, the State may use the above-described retirement mechanism or may develop a different mechanism that will achieve the required retirement of excess allowances.

3. Requirements for States Choosing to Control Sources Other Than EGUs

a. Overview of Requirements

As noted in both the CAIR NPR and the CAIR SNPR, if a State chooses to require emissions reductions from non-EGUs, the State must adopt and submit SIP revisions and supporting documentation designed to quantify the amount of reductions from the non-EGU sources and to assure that the controls will achieve that amount. Although EPA did not propose in the CAIR NPR that States be required to impose an emissions cap on those sources, but instead solicited comment on the issue, EPA proposed in the CAIR SNPR that States be required to impose an emissions cap in certain cases on non-EGU sources. (See discussion in VII.A.1 of today's preamble.)

If a State chooses to obtain some, but not all, of its required reductions for SO 2 or NO X emissions from non-EGUs, it would still be required to set an EGU budget for SO 2 or NO X respectively, but it would set such a budget at some level higher than shown in Tables V-1, V-2, or V-4 in today's preamble, thus allowing more emissions from EGUs. The difference between the amount of a State's SO 2 budget in Table V-1 and a State's selected higher EGU SO 2 budget would be the amount of SO 2 emissions reductions the State demonstrates it will achieve from non-EGU sources. By the same token, the difference between the amount of a State's annual NO X budget in Table V-2 and a State's selected higher annual EGU NO X budget would be the amount of annual NO X emissions reductions the State demonstrates it will achieve from non-EGU sources. [109] Further, the difference between the amount of a State's seasonal NO X budget in Table V-4 and a State's selected higher ozone season EGU NO X budget would be the amount of ozone season NO X emissions reductions the State demonstrates it will achieve from non-EGU sources.

Special Concerns About SO 2 Allowances

In the case where a State requires a portion of its SO 2 emissions reductions from non-EGU sources and a portion from EGUs, there remains a concern about the impact of excess title IV allowances above a State's EGU cap, particularly on the operation of the title IV SO 2 cap and trade program. Consequently, today, we are adopting the requirement that these States include a mechanism for retirement of the allowances in excess of the State's SO 2 budget.

Like a State choosing to control only EGUs but not to participate in the trading program, a State that chooses to control non-EGUs and EGUs must adopt a mechanism for retiring surplus title IV allowances. The number of title IV allowances that must be retired is equal to the difference between the number of title IV allowances allocated to EGUs in that State and the SO 2 budget the State sets for EGUs under this rule. If the State uses a retirement mechanism (as discussed in VII.A.2.d.) in which a source surrendering allowances under the title IV SO 2 cap and trade program surrenders more allowances than otherwise required under title IV, the total number of allowances surrendered per ton of emissions in this case will be less than 2 to 1 in Phase 1 and less than 2.86 to 1 in Phase 2. This is because the non-EGUs will control to achieve a portion of the CAIR SO 2 reduction required, and so there will be a smaller surplus of title IV allowances than if all the required reductions were achieved by EGUs. The appropriate retirement factor will equal two times the State's SO 2 budget in Phase I or 2.86 times the State's SO 2 budget in Phase II as noted in Table V-1 of the budget section, divided by the State's selected higher EGU SO 2 budget (taking into account non-EGU reductions). The factor could then be used as the EGU retirement ratio for compliance purposes in a scenario where a State has decided to control SO 2 emissions from EGUs through a mechanism other than the EPA-administered trading program.

A simplified example can help illustrate this. Let us assume a State's sources were allocated a total of 200 allowances under title IV. Under CAIR, in Phase I, the State's reduction requirement would thus be 100 tons. Suppose this State decided that 25 tons would be reduced by non-EGUs and the remaining 75 tons would be reduced by the EGUs. (The State's budget for EGUS would increase to 125 tons.) The State would also need to retire 75 excess title IV allowances. This could be accomplished by requiring each Acid Rain source to surrender a total of 1.6 vintage 2010 through 2014 allowances (200 allowances allocated in the State/125 tons in State EGU budget) per ton of SO 2 emissions. The allowances surrendered would satisfy the Acid Rain Program requirement of surrendering one allowance per ton of emissions, as well as achieving the additional retirement requirement under CAIR since 200 allowances would be used for EGUs to emit the EGU budget of 125 tons of SO 2. (Pre-2010 allowances continue to be available for use on a one-allowance-per-ton-of-emissions basis here as in other situations.)

This is consistent with EPA's overall approach. If this same State decided to get all reductions (i.e., 100 tons) from EGUs, the State would require EGUs to retire 100 additional allowances by surrendering a total of 2 vintage 2010 through 2014 allowances (200 allowances allocated in the State/100 tons in State EGU budget) per ton of SO 2 emissions.

The demonstration of emissions reductions from non-EGUs is a critical requirement of the SIP revision due from a State that chooses to control non-EGUs. The State must take into account the amount of emissions attributable to the source category in both (i) the base case, in the implementation years 2010 and 2015, i.e., without assuming any SIP-required reductions under the CAIR from non-EGUs; and (ii) in the control case, in the implementation years 2010 and 2015, i.e., assuming SIP-required reductions under the CAIR from non-EGUs. We proposed an alternative methodology for calculating the base case for certain large non-EGU sources, as described below, but generally the difference between emissions in the base case and emissions in the control case equals the amount of emissions reductions that can be claimed from application of the controls on non-EGUs. (See discussion later in this section for criteria applicable to development of the baseline and projected control emissions inventories.)

States that meet the lesser of their CAIR ozone season NO X budget or NO X SIP Call EGU trading budget using the CAIR ozone season NO X trading program also satisfy their NO X SIP Call requirements for EGUs. States may also choose to include all of their NO X SIP Call non-EGUs in the CAIR ozone season NO X program at their NO X SIP Call levels (i.e., the non-EGU trading budget remains the same).

To the extent EPA allows through the Regional Haze Rule and a State then chooses to use EPA analysis to show that CAIR reductions from EGUs meet BART requirements, States that achieve a portion of their CAIR reductions from sources other than EGUs and wanting to show that even with those reductions the EGUs will meet BART requirements must make a supplemental demonstration that BART requirements are satisfied.

b. Eligibility of Non-EGU Reductions

In the CAIR SNPR, EPA proposed that, in evaluating whether emissions reductions from non-EGUs would count towards the emissions reductions required under the CAIR, States may only include reductions attributable to measures that are not otherwise required under the CAA. Specifically, EPA proposed that States must exclude non-EGU reductions attributable to measures otherwise required by the CAA, including: (1) Measures required by rules already in place at the date of promulgation of today's final rule, such as adopted State rules, SIP revisions approved by EPA, and settlement agreements; (2) measures adopted and implemented by EPA (or other Federal agencies) such as emissions reductions required pursuant to the Federal Motor Vehicle Control Program for mobile sources (vehicles or engines) or mobile source fuels, or pursuant to the requirements for National Emissions Standards for Hazardous Air Pollutants; and (3) specific measures which are mandated under the CAA (which may have been further defined by EPA rulemaking) based on the classification of an area which has been designated nonattainment for a NAAQS, such as vehicle inspection and maintenance programs.

In discussing this proposal, EPA noted that States required to make CAIR SIP submittals may also be required to make separate SIP submittals to meet other requirements applicable to non-EGUs, e.g., nonattainment SIPs required for areas designated nonattainment under the PM 2.5 or 8-hour ozone NAAQS or regional haze SIPs. The EPA noted it is likely that CAIR SIP submittals will be due before or at the same time as some of these other SIP submittals. We therefore proposed that States relying on reductions from controls on non-EGUs must commit in the CAIR SIP revisions to replace the emissions reductions attributable to any CAIR SIP measure if that measure is subsequently determined to be required to meet any other SIP requirement.

Some commenters objected to the proposed exclusion of credit for measures which are mandated under the CAA based on the classification of an area which has been designated nonattainment for a NAAQS, as well as to the proposed requirement that such measures must be replaced if they are later determined to be required in meeting separate SIP requirements. These commenters reasoned that such a requirement would not be applied to EGUs and would impose unnecessary and costly burdens on non-EGUs, thus creating an incentive for States to avoid controlling non-EGUs and to impose all CAIR reduction requirements on EGUs. One commenter further objected that, as long as a measure was not included in the base case EPA used to determine a State's contribution to other States' nonattainment under CAA section 110(a)(2)(D), there is no justification for excluding CAIR credit for such measure, and that EPA's proposed exclusion of credit for any measure “otherwise required by the CAA” is inconsistent with the NO X SIP Call.

In response to these comments, EPA agrees that it is not appropriate to apply this proposed restriction inconsistently to EGUs and non-EGUs. Thus, EPA is adopting a modified form of the proposed criteria for the eligibility of non-EGU emissions reductions, eliminating the requirement that States must exclude non-EGU reductions attributable to measures otherwise required by the CAA based on the classification of an area which has been designated nonattainment for a NAAQS. Consequently, the final rule allows credit for measures that a State later adopts in response to requirements which result from an area's nonattainment classification, such as reasonably available control technology (RACT). With this change, all emissions reductions are eligible for credit in meeting CAIR except: (1) Measures adopted or implemented by the State as of the date of promulgation of today's final rule, such as adopted State rules, SIP revisions approved by EPA, and settlement agreements; and (2) measures adopted or implemented by the Federal government (e.g., EPA or other Federal agencies) as of the date of submission of the SIP revision by the State to EPA, such as emissions reductions required pursuant to the Federal Motor Vehicle Control Program for mobile sources (vehicles or engines) or mobile source fuels, or pursuant to the requirements for National Emissions Standards for Hazardous Air Pollutants.

This exclusion of credit is consistent with EPA's approach in the NO X SIP Call, although a direct comparison of the creditability requirements in the CAIR and in the NO X SIP Call is not possible due to the timing and context in which both rules were developed. The NO X SIP Call used statewide budgets for all sources as an accounting tool to determine the adequacy of a strategy, while the CAIR takes a different approach in which baseline emission inventories for non-EGU sectors will, if needed, be developed later. The NO X SIP Call did, as does the CAIR, restrict States from taking credit for any Federal measures adopted after promulgation of the rule (63 FR 57427-28). It also did not allow credit for already adopted measures, but the timing of the NO X SIP Call was such that nonattainment planning measures would have already likely been adopted as the SIP deadlines for adoption of such measures had passed. In today's action, nonattainment planning measures adopted after the promulgation of today's rule will be allowed credit under CAIR.

In order to take credit for CAIR reductions from non-EGUs, the reductions must be beyond what is required under the NO X SIP Call. That is, a reduction must be in the non-ozone season or it must be beyond what is expected in the ozone season. Non-ozone season reductions must also be beyond what is in the base case, particularly for units that have low NO X burners and certain SCRs (e.g., ones required to be run annually). The reductions must be in addition to those already expected. If ozone season reductions are considered, the non-EGU NO X SIP Call trading budget must be adjusted by the increment of CAIR reductions beyond the levels in the NO X SIP Call. This removes the corresponding allowances from the market and ensures that the emissions do not shift to other sources.

After evaluating the eligibility of non-EGU reductions in accordance with the requirements discussed here, States must exclude credit for ineligible measures by (i) including such measures in both the baseline and controlled emissions inventory cases, if they have already been adopted; or (ii) excluding them from both the base and control emissions inventory cases if they have not yet been adopted. (See discussion later in this section regarding development of emissions inventories and demonstration of non-EGU reductions.)

c. Emissions Controls and Monitoring

As noted in section VII.A.1., we modified the “hybrid” approach described in the CAIR NPR as it applies to certain non-EGUs, and adopt today the approach described in the CAIR SNPR. Specifically, for States that choose to impose controls on large industrial boilers and turbines, i.e., those whose maximum design heat input is greater than 250 mmBtu/hr, to meet part or all of their emissions reductions requirements under the CAIR, State rules must include an emissions cap on all such sources in their State. Additionally, in this situation, States must require those large industrial boilers and turbines to meet part 75 requirements for monitoring and reporting emissions as well as recordkeeping. This ensures consistency in measurement and certainty of reductions and has been proven technologically and economically feasible in other programs.

If a State chooses to control non-EGUs other than large industrial boilers and turbines to obtain the required emissions reductions, the State must either (i) impose the same requirements, i.e., an emissions cap on total emissions from non-EGUs in the source category in the State and part 75 monitoring, reporting and recordkeeping requirements; or (ii) demonstrate why such requirements are not practicable. In the latter case, the State must adopt appropriate alternative requirements to ensure that emissions reductions are being achieved using methods that quantify those emissions reductions, to the extent practicable, with the same degree of assurance that reductions are being quantified for EGUs and non-EGU boilers and turbines using part 75 monitoring. This is to ensure that, regardless of how a State chooses to meet the CAIR emissions reduction requirements, all reductions made by States to comply with the CAIR have the same, high level of certainty as that achieved through the cap and trade approach. Further, if a State adopts alternative requirements that do not apply to all non-EGUs in a particular source category (defined to include all sources where any aspect of production of one or more such sources is reasonably interchangeable with that of one or more other such sources), the State must demonstrate that it has analyzed the potential for shifts in production from the regulated sources to unregulated or less stringently regulated sources in the same State as well as in other States and that the State is not including reductions attributable to sources that may shift emissions to such unregulated or less regulated sources.

d. Emissions Inventories and Demonstrating Reductions

To quantify emissions reductions attributable to controls on non-EGUs, the States must submit both baseline and projected control emissions inventories for the applicable implementation years. We have issued many guidance documents and tools for preparing such emissions inventories, some of which apply to specific sectors States may choose to control. [110] While much of that guidance is applicable to today's rulemaking, there are some key differences between quantification of emissions reduction requirements under a SIP designed to help achieve attainment with a NAAQS and emissions reduction requirements under a SIP designed to reduce emissions that contribute significantly to a downwind State's nonattainment problem or interfere with maintenance in a downwind State. Because States are taking actions as a result of their impact on other States, and because the impacted States have no authority to reduce emissions from other States, the emissions reduction estimates become even more important. (For a complete discussion, see 69 FR 32693; June 10, 2004.)

Specifically, when we review CAIR SIPs for approvability, we intend to review closely the emissions inventory projections for non-EGUs to evaluate whether emissions reduction estimates are correct. We intend to review the accuracy of baseline historical emissions for the subject sources, assumptions regarding activity and emissions growth between the baseline year and 2010 [111] and 2015, and assumptions about the effectiveness of control measures.

Before describing the specific steps involved in this quantification process, EPA notes that a few commenters objected to the proposed requirements as arbitrary restrictions intended to discourage States' discretion in imposing control measures on non-EGUs since these requirements would use what the commenters describe as extremely conservative emissions baseline and emissions reduction estimates. No commenter refuted EPA's explanation, noted above, of the need for stringent requirements to ensure greater accuracy of emission inventories and greater certainty of reduction estimates used in SIPs addressing transported pollutants. The EPA maintains that the need for more accurate inventories and more certain reduction estimates justifies the requirements discussed below. Further, no commenter provided an alternate method of addressing EPA's concerns about the development of such inventories and reduction estimates. Thus, EPA is finalizing its proposed approach.

i. Historical Baseline

To quantify non-EGU reductions, as the first step, a historical baseline must be established for emissions of SO 2 or NO X from the non-EGU source(s) in a recent year. The historical baseline inventory should represent actual emissions from the sources prior to the application of the controls. We expect that States will choose a representative year (or average of several years) during 2002-2005 for this purpose.

The requirements for estimating the historical baseline inventory that follow reflect EPA's view that, when States assign emissions reductions to non-EGU sources, achievement of those reductions should carry a high degree of certainty, just as EGU reductions can be quantified with a high degree of certainty in accordance with the applicable part 75 monitoring requirements. Because the non-EGU emissions reductions are estimated by subtracting controlled emissions from a projected baseline, if the historical baseline overestimates actual emissions, the estimated reductions could be higher than the actual reductions achieved.

For non-EGU sources that are subject to part 75 monitoring requirements, historical baselines must be derived from actual emissions obtained from part 75 monitored data. For non-EGU sources that do not have part 75 monitoring data, historical baselines must be established that estimate actual emissions in a way that matches or approaches as closely as possible the certainty provided by the part 75 measured data for EGUs. For these sources, States must estimate historical baseline emissions using source-specific or category-specific data and assumptions that ensure a source's or source category's actual emissions are not overestimated.

To determine the baseline for sources that do not have part 75 measured data, States must use emission factors that ensure that emissions are not overestimated (e.g., emission factors at the low end of a range when EPA guidance presents a range) or the State must provide additional information that shows with reasonable confidence that another value is more appropriate for estimating actual emissions. Other monitoring or stack testing data can be considered, but care must be taken not to overestimate baselines. If a production or utilization factor is part of the historical baseline emissions calculation, a factor that ensures that emissions are not overestimated must be used, or additional data must be provided. Similarly, if a control or rule effectiveness factor enters into the estimate of historical baseline emissions, such a factor must be realistic and supported by facts or analysis. For these factors, a high value (closer to 100 percent control and effectiveness) ensures that emissions are not overestimated.

ii. Projections of 2010 and 2015 Baselines

The second step in quantifying SO 2 or NO X emissions reductions for non-EGUs is to use the historical baseline emissions and project emissions that wou