SEPA
United States
Environmental
Protection
Agency
Office of Air Quality
Planning and Standard*
Research Triangle Park, NC 27711
EPA-450/4-90-002a .
DECEMBER 1990
AIR
REGIONAL OZONE MODELING
FOR
NORTHEAST TRANSPORT
(ROMNET)
Draft Report
-------
EPA-450/4-91-0023
REGIONAL OZONE MODELING
FOR
NORTHEAST TRANSPORT
(ROMNET)
Edited by
Norman C. Possiel*
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
Lenard B. Milich
Beverly R. Goodrich
Computer Sciences Corporation
P.O. Box 12767
Research Triangle Park, NC 27709
On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
December, 1990
-------
DISCLAIMER
This report has been reviewed by the Office of Air Quality Planning and Standards, United States Envi-
ronmental Protection Agency, and has been approved for publication. Any mention of trade names or
commercial products is not intended to constitute endorsement or recommendation for use.
-------
ACKNOWLEDGEMENTS
ROMNET was a joint project involving numerous organizations from within and outside of the U.S. Envi-
ronmental Protection Agency (EPA). Individuals participating in the project are listed in Appendix B. The
authors and editors wish to express their sincere appreciation for the efforts of all participants during the
project. In addition, the following groups and individuals deserve special recognition. As a conse-
quence of the management direction and technical information provided by representives of the Sta-
te/local agencies, the Northeast States for Coordinated Air Use Management, and the EPA Regional
Offices, the results of ROMNET successfully meet the project goals for enabling States to quantify
contributions from regional and interurban pollutant transport in the development of ozone State Imple-
mentation Plans.
The cornerstone of ROMNET was the EPA Regional Oxidant Model (ROM), which was used to address
a number of long-standing strategic ozone nonattainment issues. Ms. Joan Novak, EPA, was instru-
mental in providing leadership and guidance in the use of this model. The hundreds of ROM applica-
tions necessary to simulate the ROMNET strategies, as well as the creation of graphics used to analyze
and convey the results, were skillfully performed by members of the Computer Sciences Corporation
(CSC) "ROM Team" under the direction of Mr. Richard Wayland. An important contribution to the analy-
sis and interpretation of the ROM applications was made by Mr. William Cox, EPA. Finally, the tireless
efforts of Ms. Cynthia Baines and Ms. Lillian Faison, EPA, and Ms. Christine Bullock and Ms. Christine
Maxwell, CSC, in typing this report are greatly appreciated.
in
-------
This page is intentionally left blank.
-------
TABLE OF CONTENTS
Disclaimer ii
Acknowledgements iii
Table of Contents v
Table of Figures viii
Table of Tables xii
Executive Summary ES-1
Introduction ES-3
Purpose and Intent of ROMNET ES-3
Project Scope and Organization ES-4
The ROMNET Technical Program ES-5
The Regional Oxidant Model ES-5
Procedure to Address Strategic Issues ES-6
Episode Selection ES-6
ROMNET Base Case Inventory ES-7
ROM Evaluation ES-8
Future Baseline Emissions ES-8
Strategic Issues And Simulated Control Strategies ES-9
Strategic Issues ES-9
Findings for Strategy Simulations ES-10
Caveats ES-12
Technology Transfer ES-13
Need for Technology Transfer ES-13
Gridded Model Information Support System (GMISS) ES-14
ROM/UAM Interface Program System (Interface) ES-14
Accomplishments of the ROMNET Project ES-14
1 Project Goals and Scope 1-1
1.1 Purpose and Intent of ROMNET 1-3
1.2 Structure of the Final Report 1-3
1.3 Background to ROMNET 1-4
1.4 Rationale for the Goals and Scope of ROMNET 1 -5
1.5 ROMNET Organizational Structure 1-7
1.6 Overview of the ROMNET Technical Program 1-8
2 The ROMNET Modeling System 2-1
2.1 Introduction 2-3
-------
2.2 General Characteristics of the Regional Oxidant Model (ROM) 2-3
2.2.1 Physical Processes Within Layers 1, 2 and 3 2-3
2.2.2 ROM Chemistry 2-5
2.2.3 System Components 2-6
2.2.4 Quality Assurance Procedures for ROM Data Sets 2-9
2.2.5 ROM Simulations for ROMNET 2-9
2.2.6 ROM Limitations for ROMNET 2-10
2.3 Evaluation of ROM2.1 2-11
2.3.1 Introduction 2-11
2.3.2 Description of Evaluation Episodes 2-11
2.3.3 Database Development 2-12
2.3.4 Ambient Evaluation 2-14
2.3.5 Summary 2-22
3 Ozone Episode Selection 3-1
3.1 Introduction 3-3
3.2 Technical Approach 3-3
3.2.1 Initial Ambient Data Review 3-3
3.2.2 Episode Selection Criteria 3-4
3.2.3 Episode Ranking Scheme 3-4
3.2.4 Trajectory/Meteorology Analyses 3-6
3.2.5 Flow Regimes During the Top Ten Episodes 3-8
3.2.6 Episodes Selected for ROM Simulations 3-10
4 Emissions Scenario Development 4-1
4.1 Introduction 4-3
4.2 Overview and Structure of Emissions Inventories 4-3
4.2.1 Features of the ROMNET Inventories 4-3
4.2.2 Emissions Allocation Methodologies 4-4
4.3 1985 ROMNET Base Case Emissions Development 4-6
4.3.1 Parent Emissions Inventories 4-6
4.3.2 Adjustments to the NAPAP Inventories 4-6
4.3.3 Characteristics of the 1985 Base Case Emissions 4-11
4.4 Projection Year and Control Strategy Scenarios 4-15
4.4.1 Overview of Future Baseline Scenarios 4-15
4.4.2 Overview of the Control Strategy Scenarios , 4-17
4.5 Emissions Control Measures 4-27
4.5.1 VOC Stationary-Source Controls for 2005 Baseline Scenarios 4-27
4.5.2 NOX Stationary-Source Controls for 2005 Baseline Scenarios 4-30
VI
-------
4.5.3 VOC, NOX, and CO 2005 Baseline Mobile Source Controls 4-32
4.5.4 VOC Maximum Technology Stationary Source Controls 4-33
4.5.5 NOX Maximum Technology Stationary Source Controls 4-34
4.5.6 Mobile-Source VOC, NOX, and CO Maximum Technology Controls 4-34
4.5.7 Controls Applied in the CS05 Clean Air Act Strategy 4-35
4.5.8 Development of Reactivity-Reduction Scenarios 4-37
4.6 Development of A Biogenic-Source Emissions Inventory 4-39
4.6.1 Introduction 4-39
4.6.2 Description of the System 4-39
4.6.3 Computational Processes 4-43
4.6.4 Characteristics of the Biogenic VOC Emissions Inventory 4-44
5 Regional Modeling Results and Project Findings 5-1
5.1 Introduction 5-3
5.2 1985 Base Case and 2005 Baseline Predictions 5-3
5.2.1 1985 Base Case Ozone Predictions - July 1988 Episode 5-3
5.2.2 2005 Baseline Ozone - July 1988 Episode 5-4
5.3 Results of Control Strategy Simulations for Key Issues 5-5
5.4 Summary of Major Findings To Key Strategy Issues 5-39
6 State Access and Use of ROM Databases 6-1
6.1 Introduction 6-3
6.2 The Gridded Model Information Support System (GMISS) 6-5
6.2.1 Overview of GMISS 6-5
6.2.2 GMISS Databases 6-5
6.2.3 System Functional Description 6-7
6.2.4 User Protocol for Data Retrieval 6-7
6.3 The Regional Oxidant Model - Urban Airshed Model Interface Program System 6-10
6.3.1 Introduction 6-10
6.3.2 Features and Limitations of the Interface 6-11
6.3.3 Description of the Interface System 6-13
References and Bibliography R-1
VII
-------
TABLE OF FIGURES
ES-1 The ROMNET region ES-16
ES-2 ROMNET management structure ES-17
ES-3 Components of the ROM ES-18
1-1 The Northeast Corridor and other Metropolitan Statistical Areas in the ROMNET
region 1-11
1-2 Observed maximum 1-hour ozone concentrations for July 2-17, 1988 across the
ROMNET region 1-12
1-3 ROMNET management structure 1-13
2-1 The ROMNET modeling domain 2-25
2-2 The ROM vertical layers and their functional features 2-26
2-3 Components of the ROM 2-27
2-4 Percentage of monitors with daily maximum ozone exceeding 120 ppb in the north-
eastern U.S 2-28
2-5 ROM grid points overlaying the UAM domain for the New York City metropolitan area
2-28
2-6 AIRS ozone monitoring sites divided into five geographical groups 2-29
2-7 Monitoring sites used for developing boundary conditions for the OMNYMAP domain
2-29
2-8 Quantile-quantile plots of daytime hourly ozone for the July 1985 episode 2-30
2-9 Comparison of observed and modeled maximum ozone for the July 1985 episode 2-31
2-10 Spatial patterns of maximum ozone for July 9-11,1985 2-32
2-11 Spatial patterns of maximum ozone for July 13 -15,1985 2-33
2-12 Spatial patterns of maximum ozone for July 18 - 20,1985 2-34
2-13 Spatial patterns of maximum ozone for August 13-15,1985 2-35
2-14 Mean residuals versus wind persistence for the UAM boundary 2-36
2-15 Mean residuals versus daily average wind direction for the UAM boundary 2-36
2-16 Division of the OMNYMAP boundary into eight groups 2-37
2-17 Mean daytime ozone concentrations by UAM group along the UAM boundary 2-38
2-18 Mean residuals for each UAM group experiencing incoming flow 2-38
2-19 Ozone concentrations for 0800 EST July 10,1985 2-39
2-20 Ozone concentrations for 1000 EST July 10,1985 2-40
2-21 Ozone concentrations for 1200 EST July 10, 1985 2-41
2-22 Ozone concentrations for 1400 EST July 10,1985 2-42
2-23 Ozone concentrations for 1600 EST July 10,1985 2-43
2-24 Time series of selected meteorological parameters for grid cell (45,21) on
July 10,1985 2-44
viii
-------
3-1 Box-plot time series showing the frequency distribution of daily maximum ozone
concentrations - May through September, 1988 3-11
3-2 Subregions of the Northeast Corridor used for aggregating observed data in the
episode selection process 3-12
4-1 Summary of software modules and input data used in FREDS 4-46
4-2 Effect of average daily temperature on VOC emissions 4-47
4-3 Effect of average daily temperature range on VOC emissions 4-47
4-4 Distribution of regionwide 1985 VOC emissions 4-48
4-5 Distribution of regionwide 1985 NOX emissions 4-49
4-6 1985 base case anthropogenic emissions of VOC and NOX 4-50
4-7 1985 base case anthropogenic emissions of NOx from industrial plants and utilities ... 4-51
4-8 1985 base case emissions by day type 4-52
4-9 Diurnal profiles of point, area, and mobile sources of VOC and NOX 4-53
4-10 Diurnal profiles of point, area, and mobile sources of CO 4-54
4-11 Mobile-source emissions on a "cool" day and a "warm" day 4-55
4-12 Regionwide daily total mobile-source VOC and NOX emissions, July 1988 episode 4-56
4-13 Regionwide daily total mobile-source CO emissions, July 1988 episode 4-57
4-14 VOC emissions for the Phase II scenarios and the percent reduction from the 1985
base case and the 2005 baseline 4-58
4-15 NOX emissions for the Phase II scenarios and the percent reduction from the 1985
base case and the 2005 baseline 4-59
4-16 CO emissions for the Phase II scenarios and the percent reduction from the 1985 base
case and the 2005 baseline 4-60
4-17 The Northeast Corridor and nonattainment areas outside the Corridor 4-61
4-18 Location of major NOX point sources controlled in the maximum technology strategy
4-62
4-19 Relationship between meteorological temperature and (1) isoprene and alpha-pinene
emissions, (2) monoterpenes/unidentified hydrocarbon emissions 4-63
4-20 Flowchart of the Biogenic Emissions Inventory System 4-64
4-21 Biogenic isoprene emissions for 1000 and 1500 EST on a 'cool" day 4-65
4-22 Biogenic isoprene emissions for 1000 and 1500 EST on a "warm" day 4-66
4-23 Regionwide daily total biogenic VOC emissions, July 1988 episode 4-67
5-1 Predicted 1985 base case episode maximum 1 -hour ozone concentrations for the July
1988 episode 5-41
5-2 Predicted 2005 baseline episode maximum 1-hour ozone concentrations for the July
1988 episode 5-42
5-3 Areas in the ROMNET domain used in calculating selected metrics 5-43
5-4 Anthropogenic VOC emissions and NOX emissions for the U.S. portion of the
ROMNET region 5-44
IX
-------
5-5 Percent change in VOC emissions between CS12 and the 2005 baseline for anthro-
pogenic emissions only and anthropogenic plus biogenic emissions 5-45
5-6 Percent reduction in NOX emissions between CS11 and the 2005 baseline 5-46
5-7 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
2005 baseline and CS11 5-47
5-8 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
2005 baseline and CS12 5-48
5-9 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
CS13, CS14,andCS10 5-49
5-10 Quantile-quantile frequency distributions of 1 -hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for Baltimore/Washington, DC and vicinity 5-51
5-11 Quantile-quantile frequency distributions of 1 -hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for New York City 5-52
5-12 Quantile-quantile frequency distributions of 1 -hour daily maximum ozone and episode
. mean 8-hour daily maximum ozone for Greater Connecticut 5-53
5-13 Diurnal time series of maximum hourly ozone concentrations in New York City and
Greater Connecticut for CS11 andCS12 5-54
5-14 Diurnal time series of maximum hourly ozone concentrations in Baltimore/Washing-
ton, DC and vicinity and Pittsburgh and vicinity for CS11 and CS12 5-55
5-15 Predicted daily maximum ozone concentrations across the New York City area for July
9, 1988: 2005 baseline and the change in ozone following the application of NOX
controls in CS11 5-56
5-16 Morning total VOC and NOX emissions (tons) on July 9, 1988 in the vicinity of New
York City 5-57
5-17 Population exposure to 1-hour ozone > 100 ppb in Boston and coastal New England
for selected scenarios 5-58
5-18 Population exposure to 1-hour ozone > 125 ppb in Baltimore/Washington, DC and
Philadelphia for selected scenarios 5-59
5-19 Population exposure to 1-hour ozone > 125 ppb in New York City and Greater
Connecticut/Rhode Island for selected scenarios 5-60
5-20 Population exposure to 1-hour ozone > 125 ppb in Cleveland/Detroit (combined) and
Pittsburgh and Charleston, WV (combined) for selected scenarios
5-61
5-21 Location of the boundary to the Northeast Corridor used in quantifying incoming
ozone transport 5-62
5-22 Layer 2 forward trajectories starting at 1500 EST from locations along the boundary to
the Northeast Corridor for July 9,1988 and July 10,1988 5-63
5-23 Peak and mean ozone concentrations within the three Northeast Corridor boundary
segments 5-64
5-24 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
2005 baseline and CS25 5-65
5-25 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
CS19andCS24 5-66
-------
5-26 ROM layer 2 trajectory for the transport case study 5-67
5-27 Time history of layer 2 ozone, NOX, and ROG concentrations: 2005 baseline and CS25
5-68
5-28 Time history of layer 2 ozone, NOX, and ROG concentrations: CS19 and CS24 5-69
5-29 Emissions for Phase II scenarios, and predicted highest and second-highest daily
maximum ozone concentrations for selected urban areas 5-70
5-30 Predicted episode maximum 1 -hour ozone concentrations for the June 1983 episode:
2005 baseline and CS19 5-74
5-31 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
Phase 11985 base case with 'best estimate* biogenics and 1985 base case with "low"
biogenics 5-75
5-32 Predicted episode maximum 1 -hour ozone concentrations for the July 1988 episode:
Phase 11985 base case with "best estimate" biogenics and 1985 base case with "high"
biogenics 5-76
5-33 Predicted episode maximum 1-hour ozone concentrations for biogenic sensitivity
scenarios for selected urban areas 5-77
5-34 Predicted grid-hours exceeding 125 ppb for biogenic sensitivity scenarios for selected
urban areas 5-79
5-35 Predicted population exposure to ozone exceeding 125 ppb for biogenic sensitivity
scenarios for selected urban areas 5-81
6-1 Flow diagram showing the data retrieval and interface processing steps performed to
generate data files for the (JAM preprocessors and model 6-15
XI
-------
TABLE OF TABLES
ES-1 Organizations participating in ROMNET ES-19
ES-2 ROMNET control strategies ES-20
1-1 Organizations participating in ROMNET 1-14
2-1 ROM chemical species 2-45
2-2 Initial mean tropospheric background concentrations, precursor species 2-45
2-3 Ozone monitoring sites used for developing UAM boundary conditions along the
OMNYMAP domain 2-46
2-4 Summary statistics for all daytime hourly and daily maximum ozone concentrations for
the ROMNET region (July and August 1985 data merged) 2-47
2-5 Comparison of daytime hourly and daily maximum statistics for the July and August
1985 episodes 2-47
2-6 Statistical summary by geographical group for the July 1985 episode: daytime hourly
and daily maximum concentrations 2-48
2-7 Summary statistics for UAM boundary conditions for the July and August 1985
episodes 2-49
2-8 Daily summaries of UAM boundary conditions 2-49
3-1 Candidate episodes: 1983-1988 3-13
3-2 Summary ranking of 27 candidate episodes 3-14
3-3 Wfthin-Corridor flow regimes 3-15
4-1 ROMNET emissions scenarios 4-68
4-2 Revised summer allocation factors for area-source categories 4-69
4-3 Summary of control measures in the baseline projection and maximum technology
inventories 4-70
4-4 Control measures for the draft Clean Air Act analysis 4-71
4-5 NSPS efficiencies for point and area sources 4-72
4-6 Summary of State estimates for existing area-source controls 4-73
4-7 Summary of NOX control measures for Phase I 4-74
4-8 Maximum technology for point sources 4-75
4-9 Maximum technology for area sources 4-76
4-10 Maximum technology efficiencies for NOX 4-77
4-11 Strategy 5 point-source control efficiencies by attainment category 4-78
4-12 Strategy 5 area-source control efficiencies by attainment category 4-79
4-13 Strategy 5 attainment categories and across-the-board VOC reductions 4-80
4-14 VOC speciation factors for methanol vehicles 4-81
xii
-------
4-15 Emission fluxes for vegetation types in the Biogenic Emissions Inventory System 4-81
5-1 Procedures used for calculating population exposure 5-83
5-2 Summary of emissions reductions that reduced predicted ozone to < 125 ppb 5-84
5-3 Comparison of the July 1988 and June 1983 episode maximum 1 -hour concentrations
for the 2005 baseline and CS19 5-85
5-4 Summary of ozone metrics used to assess the impacts of reactivity-based strategies
5-86
5-5 Cases tested with varying biogenic emissions 5-87
6-1 Summary of the DAM preprocessor programs 6-16
6-2 Retrieved Interface files used by interface programs 6-16
6-3 Overview of the ROM-UAM Interface programs and input/output files 6-17
XIII
-------
This page is intentionally left blank.
-------
EXECUTIVE SUMMARY
by
Edwin L Meyer
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
-------
This page is intentionally left blank.
-------
INTRODUCTION
Purpose and Intent of ROMNET
The Regional Ozone Modeling for Northeast Transport (ROMNET) Project was initiated by the U.S.
Environmental Protection Agency (EPA) and State and local agencies in the Northeast, as part of a joint
effort to address the problem of regional transport in developing effective and equitable control pro-
grams to attain the ozone National Ambient Air Quality Standards (NAAQS) in this region.1
The specific goals of ROMNET are:
1. to evaluate the relative effectiveness of regional controls on ozone levels in the Northeast;
2. to provide quantitative estimates of ozone and precursor concentrations transported between urban
areas following the application of regional control measures; and
3. to provide procedures and guidance for incorporating ozone and precursor transport in future State
Implementation Plan (SIP) development.
The intent of ROMNET is for EPA and State decision-makers to use the findings in guiding ozone-policy
development and planning for potential regional control programs, as well as in urban area-specific
strategy evaluations. The urban-scale analyses are to be conducted by the States as part of the SIP
process. The results and guidance from the project will enable States to quantify changes in future
levels of ozone and precursor transport expected to follow implementation of nationwide Federal mea-
sures, local control programs in upwind cities, and potential regional strategies.
The remainder of the Executive Summary provides a brief overview of the major components of the
project including:
a. the project scope and organization;
b. the ROMNET technical program;
c. the findings from ROM simulations conducted to address "strategic" issues regarding the effective-
ness of regional control scenarios;
d. the mechanisms and procedures developed to enable States to explicitly consider regional and
interurban transport in preparing ozone (SIP's); and
e. a summary of the major accomplishments of the project.
1. An exceedence of the NAAQS is defined as a daily maximum 1-hour ozone concentration > 0.12 ppm (124 ppb). The NAAQS
is not to be exceeded more than once per year at any location.
ES-3
-------
Readers interested in a detailed description of these topics and other aspects of ROMNET are referred
to the main sections of the final report. Figures and tables referred to herein are provided at the end of
the Executive Summary.
Project Scope and Organization
A fundamental concept of ROMNET was the active involvement of State and local agencies with the EPA
in designing and conducting the project. This began with the development of a protocol by EPA and
State representatives. The protocol established the project goals, direction, organizational structure,
technical tasks, and schedule. The ROMNET protocol is included in Appendix A of this report. The
organizations participating in ROMNET are listed in Table ES-1. The geographic area covered by the
project is shown in Figure ES-1.
The participation by States in the management and technical aspects of ROMNET was fostered by the
organizational structure shown in Figure ES-2. This structure provided a means for States and other
participants to have a major role in the project decision-making process, and for communicating infor-
mation on project activities to all participants.
The Management Review Committee consisted of senior managers from State/local agencies and EPA
Regional Offices, the Office of Air Quality Planning and Standards (OAQPS), and the Atmospheric
Research and Exposure Assessment Laboratory (AREAL). The responsibility of this committee was to
provide overall direction for the project. The three technical committees consisted of representatives
from the participating organizations having expertise in one or more of the technical areas.
To ensure close coordination among the committees, the Program Director, Technical Coordinator, and
Chair of the Management Review Committee and three technical committees were drawn from OAQPS.
The Program Director had the overall responsibility of ensuring the technical adequacy of the study,
making sure that participants were informed of the project's status, and that the project was as respon-
sive as feasible to directions from the Management Review Committee. The Technical Coordinator
assisted the Program Director by managing day-to-day activities of the project, establishing priorities for
a diverse set of technical tasks, making sure these priorities were observed, and ensuring that technical
activities were integrated appropriately.
Decisions by the committees were made through a consensus building process. For technical issues,
straw man proposals were typically formulated by the EPA for discussion by committee members.
ES-4
-------
Consensus recommendations from the technical committees on issues such as strategy selection were
made to the Management Review Committee which, through consensus, provided resolution or direc-
tion on the particular topic.
THE ROM NET TECHNICAL PROGRAM
The ROMNET technical program consisted of three major components:
1. simulations using the Regional Oxidant Model (ROM) for various emissions scenarios designed to
address 'strategic* issues relative to the impacts of regional control measures on ozone levels in the
Northeast;
2. the development of the Gridded Model Information Support System (GMISS) to manage ROM data
sets and provide a means for States to easily access and retrieve data necessary to develop trans-
port estimates and certain other inputs required for urban scale SIP modeling; and
3. the development of the ROM/Urban Airshed Model (UAM) Program System which, through a series
of computer programs, translates and reformats ROM data sets into inputs to the UAM.
The remainder of this part of the Executive Summary provides an overview of the ROM and the proce-
dures followed for simulating the impacts of control strategies. The results of strategy simulations are
presented later, as is a brief description of GMISS and the ROM/DAM Interface.
The Regional Oxidant Model
The Regional Oxidant Model (ROM), Version 2.1, (Milich et a/., 1990) was used for estimating ozone
concentrations across the Northeast under various emissions scenarios selected for evaluation as part
of ROMNET. ROM is a three-dimensional photochemical grid model having horizontal grid resolution of
1/6° of latitude by 1/4° of longitude (approximately 18.5 km x 18.5 km in the Northeast). Vertically, there
are three layers of cells between the ground and the base of a subsidence inversion serving as a vertical
lid on pollutant dispersion during fair weather conditions. Dimensions of the three vertical layers vary
diurnally in response to physical processes occurring in the atmosphere.
As shown in Figure ES-3, five broad classes of input information are needed to drive the ROM model.
The ROM Processor Network is a series of computer programs (i.e., preprocessors) that convert raw
input data into the variables required by the core model. The ROM core model performs the horizontal
and vertical transport and chemistry calculations that estimate ozone concentrations (and concentra-
tions for 34 other chemical species) for each grid cell in the domain. Chemistry assumed in the core
model is described by the Carbon Bond 4 (CB-IV) mechanism. This mechanism is identical to the one
ES-5
-------
contained in the Urban Airshed Model, recommended by the EPA for use in SIP attainment demonstra-
tions. The ROM generates hourly concentrations that represent grid cell average values. A more
detailed discussion of the ROM is contained in Section 2.
Procedure to Address Strategic Issues
The following steps were followed to address strategic issues on the relative effectiveness of control
strategies:
1. meteorological episodes were selected for simulation by the ROM;
2. a base case inventory was developed, and an evaluation of the ROM was performed by comparing
predicted base case ozone levels with observed values;
3. base case emissions were projected to an agreed upon future year, and "future baseline0 predic-
tions were simulated;
4. control measures were applied to future baseline emissions to derive control strategy (CS) scenar-
ios, and postcontrol ozone levels were simulated; and
5. predictions for the different strategies and the future baseline scenario were analyzed and inter-
compared relative to the strategic issues.
Steps 1 -3 are described next, followed by the strategic issues, the associated emissions control strate-
gies, and the conclusions drawn from ROM simulations.
Episode Selection
Two factors were considered in selecting episodes to simulate with ROM: ambient ozone concentra-
tions and wind flow patterns. Since the primary focus of the study was the Northeast Corridor, the
selection of episodes focused on conditions in this portion of the domain. First, ozone data reported in
the EPA's Aerometric Information and Reporting System (AIRS) were reviewed for the years 1983-88 to
identify episodes having at least three consecutive days with ozone levels > 125 ppb within the Corri-
dor. The 27 episodes identified by this process were then ranked in terms of the magnitude of daily
maximum ozone levels and the frequency of high concentrations.
The ten top-ranked episodes were next examined to ascertain the occurrence of key wind flow patterns.
The goal was to choose episodes featuring a variety of wind flow regimes typically associated with high
ozone including, in order of priority, along Corridor transport, stagnation/recirculation, and westerly wind
flow.
ES-6
-------
The period July 2-17, 1988, was identified by the above process as the most severe in terms of ozone
i
concentrations and as having a high frequency of along Corridor and westerly wind flow patterns. Thus,
all control strategies were simulated with this episode. A second episode, June 9-20, 1983, also con-
tained high ozone levels but was characterized by recirculation conditions. This period was chosen for
simulating with a subset of emissions scenarios in order to test the robustness of conclusions drawn
from simulations with the July 1988 episode. In addition, the two most highly ranked 1985 episodes,
August 7-16,1985, and July 7-22,1985, were chosen for a ROM performance evaluation. A more com-
plete discussion of the episodes selected and the selection process is given in Section 3.
ROMNET Base Case Inventory
Inventories for volatile organic compounds (VOC), nitrogen oxides (NOX), and carbon monoxide (CO)
used in ROMNET were derived from the 1985 inventories compiled as part of the National Acid Precipi-
tation Assessment Program (NAPAP) (see Section 4). The following modifications were made to the
NAPAP 1985 inventories as part of ROMNET:
1. Biogenic emissions were calculated and included in ROM simulations. These are of greatest
importance for VOC, and reflect emissions from vegetation. The EPA Biogenic Emissions Inventory
System (BEIS) was used to derive hourly biogenic emissions estimates for the ROMNET domain.
2. The MOBILE4 model was used to obtain estimates for mobile-source exhaust and evaporative
emissions, as opposed to an earlier version of the MOBILE model used in NAPAP. MOBILE4 con-
siders such recently-identified phenomena as evaporative running losses, and also contains an
improved treatment of mobile-source emissions as a function of temperature.
Both mobile evaporative emissions and biogenic emissions can be very temperature sensitive. Emis-
sions of certain biogenic species are also sensitive to sunlight intensity. Since mobile and biogenic
emissions constitute major portions of the VOC inventory, day-specific meteorological data were used to
derive gridded hourly emissions for these categories. In addition, emission factors and/or activity levels
from six stationary source VOC categories were adjusted to reflect higher temperatures and enhanced
activity levels likely on typical sunny and warm episode days.
A more complete description of the 1985 base ROMNET inventory is contained in Section 4.
ES-7
-------
ROM Evaluation
Predictions obtained with the ROM using the 1985 base inventory described above were compared with
ozone observations made on 26 days during the two selected 1985 episodes. Three types of tests were
performed:
1. unpaired aggregate distributions of observations and predictions were compared for each of five
subregions of the domain;
2. predicted and observed spatial patterns for daily maximum ozone were compared through inspec-
tion of contours of observed and predicted concentrations; and
3. the model's ability to specify boundary concentrations for an urban scale domain was assessed by
comparing boundary condition ozone concentrations calculated from ROM predictions with con-
centrations derived subjectively from ambient ozone measurements.
Results of the first test indicate agreement within ±10 percent between the predicted upper 50th per-
centile of the distributions (except for the very highest values) and the observed distributions, except in
the southwestern portion of the domain. In this area, there was a tendency to underpredict the higher
end of the observed distributions. Also, there was a uniform tendency to overpredict the lowest end of
the observed distributions. A more detailed discussion of the ROM evaluation is provided in Section 2.
Based on the findings for all of the tests, ROM performance appears satisfactory for estimating episodic
ozone levels for use in assessing regional scale concentrations and patterns. Results tend to confirm a
previously held view that the simulation of peak urban ozone concentrations and strategies to reduce
them are best performed with urban scale models.
Future Baseline Emissions
The year 2005 was selected for the "future baseline* scenario. This year represents a period far enough
in the future to allow for widespread penetration of new control measures, and it is within the bounds of
credible growth forecasts.
The 2005 baseline reflects composite effects of growth, retirement of old sources, and the impact of
control programs in place or committed to as of 1988. Emissions projected to 2005 also consider the
best assessment of State participants concerning rule effectiveness and penetration for existing control
measures. In essence, the 2005 baseline addresses the question: 'What might happen in 2005 under
existing or already planned control programs, if no new measures were added?'
ES-8
-------
Regionwide reductions in VOC, NOXl and CO emissions between 1985 and the 2005 baseline were 35
percent, 4 percent, and 45 percent, respectively. Within the Corridor, the net change in VOC and NOX
emissions between 1985 and 2005 was 34 percent and 6 percent, respectively. Section 4 contains more
detailed specifications on the development of the 2005 baseline emissions scenario.
Results from applying the ROM for the 2005 baseline using the July, 1988, episode indicate that present
control measures (i.e., pre-1990 Clean Air Act amendments) will be insufficient to reduce maximum daily
ozone to below 125 ppb everywhere in the domain. In the Corridor, episode maximum ozone concen-
trations are reduced by 5 to 15 percent, with predicted peak values remaining over 200 ppb in and near
New York City. In the western part of the domain, ozone is also reduced by 5 to 15 percent near several
major cities. However, the areal extent of predicted episode maxima > 125 ppb has been reduced (see
Section 5 for greater detail).
STRATEGIC ISSUES AND SIMULATED CONTROL STRATEGIES
Strategic Issues
The ROMNET control strategies were formulated to examine five strategic issues:
1. What are the relative benefits of VOC and NOX controls in reducing required ozone levels?
2. What is the impact of reducing regional transport on ozone concentrations in the Northeast Corri-
dor?
3. What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted ozone lev-
els in the Northeast to below 125 ppb?
4. What are the effects of reactivity-based strategies in reducing regional ozone levels?
5. How does uncertainty in biogenic emissions affect conclusions drawn about the benefits of control-
ling anthropogenic sources?
Twenty-four control strategies were simulated to address these issues. Table ES-2 summarizes the
strategies, and notes the corresponding strategic issues.
Simulations for the 2005 baseline and CS19 were repeated with the June 9-20,1983, episode to note the
sensitivity of conclusions relative to differing meteorological conditions. Details of the control measures
in each strategy, as well as the resulting net changes in VOC, NOX, and CO emissions, are provided in
Section 4.
ES-9
-------
Findings for Strategy Simulations
Conclusions drawn relative to the five strategic issues addressed by ROMNET control strategies are
summarized below. The analyses which led to these findings are contained in Section 5 of the report.
Issue #1
What are the relative benefits of VOC controls versus NOX controls in reducing ozone levels across
the region?
Maximum technology NOX controls appear to produce larger ozone reductions than stringent
VOC controls in many areas of the Northeast. This is particularly notable in the western por-
tion of the region, where biogenic VOC emissions are highest and a large component of NOX
emissions are rural point sources. In the Corridor, except for New York City, NOX controls tend
to reduce the spatial extent of high ozone levels, whereas VOC controls are more effective in
reducing the peak values. Combining VOC and NOX controls provides both benefits by
reducing the magnitude and spatial extent of high ozone concentrations.
In the presence of stringent VOC controls, peak ozone levels appear to be more sensitive to
mobile- rather than point-source NOX emissions reductions in several cities (e.g., Philadelphia,
Boston, Detroit, and Cleveland). In areas dominated by NOX point sources, (e.g., Pittsburgh
and Charleston, WV), point-source controls are more effective than mobile-source reductions.
In New York City, NOX controls alone or with VOC controls are counterproductive relative to
VOC controls for short-term peak concentrations and population exposure. This effect is
most prominent in the core of the urban plume and appears to be associated with high NOX
emissions in the center of the urban area. Outside the main plume, ozone levels are actually
reduced by NOX controls. For longer averaging times (e.g., episode mean 8-hour daily maxi-
mum averages) NOX plus VOC controls produce lower ozone levels than VOC controls alone.
. The relative benefit of VOC versus NOX controls varies by day (i.e., as a function of meteorol-
ogy) in some cities. This was particularly evident in the Baltimore/Washington, DC area.
Issue #2
What is the impact of reducing regional transport on Northeast Corridor ozone levels?
In the precontrol scenario, 2005 baseline ozone levels are not substantially impacted by
reducing transport into the Corridor except in Baltimore/Washington, DC, and in southeast
Pennsylvania. These areas are closest to the portion of the Corridor boundary having the
greatest incoming ozone levels during this episode. It is possible that conditions in other
episodes might produce a greater impacts in other portions of the Corridor. Still, even in
these two areas, the impacts are small relative to the level of predicted ozone for the 2005
baseline.
ES-10
-------
In the postcontrol scenario, the results suggest that without stringent upwind controls, ozone
levels in parts of the Corridor may not be reduced to below the concentration specified in the
NAAQS even with stringent controls along the entire length of the Corridor. Again, the effects
are most pronounced for cities near the upwind boundary of the Corridor.
\
Issue #3
What levels of VOC and/or NOx emissions reduction are necessary to reduce predicted ozone
concentrations in the Northeast to below 125 ppb?
• Stringent maximum technology VOC and NOX controls may be necessary in all areas of the
Northeast Corridor.
. Additional reductions of VOC on the order of 64 percent in New York City and 54 percent in
Baltimore/Washington, DC also may be needed. At present, this level of emissions reductions
is beyond known or envisaged control technologies.
The effectiveness of the most stringent control strategy in reducing ozone to < 125 ppbwas
confirmed using an alternate meteorological episode.
Application of stringent NOX controls in New York City is counterproductive. Ozone levels
approach 125 ppb as VOC controls are increased. It appears that even without NOX controls,
stringent VOC technology and reactivity-based measures may be needed in this urban area.
. Considering rule effectiveness and a more realistic representation of control programs (e.g.,
fleet turnover), results show predicted episode maximum ozone levels of just above 125 ppb
in most sections of the Northeast Corridor with the most stringent VOC/NOx/reactivity-
reduction strategy simulated. The exception is New York City, where the peak ozone level is
predicted to be 140 ppb.
Issue #4
What are the effects of reactivity-based strategies on regional ozone levels?
Reactivity-based strategies similar to those simulated in ROMNET may provide the greatest
benefit in large urban areas that are VOC-limrted, and thus are more responsive to changes in
VOC emissions. In such areas, reactivity-based measures could provide reductions in ozone
levels comparable with that provided by a stringent technology-based approach. Also, reac-
tivity measures may counterbalance the negative impact of NOX controls in New York City and
other areas that show a similar response.
In other Northeast Corridor cities, and by extension, cities that respond to both VOC and NOX
controls, reactivity measures may produce a notable reduction in daily maximum ozone levels
> 125 ppb. However, for the Corridor cities, the reduction provided by the reactivity mea-
sures was only half of that from the technology-based VOC controls, and four times less than
that from the VOC plus NOX controls.
ES-11
-------
In other, more peripheral sections of the Corridor, the relative benefits of VOC technology
controls alone or with NOX controls far outweigh the benefits of reactivity-based controls.
In cities outside of the Corridor that were found to be most sensitive to NOX controls, there was
less reduction in ozone from the reactivity measure compared with the technology-based
controls.
Issue #5
How does the large uncertainty in biogenic emissions alter conclusions regarding the effective-
ness of control measures?
ROM simulations with 'best estimate" biogenic emissions yield predicted ozone concentra-
tions closer to observed values than biogenic emissions at either end of the uncertainty range
(± a factor of three), which provides some added confidence in the emissions rates used in
ROM strategy simulations.
• The sensitivity of predicted ozone levels and population exposure to biogenic uncertainty
varies considerably from city to city. In general, there is a greater sensitivity with increasing
biogenics at the high end of the uncertainty range than at the lower end.
Grid hours > 125 ppb are much more sensitive to an increase in biogenic emissions than the
other metrics. If biogenics were at the "low" end of the range, predicted ozone in Pittsburgh
and Detroit would fall below 125 ppb with the controls implied by proposed 1989 Clean Air Act
legislation (CS05); all other cities would still be above this level (note that predicted ozone in
Charleston, WV, is below 125 ppb with "best estimate" biogenics).
• With the application of stringent controls that reduced peak ozone to below 125 ppb with
"best estimate" biogenics (CS19), ozone levels rise well above this level in the "high" biogenics
scenario in all cities except Pittsburgh and Charleston. Thus, if biogenics are actually near the
high end of the uncertainty range, then additional controls providing reductions in emissions
beyond those in CS19 will be needed in many of the Northeast cities.
Caveats
There are three broad caveats to the findings and conclusions discussed above:
1. The relatively coarse grid cell size in the ROM (a horizontal grid size of ~ 18.5 x 78.5 km) and limited
vertical differentiation (two layers within the daytime atmospheric boundary layer). Emissions are
injected as a total flux by layer for each grid, and thus the fate of emissions from individual point
sources is not treated explicitly. Concentration estimates are grid cell (horizontal and vertical) aver-
age values for each layer. This may in part explain why the model tends to underpredict observed
peak ozone concentrations associated with sharp concentration gradients and overpredict
minimum values affected by titration from local sources of NO. In ROMNET, the large grid size may
affect conclusions regarding NOX control, since a large fraction of NOX emissions emanate from
ES-12
-------
individual point sources, particularly in areas outside the Northeast Corridor. Also, the tendency to
smooth out peak values may affect findings on the level of emissions controls necessary to reduce
ozone to < 125ppb.
2. The emissions inventories used in ROMNET. As indicated previously, there are large uncertainties in
biogenic emissions. Also, even though the NAPAP emissions data were subjected to extensive
quality assurance, there remains some unqualified level of uncertainty in anthropogenic emissions.
In this regard, the findings from ROMNET should be used to help establish control directions and
approximate starting points for States to begin strategy evaluations using more recent (1990),
locale-specific, quality assured inventories.
3. This study concentrated heavily on one episode (July 2-17,1988). This episode was the most severe
of all episodes in the Northeast, at least as far back as 1980. Although the episode included several
of the meteorological conditions typical of ozone episodes in this region, consideration of a wide
range of meteorological scenarios was not possible for the strategy simulations.
TECHNOLOGY TRANSFER
Need for Technology Transfer
In ROMNET, the use of the ROM to estimate effects of regional control strategies on ozone levels was
demonstrated. There are four reasons why using ROM-system data to assist in developing required
ozone NAAQS attainment demonstrations would be advantageous.
1. The ROM data provide a means for ensuring consistency among States in the procedures used to
derive transport assumptions in the design of urban strategies.
2. The ROM provides a means for simulating the combined effects from numerous urban areas across
the region.
3. The model provides the most technically defensible way to estimate future ozone/precursor con-
centrations transported into urban areas.
4. The ROM system provides data for use in SIPs which may be very difficult or expensive to obtain
otherwise.
For reasons noted previously, using the ROM as the sole modeling tool to demonstrate attainment of the
ozone NAAQS may not be the best approach. The Urban Airshed Model (DAM) has been designated as
the preferred method for attainment demonstrations. However, the ROM is most useful in conjunction
with "nesting" the UAM within a ROM domain. In this approach, the ROM system is used to provide initial
and boundary conditions estimates, as well as other inputs, for UAM applications.
ES-13
-------
Although the concept of nesting is straightforward, its implementation is not. Part of the difficulty lies in
managing the large quantity of data generated as part of ROM simulations, and in providing an easy
means for States to access and retrieve selected data sets needed in UAM applications. Also, although
ROM and UAM are both three-dimensional photochemical grid models, they are dissimilar in many
aspects. For example, the horizontal and vertical structure and resolution of the two models differ. Also,
the UAM preprocessors require input data formats which are unlike those used for ROM data sets.
These technical problems have been overcome as part of ROMNET through the development of two
computer systems. These are the Gridded Model Information Support System (GMISS) and the
ROM/UAM Interface Program System (Interface).
Gridded Model Information Support System (GMISS)
GMISS contains an archive of ROM-system databases, as well as utilities for retrieving these data, and
can be easily accessed by the States. The data types in GMISS include ROM-predicted pollutant con-
centrations, selected meteorological inputs to the ROM, biogenic emissions calculated for ROM simu-
lations, and topographic information. The UAM subsystem of GMISS is designed to provide all of the
ROM-system data necessary to use the ROM/UAM Interface for developing UAM inputs. GMISS resides
on the EPA IBM 3090, and is accessible by States from their local terminals via a user-friendly menu
system. Data can be retrieved for user-specified dates and subregions of the ROM domain. Additional
information on GMISS is provided in Section 6 of this report.
ROM/UAM Interface Program System (Interface)
The ROM/UAM Interface is a set of computer programs that convert ROM data files accessed from
GMISS into input files used directly by the UAM and four of its 13 preprocessor programs. The Interface
does not handle mixing heights or anthropogenic emissions, both of which must be supplied by the
user. The Interface is sufficiently flexible to allow the user to substitute user-derived inputs (e.g., wind
fields) for ROM-system data. Interface programs prescribe a series of algorithms that match and inter-
polate ROM data to the resolution used by UAM. The structure of the ROM/UAM Interface Program
System is described in greater detail in Section 6.
ACCOMPLISHMENTS OF THE ROMNET PROJECT
The ROMNET project was successful in achieving the goals set forth at its inception. This success was
attained through the accomplishment of several major milestones. First, ROMNET brought together
ES-14
-------
numerous State/local agencies and EPA offices to focus their energies in a cooperative manner to
address long-standing issues concerning problems of ozone transport in the Northeast. During
ROMNET, State agency representatives played an active role along with the EPA in the management
and technical decision-making aspects of the project. The organizational structure and communica-
tions processes adopted in ROMNET may be useful as a framework for future joint efforts among States
and the EPA.
Second, the Regional Oxidant Model (ROM) was demonstrated to be a practical tool for addressing
policy-related issues concerning the efficacy of various regional control strategies. The ROMNET pro-
gram has provided insights to Federal, State, and local policymakers on a number of such issues,
including: (a) the relative effects of VOC and NOX controls; (b) the role of transport into and within the
Northeast Corridor; (c) the magnitude of controls potentially needed to reduce ozone to the level speci-
fied in the NAAQS; (d) the usefulness of strategies that incorporate changes in the photochemical
reactivity of VOC emissions; and (e) the effects of the large uncertainty in biogenic emissions on
conclusions regarding the efficacy of control strategies.
Finally, as a result of ROMNET, computer systems and databases have been developed which will
enable State agencies to explicitly consider regional and interurban transport in preparing ozone SIPs.
The infrastructure containing these systems gives States access to ROM-system databases and pro-
vides methodologies and guidance for using these data to support urban-level attainment demonstra-
tions. These systems are generic in nature and can be used by States in other portions of the United
States where ROM simulations are planned as part of future EPA-State regional modeling projects.
ES-15
-------
Figure ES-1. The ROMNET region.
ES-16
-------
ADVISORY
COUNCIL
EMISSIONS
COMMITTEE
MANAGEMENT REVIEW COMMITTEE
PROGRAM DIRECTOR
TECHNICAL COORDINATOR
STRATEGY
COMMITTEE
MODELING
COMMITTEE
Figure ES-2. ROMNET management structure.
ES-17
-------
RAW INPUT DATA
l
| ("AIR QUALITY) ,
i
i
j (METEOROLOGY)
i
i
i ("EMISSIONS ^)
1 ^ -s
i
| C LAND USE ")
1
{ (JOPOGRAPHY) 1
n nuivi
CORE MODEL
1
ROM •*
^ PROCrSSOR 1 L
1 ^ta. HTHMF
CONCENTRATIONS
i _ . . .. .
Figure ES-3. Components of the ROM.
ES-18
-------
TABLE ES-1. ORGANIZATIONS PARTICIPATING IN ROMNET
State/local air pollution control agencies:
Connecticut New York
Delaware Ohio
Kentucky Pennsylvania
Maine Philadelphia
Maryland Rhode Island
Massachusetts Vermont
Michigan Virginia
New Hampshire Washington, DC
New Jersey West Virginia
Northeast States for Coordinated Air Use Management (NESCAUM)
EPA Regions I through V
EPA Headquarters Offices:
Office of Air Quality Planning and Standards (OAQPS)
Atmospheric Research and Exposure Assessment Lab (AREAL)
Office of Policy Analysis and Review
Office of Policy Planning and Evaluation
Office of Mobile Sources
Contractors:
Computer Sciences Corporation (CSC)
Alliance Technologies
ES-19
-------
TABLE ES-2. ROMNET CONTROL STRATEGIES
Control
Strategy Number
Controls
Issue(s)
Addressed
Rationale
CS01 (MOBILE 3.9)
CS02 (MOBILE 3.9)
CS03 (MOBILE 3.9)
CS04
CS05
CS06-CS09
CS10
CS11
CS12
CS13-CS14
Maximum technology VOC 2,3
regionwide; NOX at 2005 base-
line levels
Maximum technology VOC in 2
NE Corridor only; NOX at 2005
baseline
Maximum technology VOC in 2
NE Corridor and in other nonat-
tainment areas; only 2005 base-
line elsewhere
Not performed
VOC and NOX controls pre- 3
scribed in HR3030 with October
1989 Waxman-Oingell tailpipe
standards
CS01 and CS05 varying bio- 5
genie emissions by ± a factor
of 3
Enhanced maximum technol- 1, 3
ogy VOC and NOX controls
applied regionwide
Maximum technology NOX; VOC 1, 3
at 2005 baseline
Maximum technology VOC; NOX 1,3
at 2005 baseline
Maximum technology VOC with 1, 3
NOX point source controls
(CS13) or NOX mobile source
controls (CS14)
Test effects of strategies
emphasizing VOC
reductions by compari-
son with 2005 baseline.
Test effect of VOC con-
trols outside Corridor on
ozone in Corridor by
comparison with CS01.
Test effect of "rural" VOC
controls outside Corri-
dor on ozone Corridor
by comparison with
CS01 and CS02.
Test effects of proposed
Clean Air Act legislation.
Test sensitivity of con-
clusions to uncertainty
in biogenics.
Test effects of VOC and
NOX controls.
Compare effects of NOX
controls with VOC and
NOX(CS10).
Compare effects of VOC
controls with VOC and
NOX controls (CS10) and
NOX controls above.
(CS11). Test whether it
is appropriate to focus
controls on certain NOX
source types.
continued
ES-20
-------
TABLE ES-2 (continued)
Control
Strategy Number
Controls
Issue(s)
Addressed
Rationale
CS15
CS16
CS17
CS18
CS19
CS20
CS21.CS22
Maximum technology VOC and 3, 4
NOX controls but with methanol
(M100) vehicles in the NE Corri-
dor and low-reactivity solvent
substitution regionwide
CS10 controls plus across-the- 3
board VOC reductions in New
York City and
Baltimore/Washington, DC.
Also NOX point source emis-
sions at 2005 baseline in New
York City
Maximum technology NOX con- 5
trols (CS11) with "low" biogenics
CS15 with additional across- 3
the-board VOC reductions in
New York City and
Baltimore/Washington, DC; also
NOX point-source emissions at
2005 baseline in New York City
CS18 but with NOX point source 2, 3
emissions at 2005 baseline in
Baltimore/Washington, DC
2005 baseline with methanol 4
(M100) vehicles in NE Corridor
and low-reactivity solvents
regionwide
Biogenics sensitivity tests with 5
CS19 varying biogenic emis-
sions by ± a factor of 3
Test implications of
reducing reactivity.
Reduce maximum ozone
to < 125 ppb through-
out the U.S. portion of
the region.
Examine effectiveness of
NOX control within
uncertainty range in bio-
genics.
Test relative importance
of reactivity reduction
and VOC technology
controls by comparison
withCS12andCS16.
Reduce maximum ozone
to < 125 ppb through-
out the U.S. portion of
the region.
Compare effectiveness
of reactivity reductions
in base vs. postcontrol
scenarios and with VOC
technology controls.
Test effect of biogenic
emissions uncertainty
on a strategy that
reduces ozone to
< 125 ppb.
continued
ES-21
-------
TABLE ES-2 (concluded)
Control Issue(s)
Strategy Number Controls Addressed Rationale
CS23 CS19, but with more realistic 3 Test sensitivity of results
"rule effectiveness' assumptions to limitations in control
and vehicle fleet penetration program effectiveness.
estimates
CS24 CS19 in NE Corridor; 2005 2 Assess relative impor-
baseline elsewhere tance of transport from
outside the Corridor vs.
controls on emissions
within the Corridor.
CS25 CS19 outside Corridor; 2005 2
baseline in NE Corridor
ES-22
-------
SECTION 1
PROJECT GOALS AND SCOPE
by
Norman C. Possiel*
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
* On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
1.1 PURPOSE AND INTENT OF ROMNET
The Regional Ozone Modeling for Northeast Transport (ROMNET) Project was initiated by the U.S.
Environmental Protection Agency (EPA) and State and local agencies in the Northeast, as part of a joint
effort to address the problem of regional transport in developing effective and equitable control pro-
grams to attain the ozone National Ambient Air Quality Standards (NAAQS)1 in this region. The specific
goals of ROMNET are:
(1) to evaluate the relative effectiveness of regional controls on ozone levels in the Northeast;
(2) to provide quantitative estimates of ozone and precursor concentrations transported between
urban areas following the application of regional control measures; and
(3) to provide procedures and guidance for incorporating ozone and precursor transport in future
State Implementation Plan (SIP) development.
The intent of ROMNET is for EPA and State decision-makers to use the findings in guiding ozone-policy
development and planning for potential regional control programs, as well as in urban-area specific
strategy evaluations. The urban-scale analyses are to be conducted by the States as part of the SIP
process. The results and guidance from the project will enable States to quantify changes in future
levels of ozone and precursor transport expected to follow implementation of nationwide Federal mea-
sures, local control programs in upwind cities, and potential regional strategies.
1.2 STRUCTURE OF THE FINAL REPORT
This report was prepared to document the activities and findings of ROMNET. The following is a brief
description of the contents of each section:
Executive Summary - presents an overview of ROMNET and the principal findings and con-
clusions of the project;
• Section 1, Project Goals and Scope - provides the background to ROMNET, the goals and
structure of the project, and the technical approach adopted to meet the project objectives;
• Section 2, The ROMNET Modeling System - describes the Regional Oxidant Model (ROM)
and the evaluation of the ROM conducted as part of ROMNET;
• Section 3, Ozone Episodes Selection -- describes the episode selection process, and the
ozone levels and meteorological conditions during episodes simulated;
1. An exceedance of the NAAQS is defined as a daily maximum 1-hour ozone concentration greater than 0.12 ppm (124 ppb).
The NAAQS is not to be exceeded on average more than once per year at any location.
1-3
-------
. Section 4, Emissions Scenario Development - describes the base case emissions scenario,
future baseline scenario and control strategies, as well as the procedures used to develop
inventories that reflect these scenarios;
Section 5, Regional Modeling Results and Project Findings - presents the results of ROM
simulations conducted to evaluate the effectiveness of control strategies;
• Section 6, State Access and Use of ROMNET Databases - describes the Gridded Model
Information Support System (GMISS) and the ROM-Urban Airshed Model (UAM) Interface;
and
Appendices - includes supporting information for sections of this report.
Note that all figures and tables are placed at the end of the Section in which they are referenced.
1.3 BACKGROUND TO ROMNET
Since the early I970's, there has been a growing awareness that ozone and ozone precursors are
transported beyond the political jurisdiction of source areas to impact air quality levels at considerable
distances downwind. This phenomenon is now well documented by the scientific community and has
been recognized by policymakers at the EPA. Transport effects are especially acute where several
urban areas are in close proximity to one another, as in the Northeast. There, five major urban areas,
from Washington, DC to Boston and adjacent medium-size cities and suburbs, nearly overlap to provide
an almost continuous corridor of sources emitting ozone precursors, as shown in Figure 1 -1. Aside from
these source areas, emissions in other less populated areas of the Northeast and/or cities such as
Pittsburgh, Buffalo, Cleveland, Detroit, etc., are likely to contribute to the regional ozone burden trans-
ported within the Northeast. A compounding factor is the meteorological conditions that frequently exist
in this region during the summer. In particular, the wind flow often favors ozone and precursor transport
both between urban areas in the Corridor, and into the Corridor from other parts of the Northeast. These
conditions, coupled with the time scales of ozone photochemistry, can result in multiday episodes of
high ozone concentrations across broad areas of the region, as shown in Figure 1 -2.
As a consequence of the prevalence of interurban and intraregional transport, a regional perspective is
necessary for the development of control programs in the Northeast. At the request of States in this
region, the EPA initiated ROMNET in October I987 as a three year effort to provide a quantitative tech-
nical means to support such a regional approach. The State/local agencies, EPA organizations, and
EPA contractors participating in ROMNET are listed in Table 1-1.
1-4
-------
The first task of the ROMNET project was the development of a protocol which sets forth the goals,
scope, technical approach, schedule, and organizational structure agreed to by the participants. This
document is included as Appendix A. The remainder of this Section describes the rationale for the goals
and scope of ROMNET and the objectives designed to address these goals; the project organizational
structure; and the major technical tasks conducted to meet the objectives. As indicated in Section 1.2,
details on the individual technical tasks are provided in other portions of this report. More information on
the organizational structure can be found in the ROMNET Protocol in Appendix A.
1.4 RATIONALE FOR THE GOALS AND SCOPE OF ROMNET
This Section describes the rationale for each of the major goals of ROMNET identified in Section 1.1.
The rationale for the first goal of ROMNET follows from the premise that equitable and effective strate-
gies for attainment in both urban and rural areas of the Northeast will require relatively stringent control
measures along the Corridor and potentially throughout the entire region. Such a broad strategy may
be needed to reduce interurban and intraregional impacts associated with ozone and precursor trans-
port. Thus, it is important to identify strategies that are most effective in reducing regional ozone con-
centrations.
The rationale for the second goal is based upon prior urban scale modeling analyses, which indicate
that the effectiveness of emissions controls is strongly sensitive to assumptions about the levels and
mixture of ozone and precursor concentrations transported into the modeling area. Present EPA policy,
as well as provisions of the 1990 Clean Air Act legislation, require that urban-scale modeling be used in
identifying and evaluating control measures for certain nonattainment areas. Therefore, it is critical that
meaningful representative estimates of urban boundary conditions be developed and used in SIP mod-
eling.
The third goal reflects the practicality of deriving urban boundary conditions from transport concentra-
tions. Procedures and guidance are needed on how to link regional and urban models so that data from
the large scale model will mesh in a cohesive manner with the smaller scale model. For this, an
infrastructure or system is required to store, manage, manipulate and transfer regional databases to
States for their use in urban modeling. These databases provide estimates of transport under various
regional emissions scenarios.
1-5
-------
Meeting the three project goals is not straightforward because of the complexity of the transport prob-
lem and the structure of available models. For example, regionally-high ozone concentrations that make
transport considerations important are not the result of a few easily identifiable plants or source types.
Rather, ozone levels result from the combination of various sources, including numerous, but individu-
ally small, sources of volatile organic compounds (VOC), such as dry cleaners, gasoline service stations,
bakeries, etc.; mobile source emissions of VOC, oxides of nitrogen (NOX), and carbon monoxide (CO)
from passenger vehicles, trucks etc.; and large point sources of NOX, VOC, and/or CO, such as power
plants, industrial facilities, and gasoline terminals, etc. Thus, evaluating regional strategies requires
specification of both emissions and controls for multiple pollutants from numerous source types of dif-
ferent sizes, some of which are clustered in urban areas while others tend to be in rural areas or spread
throughout the region. Also, the specification of boundary conditions for urban modeling is difficult, in
part because of a lack of spatial and temporal resolution of ozone and precursor measurements at the
surface and aloft. This is further complicated by the need to estimate expected future-year boundary
conditions. Such future-year scenarios account for the effects of anticipated changes in emissions
between the current base year and the projected or mandated attainment date. Factors that are con-
sidered in these scenarios include population growth, existing control programs, and strategies
required but not yet fully implemented in upwind areas. Finally, fully-nested regional-to-urban scale
photochemical models are presently unavailable. As a consequence, procedures are needed for pro-
viding State agencies with standard methods and access to regional modeling results, from which the
regional databases can be applied to urban-scale modeling.
The EPA Regional Oxidant Model (ROM) was selected as the tool to quantify transport and evaluate the
effectiveness of regional strategies. The ROM is appropriate for such use for the following reasons: (1) it
is designed to simulate the photochemistry and meteorology associated with ozone formation and pol-
lutant transport across regional spatial scales for multiday summertime episodes; (2) emissions from
ozone-precursor source types are explicitly input to the ROM, and thus growth factors and control
measures can be specified by source category, pollutant, and location; and (3) the ROM produces
gridded three-dimensional estimates of ozone and precursor concentrations with a spatial resolution
and time step interval suitable for providing boundary condition estimates for urban scale models.
Three primary objectives were identified for using ROM in ROMNET in order to meet the project goals.
They are:
(1) to apply ROM to selected base and control strategy scenarios for various ozone episodes;
(2) to interpret the predictions in order to evaluate the relative effectiveness of simulated regional
strategies for reducing ozone levels; and
1-6
-------
(3) to provide the ROM predictions and ancillary input data sets to States for purposes of urban
modeling, along with methods and guidance on how to use this information.
The scope of ROMNET was limited to the organizational interactions and technical tasks required to
meet these objectives. As noted in Section 1.1, ft is intended that the databases, methodologies, and
guidance developed in ROMNET be used by States in subsequent urban scale modeling, as part of the
SIP development process. The project findings and conclusions concerning the efficacy of regional
strategies are to be used by the EPA, States, and a possible future regional commission to support
development of the policies and legislation needed to provide consistent, coordinated SIPs along the
Northeast Corridor, and for the potential implementation of regional control measures.
1.5 ROMNET ORGANIZATIONAL STRUCTURE
A multilayered organizational structure was adopted to achieve the project objectives given in
Section 1.4. The management and technical components of this structure are shown in Figure 1 -3. This
concept was designed to involve as many of the interested agencies as possible in the activities and
decision-making process, and to allow individual expertise from both States and the EPA to be utilized in
the technical tasks. Members of the various committees, the Program Director, Technical Coordinator
and contractor teams are listed in Appendix B.
The Management Review Committee was formed to provide a forum to ensure that the focus of
ROMNET was maintained within the interests of the multiple agencies involved. Specific functions
included:
(1) review of the status of technical tasks to ensure that the project goals were being met on sched-
ule, and
(2) the review and approval of proposed strategies and episodes selected by the technical commit-
tees for simulation.
The Program Director was given responsibility for the overall technical adequacy of the program. Part of
this function included informing and consulting with the Management Review Committee on the prog-
ress of technical activities and ensuring that consensus recommendations of the Management Review
Committee were integrated into the technical program. The Program Director was assisted by the
Technical Coordinator, who was responsible for managing the day-to-day activities of the project and
coordinating activities of the technical committees, EPA participants, and contractors.
1-7
-------
Three technical committees were established as an avenue for participants to provide input and review
of the design and implementation of the technical program. All three committees were chaired by a staff
member of the EPA's Office of Air Quality Planning and Standards (OAQPS). Functions of the Modeling
Committee included episode selection, upgrades to the ROM, development of the ROM-Urban Airshed
Model (UAM) Interface, the supplemental evaluation of the ROM, ROM simulations, and development of
the Gridded Model Information Support System (GMISS). The Emissions Committee was responsible
for oversight of regional emissions inventory development for base case, future baseline, and control
strategy scenarios. The Strategy Committee was charged with the design of future baseline and control
strategies. Finally, the Advisory Council was established to advise the Program Director on major issues
involving resources, schedules, cooperation, technical activities, and program directives. The Program
Director was to use this group as a platform for discussion of options and alternative recommendations
to be made to the Management Review Committee on issues involving these topics. Members of the
Advisory Council are listed in Appendix B. In reality, there was little consultation with this group largely
because no major problems were encountered relative to these issues. Minor problems were either
resolved at the technical committee level or were brought directly to the attention of the Management
Review Committee for consideration.
1.6 OVERVIEW OF THE ROMNET TECHNICAL PROGRAM
The technical tasks conducted under the purview of each of the Committees are listed below.
Modeling Tasks
• An objective ranking procedure was developed and implemented to identify candidate
episodes for ROM simulation.
The ROM was upgraded to version 2.1 in order to:
(1) expand the domain to include all of Ohio and Virginia,
(2) upgrade components related to the chemical mechanism,
(3) improve the efficiency of emissions processing, and
(4) alter several meteorological processors to produce closer representation of observed
ozone plume orientation.
ROM was applied for over thirty combinations of episodes and emissions scenarios.
• A supplemental evaluation of ROM was conducted using selected ROMNET episodes.
• The ROM-Urban Airshed Model (UAM) Interface was developed to provide a means for using
ROM predictions to derive UAM boundary conditions. The Interface also contains proce-
dures for using ROM system data sets to generate meteorological, air quality, topographical,
and biogenic emissions input for UAM.
1-8
-------
• The Gridded Model Information Support System (GMISS) was developed to:
(1) archive ROM predictions and system data sets needed for the ROM-DAM Interface, and
(2) provide a user-friendly menu-based system to enable States to retrieve data sets directly
via their local terminals from the EPA National Computer Center.
OAQPS was responsible for episode selection and, together with Computer Sciences Corporation
(CSC), the development of GMISS. AREAL and their contractor, CSC, designed and implemented the
upgrades to the ROM, the development of the ROM-UAM Interface, and the supplemental evaluation of
the ROM. The ROM simulations, including processing ROM inputs, execution of ROM simulations, and
post-processing of output for statistical summaries and graphical displays, were performed by CSC as
specified by the OAQPS and AREAL State and Regional Office Committee members were instrumental
in the review and analysis of ozone and meteorology data to select episodes for simulation. In addition,
the New York State Department of Environmental Conservation provided review and input to the devel-
opment of the ROM-UAM Interface.
Emissions Inventory Development
• The Biogenic Emissions Inventory System was developed to provide estimates of biogenic
emissions for VOC and NOX. As part of this task, a canopy module was implemented to
provide more representative treatment of the effects of meteorological conditions on VOC
emissions from forested areas.
. The 1985 National Acid Precipitation Assessment Program (NAPAP) anthropogenic emis-
sions inventory was obtained for use in ROMNET and modified to:
(1) upgrade mobile source emissions to MOBILE 4,
(2) provide procedures for altering mobile source emissions to reflect grid and day-specific
temperatures, and
(3) adjust seasonal emissions from selected VOC stationary sources to represent temper-
ature and activity levels typical of ozone episodes.
Procedures were developed to create emissions inventories for projected future baseline
scenarios and control strategies. Estimates of expected emissions growth rates, the effi-
ciency, effectiveness, and penetration of existing control measures were developed and used
in generating these inventories, as were estimates of the efficiencies of maximum technology
controls.
Inventories of anthropogenic emissions were created for two future baseline scenarios and six
technology-based scenarios.
A majority of the emissions inventory work was conducted by Alliance Technologies, Inc., under contract
to EPA OAQPS. The Biogenic Emissions Inventory System was developed by the AREAL and CSC. The
EPA Office of Mobile Sources provided inputs for treating evaporative emissions, temperature effects,
and clean-fuel vehicles to support development of mobile emissions estimates. Principal activities
1-9
-------
involving the State and Regional Office committee members were (1) the review of the 1985 base case
emissions, and (2) the development of growth rates and existing control efficiency, rule effectiveness,
and rule penetration estimates for use in the projected future baseline scenario.
Strategy Design and Interpretation
Twenty-four control strategies were designed and simulated using various combinations of
the future baseline and technology-based scenarios together with across-the-board and
reactivity reductions, and biogenic-uncertainty sensitivity scenarios.
• Strategies were designed for simulation in various phases. Results of each phase were
examined and interpreted to reveal the effectiveness of control measures in order to prescribe
strategies for the subsequent phase.
• Participants in ROMNET, and EPA management were briefed on the results of strategy simu-
lation following the completion of each phase.
. The findings of ROMNET strategy simulations are documented in this report.
For the most part, straw man proposals of strategies were prepared by the OAQPS for discussion and
development by Strategy Committee members. Strategies formulated by the Committee were pres-
ented to the Management Review Committee for approval. Translation of strategies into emissions
reductions were performed by Alliance Technologies, under contract to the OAQPS, and by CSC, under
contract to the AREAL
1-10
-------
[NORTHEAST
'CORRIDOR
|BOSTON
GREATER CONNECTICUT
^NEW YORK CITY
^PHILADELPHIA
^PITTSBURGH
H| BALTIMORE
^WASHINGTON, DC
^CLEVELAND
g| CHARLESTON
p^| DETROIT
I—3 OTHER MSAs
Figure 1-1. The Northeast Corridor and other Metropolitan Statistical Areas (MSA's) in the ROMNET
region.
1-11
-------
Episode Maximum 1-Hour Ozone Concentrations - 7/2/88 Through 7/17/88
Concentration (ppb): A < 125 o>= 125 A >= 150 • >= 175 •*• >= 200
Figure 1 -2.. Observed maximum 1 -hour ozone concentrations for July 2 -17,1988 across the ROMNET
region.
1-12
-------
ADVISORY
COUNCIL
EMISSIONS
COMMITTEE
MANAGEMENT REVIEW COMMITTEE
PROGRAM DIRECTOR
TECHNICAL COORDINATOR
STRATEGY
COMMITTEE
MODELING
COMMITTEE
Figure 1 -3. ROMNET management structure.
1-13
-------
TABLE 1 -1. ORGANIZATIONS PARTICIPATING IN ROMNET
State/local air pollution control agencies:
Connecticut New York
Delaware Ohio
Kentucky Pennsylvania
Maine Philadelphia
Maryland Rhode Island
Massachusetts Vermont
Michigan Virginia
New Hampshire Washington, DC
New Jersey West Virginia
Northeast States for Coordinated Air Use Management (NESCAUM)
EPA Regions I through V
EPA Headquarters Offices:
Office of Air Quality Planning and Standards (OAQPS)
Atmospheric Research and Exposure Assessment Lab (AREAL)
Office of Policy Analysis and Review
Office of Policy Planning and Evaluation
Office of Mobile Sources
Contractors:
Computer Sciences Corporation (CSC)
Alliance Technologies
1-14
-------
SECTION 2
THE ROMNET MODELING SYSTEM
by
Kenneth L Schere*
Thomas E. Pierce*
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Dennis Doll*
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
Lenard B. Milich
Jeffrey O. Young
Computer Sciences Corporation
P.O. Box12767
Research Triangle Park, NC 27709
On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
2.1 INTRODUCTION
The Regional Oxidant model (ROM) was used as the tool for estimating regional ozone concentrations
expected to result from applying the various ROMNET emissions control strategies. This Section of the
report describes the characteristics of the ROM, the associated input-processor system, quality assur-
ance procedures followed for ROM simulations, and limitations to the use of the model. An evaluation of
the ROM, conducted as part of ROMNET, is also summarized in this Section.
2.2 GENERAL CHARACTERISTICS OF THE REGIONAL OXIDANT MODEL (ROM)
The ROM was designed to simulate most of the important chemical and physical processes that are
responsible for the photochemical production of ozone over a domain of 1000 km and for episodes of up
to 15 days in duration (Lamb, 1983). These processes include (1) horizontal transport, (2) atmospheric
chemistry and subgrid-scale chemical processes, (3) nighttime wind shear and turbulence associated
with the low-level nocturnal jet, (4) the effects of cumulus clouds on vertical mass transport and photo-
chemical reaction rates, (5) mesoscale vertical motions induced by terrain and the large-scale flow,
(6) terrain effects on advection, diffusion, and deposition, (7) emissions of natural and anthropogenic
ozone precursors, and (8) dry deposition. The processes are mathematically simulated in a three-
dimensional Eulerian model with three vertical layers, including the boundary layer and the capping
inversion or cloud layer. Horizontal grid resolution is 1/4° longitude by 1/6° of latitude, or about 18.5 km
x 18.5 km. The modeling domain adopted for use in ROMNET is shown in Figure 2-1.
2.2.1 Physical Processes within Layers 1. 2 and 3
In general, the three model layers are free to locally expand and contract in response to changes in the
physical processes occurring within them. During an entire simulation period, horizontal advection and
diffusion and gas-phase chemistry are modeled in all three layers. ROM predictions from layer 1 are
used as surrogates for surface concentrations. Although the time-scale of output concentrations is
30 minutes, ROM predictions are aggregated into 1-hour average concentrations. Figure 2-2 shows the
ROM layers during the day and at night, and describes some of their features.
Layers 1 and 2 model the depth of the well-mixed layer during the day. Some special features of layer 1
include the modeling of (1) the substantial wind shear that can exist in the lowest few hundred meters
s
above ground in local areas where strong winds exist and the surface heat flux is weak, (2) the thermal
internal boundary layer that often exists over large lakes or near sea coasts, and (3) deposition onto
2-3
-------
terrain features that protrude above the layer. At night, layer 2 represents what remains of the daytime
mixed layer. As stable layers form near the ground and suppress turbulent vertical mixing, a nocturnal
jet forms above the stable layer and can transport aged pollutant products and reactants considerable
distances. At night, emissions from tall stacks and warm cities are injected directly into layers 1 and 2.
Surface emissions are specified as a mass flux through the bottom of layer 1.
During the day, the top model layer, layer 3, represents the synoptic-scale subsidence inversion char-
acteristic of high ozone-concentration periods; the base of layer 3 is typically 1 to 2 km above the
ground. Relatively clean tropospheric air is assumed to exist above layer 3 at all times. If cumulus
clouds are present, an upward flux of ozone and precursor species is injected into the layer by pene-
trative convection. At night, ozone and the remnants of other photochemical reaction products may
remain in this layer and be transported long distances downwind.
When cumulus clouds are present in a layer 3 cell, the upward vertical mass flux from the surface is
partially diverted from injection into layer 1 to injection directly into the cumulus cloud of layer 3. In the
atmosphere, strong thermal vertical updrafts, primarily originating near the surface in the lowest portion
of the mixed layer, feed growing fair weather cumulus clouds with vertical air currents that extend in one
steady upward motion from the ground to well above the top of the mixed layer. These types of clouds
are termed "fair weather cumulus" since atmospheric conditions are such that they do not grow to the
extent that precipitation forms. The dynamic effects of this transport process and daytime cloud evolu-
tion can have significant effects on the chemical fate of pollutants. For example, fresh emissions from
the surface layer can be injected into a warm thermal and rise, essentially unmixed, to the top of the
mixing layer where they enter the base of a growing cumulus cloud. Within the cloud, the chemical
processes of ambient pollutant species are suddenly altered by the presence of liquid water and the
attendant attenuation of sunlight. The presence of fair weather cumulus clouds implies that the atmo-
sphere above the earth's boundary layer is too stably stratified for thermals to penetrate higher. In this
case the air comprising the tops of these clouds returns to the mixed layer and is heated on its descent
since it is being compressed by increasing atmospheric pressures. Ultimately, the air again arrives at
the surface level where new emissions can be injected into it and ground deposition may occur, and the
process may begin again. The time required for one complete cycle is typically 30 to 50 minutes with
perhaps one-tenth of the time spent in the cloud stage.
Within the ROM system, a submodel parameterizes the above cloud flux process and its impact on mass
fluxes among all the model's layers. In the current implementation of the chemical kinetics, liquid-phase
chemistry is not modeled, and thus part of the effects from the cloud flux processes are not accounted
for in the simulations. Future versions of the chemical kinetics may include liquid-phase reactions. The
2-4
-------
magnitude of the mass flux proceeding directly from the surface layer to the cloud layer is modeled as
being proportional to the observed amount of cumulus cloud coverage and inversely proportional to the
observed depth of the clouds.
Horizontal transport within the ROM system is governed by hourly wind fields that are interpolated from
periodic wind observations made from upper-air soundings and surface measurements. During the
nighttime simulation period, the lowest few hundred meters of the atmosphere above the ground may
become stable as a radiation inversion forms. Wind speeds increase just above the top of this layer,
forming the nocturnal jet. This jet is capable of carrying ozone, other reaction products, and emissions
injected aloft considerable distances downwind. This phenomenon is potentially significant in modeling
regional-scale air quality and is implicitly treated by the model, where the definition of layer 1 attempts to
account for it.
Because standard weather observations do not have the spatial or temporal resolution necessary to
determine with confidence the wind fields in layer 1, a submodel within the ROM system was developed
to simulate the nighttime flow regime in layer 1 only. This prognostic flow submodel is activated only
when a surface inversion is present over most of the model domain. At all other times the flow in layer 1
is determined from interpolation of observed winds. The nighttime flow regime within layer 1 is
influenced by buoyancy, terrain, warm cities, pressure-gradient forcing, and frictional forces, all of which
are accounted for in the model's flow formulation. Solution of the wind submodel equations produces
estimates of the wind components as well as the depth of the inversion layer for all grid cells in layer 1.
2.2.2 ROM Chemistry
The chemical kinetic mechanism embedded in the current version of the ROM is the Carbon Bond IV
(CB-IV) set of reactions (Gery et a/., 1989). This mechanism simulates the significant reaction pathways
responsible for gas-phase production and destruction of the constituents of photochemical smog on
regional scales. The mechanism consists of 82 reactions encompassing 35 individual species; these
species are listed in Table 2-1. The ROM's chemical solution scheme makes no a priori assumptions
concerning local steady states. Therefore, all species are advected, diffused, and chemically reacted in
the model simulations.
The CB-IV contains a standard set of reactions for atmospheric inorganic chemical species, including
O3, NO, NO2, CO, and other intermediate and radical species. Organic chemistry is partitioned along
j1
reactivity lines based on the carbon structures of the organic molecules. Ten individual categories of
organics are represented to account for the chemistry of the hundreds of organic molecules existing in
2-5
-------
the ambient atmosphere: ETH, an explicit representation of ethene; FORM, an explicit representation of
formaldehyde; OLE, a double-bonded lumped structure including two carbons (e.g., olefins); PAR, a
single-bond, single-carbon structure (i.e., paraffins); ALD2, the oxygenated two-carbon structure of the
higher aldehydes; TOL, the aromatic structure of molecules with only one functional group (e.g., tolu-
ene); XYL, the structure of molecules with multifunctional aromatic rings (e.g., xylene); ISOP, the five-
carbon isoprene molecule; and NONR, a single-carbon organic structure not significantly participating
in the reaction sequence. In addition, MTHL, (methanol) was included in the mechanism for use with
control strategies that include methanol-powered vehicles.
2.2.3 System Components
The three components of the ROM system are shown in Figure 2-3. The first component specifies the
five types of 'raw0 data that are required for input to the ROM: air quality, meteorology, emissions, land
use, and topography.
Raw Data
Air quality data consist of hourly ozone observations obtained from the U.S. Environmental Protection
Agency's National Air Data Branch. These hourly observations are used to specify the initial and
upwind-boundary ozone concentrations required by the ROM. Initial conditions are derived from the
mean tropospheric background concentrations listed in Table 2-2 (Killus and Whitten, 1984) using the
temperature-dependent rate constants calculated by the ROM system. Background concentrations are
used since ROM simulations begin on relatively-low ozone days, several days prior to observations of
ozone levels > 120 ppb. This approach is designed to minimize the impact of initial conditions on ozone
levels during the period of most interest (i.e., high ozone days).
The twice-daily (daytime and nighttime) gridded equilibrated concentrations for the 35 species used in
the ROM, for the north, south, east, and west boundaries of each model layer are derived as follows:
each boundary is assigned a single value for ozone, based on average ambient measurements at rural
monitoring sites. The remaining 34 species are then equilibrated to this ozone concentration, generat-
ing the set of concentration values for that boundary.
Meteorology input data consist of the regular hourly surface and upper-air observations from the U.S.
National Weather Service and Environment Canada. Surface-weather station reports include wind
speed and direction, air temperature and dew point temperature, atmospheric pressure, and cloud
amounts and heights. Twice-daily sounding data from the upper-air observation network are included in
2-6
-------
the meteorology database, and consist of atmospheric pressure, wind speed and direction, and air
temperature and dew point temperature. Additional meteorology data are obtained from the National
Climate Data Center and consist of buoy and Coastal Marine Automated Station data; parameters typi-
cally reported are wind speed and direction, and air and sea temperatures. Meteorology data are used
to specify the meteorological parameters in ROM simulations. Detailed descriptions of meteorology
data and meteorology data preprocessing can be found in Milich era/. (1990).
Emissions input data consist of annual point-source emissions, with stack parameters and seasonal,
day-type factors; area-source emissions for typical summertime Saturdays, Sundays, and a °generic"
weekday; and mobile-source emissions. Biogenic emissions data are obtained and preprocessed prior
to inclusion in the emissions database. Detailed descriptions of emissions data used in ROMNET are
included in Section 4.
Land use input data consist of 11 land use categories in 1/4° longitude by 1/e° latitude grid cells. Data
are provided for the U.S. and Canada as far as 55 °N. The land use categories are (1) urban land,
(2) agricultural land, (3) range land, (4) deciduous forests, (5) coniferous forests, (6) mixed forest with
wetlands, (7) water, (8) barren land, (9) nonforested wetland, (10) mixed agricultural land and range
land, and (11) rocky, open places occupied by low shrubs and lichens. The data were obtained from the
EPA Environmental Monitoring Systems Laboratory, Las Vegas, NV. Land use data are used to obtain
biogenic emissions estimates as a function of the area of vegetative land cover, and for the determina-
tion of surface heat fluxes.
Topography input data consist of altitude matrices of elevations for 30"x 30"cells in a 71/2° x 71/2° grid.
The data are obtained from the GRIDS database operated by the U.S. Environmental Protection Agen-
cy's Office of Information Resources Management. Topography data are used in the calculation of layer
heights.
Processor System
The second component of the ROM system is the ROM processor network. The raw input data to the
ROM are manipulated by a hierarchical network of 35 processors that range in function from simple
reformatting of emissions data to generating the complex wind fields that drive the atmospheric
transport. These processors are interconnected by their requirements for and production of data. The
ultimate product of the processor network is a collection of data files that can be categorized into two
types: processor files (PF) and model files (MF). PF's contain partially processed data required as input
2-7
-------
to higher level processors. MF's contain the parameter fields that are transformed into the variables
required by the model algorithms; however, they also provide input to a number of higher level
processors.
The processors are organized into nine distinct hierarchical stages, numbered 0-8. Stage 0 processors
produce output files such as the gridded land use data. Stage 1 processors interface directly with the
preprocessed and extensively quality-reviewed meteorological and emissions input data sets. Subse-
quent stages transform the input data into the gridded fields of temporally and spatially varying param-
eter values needed by the highest stages of the processing network. Processors at any stage can
interface directly with the B-Matrix compiler, described below, through the production of model input
files (MF's). This multistage organization is important to the network because it clearly delineates the
sequence of program execution. Processors at the same stage may execute simultaneously. A pro-
cessor at any given stage, however, must wait until all processors from lower stages along its input data
paths have been completed. Formal definition of all data/processor relationships and automation of
processor executions are essential to ensure consistency and validity of MF's. Appendix C contains a
description of the function(s) of each processor.
The B-Matrix Compiler (BMC) serves as the interface between the model input files and the algorithms
that solve the coupled set of finite difference equations describing the governing processes in each
layer of the model. The program functions similarly to a computer language compiler that transforms
high-level language commands into a machine or algorithm-specific representation. The BMC mathe-
matically combines physical parameters such as layer thicknesses, air densities, etc., into the complex
coefficients required for solution of the governing equations. These coefficients can no longer be
equated with physical quantities; they are purely mathematical entities related specifically to the form of
the finite difference algorithms used by the ROM.
Core Model
The third component of the ROM system is the Core Model. The governing equations in the Core Model
are expressed in a form that allows the chemical kinetics, advection, and vertical flux to be treated
independently. The chemistry module exchanges information with algorithms of the governing equa-
tions via two vectors: (1) a vector that contains the net production rate of each species, and (2) a vector
that contains the net destruction rate. Such design simplifications enhance the flexibility of the model
and are not limited to the interchanges of the chemical mechanism; they apply to all theoretical.
formulations of the physical and meteorological processes (i.e., to all the processors).
2-8
-------
2.2.4 Quality Assurance Procedures for ROM Data Sets
Quality assurance on inputs to the ROM begin with a review of the "raw" data elements. Meteorological
data are examined to ensure that data sets are complete for the entire episode being modeled. Missing
values are filled in through interpolation and extrapolation from available data and knowledge of the
expected variation in meteorological conditions. In addition, surface and upper-air data are screened
against a typical range of values for each parameter to identify possible outliers. Calculated and
time/space-interpolated values, such as layer heights, heat fluxes, wind fields, temperature, etc., are
reviewed to ensure that representative, consistent, and coherent spatial fields and temporal patterns are
produced.
Emissions input data are quality-assured by review of (1) State-level and urban-level emissions summa-
ries by point, area, and mobile category for VOC, NOX and CO; (2) regional aggregate diurnal profiles;
and (3) spatial patterns of gridded data. This information is then compared against the expected
changes in emissions given the specification of growth and/or controls in the scenario under review.
Comparisons are also made between the processed data and the precursor "raw" emissions data.
Quality assurance of the ROM predictions focuses on a review of daily maximum ozone concentrations
from layer 1 for each day simulated. Spatial contours and fields of gridded ozone values are reviewed
shortly after the model runs are completed. For base-case scenarios, predictions are examined for
consistency with measured concentration levels and patterns. Predictions for future baseline and strat-
egy scenarios are compared against base case concentrations while considering the changes in emis-
sions from the base scenario.
2.2.5 ROM Simulations for ROMNET
The ROM was typically applied over multiday periods that contain "episodes' consisting of periods when
high ozone concentrations were measured across broad areas of the ROMNET region. The selection of
episodes for simulation in ROMNET is discussed in Section 3.
The ROM system is constructed for simulating episodes in 3-day (72-hour) blocks starting at noon EST.
The ROM processor network is run for each episode, and specifies inputs of meteorological data and
biogenic and anthropogenic emissions. Anthropogenic emissions processors output data for each
emissions scenario and episode since hourly mobile-source emissions are adjusted by grid cell for
day-specific temperatures.
2-9
-------
The ROM processors are run on the EPA-NCC VAX ™ cluster, which consists of a VAX 8650, a VAX 6420,
and a VAX 8600.1 The CPU time for a typical 72-hour run is approximately 12 hours on the VAX 8650.
Data from the processor network are used to create the B-matrix data files, which are then transferred to
the EPA's IBM ® 3090 where the Core Model is run.2 The Core Model generates 35 species concentra-
tions for the three ROM prognostic layers for each of the 3,328 grid cells. The CPU time for a typical
72-hour run is approximately 9.5 hours. A total of 1.26 x108 predicted 1 -hour data values are generated
for each episode/scenario simulated. Following completion of each simulation, predictions are archived
and selected species extracted for subsequent review and analysis.
2.2.6 ROM Limitations for ROMNET
There are several limitations inherent in the use of the model for ROMNET. Among the most important of
these are: (1) the ROM, with its current 18.5 km x 18.5 km grid resolution, is not designed to provide
detailed information at local scales that are significantly influenced by local source distributions or by
mesoscale meteorological conditions that are not reflected in the space/time scales of surface and
upper-air meteorology measurements; (2) the model is designed to represent only fair-weather condi-
tions, and does not take into account any aqueous-phase chemistry; and (3) cumulus cloud processes
are such that when a cloud is created in a grid cell, it remains there for a full hour (i.e., cloud physics are
not considered), and the cloud is not advected.
1. VAX is a trademark of the Digital Equipment Corporation.
2. IBM is a registered trademark of the International Business Machines Corporation.
2-10
-------
2.3 EVALUATION OF ROM 2.1
2.3.1 Introduction
An evaluation of the ROM was conducted as part of ROMNET to supplement the extensive evaluation
performed on a predecessor version of the model by Schere and Wayland (1989). The primary differ-
ence between the two evaluations is that Schere and Wayland used a special field-study data set from
1980, and 1980 NAPAP emissions for the ROM simulations. The ROMNET evaluation used the ROM
version 2.1 applied to the ROMNET 1985 base case emissions scenario (see Section 4.3). Routinely-
measured ozone data from 1985 were compared with predicted values. Schere and Wayland had
access to extensive field measurement data for ozone, hydrocarbons, and nitrogen oxides as well as
aircraft transects, whereas the ROMNET evaluation was limited to routine data stored in the EPA's
Aerometric and Information Retrieval System (AIRS). After surveying AIRS, it was concluded that there
was insufficient nitrogen oxide and hydrocarbon measurements for a rigorous model evaluation.
Therefore, the evaluation of ROM2.1 focused on hourly observations of surface ozone from State and
local agency monitoring sites.
This section compares observed and modeled ozone concentrations for selected episodes of high
ozone observed during the summer of 1985. Episodes from 1985 were chosen because they corre-
spond to the base year emissions inventory. The objectives of the ambient evaluation were '(1) to
examine overall evaluation statistics to determine whether a general bias exists in the model
calculations, (2) to look at spatial patterns of maximum concentration to determine whether a spatial bias
exists, and (3) to examine the model's applicability for determining DAM boundary conditions.
2.3.2 Description of Evaluation Episodes
Two ozone episodes during the summer of 1985 were selected for evaluating ROM. Two episodes,
July 7-22 and August 7-16, were identified as the most 'severe' during the 1985 summer season in terms
of the magnitude and spatial extent of ambient ozone concentrations in the Northeast. This is evident
from the seasonal distribution of ozone > 120 ppb as shown in Figure 2-4. The procedures used to
select these episodes are described more fully in Section 3.2. Brief descriptions of the meteorological
conditions and observed ozone concentrations are given in Appendix D.
2-11
-------
2.3.3 Database Development
Two types of databases were needed for the ambient evaluation: hourly concentrations of ozone from
model estimates and ambient observations. Development of these two data bases is described below.
Modet Estimates
To produce the model-predicted database, the ROM was run for the two 1985 episodes. Ozone
concentrations predicted for layer 1 were extracted for use in this evaluation. Three different model
databases were developed for the evaluation:
1. Point estimates from gridded data: For the portion of the evaluation concerned with general
statistics, gridded ozone values were interpolated to actual monitoring locations using a
biquintic interpolation scheme (Lamb, 1985) that is consistent with the methodology used in
the model.
2. Contoured values of gridded data: For analyzing spatial patterns, an objective contouring
algorithm was used to produce computer graphics depicting concentration fields based on
the gridded ROM data.
3. Interpolated values derived using the ROM/UAM Interface methodology: For the portion of
the evaluation concerned with boundary conditions, an interpolation scheme was used that is
described below; the scheme is consistent with the ROM/UAM Interface described in Section
6. This procedure allowed the transformation of boundary conditions from a 18.5-km grid size
to the grid size of the selected UAM domain, 8 km.
a. A band of three ROM grid cells surrounding the UAM New York Metropolitan area domain
(OMNYMAP: Rao, 1987) was defined (Figure 2-5).
b. Three-hour running averages, centered on the hour, were calculated for each hour for
each ROM grid cell noted in the three-cell band.
c. Using the hourly concentration averages from step b, an hourly spatial average was taken
of each three-cell set normal to the UAM outer boundary and the result noted on the UAM
boundary.
d. These concentrations were then spatially interpolated (using linear averages) to the UAM
grid cell centers along the boundary.
2-12
-------
Ambient Observations
The following three observational databases, which are analogous to the three model databases, were
developed: (1) a set of observations used for developing overall statistics; (2) a set of observations used
for creating contour plots; and (3) a set of observations used for developing UAM boundary conditions.
Ozone concentrations were obtained from monitoring data archived in AIRS. Hourly ozone concentra-
tions were selected for sites located in the U.S. portion of the ROMNET domain for the two episodes.3
An extensive review and screening of the data was performed. Only daytime values (0800 LST to
1900 LST) were included in the evaluation because nighttime observations are influenced by localized
processes that often include scavenging of ozone by NO emissions and therefore do not reflect
vertically-integrated ozone concentrations in layer 1 of the ROM (Schere and Wayland, 1989). Further-
more, data for sites on days missing more than 25 percent of their observations were excluded. The
data were examined for extremely high or low values; several sites, such as Poughkeepsie, NY, were
eliminated because mean daytime ozone concentrations were consistently below 50 ppb and may have
reflected local NO scavenging. Of the more than 200 sites in the original database, 187 of these were
used in computing general statistics. For portions of the analysis, the data were divided into five
geographical groups (Figure 2-6).
For the spatial plots, little effort was made to eliminate sites because only maximum concentrations were
considered. As a result, the plots showed a few locations with extremely low values (due either to poor
data recovery or local NO scavenging). These sites were ignored when manually contouring the data.
The monitoring data used for developing the UAM boundary condition database were given special
consideration. The approach followed is consistent with the guidance given by Rao et al. (1987).
Because so few monitoring sites were available for developing boundary concentrations, the monitored
data were distributed among six locations along the UAM boundary: south, southwest corner, west,
northwest corner, north, and east (Table 2-3). The assignment of particular sites to a boundary location
depended on the prevailing wind direction for that day. If more than one site was available for a location,
the hourly concentrations were averaged. Monitors used in this analysis of boundary conditions are
shown in Figure 2-7. After averaging at the six locations, concentrations were spatially interpolated
(using linear averaging) along the UAM boundary. To be consistent with the ROM/UAM Interface meth-
odology, the hourly concentrations were used to create three-hour running averages as described ear-
lier.
3. Canadian ozone monitoring data are not included on AIRS and were not readily available for this analysis.
2-13
-------
2.3.4 AMBIENT EVALUATION
To emphasize different aspects of model performance, the ambient evaluation of ROM2.1 is separated
into three parts: overall statistics, spatial patterns, and UAM boundary conditions.
Overall Statistics
Using the separated modeled and observed databases of hourly ozone concentrations, descriptive
statistics were computed using SAS (1985) procedures to assess the ROM's overall temporal and
spatial performance. Summary statistics for all modeled and observed daytime hourly concentrations
(July and August episodes combined) are shown in Table 2-4, together with statistics for daily maxima at
all sites. The mean observed and modeled values agree fairly closely: the hourly means agree to within
10 ppb and the daily maximum means agree to within 1 ppb, which are not significant differences at the
99 percent confidence level. Although the model overestimated the mean hourly value, it tended to
underestimate the hourly and daily maximum values in the upper extremes of the frequency distrib-
utions. The 95th-percentile hourly value was underestimated by 1 ppb, and the maximum was
underestimated by 50 ppb; daily maximum values show a similar tendency.
Statistics for the individual July and August episodes, shown in Table 2-5, are similar to those for the
combined episodes. Because results for the two episodes are similar, discussion is limited primarily to
the July episode.
To examine spatial trends in the data, the monitoring sites were grouped into five geographical clusters:
(1) Northern Corridor, (2) Southern Corridor, (3) Ohio Valley-Middle Atlantic, (4) Interior Northeast, and
(5) Great Lakes. The spatial grouping is shown in Figure 2-6. Statistics for each group are summarized
in Table 2-6. Depending on the statistic, model performance was better in some groups than in others.
The preferred statistic examined was the 95th-percentile value from the daily maximum concentrations,
because it is more robust than the absolute maximum value and more appropriate for use in evaluating
predictions from a regional-scale model. As shown in this table, all concentrations at the 95th percentile
were underestimated except for group 1, which was overestimated by 5 percent. It is encouraging to
note that, except for group 3 which was underestimated by 19 percent, all estimated and observed
concentrations at the 95th percentile agreed to within 10 percent. The large underprediction for the
absolute maximum concentrations, while significant, is not unexpected given the sharp spatial gradients
typical of ambient peak concentrations when compared to the large grid size of the model.
2-14
-------
A valuable tool for assessing model performance is a quantile-quantile plot, which can be used to com-
pare the frequency distributions of sorted observed and sorted estimated concentrations. The concen-
trations are sorted from highest to lowest and then plotted on an x-y plot. The x-axis depicts observed
data, and the y-axis depicts estimated data.
Quantile-quantile (hereafter referred to as QQ) plots of hourly daytime concentrations for each of the five
geographical groups during the July episode are shown in Figure 2-8. In each of these plots, the solid
line denotes a perfect fit, and the dashed lines show a ten percent deviation from this line. The findings
from the QQ plots are consistent with overall statistics shown previously. In general, except for group 3,
predictions in the upper portion of the distributions were within 10 percent of observations. Model esti-
mates from groups 2, 3, and 4 tended to be lower than observations in the higher quantiles; underesti-
mates were evident in the top 30 percent of group 2, top 50 percent of group 3, top 20 percent of
group 4. Group 1 showed general overestimates in the upper quantiles (except for the top 1 percent),
and group 5 ranged from good agreement to overestimates at the upper values. Overestimates were
evident at the lower ends of the frequency distribution for all five groups. Median values in general
agreed quite well. In groups 1 and 5, median values were overestimated by about 20 percent (around
10 ppb); in groups 2, 3, and 4, the median was overestimated by 10 percent or less. In the Northeast
Corridor, differences between the northern portion of the Corridor (group 1) and the southern portion
(group 2) were evident. Group 2 estimates agreed much better with observations than did group 1.
Similar to what Schere and Wayland (1989) noted in their evaluation of ROM2.0, estimates in the south-
ern part of the Corridor tended to be lower than the observations at the upper end of the frequency
distribution, although estimates were within 10 percent of observations except for the maximum value.
The tendency to underestimate peak concentrations in group 2 is discussed below in the spatial analy-
sis section. The region showing the greatest underestimate in the upper values was group 3, an area
removed from the extensive metropolitan area along the Northeast Corridor. In groups 4 and 5,
agreement was quite good.
In an effort to understand how well the ROM tracked daily maximum concentrations, frequency distrib-
utions of maximum (observed and estimated) ozone concentrations from the data in each group for
each day were computed. Figure 2-9 compares box plots of modeled and observed daily maximum
concentrations for all but the first two days of the July episode. The first two days are not shown
because the model is strongly influenced by initial conditions and tends to underestimate ozone con-
centrations during this "start-up" period. The box plots denote the maximum, 75th percentile, 50th per-
centile (or median), 25th percentile, and minimum value. These plots show similar results to those noted
in the QQ plots for hourly concentrations; however, they also provide temporal information.
2-15
-------
On a daily basis, the model tended to track the observations fairly well. However, due to the factors
described previously, the highest daily maximum values are largely underpredicted on several days. For
all of the regions combined, there were about equal numbers of days with underestimates and overes-
timates. As discussed for the QQ plots, group 1 tended to experience the most overestimates (12 out of
14 daily medians were overestimated) while group 3 tended to experience the most underestimates
(medians for 12 out of 14 days were underestimated). The daily maximum plots also show that west of
the Corridor (i.e., west of groups 1 and 2), median daily maximum concentrations tended to approach
background values.
One of the ROM's most important uses is to simulate days having ozone exceedances (hourly concen-
trations greater than 120 ppb). In group 1, eight exceedance days were observed while ten exceedance
days were estimated, thus indicating that the ROM tends to slightly overestimate in the northern portion
of the Corridor. Further to the south in group 2, however, ten exceedance days were observed but only
six were estimated. In group 3, six exceedance days were observed but none were estimated. The
monitors reporting these six exceedances were in smaller urban areas such as Richmond, VA, Norfolk,
VA, and Charleston, VW, than those found in the Northeast Corridor. This suggests that the coarse grid
resolution (19 km) of the ROM may be too large to adequately resolve smaller urban plumes. Also,
naturally-occurring hydrocarbon and NOX emissions are probably more important in these areas than in
the Corridor. Thus, uncertainties in estimating naturally-occurring emissions may have contributed to
poor performance in group 3. Despite the relatively poor model performance in the southern portion of
the domain, performance in the Great Lakes, Interior Northeast, and the Northern Corridor groups
seems quite good.
Spatial Patterns
For this portion of the evaluation, spatial patterns of maximum hourly ozone were examined for four
distinct three-day episodes that represent a range of meteorological conditions and model performance.
July 9-11
This was a period of unsettled weather with weak pressure gradients. For much of the period, a weak
surface trough was draped from Long Island to Virginia to northeastern Ohio to Illinois. By the 10th,
southwesterly flow became better defined in advance of a cold front moving eastward through the Great
Lakes. Of special interest on this day was a mesoscale high pressure area associated with a line of
thunderstorms in central Pennsylvania, which eventually moved through the southern part of the
2-16
-------
Corridor and significantly reduced ozone concentrations. This feature will be discussed below in the
section on background concentrations. By the morning of the 11th, less humid conditions and
northwesterly flow prevailed as the cold front had pushed off the East Coast.
During this period, monitors picked up ozone plumes above 120 ppb from Wilmington, DE, to Boston.
Observed and estimated maximum hourly concentrations greater than 160 ppb were observed near
central New Jersey and coastal Connecticut (Figure 2-10). Agreement over southern Connecticut was
good, although the modeled plume appears to be shifted slightly east and south of observed values.
The model also shifted the high concentrations just off the coast of New Jersey near Atlantic City. Sim-
ulating small-scale features during this period was particularly difficult because of the weak wind flow,
the occurrence of the mesoscale high, and the persistence of a coastal trough. Overall, however,
estimates of maximum ozone for this period agreed fairly well with observed data.
July 13-15
During this period, the meteorological scenario typically associated with elevated ozone concentrations
in the Northeast occurred. A weak cool high pressure system centered over New York on the 13th gave
way to a warm front on the 14th. By the 15th, southwesterly flow was fully established. Unsettled
conditions prevailed in the western portion of the model domain. Showers were reported on the 14th
and the 15th in western New York and Pennsylvania in advance of a slow moving cold front.
During this period, two separate areas of high ozone values were observed. A modest area of excee-
dances (maximum of 132 ppb) was noted from Boston to the southern coast of Maine. High concen-
trations were also noted in distinct plumes near Philadelphia and New York City. Concentrations greater
than 200 ppb were observed near Wilmington, DE, (211 ppb) and Bayonne, NJ (218 ppb). Although an
area of observed exceedances stretched from Washington, DC, to central Long Island, no well-defined
plume downwind of New York City was apparent.
Except for the area north of New York City, estimated ozone concentrations were lower than those
observed during this episode. However, Figure 2-11 shows that the spatial patterns of the high ozone
concentration areas were well represented. Near Boston, the modeled and observed concentrations
agreed closely as noted by the placement of the 140 ppb contour on both plots. Downwind of Wil-
mington, DE, the model underestimated the peak observed concentration by about 80 ppb. Part of the
poor performance may be due to a coastal trough that existed during this episode. Pagnotti (1987)
points out that small-scale features such as this trough tend to produce localized pockets of high ozone
that would not be well-represented by a regional model such as the ROM.
2-17
-------
July 18-20
High pressure held firmly over the Northeast, gradually shifting southward and weakening. By the 19th,
southerly flow had become fully established over the Corridor. On the 20th, a coastal trough was evi-
dent along the Atlantic Coast. Under light wind conditions, high concentrations were once again posi-
tioned in small, distinct areas. Values above the standard stretched from the Delmarva peninsula to
central Massachusetts. The highest concentration (164 ppb) was observed near Trenton, NJ. A small
area of high concentrations (maximum of 152 ppb) was observed along the coast of Maine. With minor
exceptions, the model showed reasonable agreement to the peak observed values during this period
(Figure 2-12). A plume over 120 ppb extended from near Baltimore northward to off the coast of Maine.
Pockets over 160 ppb were estimated near New York, which generally agreed with the observed values.
Except for missing the high concentrations over the Delmarva peninsula, ROM appears to do a good job
predicting the magnitude and placement of the maximum ozone plume during this period.
August 13-15,1985
This period was similar to July 13-15. Weak cool high pressure was positioned over the Northeast on the
13th. The system had moved eastward by the 14th allowing weak southwesterly flow to ensue. By the
15th, a cold front began moving through the western portion of the domain. Southwesterly flow
strengthened, but a hint of a coastal trough was evident in the surface weather map.
Monitoring data revealed an extensive area of ozone exceedances that stretched from Wilmington, DE,
to coastal Maine. Highest concentrations were observed near Hartford, CT (219 ppb). An isolated area
of high concentration (maximum of 187 ppb) was observed just north of Baltimore.
The general orientation of the modeled ozone plume is excellent when compared with observations
(Figure 2-13). The highest estimates, however, only approached 200 ppb, more than 20 ppb less than
the highest observation. The model did simulate an area of high concentrations (greater than 180 ppb)
off the coast of Maine, close to where relatively high concentrations were observed. While the model did
not replicate the extreme peaks, it satisfactorily estimated the pattern and shape of the ozone plume
near the Northeast Corridor.
Boundary Conditions
One of the major uses of the databases generated by ROM is to estimate boundary conditions for the
Urban Airshed Model (UAM). As part of the evaluation, ROM-generated boundary conditions were
2-18
-------
compared to measured ozone concentrations near the New York City OMNYMAP domain. The
procedures used to aggregate and match observed and predicted data sets is described in
Section 2.3.3.
This analysis was limited to near-surface concentrations because measurements were available only
from surface-based monitors. The 73 DAM grid points used stretch from the coast of New Jersey
clockwise to the coast of Connecticut. Grid boundaries over the Atlantic Ocean were excluded because
of the lack of oceanic ozone monitoring data.
The analysis was started by looking at the overall performance of the ROM in estimating daytime
boundary conditions. Table 2-7 shows summary statistics for the 20,155 data points available. These
data were taken from valid combinations of estimated and observed daytime (0800 LST to 1900 LST)
hourly concentrations for 73 grid locations for 24 days of data. The model slightly overestimated the
mean concentrations (by 5 ppb or 7.6 percent). At higher quantiles, however, the model tended to
underpredict in a manner consistent with the statistics presented earlier. In addition, the larger standard
deviation of the observed concentrations shows that observations exhibited more variability than the
model estimates.
Because there was no overwhelming bias apparent in the model, the boundary condition calculations
were examined in further detail. Table 2-8 compares estimated with observed mean concentrations on
a daily basis. While the mean values summed over all the days agreed to within 10 percent, there was
some day to day variability. Percent differences for individual days ranged from -24 percent to
+49 percent. However, estimated and observed mean values were within +10 ppb on 14 of the 22
days. Other pertinent information on mean wind persistence, wind speed, and wind direction is also
noted in this table.
Figure 2-14 displays model performance as a function of wind persistence. Surprisingly, the more per-
sistent flow conditions were associated with greater overestimates, as noted by the mean fractional
errors which dropped from 23 percent percent during very persistent conditions to -18 percent during
low persistence.
Next, the relationship between model performance and daily average wind direction was examined.
Figure 2-15 shows that southerly and southwesterly winds were most prevalent during the model simu-
4
lations. In terms of model performance, model estimates and observations agreed more closely when
southerly winds prevailed. With southerly winds, there were 10 days of overestimates and 8 days of
underestimates, for an average error of 5 percent. With northerly winds, the model overestimated on all
five days, and the average error was +26 percent.
2-19
-------
The data set was then geographically stratified to look for spatial biases in the estimates. Figure 2-16
shows the eight groups of grid cells formed, with group 1 located in the south and group 8 located in the
east. Mean observed and estimated concentrations for the eight groups are shown in Figure 2-17. The
highest observations and estimates occurred in group 1, which is located in the center of the Northeast
Corridor and would experience high ozone concentrations typical of southwesterly flow conditions. It is
noteworthy that the observed and modeled patterns in Figure 2-17 resemble each other: Both show a
minimum along the northern boundary and a maximum for group 1. Best agreement occurred for
groups 3 and 5.
The above statistics include all grid points regardless of the wind direction. Using the same eight grid
cell groups, model performance for individual groups experiencing incoming flow on a given day were
next examined. For example, if southwesterly winds predominated, the daytime concentrations for
group 2 were then analyzed. Figure 2-18 shows the mean residuals for each group for days having
upwind flow. As in Table 2-8, residuals shown in Figure 2-18 varied appreciably from day-to-day, rang-
ing from -15 percent to +25 percent. However, 8 of 13 days experienced mean residuals less than 10
percent.
There were a few days when the model performance was admittedly disappointing. In particular, day
191 showed a 25 percent overestimate in group 2. Although resources to analyze every day in detail
were lacking, this day was examined in detail in order to gain some insight on model performance.
The 1200 GMT surface weather map for 10 July 1985 (Julian day 191) gave some indication as to why
observed concentrations were lower than expected in eastern Pennsylvania and western New Jersey.
This map shows a mesoscale high over south central Pennsylvania which was apparently induced by a
cluster of thunderstorms. A stationary front extending from a low over central Indiana stretched east-
ward to northern Virginia and merged into a surface trough that stretched northeastward along the
Atlantic seaboard. The main weather feature was a combined cold and occluded front positioned along
the St. Lawrence Valley southward through the Ohio Valley.
To examine why the daytime ozone observations were underestimated in group 2, hourly plots were
generated from both the observations and the estimates. Figure 2-19 shows the observed and esti-
mated concentration fields for 0800 EST on July 10,1985 (one hour after the 1200 GMT surface weather
map). Two features are given on the observed plot: weather data reported at each National Weather
Service surface observing site, and measured hourly ozone concentrations. The circled ozone mea-
surements denote the sites used in generating the observed UAM boundary conditions for group 2. The
18 gridded model values used in defining group 2's boundary conditions are enclosed by a rectangle.
2-20
-------
At 0800 EST, estimates averaged more than 20 ppb higher than the observed values. The weather data
indicate that the surface trough is still located along the Atlantic Seaboard. Ahead of this trough, winds
were generally from the south; behind the trough, winds were generally from the west.
Moving to 1000 EST, estimates were still about 20 ppb higher than the observations (Figure 2-20).
Ozone concentrations in northeastern Pennsylvania were particularly low. This area was experiencing
overcast skies, temperatures around 22°C, and high relative humidity. This suggests the likelihood of
shower activity which may have caused lower ozone concentrations by allowing increased vertical
transport and reduced photochemical production.
By 1200 EST, thunderstorm activity had begun (Figure 2-21) as indicated by the Wilmington, DE,
observation. Overestimates by the model continued throughout group 2. Although overcast and rela-
tively cool conditions predominated in eastern Pennsylvania, Atlantic City experienced hot, sunny con-
ditions as indicated by a temperature there of 93° F. The estimated concentration field shows a strong
gradient across group 2, ranging from 79 ppb in the northwest to 102 ppb in the southeast. The
observed concentrations, in contrast, range from 40 ppb to 69 ppb. Observed concentrations as high
as 111 ppb were reported in central New Jersey but these were not prescribed for use in the UAM/ROM
Interface methodology. After investigating additional sources of meteorological data, it was found that
Philadelphia reported a wind gust of 45 miles per hour. This information, which was not available in the
routine modeling database, strongly suggests the nearby presence of organized convective activity
such as a squall line.
Figure 2-22 shows that by 1400 EST, shower activity had moved into Atlantic City. The temperature in
Atlantic City dropped more than 9°F in two hours and observed ozone concentrations in the area
declined over 50 ppb. Easterly winds reported at both Philadelphia and Wilmington are probably indic-
ative of outflow from thunderstorm activity in central New Jersey. Meanwhile over northeastern PA, skies
were beginning to clear although ozone concentrations remained low. Predictions in group 2 seem less
affected by the storm activity and are as high as 112 ppb.
By 1600 EST (Figure 2-23), observed ozone concentrations near Scranton had rebounded to near
modeled values (65 ppb versus 75 ppb). In the area near Atlantic City, which was still experiencing
shower activity, estimated concentrations were as high as 125 ppb but observed concentrations were
only around 30 ppb.
The meteorological data used in the ROM offer some clues as to why the ROM overestimated concen-
trations in group 2. Figure 2-24 shows the time series of selected meteorological parameters that were
calculated for grid cell (45,21), the center cell of the third row from the top of the rectangle outlined in the
2-21
-------
previous five figures. The simulated cloud cover and solar fluxes in Figure 2-24 did show some effect of
the observed cloudy conditions but probably not enough. Cloud cover increased to more than 90 per-
cent by 1300 EST but only slightly reduced the computed solar flux. The model layer heights
(Figure 2-24) show some increase, with layer 3 growing to 1700 m by 1100 EST. However, the localized
convective activity probably caused more vigorous vertical and horizontal mixing than was simulated in
the model. Horizontal dilution in the model also was probably smaller than actual dilution as indicated
by the relatively low modeled wind speeds in Figure 2-24. Apparently the small-scale (relative to the
model) cluster of thunderstorms was not captured in the overall interpolation of meteorological data.
This case study is noteworthy because it highlights some of the ROM's limitations. ROM is a regional-
scale model that was designed for application in relatively benign, steady-state summertime conditions.
The occurrence of a mesoscale surface trough and localized thunderstorm activity resulting in dynamic
subgrid-scale atmospheric processes seems to have affected the ROM's ability to estimate boundary
conditions along portions of the UAM domain on day 191. The results indicate that development of DAM
boundary conditions from either observed or modeled data should include careful examination of the
effects of mesoscale meteorological conditions, which as has been shown, can cause localized pertur-
bations in ozone concentrations.
2.3.5 Summary
The Regional Oxidant Model, Version 2.1, (ROM2.1) has been evaluated with ozone monitoring data
collected during 1985 in the northeastern United States.
In the study, good overall agreement between observed and modeled ozone concentrations was
observed. The mean concentrations of the estimated and observed daily maxima agreed to within one
percent. Concentrations at the higher ends of the frequency distribution were slightly underestimated;
at the 95 percent level, observed concentrations were 127 ppb while estimated concentrations were
119 ppb. As noted previously, the tendency to underestimate peak concentrations is to be expected
with a coarse-grid model such as the ROM because of the spatial averaging that occurs with Eulerian
grid computations. Given this caveat, model performance was particularly good in the northern portion
of the Northeast Corridor, especially around New York City. The model also did an excellent job cap-
turing high ozone levels along coastal sections of Maine. The improvement in model performance in
these areas relative to the previous version of ROM (Version 2.0), seems to result from the correction of
westerly bias in low level wind flows.
2-22
-------
Model performance was not as good in Ohio, West Virginia, and Virginia. Although estimated and
observed median daily concentrations in these areas agreed to within 5 ppb, concentrations at the
95 percent frequency level were underestimated by about 15 ppb. Underestimates of the peak con-
centrations in these areas might be attributed to the relatively small-scale urban plumes and uncertain-
ties in estimating naturally-occurring emissions of NOX and hydrocarbons. However, deficiencies in
anthropogenic emission inventories may still exist despite efforts that have been made to improve them.
Spatial patterns of the three-day maximum concentrations usually showed reasonable model perform-
ance. Differences in model and estimated plumes were most evident during episodes experiencing sea
breezes, coastal troughs, and squall lines. Underestimates that were reported in the ROM2.0 evaluation
by Schere and Wayland (1989) were again noted in the Washington, DC, area. In addition, ROM2.1
tended to underpredict peak rural concentrations of ozone by about 20 ppb. More monitoring data are
needed for evaluating model performance in rural areas because of the current sparseness of these
data. It is hoped that measurements being taken under the auspices of the National Acid Precipitation
and Assessment Program (NAPAP) will aid in future evaluation of ozone concentrations. However,
agreement of the magnitude and orientation of ozone plumes in the northern portion of the Northeast
Corridor was good.
The most rigorous portion of this analysis was an evaluation of the model to estimate boundary condi-
tions for UAM application. Although the application of ROM in a deterministic manner is not recom-
mended, some type of reliable estimation scheme is needed for prescribing boundary conditions when
the UAM is to be applied for future-year emission-control strategies. As was shown in the analysis, using
monitoring data for estimating boundary conditions requires a great deal of subjectivity and is fraught
with uncertainty.
Overall, the model compared quite well with the monitoring data for estimating UAM boundary condi-
tions in the OMNYMAP domain. Mean concentrations for all daytime hours agreed to within 5 ppb (less
than 10 percent error). However, significant day-to-day variability in model performance was noted. A
case study performed on July 10, 1985, demonstrated that small-scale features can cause dramatic
effects on model performance. On this day, it was shown that a squall line produced a significant over-
prediction in ozone concentrations along the western boundary (the inflow boundary on this day) of the
UAM domain. Thus, users of the UAM should carefully screen ROM estimates and be aware of
mesoscale flow conditions. On most days, however, the model did a reasonable job estimating bound-
ary conditions.
2-23
-------
This evaluation has suggested that further improvements to the ROM are warranted. Improving the
specification of layer thicknesses and the computation of naturally-occurring emissions are topics cur-
rently being examined. In future years, it is hoped that a dynamic meteorological processor can be
incorporated that will simulate nonsteady-state flows. To continue making advances in model
development, additional monitoring data are needed to examine other chemical species (such as NOX,
isoprene, formaldehyde, and HNO3) and to fully evaluate model performance in rural areas.
2-24
-------
Figure 2-~\. The ROMNET modeling domain; points represent the southwest corner of each grid.
2-25
-------
Subsidence inversion
Layer Functions
Cloud layer
:.-;t\.
Uvv3
3a?,' £\?*
* «j° v. ...y
Mixed layer
Marine layer
Subsidence inversion
Old cloud layer
.^NiGHT
•% * ~.
"^'•i
•*••.-:
...V..V-*;- > —
^—^^^^ *•
'•"y7. ';.';'-''. Oil rnixed layer
Cloud layer
- downward transport of stratospheric ozone
- upward transport by cumulus clouds
- gas-phase photochemistry
- long-range transport by the free atmosphere
Mixed layer
- gas-phase photochemistry
- turbulence and wind shear effects on
transport and diffusion
- deposition on mountainous terrain
Marine layer
- lake- and marine-layer effects
Old cloud layer
- downward transport of stratospheric ozone
- dark gas-phase chemistry
Old mixed layer
- transport of aged gas-phase reactants
and products
- dark gas-phase chemistry
Radiation inversion/nocturnal jet layer
"•• V".-".*"-.• - transport ol aged pollutants and reactants
LI by the nocturnal jet
•' ' Radiation inversion/ - transport of nighttime emissions from
' -'nocturnal jet layer tall stacks and warm cities
s - deposition on mountainous terrain
- downward transport during jet breakdowns
- nighttime shallow mixed layer over
urban heat islands
Figure 2-2. The ROM vertical layers and their functional features.
2-26
-------
RAW INPUT DA
I
l
1 (AIR QUALITY) ,
V J
i
i
j (METEOROLOGY)
i
1 (EMISSIONS )
1 ^ J
1
|
1 ( LAND USE )
1
I
j (JOPOGRAPHY)
TA
ROM
CORE MODEL
1
I
i
i
l
_ TRANSPORT |
ROM *" . !
^ PROCF55SOR 1 i j
ir !
NETWORK _^ |
^
| CONCENTRATIONS
l
1
i
i
Figure 2-3. Components of the ROM.
2-27
-------
OZONE
c
I
03 40 -
Q.
O.
C\J
(— t
A 30 -
cc
i 20 ~
u-
o
o
< 10 -
H
2
O
£t
a.
o -1
•1.1 *
J
JLY 7-22
1
• 1 II 1 • III
•
1,1
1,
,.
29-MAY 12-JUN 26-JUN 10-JUL 24-JUL
/
1.
\UG 7-16
III
7-AUG
_-,-_., _ ..-_!_. ___•!..-.__ i^^i • •
21-AUG 4-SEP
Figure 2-4. Percentage of monitors with daily maximum ozone exceeding 120 ppb in the
northeastern U.S.
Figure 2-5. ROM grid points overlaying the DAM domain for the New York City metropolitan area.
2-28
-------
9O
•08
B2
70
66
73
-Longitude
Figure 2-6. AIRS ozone monitoring sites divided into five geographical groups.
Figure 2-7. Monitoring sites used for developing boundary conditions for the OMNYMAP domain.
REN = Rensselaer, NY; PTS = Pittsfield, MA; WAR = Ware, MA; CCP = Chocopee, MA;
AGA = Agawam, MA; PRV = Providence, Rl; PRK = Kent County, Rl; GRT = Groton, CT;
CRB = Carbondale, PA; SCR = Scranton, PA; NJ1 = Morris County, NJ; ETN = Easton,
PA; ALT = Allentown, PA; FLM = Flemington, NJ; BST = Bristol, PA; TR1 and TR2 =
Trenton, NJ; MCG = McGuire AFB, NJ; CLM = Ocean County, NJ.
2-29
-------
O
UJ
o:
z
o
o
UJ
or
Q.
o
SO 100 ISO 200 250 MO
OZONE (OBSERVED), ppb
100 ISO 700 25O
OZONE (OBSERVED), ppb
(a) Group 1: Northern Corridor
(b) Group 2: Southern Corridor
O 100
30 100 I SO 200 750
OZONE (OBSERVED), ppb
z
O
3474 C6S. DATA PTS
2969 PRCO. DATA PTS.
50 too 150 700 250 JOO
OZONE (OBSERVED), ppb
(c) Group 3: Ohio Valley-Mid Atlantic
(d) Group 4: Interior Northeast
y
o
z
o
0 50 100 150 TOO 2SO >»
OZONE (OBSERVED), ppb
(e) Group 5: Great Lakes
Figure 2-8. Quantile-quantile plots of daytime hourly ozone for the July 1985 episode.
2-30
-------
300
280
260
2<0
220
200
> IBO
8 120
100
80
60
Jeo »i 101 hu IM
300
280
260
240
220
200
£
-------
OBSERVED
PREDICTED
Figure 2-10. Spatial patterns of maximum ozone for July 9-11, 1985.
2-32
-------
OBSERVED
PREDICTED
Figure 2-11. Spatial patterns of maximum ozone for July 13-15,1985.
2-33
-------
OBSERVED
PREDICTED
Figure 2-12. Spatial patterns of maximum ozone for July 18 - 20,1985.
2-34
-------
OBSERVED
PREDICTED
Figure 2-13. Spatial patterns of maximum ozone for August 13-15,1985.
2-35
-------
MEAN DAILY RESIDUALS
30
20
10
•10
-20
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
PERSISTENCE
Figure 2-14. Mean residuals versus wind persistence for the (JAM boundary; positive values are over-
prediction, negative values are underprediction.
MEAN DAILY RESIDUALS
30
20
10
-10
-20
t
•
180 210 240 270 300 330 360 30 60 90 120 ISO 180
WIND DIRECTION
Figure 2-15. Mean residuals versus daily average wind direction for the UAM boundary; positive values
are overprediction, negative values are underprediction.
2-36
-------
Figure 2-16. Division of the OMNYMAP boundary into eight groups.
2-37
-------
CONCENTRATIONS (PPB)
100
90
80
70
60
SO
40
30
20
10
0
MEAN
OBSERVED
MEAN
PREDICTED
123
4567
UAM GROUP
Figure 2-17. Mean daytime ozone concentrations by UAM group along the UAM boundary.
Q
20
10
-20
MEAN OVER ALL DAYS
23456
UAM GROUP
Figure 2-18. Mean residuals for each UAM group experiencing incoming flow.
2-38
-------
/ Ja 2G
/ . " "'
^/\ ( "!!
10 <^><3> ^
2|0;P 2533 _J .
L/W1 '^ «*/
41 45 4] 1] 1] 11 14 44 41 4B «l 97 II 7] 70 II 91
47 44 41 11 11 10 II 11 45 17 /SI li II 71 II II II
II 1! 11 II 11 11 II 1] 1} II tl |j || |I || |7 ||
10 11 4!\_IO 11 It It 11 17 91 91 10 II Wl 79 10 77
41 41 40 }lj 11 40 41 IS 91 91 91 SI 97 11 1! 11 II
II 41 41 4P, II 47 9C !l SI 99 SI SI 31 II 17 II II
It _SO 51 II l7~\9! SI SI 51 51 II II 71 IS 11 II I!
O r^\ „ \ wU-~i
II 51 51 55 S/ SS\i7 10 51 10), 11 H/>J — 'tT'^ifs'" 17 ''"•O
II 91 II SI/ 9 4 91 91 SX^II \IO .^10 11 17 H/'JlH f) .^J«
II 91 97 91 99 91 91 91 ll ,''9IJLi''h~- 71 19 II IS 71 97 IS 17 17
r f:|
57 57 ll_^.3l 5! II 1! Iff 71 11 II 11 71 II IS II 10
-11-/5I _ !! 5< 1> 11 1! '£ 71 1! 1! 71 . II 11 II 17 11
"U 51' 1) SI 1! 7L. If'y'ti 17 II II 70 91 91 91 91 II
*"
33
:o I
4
''fli^ ^
a
91 SI 51 .
I
11 1! 71
11 71 II
II M 70
ii n^r/i
II Iffju
jj.Vie li
n& „ s,
II 55 51
91 !C II
II II 15
11 11 10
10 11 11
11 11 11
H 17 11
11 I 1)
17 I! 10
'
Figure 2-19. Ozone concentrations for 0800 EST July 10,1985; observed (top), predicted (bottom).
2-39
-------
S4 41 41 It IS 42
SI 41 49 17 (I 31
JO 41 4) 1] }l JI
40 41 41\_tl 11 31
II 49 IT IS] 10 10
11 91 97 51\ II If
11-91 It II
O
31 99 11 II It 19
44 91 10 ll,/70 17
41 99 SI IV It II
51 51 51 II 71 71 77 71 71 71 71
/
II /!! 51 II 71 71 71 71 74 77 II
il U 70^ 71 71 II II 7t II 10
7.t J 7 1 71 71 71
V
t\}\ Vtft 71 71 7!
SI 91 41 41
II 41 41 49
loi 101 104 l! " " »' "
-------
l_
3G
21 ©
f — -.
n" to'-}
37
3$
SI SJ II
SI Jl Si
II 90 II
^
\
V
"x
y
"?
i
108 in
n it n
1,1 IS < 0
II 11 17
31 41 4I'_4« 1» J7
41 44 4!
11 II I!
"o" !1
13 IS 17
II Ii II
II 17 II
II II _7I
II II ~70
IS 71 71
10 17 1C
10 71 IS
71 71 71 "
-•4-' IS c-'l?
;' ' • I !' . tc
II 1! 11
\
ll\ 51 II
15
<$J^
30 ^JIJ^VCS
^PyBf-
,•7
J
JIB
II 10 S3 SI
17 1! 44 47
17 11 41 41
11 41 41 SI
40 11 50 SI
41 41 70 71
II io"~^SJ Ii II 71 IS
II II 71\ll 7C 10 II
H /It 71
r
7 > 71 10
)7i 11 ii
Ui n 11
|
77-\ II II
11 11. 100
n^r ,8J
.JJ II ,0,
10 IS lt!
•" 1' 101
10 ^b . 11 ^^n
" " / "
«' 101 HI 101 •' »' •»
101 III III 105 1,' M ^
;>ios j«j,.'ut~~n^irr t^n
^•^I'ft t 111 13 " X^O^ir. O -'J?
_/ ^ ^J4-^ ^
^ViT-Vi'i ics •^i*-xi> 55
J^HQ^ff" " " >'
III lit 101 " so '3 31
U, 13 II 77 51 17 IS
13 II 17 71 51 11 11
IS II 10 70 55 II 11
" " " " S! ((
-------
SI
SI
41
41
11
40
si si
SI SO
41 42
11 41 '
41 41
SI SI
l,n,3 ,1
SI
II
S7
II
71
II
17
71
II
-«,
•n
II II
17 11
14 IS
1! 11
n
77 ~7J
13 II
II II
11 11
11 17
JJ,/I"
l»f? II
41 41 41
41 41 40
41 11 17
_4_I 40 11
4tj 47 41
II Y 55 II
si sT~\ss
X/-J\^
II S J 5 7^
ITy^ll 10
(
71, 71 71
III 12 II
Uj 77 71
1
I4-\7I_ 10
13 |V. 17
X1""1
:>»i •• id
101 100 101
•I " IOJ-
41 11 41 SI
31 17 40 13
37 J! 41 41
11 11 44 47
31 11 SI S4
41 IS IS 71
11 1! 72 71
/I SI S4 11 71 71 12
/
/I! SO SI 77 II II 4S
(-Ji SI U 11 II 11 IS
II S3 II Qs II 11 14
II 17 10 IS ,jj 10 11
I! 11 II ,g, II II 10
" 111 no 101 V M "T
*1 -^Jw~ai-
xo 7! 17 ir^jj ^^{'jp-n-'Tr^i^i
" l>^100 W
1 r*
" " /£$•*
lt itjlff^T
IK I'tj' ^~
" /in fnni'
•"^ .
•' IOMIIO 12!
Itl 101/111 101
^7
itl tor 101 *J
101 US 101 I0
101 J0 " "
UJ^IOI " "
-flS 111 101 1 0 J.x^l^M(l' ^2^
._ — *s^ •*" ^^^
-SriT^n in \*\f}>" !0
y(^~ilipis 12 40 is
-^^^^^"^
111 112 l! '! !! '" lf
11, II II 71 SS 11 11
10 II 10 II 55 12 17
10 71 7S 11 SI 11 11
71 71 71 11 II 11 11
71 7! 71 S7 17 II 4!
77 7! II S3 1! II 10
11 17 10
i
14 17 14
<3 1C 10
12 1M 71
10 ("WV'1
" JW
^JM/si "11
•4 " 41
47 13 40
40 It 17
17 17 11
li 14 31
31 31 31
11 11 11
40 11 17
4! 41 40
42 41 40
It 31 31
Figure 2-22. Ozone concentrations for 1400 EST July 10, 1985; observed (top), predicted (bottom).
2-42
-------
/' ' 50
70 5?
V _ Wa>
58 <. 70
27 ft 6' "d'0*5''J Mj&s^
3Jo 'P 33n J
^ *R
• (^j*.' -V
r< oM
41 41 41 41 17 41 4! 11 41 51 51 SO 50 57 71 77 71
47 41 41 47 44 41 11 17 11 41 /II 45 41 51 7! 14 II
44 44 44 41 >t Jl )( 11. '1 " (-" (7 il 51 74 II II
41 40 ll\_40 41 11 41 40 4! 47 SI 55 SI -14 1! 15 10
41 41 45 4I| 47 41 45 41 45 SI IS 71 II 17 17 II II
41 41 SO 57\ 5t 51 SC 41 55 II 10 11 ,ot ,g, m It II
54 47 54 I! iT"1 Jl !! !! II II 10 IOI no 111 107 »1 U
U S\ n <--^-l'^'J-
ii li ii is 1,4 siXsi ii 7i m). ,01 jii.yi3i~TTo^ioi ,," "
II 11 II I4_/II II 17 7J-^IO ^{.-YlT 114 151 l»t'5^,j? i , i*
7i ii ii i7i 7i i! «7 wj/ii^ .in n:. -iW^lf'Tio •> ss 17
10 71 ~IS '41 1! 71 IJ ^HJ^flf^ti'^TlJIl! 110 's «' SI 11
II 10 II !<-,!! 1C DJ Vl^lll 117 122 IC7 " lo " SI 11
II 11 II 14 II.. m m 117/111 lit » '• '« '1 " •' )•
11 II ,o, 17^,^115 III 1HJ1 111 10! " 'S '1 " 57 44 40
100 iC!_J»L'n! Ill II! Ill U? 107 " II 'J " «4 SO 4! 4!
-m/I^Xo. 101 101 10. 10! ftt H l« '< " •' SI 15 1! II
,'U l"it; It II II 11. i«»'/io; II )< '} 17 51 41 I! II 11
66 ;
CO i
C
26, ,$? *
&
II II 7!
14 11 77
14 11 75
14 M 7 1
" ]|)V-AA
7)^ 1 ! 41
IJ, 47 17
II 11 11
1! 14 11
1! 1! 1!
1! 1! 1!
14 14 11
11 17 17
41 1C 11
41 1C 11
!! !I 11
11 Jl 17
1
-
Figure 2-23. Ozone concentrations for 1600 EST July 10,1985; observed (top), predicted (bottom).
2-43
-------
25(X)
2000
I 1500
X
I
I 1000
500
1800
1600
1400
1200
1000
800
600
400
200
0
lO IS
HOUR
20
10 15 20
HOUR
20
HOUR
solar flux
cloud cover
0.8
0.6
0.4
0.2
§
O
lop of Layer 3
top of Layer 2
top of Layer 1
Layer 1 wind speed
Layer 2 wind speed
Figure 2-24. Time series of selected meteorological parameters for grid cell (45,21) on July 10,1985.
2-44
-------
TABLE 2-1. ROM CHEMICAL SPECIES
Symbol
Description
Symbol
Description
ALD2 High molecular weight aldehydes O1D
C2O3 Peroxyacetyl radical O3
CO Carbon monoxide OH
CRES Cresol and high MW phenols OLE
CRO Methylphenoxy radical OPEN
ETH Ethene
FORM Formaldehyde PAN
H2O2 Hydrogen peroxide PAR
HNO2 Nitrous acid PNA
HNO3 Nitric acid ROR
HO2 Hydroperoxy radical TO2
ISOP Isoprene structures TOL
MGLY Methylglyoxal XO2
N2O5 Dinitrogen pentoxide XO2N
NO Nitric oxide XYL
NO2 Nitrogen dioxide MTHL
NO3 Nitrogen trioxide NONR
O Q3P atom TRAC
O1 D atom
Ozone
Hydroxyl radical
Olefinic carbon bond
High molecular weight aromatic
oxidation ring fragment
Peroxyacetyl nitrate
Paraffinic carbon bond
Peroxynitric acid
Secondary organic oxy radical
Toluene-hydroxyl radical adduct
Toluene
NO to NO2 reaction
NO to nitrate (N O 3) reaction
Xylene
Methanol
Nonreactive hydrocarbons
Tracer species
TABLE 2-2. INITIAL MEAN TROPOSPHERIC BACKGROUND CONCENTRATIONS, PRECURSOR
SPECIES
Species
Concentration (ppm)
Species
Concentration (ppm)
CO
NO2
NO
Ethanol
Olefins
0.1
1.0x10-3
1.0x1fr3
3.5x10^
2.1 x1(H
Aldehydes
Formaldehyde
Toluene
Xylene
Paraffins
All others
1.12x10-3
1.4x10-3
1.4x10^
1.05x10-4
7.42x10-3
1.0x10-16
2-45
-------
TABLE 2-3. OZONE MONITORING SITES USED FOR DEVELOPING UAM BOUNDARY CONDI-
TIONS ALONG THE OMNYMAP DOMAIN 1
Monitoring site
Ocean Co, NJ
McGuire AFB, NJ
Trenton, NJ
Trenton, NJ
Flemington, NJ
Bristol, PA
Morris Co, NJ
Allentown, PA
Easton, PA
Scranton, PA
Carbondale, PA
Rensselaer, NY
Agawam, MA
Pittsfield, MA
Chicopee, MA
Ware, MA
Groton, CT
Kent Co, Rl
Providence, Rl
07
s
s
w
w
w
N
N
N
E
08
s
sw. s
w
sw
w
w
w
N
E
09
s
s
s
sw, w
sw
w
w
w
N
E
E
10
s
sw, s
sw,w
sw
w
w
N
E
11
S
s
w
w
w
w
N
N
N
E
JULY
12 13
s
s sw
s s
sw, s sw
w w
sw
w
w
w
N N
E E
14
s
s
w
sw
w
w
NW
NW
N
E
15
s
s
sw, s
w
sw
w
w
NW
NW
N
E
16
s
s
sw,w
sw
w
w
w
N
E
17
s
w
w
N
N
N
N
E
E
E
18
s
s
w
w
N
N
N
N
E
E
E
19
s
sw, s
w
sw
w
w
NW
NW
N
E
20
s
sw, w
w
w
w
N
N
N
E
21
s
sw, s
sw, s
w
sw
w
w
NW
NW
N
E
22
s
sw, w
w
w
w
N
N
N
E
Monitoring site
07 08 09 10 11
AUGUST
12 13 14 15 16
Ocean Co, NJ
McGuire AFB, NJ
Trenton, NJ
Trenton, NJ
Flemington, NJ
Bristol, PA
Morris Co, NJ
Allentown, PA
Easton, PA
Scranton, PA
Carbondale, PA
Rensselaer, NY
Agawam, MA
Pittsfield, MA
Chicopee, MA
Ware, MA
Groton, CT
Kent Co, Rl
Providence, Rl
s
s
sw, s
w
sw
w
w
NW
NW
N
E
S
S
sw, s
w
sw
w
w
N
E
S
w
w
N
N
N
E
E
S
S
SW, S S
sw s
w w
sw
w
w
w
NW
NW
N N
N
N
E E
S
S
SW, W
sw
w
w
w
N
N
N
E
S
S
S
s
sw, w
sw
w
w
N
N
N
E
S
S
S
w
sw
w
w
NW
NW
N
N
E
S
S
W
sw
w
w
NW
NW
N
E
S
SW, W
w
w
w
N
N
N
E
Compass points indicate which sites were used on each day of the episode; the direction indicates the location (s) along the
boundary where the data were used.
2-46
-------
TABLE 2-4. SUMMARY STATISTICS FOR ALL DAYTIME HOURLY AND DAILY MAXIMUM OZONE
CONCENTRATIONS (ppb) FOR THE ROMNET REGION (JULY AND AUGUST 1985
DATA MERGED)
Cone.
type
n
Mean
obs. model
Std. dev.
obs. model
95th percentile Maximum
obs. model obs. model
Daytime
hourly
Daily
max.
40,534 55.2 64.5
3,707 78.6 77.5
26.8 19.2
26.7 22.2
102.0 101.1 219.0 169.2
127.0 118.7 219.0 169.2
TABLE 2-5. COMPARISON OF DAYTIME HOURLY AND DAILY MAXIMUM STATISTICS (ppb) FOR
THE JULY AND AUGUST 1985 EPISODES
Episode
July
August
July
August
n
25,129
15,405
2,271
1,436
obs.
56.8
52.7
79.9
76.6
Mean
model
66.6
61.1
79.6
74.1
Std. dev.
obs. model
Daytime hourly
26.0 19.3
28.0 18.7
Dally maximum
25.1 22.5
29.0 21.3
95th percentile
obs. model
102.0 103.2
102.0 97.0
125.0 120.9
131.0 114.1
Maximum
obs. model
218.0
219.0
218.0
219.0
169.2
161.3
169.2
161.3
2-47
-------
TABLE 2-6. STATISTICAL SUMMARY BY GEOGRAPHIC GROUP FOR THE JULY 1985 EPISODE:
DAYTIME HOURLY AND DAILY MAXIMUM CONCENTRATIONS (ppb)
Group
n
Mean
obs. model
Std. dev.
obs. model
95th percentile
obs. model
Maximum
obs. model
Daytime hourly
1
2
3
4
5
1
2
3
4
5
6,206
6,776
5,902
3.604
2,641
558
609
537
328
239
60.2
65.2
54.3
47.8
45.2
88.0
90.7
75.1
66.1
63.4
78.4
72.6
57.3
56.4
57.9
97.8
87.9
65.9
64.7
67.2
27.8
27.5
23.7
20.1
19.5
Dally
28.9
23.8
19.1
17.9
17.8
21.9
19.0
11.8
12.8
11.3
maximum
23.1
19.5
12.1
14.5
11.7
111.0
110.0
95.0
83.0
80.0
138.0
130.0
109.0
96.0
94.0
118.7
107.6
79.4
81.3
78.5
144.9
120.8
88.0
91.7
89.3
218.0
211.0
155.0
126.0
108.0
218.0
211.0
155.0
126.0
108.0
169.2
159.8
103.5
121.4
108.0
169.2
159.8
103.5
121.4
108.0
2-48
-------
TABLE 2-7. SUMMARY STATISTICS FOR UAM BOUNDARY CONDITIONS (HOURLY DAYTIME
OZONE, ppb) FOR THE JULY AND AUGUST 1985 EPISODES
Statistic
Modeled
n 20,155
Mean 60.9
Std. deviation 17.7
95th percentile 92.9
Maximum 146.7
TABLE 2-8.
Observed
Percent difference
20,155
56.6
25.2
102.0
179.3
+7.6
-29.8
-8.9
-18.2
DAILY SUMMARIES OF UAM BOUNDARY CONDITIONS 1
Mean ozone
Date
(1985)
July 8
July 9
July 10
July 1 1
July 12
July 13
July 14
July 15
July 16
July 17
July 18
July 19
July 20
July 21
August 8
August 9
August 10
August 1 1
August 12
August 13
August 14
August 1 5
Modelled
(PPb)
46.4
59.3
66.8
63.5
61.0
72.6
64.5
67.0
61.7
57.8
56.2
74.2
70.8
53.8
46.2
54.9
55.9
52.5
56.6
59.7
69.2
80.1
Observed
(PPb)
35.1
70.1
57.1
45.5
56.5
70.0
58.9
44.9
51.1
46.1
48.9
79.6
74.1
60.2
42.3
72.0
52.1
54.5
40.3
62.5
69.7
72.7
Percent diff.
+32.2
-15.4
+ 17.0
+40.0
+8.0
+3.7
+9.5
+49.2
+20.7
+25.4
+ 14.9
-6.8
-4.5
-10.6
+9.2
-23.8
+7.3
-4.0
+40.4
-4.5
-0.7
+ 10.2
Mean wind
Speed
(ms-1)
5.2
3.8
4.6
4.7
4.6
3.1
5.9
5.2
3.0
4.4
3.7
5.9
4.4
4.1
4.7
8.7
3.6
4.0
4.4
4.1
3.8
5.4
Direction
(°)
250
171
233
320
165
187
215
200
169
35
70
222
265
177
221
110
168
200
2
159
222
331
Persistence 2
0.89
0.72
0.80
0.91
0.80
0.52
0.89
0.94
0.73
0.83
0.22
0.94
0.85
' 0.55
0.87
0.48
0.80
0.80
0.88
0.68
0.70
0.94
1. All land-based UAM boundary grid cells were used, during daytime hours only.
2. Persistence = (vector mean wind speed)/(scalar mean wind speed). Data are based on all daytime hourly wind observations
from National Weather Service stations near the OMNYMAP domain.
2-49
-------
This page is intentionally left blank.
-------
SECTION 3
OZONE EPISODE SELECTION
by
Dennis Doll*
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
Lenard B. Milich
Computer Sciences Corporation
P.O. Box12767
Research Triangle Park, NC 27709
* On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
3.1 INTRODUCTION
This Section summarizes the procedures adopted for selecting ozone episodes for simulating the
ROMNET emissions scenarios presented in Section 4. Two episodes were selected to examine the
effectiveness of various regional strategies. Two other episodes were selected for the ROMNET
supplemental evaluation of the ROM. In choosing these episodes particular attention was given to high
ambient ozone concentrations and the meteorological flow regimes in the Northeast Corridor. The
intent was to simulate conditions that represent a variety of typical meteorological transport patterns
associated with high ozone levels in the Corridor. Included in Appendix D are descriptions of the mete-
orological conditions and observed ozone concentrations during the selected episodes.
3.2 TECHNICAL APPROACH
The selection of ozone episodes for use in the ROM simulations was one of the principal objectives for
the ROMNET Modeling Committee. The following five step process was followed by the Committee to
select episodes for modeling:
1. 1983-1988 ambient ozone monitoring data in the Aerometric Information Retrieval System
(AIRS) were reviewed to identify candidate episodes with ambient ozone levels > 125 ppb;
2. quantitative episode selection criteria were identified;
3. a ranking procedure based on these criteria was designed and applied to establish the rela-
tive priority of each episode for simulation;
4. meteorological conditions within the ten top-ranked episodes were examined to characterize
transport flow regimes; and
5. the highest ranked episodes with the flow regimes of importance were selected for modeling.
In addition to candidate episodes identified from 1983-1988, one 1980 episode (July 12-23,1980) was
included in the list of candidate episodes. Prior analyses showed that this episode contained meteoro-
logical conditions of potential interest for the ROM modeling.
3.2.1 Initial Ambient Data Review
All hourly ozone data for the period May through September of 1983 through 1988 from over 150 Sta-
te/local monitoring sites in the ROMNET domain were extracted from AIRS for review. The resulting
3-3
-------
database was summarized into a time series of the frequency distribution of maximum ozone
concentrations by day for each year. An example is shown in Figure 3-1. The plots were then examined
to identify periods of at least three consecutive days when ozone levels were > 125 ppb (an excee-
dance day) within the domain. From these summaries, 27 candidate episodes were identified as listed
in Table 3-1.
3.2.2 Episode Selection Criteria
The criteria developed for prioritizing these candidate episodes focused on (1) the magnitude and spa-
tial distribution of high ozone concentrations with emphasis on areas along and adjacent to the North-
east Corridor; (2) the frequency of ozone exceedances; and (3) the representativeness of
meteorological conditions to those typically associated with high ozone within the Northeast Corridor.
The following approach was adopted for evaluating the candidate episodes against these criteria and
for ultimately selecting the episodes for modeling. First, the 27 candidate episodes were ranked based
on an objective analysis of ambient ozone data. This established a relative priority of the episodes for
model simulation. Next, for the ten top-ranked episodes, meteorological analyses were conducted to
characterize the flow regimes within each episode. The frequency of key flow regimes (described
below) were determined for each episode to assess whether a reranking of the episodes was needed to
capture the meteorological conditions of importance.
3.2.3 Episode Ranking Scheme
An objective ranking scheme was developed to prioritize the candidate episodes based on ambient
ozone concentrations. This scheme consisted of two independent parts. Together, these components
provided a comparison between episodes of (1) the magnitude of daily maximum hourly ozone con-
centrations; and (2) the number of days with an observed exceedance.
The ranking scheme was applied for ozone measurements spatially aggregated within three subregions
along the Northeast Corridor and adjacent areas as shown in Figure 3-2. In this scheme, each subre-
gion was weighted equally. The ranking for each part and subregion were combined to obtain a final
overall ranking score. A description of the decision rules applied to rank the candidate episodes starts
on the next page.
3-4
-------
Part 1 - Maximum Hourly Ozone
First, a comparison was made between episodes by subregion (i.e., comparisons were made for
subregion 1 for each episode, etc.) as follows:
1. The number of monitoring sites reporting maximum hourly ozone >150 ppb over the episode
was compared. The episode with the highest reported number of sites was ranked first; the
lowest was ranked last.
2. If the number of reporting sites was the same for two or more episodes, the episode with the
highest maximum hourly concentration was ranked first; the episode with the lowest concen-
tration was ranked last.
3. If the concentrations were < 150 ppb, then the number of sites with concentrations > 125 ppb
were compared. The episode with highest number of sites reporting >125 ppb was ranked
first; the lowest number was ranked last.
4. If the number of sites with ozone >125 ppb were the same for two or more episodes, the
episode with the highest concentration was ranked first; the lowest was ranked last.
5. If concentrations were < 125 ppb, the episode with the highest concentration was ranked
first; the lowest was ranked last.
6. If two or more regions had identical characteristics, each episode was ranked equally.
Note: a first ranking was assigned a value of 1, a last ranking was assigned a value of 27. Once all
episodes were ranked for each subregion, the rankings assigned to each of the three regions were
summed to obtain a total score for each episode (e.g., for episode 7/4/88 - 7/18/88, subregion 1=2,
subregion 2=1, subregion 3=1, total=4).
Part 2 - Number of Exceedance Days (>125 ppb)
In this part of the process a comparison was made by subregion for the number of days with ozone
>125 ppb, as follows:
1. The number of days with ozone > 125 ppb at all monitoring sites was compared. The episode
with the most number of monitors reporting > 3 exceedance days was ranked first. The epi-
sode with the least number of monitors reporting > 3 days was ranked last.
3-5
-------
2. If the number of monitoring sites reporting >3 exceedance days was the same for two or
more episodes, the episode with the highest reported number of exceedance days (e.g., 5
days) was ranked first; the second-highest number of days was ranked second, etc.
3. If the number of monitoring sites reporting >3 exceedance days was the same for two or
more episodes and the highest reported number of exceedance days was the same for two or
more episodes, then the total number of exceedance days reported among all monitoring
sites in the subregion was compared. The episode with the highest total was ranked first, etc.
4. For those episodes in which the number of exceedance days reported was < 3 days for all
monitors in the subregion, the total number of exceedance days reported among all monitors
in the subregion was compared. The episode with the highest total was ranked first, etc.
5. If two or more subregions had identical characteristics, the episodes were ranked equally.
Once all regions and episodes were ranked, the rankings assigned to each of the three subregions were
summed to obtain a total score for the episode (e.g., for episode 7/4/88 - 7/18/88, subregion 1 = 1, sub-
region 2=1, subregion 3=1, total=3).
The final step in the ranking procedures was to obtain an overall ranking for each episode by combining
the Part 1 and Part 2 scores. This is illustrated as follows:
For the episode 7/4/88 - 7/18/88 Score
Part 1 - maximum hourly ozone: 4
Part 2 - number of exceedance days: 3
Total = 7
The total scores were compared for each episode. The episode with the lowest score was ranked first,
etc. Table 3-2 shows the total score and ranking for each of the 27 candidate episodes.
3.2.4 Trajectory/Meteorology Analyses
After ranking the candidate episodes, detailed meteorological analyses were conducted on the ten top-
ranked episodes. This was done to characterize the meteorological conditions and transport flow
regimes associated with each episode to (1) rerank the order of the top ten episodes if necessary based
on meteorology, and (2) select those episodes that best represent the key transport flow regimes and
meteorological conditions considered most useful for ROM simulation.
3-6
-------
The Modeling Committee determined that the key transport regimes of interest for the modeling were:
along-Corridor (southwesterly) flow;
interurban recirculation conditions; and
westerly flow.
These flow regimes have been historically associated with episodes of high ozone. Because of the
geographic location of major ozone precursor source areas in the Northeast, these regimes favor inter-
urban and/or intraregional transport of ozone and its precursors. Thus, it is important to simulate such
episodes in order to assess the effectiveness of control strategies during transport conditions.
The analysis to identify the occurrence of key flow regimes consisted of two parts. The first was
designed to describe the daytime transport flows which actually occurred within the Corridor on days
with observed exceedances. Each exceedance day was classified according to one of seven synoptic
meteorological patterns. These patterns, and the corresponding within-Corridor flow regimes, are listed
in Table 3-3. These patterns were identified based upon a review of midday surface weather maps and
summaries of individual National Weather Service surface meteorological observations. Additionally,
three-day forward trajectories were constructed using the Atmospheric Transport and Dispersion Model
(ATAD) from four Corridor cities to facilitate the determination of the flow regimes within the Corridor.
The second part of the analysis focused on the direction of transport into the Corridor from upwind
areas. Each exceedance day was classified according to one of four incoming flow regime classes as
follows:
(A) southwesterly (along-Corridor);
(B) westerly;
(C) interurban recirculation;
(D) northwesterly.
For this part, three-day backward trajectories were constructed from five Corridor cities: Portland,
Providence, New York City, Philadelphia, and Washington, DC. Additionally, forward trajectories were
constructed from four cities in the western part of the domain: Detroit, Pittsburgh, Charleston, and
Danville.
Each day of the top ten episodes was classified according to both the synoptic pattern/within-Corridor
flow regimes shown in Table 3-3 and the transport flow direction into the Corridor (A through D above).
This information was then reviewed to identify the frequency of days during each episode that were
characterized by the key transport regimes.
3-7
-------
In addition to flow regimes, meteorological data summaries were prepared for each day of the top ten
episodes. These summaries provided information on temperature, cloud cover, wind speed, and direc-
tion. This information was used in conjunction with the flow regimes to better understand the meteoro-
logical conditions characteristic of each episode.
The flow regimes, meteorology conditions, and ozone concentration data were summarized for each
exceedance day of each top ten episode. This information was condensed and submitted to the Mod-
eling Committee for review and final selection of episodes for simulation.
3.2.5 Flow Regimes During the Top Ten Episodes
This section summarizes the occurrence of the key flow regimes during the ten top-ranked episodes.
Each episode is described below in rank order as given in Table 3-2.
Episode #1: July 4 - July 18,1988
This episode was characterized by a persistent trough along the Corridor. Flow into the Corridor was
generally from the west to southwest on most days. Within the Corridor, along-Corridor flow occurred on
10 of the 15 episode days with weak westerly or recirculation flow on other days.
Episode #2: June 13 - June 23,1988
A persistent surface trough along the Corridor was also the main feature of this episode. The flow
coming into the Corridor was from the west on eight of the episode days. Within the Corridor,
along-Corridor flow was noted on only three days with recirculation conditions on two days. The
remainder of the days had westerly flow.
Episode #3: July 12 • July 23,1980
High pressure systems over the east coast and just offshore dominated this episode. Along-Corridor
flow occurred on eight of nine days during this episode. The flow into the Corridor was from the
southwest or west on eight days.
Episode #4: June 9 - June 20,1983
This episode was characterized by fronts and a weak pressure pattern within the Corridor. Recirculation
conditions occurred on six of the ten days. A west to northwest flow occurred on other days.
3-8
-------
Episode #5: August 9 - August 17,1988
A high pressure system off the east coast was the main meteorological feature during this episode with
a weak trough along the Corridor on several days. Flow into and along the Corridor was from the
southwest on seven days. A northwest flow occurred on other days.
Episode #6: July 26-August 5,1988
A trough along the east coast with weak high pressure over western Pennsylvania were the main
meteorological features of this episode. Flow into much of the Corridor was from the southwest on eight
days and from the northwest on other days. Within the Corridor, along-Corridor flow occurred on seven
days with variable flow on other days.
Episode #7: June 4-June 13, 1984
A high pressure system was located off the east coast during much of this episode. Flow into the
Corridor was from the west on eight of the nine days. Within the Corridor, along-Corridor flow occurred
on six days with westerly flow on other days.
Episode #8: July 16-July 26,1987
Weak surface flow and fronts were the main meteorological features in the Corridor during this episode
resulting in recirculation conditions on most days. Flow into the Corridor was generally from the west or
northwest.
Episode #9: August 6 - August 16,1985
High pressure systems moved across the east coast during much this episode. Flow into and within the
Corridor was from the west and/or northwest on three days. Recirculation conditions occurred on four
days and along-Corridor flow on four days.
Episode #10: July 26 - August 1,1983
A high pressure system moved across the Corridor during the first part of the episode followed by a cold
front which remained near the east coast for several days. Flow into the Corridor was from the
southwest on three days and from the west to northwest on three days. Within the Corridor,
along-Corridor flow occurred on three days with variable flow associated with the front on other days.
3-9
-------
3.2.6 Episodes Selected for ROM Simulations
The July 4 - July 18, 1988 (July 88) episode produced some of the highest ozone concentrations
observed throughout the Corridor since 1980, as indicated from the episode ranking analysis. Addi-
tionally, exceedances were observed in rural areas and across much of the western portion of the
domain with peak values from 150 to 175 ppb. None of the other top ten episodes displayed a similar
spatial extent of high ozone levels. Also, from the meteorological analysis, this episode contains the
greatest number of days with along-Corridor flow and transport flow into the Corridor from the western
portion of the region. Given the magnitude and frequency of high ozone levels and the frequency of the
most important transport flow regimes, this episode was selected for use in simulating the ROMNET
control strategy scenarios.
Another episode, representing one of the other key flow regimes, was selected to test the robustness of
conclusions drawn from the July 88 episode. The intent was to simulate the final strategy, which
reduced predicted ozone to below 125 ppb, with an alternate type of flow regime also associated with
high ozone in the Corridor. However, examining the remaining top ten episodes in order of ranking
indicates that the second and third ranked episodes (June 88 and July 80) contained flow regimes that
were, to a large extent, represented in the July 88 episode (i.e., along-Corridor and westerly flow).
Additionally, there was some concern that the dominating westerly flow in the June 88 episode would
transport the predicted Corridor ozone plumes offshore over water, and thus complicate the interpreta-
tion of model predictions.
The fourth-ranked episode (June 83) was characterized by interurban recirculation conditions, which
was quite distinctive from that in other top ten episodes. Thus, June 83 was chosen to complement the
July 88 episode for strategy simulations.
In addition to the July 88 and June 83 episodes, two 1985 episodes were selected for use in the
ROMNET evaluation of the ROM. Episodes from 1985 were needed since 1985 was the year of the
ROMNET base emissions inventory. For this, the highest and second highest 1985 episodes were
selected. These episodes are: August 6-16, 1985 and July 7 - 22, 1985. The August episode was
ranked 9th and the July episode was ranked 23rd.
A detailed description of the meteorological conditions and ambient ozone levels during the four epi-
sodes selected for simulation is provided in Appendices D, E, and F. Note that an extra day was added
on to the front of three of the episodes in order to initiate ROM simulations on a "clean" day, as described
in Section 2.
3-10
-------
Doily Maximum Ozone Percentiles — 1988
250-1
Q_
0-^200-
C
"O 150-i
L-
-+—'
c
£ 100-
o
o
C 50-
O
N
O
0-
I
01 MAY
01JUN
01JUL 01AUG
Date
01SEP
010CT
Figure 3-1. Box-plot time series showing the frequency distribution of daily maximum ozone concen-
trations - May through September, 1988.
3-11
-------
Figure 3-2. Subregions of the Northeast Corridor used for aggregating observed data in the episode
selection process. R1: subregion 1; R2: subregion 2; R3: subregion 3.
3-12
-------
TABLE 3-1. CANDIDATE EPISODES: 1983 -1988
1983
June 9 - June 20
July 26 - August 1
August 13-August 21
August 31 - September 7
September 8 - September 14
1984
June 4-June 13
July 9-July 16
July 29 - August 5
1985
June 27 - July 6
July 7-July 14
July 15-July 23
August 6 - August 16
September 13 - September 23
1986
May 25 - June 2
June 18-June 28
July 3 - July 9
1987
May 27 - June 2
June 14-June 21
July 6-July 14
July 16-July 26
August 13 - August 18
1988
May 28 - June 1
June 13-June 23
July 4- July 18
July 26 - August 5
August 9 - August 17
3-13
-------
TABLE 3-2. SUMMARY RANKING OF 27 CANDIDATE EPISODES
Episode
July 4 -July 18, 1988
June 1 3 - June 23, 1 988
July 12 -July 23, 1980
June 9 - June 20, 1983
August 9 - August 1 7, 1 988
July 26 - August 5, 1988
June 4 -June 13, 1984
July 16 -July 26, 1987
August 6 - August 1 6, 1 985
July 26 - August 1 , 1 983
August 31 - September 7, 1 983
July 29 - August 5, 1984
July 6 -July 14,1987
August 1 3 - August 21,1 983
June 14 -June 21, 1987
July 9 -July 16, 1987
September 8 - September 14, 1983
May 28- June 1, 1988
July 15 -July 23, 1985
May 27 -June 2, 1987
August 13 - August 18, 1987
September 13 - September 23, 1987
July 7 -July 14, 1985
May 27 -June 2 1986
July 3 -July 9, 1986
June 27 - July 6, 1985
June 18 -June 28, 1986
Total
Score
7
16
19
26
28
43
47
57
70
74
76
80
80
84
86
90
91
98
103
107
116
120
125
131
133
138
140
Ranking
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
3-14
-------
TABLE 3-3. WITHIN-CORRIDOR FLOW REGIMES
Type Synoptic meteorological pattern and regime description
1 Dominated by high pressure; no fronts/troughs near Corridor:
generally weak flow;
within-Corridor flow generally light northerly (front side of HIGH);
flow in western part of modeling domain generally light southerly (back side of HIGH).
2 Transition regime - high pressure along east coast extending SE and NW:
weak surface flow.
3 High pressure system off the east coast:
westerly or along-Corridor flow (back side of HIGH).
4 Pre-frontal: along Corridor flow.
5 Cold front or trough along east coast:
variable or southwesterly flow.
6 Frontal pattern between New York and Boston:
easterly flow north of front;
along-Corridor flow south of front.
7 Frontal pattern between New York and Philadelphia;
frontal pattern around Baltimore/Washington, DC, and south;
low or weak pressure pattern: variable or recirculating flow.
3-15
-------
This page is intentionally left blank.
-------
SECTION 4
EMISSIONS SCENARIO DEVELOPMENT
by
William H. Battye
EC/R Incorporated
107 Highland Drive
Chapel Hill, NC 27514
John E. Langstaff
Mark G. Smith
Alliance Technologies Corporation
100 Europa Drive, Suite 150
Chapel Hill, NC 27514
and
Keith A. Baugues
Norman C. Possiel*
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
* On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
4.1 INTRODUCTION
During ROMNET, three types of emissions scenarios were simulated by the ROM: base year, projected
future baseline, and control strategies (CS) applied to the future baseline. The base year was 1985 and
the future baseline selected was 2005. Two variations of the 2005 baseline were examined along with 24
strategies. All of the ROMNET scenarios are listed in Table 4-1.
This Section of the report describes each of these scenarios and the development of the corresponding
anthropogenic and biogenic emissions inventories. An overview of the basic emissions inventory
structure common to all scenarios is presented first, along with a description of the methodologies used
for spatial, temporal, and species allocation of the emissions data (Section 4.2). This is followed by a
discussion of the development and characteristics of the 1985 base year scenario (Section 4.3). Pres-
ented next is the development of the 2005 scenarios and control strategies (Section 4.4). Control effi-
ciencies and details on specific control measures are provided in Section 4.5. Finally, biogenic
emissions inventory development is explained in Section 4.6.
4.2 OVERVIEW AND STRUCTURE OF EMISSIONS INVENTORIES
4.2.1 Features of the ROMNET Inventories
In general, ROMNET inventories provide emissions data for volatile organic compounds (VOC), nitrogen
oxides (NOX), and carbon monoxide (CO), which are precursors in tropospheric ozone formation. NOX
emissions are partitioned into NO and NO2. VOC emissions are assigned into one of ten reactivity
classes: olefin (OLE), paraffin (PAR), toluene (TOL), xyiene (XYL), formaldehyde (FORM), other alde-
hydes (ALD2), ethene (ETH), isoprene (ISOP), nonreactive (NONR), and methanol (MTHL). VOC clas-
sifications are based on the Carbon Bond IV system, discussed in depth by Gery etal. (1988). Emission
rates have been adjusted to reflect activity levels and temperatures characteristic of episodes with
exceedences of the ozone NAAQS in the Northeast.
Anthropogenic emissions data are provided separately for point sources, stationary area sources, and
mobile sources. Point sources include all plants emitting 100 tons per year or more of any criteria pol-
lutant (VOC, NOX, or CO) that were handled on an individual point basis in the EPA's National Emissions
Data System (NEDS).
4-3
-------
The area-source inventory does not address individual sources, but instead contains emissions for
aggregated groups of sources that are too small or numerous to be covered by the point source inven-
tory. Area sources in the ROMNET inventory include minor fuel combustion sources; open burning and
solid waste disposal; structural and forest fires; non-highway transportation sources, such as trains,
airplanes and off-highway vehicles; solvent evaporation from paints and other solvent uses; and some
industrial fugitive emissions and process vent emissions. The mobile source inventory includes aggre-
gates of highway vehicle emissions. The mobile-source data further classify VOC emissions into evap-
orative emissions from gasoline vehicles, exhaust emissions from gasoline vehicles, and exhaust
emissions from diesels (evaporation emissions from diesels are negligible).
Both the mobile- and stationary area-source inventories include emissions for three separate day types:
typical weekdays, Saturdays, and Sundays. Each of these contains hourly emission profiles for the 13
pollutant species categories listed above. Emissions in these inventories are spatially allocated into the
ROM grid system.
The point-source inventory contains hourly allocation factors for the three day types, as well as the
location of each source, along with other pertinent data such as stack height, emission temperature,
operating rate, source classification code (SCO), and Standard Industrial Classification (SIC).
The point-, area-, and mobile-source emissions inventories representing the various ROMNET scenarios
were developed by Alliance Technologies and delivered to Computer Sciences Corporation for pro-
cessing and preparation for input to the ROM. An overview of the methodologies used by Alliance to
generate these inventories is discussed next. Emissions input processing procedures followed by CSC
are described in Section 2.2.
4.2.2 Emissions Allocation Methodologies
The ROMNET inventory development effort relied heavily on a data management software system
developed under the National Acid Precipitation Assessment Program (NAPAP). This software, known
as the Flexible Regional Emissions Data System (FREDS), is used to prepare a gridded hourly database
suitable for input to the ROM emissions-processing system. Figure 4-1 summarizes the modules used
in FREDS and the input data required by each module.
The initial annual inventory has only one field for organic compound emissions, in which either VOC or
THC is entered, depending on the source category. The Hydrocarbon Preprocessor uses methane
emission data to calculate emissions of the missing pollutant, so that the output from this step contains
4-4
-------
both THC and VOC emissions. Other nonreactive pollutants are treated as VOC in this step. For some
source categories, the Hydrocarbon Preprocessor adjusts THC and VOC to add formaldehyde, which is
not detected by some hydrocarbon measurement techniques.
The annual inventory also contains data on a number of pollutants besides VOC, THC, NOX, and CO.
Entries for these other pollutants are removed in the Model Data Extraction step, in order to reduce
computer processing time.
The Speciation Module divides THC and VOC into reactivity classes, and NOX into NO and NO2, using
speciation factor files compiled by the PSPLIT program. The Spatial Allocation Module assigns emissions
to modeling grids, and the Temporal Allocation Module resolves annual emission rates into hourly rates
for different seasons and day-types. The order in which these three modules are run depends on the
inventory subset being processed. For mobile- and area-source inventories, the modules are run in the
above order. For point sources, spatial allocation is performed first, followed by temporal allocation and
speciation. The last module of FREDS is the Model Input Preprocessor, which produces a final resolved
inventory that is input to the ROM emissions-processing system.
For point sources, spatial allocation to modeling grids is based on point-specific UTM coordinates in the
NEDS inventory. Temporal allocation for point sources is based mainly on operating data provided in
the inventory. State- and category-specific operating profiles were developed for certain utility groups
based on data provided by the Department of Energy and the Tennessee Valley Authority.
Spatial and temporal allocation for area and mobile sources is performed using average factors, since
these sources are handled as county and source category aggregates in the NAPAP annual inventory.
Spatial allocation factors are based on land use patterns, population distributions, and spatial distrib-
utions of other emission surrogates; thus, they are specific to counties and source categories.
Speciation factors for point sources as well as area and mobile sources are source-category averages
based on average emission composition. Additional detail on FREDS and on the various allocation fac-
tors can be found elsewhere (EPA, 1989; Modica era/., 1988; and Walters etal., 1988).
4-5
-------
4.3 1985 ROMNET BASE CASE EMISSIONS DEVELOPMENT
4.3.1 Parent Emissions Inventories
The ROMNET base year inventory is derived from the NAPAP 1985 emissions database, which in turn is
derived from the NEDS. The NAPAP database provides a number of benefits as a starting point for
ROMNET. Chief among these benefits is the extensive data gathering and quality control effort that went
into the NAPAP database. The NAPAP database also provides a consistent inventory for the entire
ROMNET domain, including Canada. Finally, data management software developed under NAPAP
allows the flexibility to make the modifications needed for ozone modeling, including VOC speciation
and adjustments for temperature effects.
The 1985 NAPAP emissions database covers the 48 contiguous United States, the District of Columbia,
and Canada to 60 °N latitude. Emissions data are included for VOC, THC, CO, SO2, NOX, NO, NO2, four
other acid gases, 32 hydrocarbon classes, and 15 classes of paniculate matter. The overall database
includes two basic inventory types: an annual inventory, and a set of 12 temporal scenario inventories
representing a typical weekday, Saturday, and Sunday in each of the four seasons. Further information
on the NAPAP inventories is provided by Saeger etal. (1989).
4.3.2 Adlustments to the NAPAP Inventories
A number of modifications were made to the 1985 NAPAP database in order to produce the ROMNET
1985 base year anthropogenic inventories. The key changes included (1) adjustment of VOC emissions
for several stationary source categories to reflect meteorological conditions and activity patterns typical
of high ozone episodes, and (2) upgrades to mobile source emissions to reflect the latest version of the
MOBILE model and to adjust mobile emissions for grid-specific daily temperatures. Procedures for
implementing these adjustments and upgrades are described in this Section. Other changes made to
the NAPAP inventory, including quality control review of both major CO point sources and VOC emis-
sions from hazardous waste treatment, storage, and disposal facilities (TSDFs) are described in Alliance
Technologies Corp. (1989). The outcome of the adjusted VOC emissions combined with other portions
of the NAPAP inventories resulted in the 1985 ROMNET base case emissions inventories for VOC, NOX,
and CO.
4-6
-------
Stationary-Source Adjustments
In the NAPAP seasonal allocation methodology, calculation of area-source summer emissions is based
on typical seasonal "activity" patterns and average temperatures. "Activity" in this sense is similar to a
throughput for a point source. Thus, the NAPAP allocation factors do not take into account increases in
emissions due to the greater activity and higher temperatures that accompany fair-weather conditions
associated with ozone episodes. In addition, data are lacking on seasonal activity patterns for a number
of VOC area source categories; therefore emissions from these categories are assumed in NAPAP to be
uniform throughout the year.
In order to obtain emissions levels more representative of ozone episodes, VOC emissions from six dif-
ferent area source categories were adjusted for temperature and expected enhanced activity patterns
during episodes. Three categories were adjusted for temperature effects on vapor pressure:
Gasoline marketing (Category 54)
Bulk gas terminals and plants (Category 103)
Hazardous waste treatment, storage, and disposal facilities (Category 109).
Another three categories were adjusted for effects of weather on activity patterns:
Gasoline vessels (Category 52)
Architectural coating (Category 82)
Miscellaneous solvent use (Category 95).
The adjustment for gasoline vessels accounts for increased activity levels on the use of pleasure boats,
which make up the bulk of this category. The adjustment for miscellaneous solvent use accounts for
activity effects on pesticide use. For architectural surface coating, one might expect the main influence
to be the effect of temperature on evaporation rate. However, temperature effects on evaporation rate
are small in comparison with the uncertainty of hourly solvent usage profiles. The main activity effect for
this category is higher usage of external oil-based paint on rain-free days.
These six categories contribute a total of about 6 percent to the overall VOC inventory for the ROMNET
region. The maximum contribution for any single category is 2 percent (for gasoline marketing). Since
none of the six categories rivals mobile sources for VOC emissions, a day-specific adjustment was
judged to be unnecessary. Rather, emissions were adjusted to reflect conditions on a typical episode
day. The adjustments were accomplished by changing the summer temporal allocation factors for these
categories. The values adopted for the ROMNET program are shown in Table 4-2. The procedures
4-7
-------
used to calculate these factors are described in Alliance Technologies Corp. (1989). The effect of these
adjustments was to increase total 1985 anthropogenic VOC in the ROMNET region by 6.5 percent
compared with the NAPAP inventory.
Adjustments were also considered for a number of point-source categories and other area-source
categories, including degreasing and other industrial solvent use categories. However, these opera-
tions are conducted indoors in controlled environments, and hence are not influenced by outside tem-
peratures or weather effects. Adjustments were also considered for point-source volatile organic liquid
storage, because emissions from this category are sensitive to temperature in the same way as bulk
gasoline terminals and bulk plants. However, adjustments of point-source emissions would have been
much more difficult than area sources, and the combined contribution from point source storage cate-
gories was only about 0.4 percent of the overall VOC inventory. Thus, no adjustments were made to
VOC point source emissions.
Mobile Source Adjustments
The purpose of mobile source adjustments in ROMNET was to produce emissions estimates that reflect
the EPA's latest revisions to the MOBILE emission factor model (EPA, 1989), and that are representative
of the high temperatures during ozone episodes. In addition, emission estimates in the mobile inventory
were revised to reflect the sensitivity to temperature of motor vehicle inspection and maintenance (I/M)
program control efficiency estimates.
MOBILE is an empirical model used by the EPA to calculate emission factors from mobile sources under
different conditions. Maintained by the EPA's Office of Mobile Sources (OMS), it calculates emission
factors on a gram per mile basis for different types of vehicles at different speeds. These emission fac-
tors depend on average daily temperature, diurnal temperature variation, and other model inputs such
as vehicle age distribution and operating conditions.
Emission rates in the NAPAP inventory were generated with MOBILE3.0 using annual average ambient
temperatures at the State level. However, following the completion of the NAPAP inventories, upgrades
to the MOBILE model were made by EPA. In order to use the most current methodologies in ROMNET,
mobile emissions were revised twice during the program as new versions of the MOBILE model became
available. At the start of ROMNET MOBILE4 had not been released, but a predecessor version, MOBI-
LE3.9, was available. Thus, the NAPAP emissions were first revised using MOBILE3.9 for the original
ROMNET 1985 base case inventory. This inventory was later upgraded with MOBILE4.
4-8
-------
The major differences between the three versions of MOBILE pertain to the treatment of evaporative
emissions. MOBILE3.0, used in NAPAP, includes temperature-dependence functions for exhaust emis-
sion factors, but not for evaporative VOC. MOBILE3.9 relates evaporative emission factors to average
ambient air temperature, diurnal temperature variation, and fuel volatility expressed in terms of Reid
vapor pressure (RVP). MOBILE4 adds calculations for running loss and excess evaporation based on
recent test data. Both of these latter terms refer to evaporative emissions caused by the inability of
existing carbon adsorption fuel vapor control systems to handle recent increases in fuel volatility. Run-
ning losses occur while the vehicle is running and excess evaporation occurs while the vehicle is
standing.
Figure 4-2 and Figure 4-3 show the impacts of average daily temperature and diurnal temperature vari-
ation, respectively, on VOC emission factors predicted by MOBILE4. The figures also illustrate the esti-
mated additional contribution of running loss and excess evaporation to total VOC. NOX and CO
emissions also vary with average ambient temperature, although to a lesser extent than VOC. These
latter two pollutants are not affected by the range in diurnal temperature changes.
As Figure 4-2 shows, the effect of average daily temperature on VOC emissions is minimal up to about
70° F but is substantial at the higher temperatures representative of ozone episodes. Because of this
strong dependence on temperature, and because the mobile-source inventory accounts for a substan-
tial portion of the overall inventory, the ROMNET mobile methodology was designed to adjust mobile
emissions based on grid-specific, daily average temperature and diurnal variation for each ozone
episode day that is modeled.
Development of the mobile-source adjustment methodology was complicated by the volume of data in
the mobile-source inventory. The general goals in developing the adjustment methodology were to: (1)
limit data processing requirements, and (2) minimize the errors introduced by any simplifying assump-
tions. There are twelve categories of highway mobile sources for each county in the inventory, made up
of a matrix of four vehicle types by three roadway types. The vehicle types included in the inventory are
light duty gasoline vehicles, light-duty gasoline trucks, heavy-duty gasoline vehicles, and heavy-duty
diesel vehicles. The roadway types are limited-access highways (50 mph), rural roads (45 mph), and
urban roads (19.6 mph). Each combination of vehicle type and road type is derived from a separate
MOBILE emission factor and is entered separately in the inventory.
Day- and grid-specific temperature adjustment on a category-by-category basis would have required
production of a set of gridded, hourly inventories for each of the twelve mobile source categories. On
4-9
-------
the other hand, lumping the twelve categories together prior to the adjustment would not allow county-
level variations in vehicle- and road-type distributions to be reflected in the adjustment factors, causing
errors in the county- and grid-level emissions.
Other factors also complicated the design of an adjustment methodology. For example, the use of
mobile-source inspection and maintenance (I/M) programs varies across the region, and the effective-
ness of these programs in reducing emissions has been shown to vary widely from State to State. In
addition, evaporative emission factors, exhaust emission factors for gasoline vehicles, and diesel
exhaust emission factors respond differently to temperature and have significantly different VOC species
compositions. Thus, the distribution of VOC across reactivity classes varies with temperature.
The final methodology adopted for ROMNET involved a two-step approach that satisfied both of the
above goals. The basic approach was to make most of the necessary adjustments on a category-
specific basis with the annual mobile inventory.
In Step 1, each mobile source category in the annual inventory was adjusted separately to reflect emis-
sion factors from new versions of the MOBILE model. For MOBILE3.9, running loss and excess evapo-
rative emissions are added in this step as described by Alliance Technologies Corp. (1989). The
temperature used in generating the inventory was changed from the State-specific annual averages
used in NAPAP (ranging from about 45°F to 58°F for the ROMNET region) to a common daily average
temperature of 85°F and a diurnal variation of 20°F. These values yield the median emission factor for
the expected diurnal temperature range over the ROMNET region in an ozone episode (approximately
75°F to 95°F). VOC emissions in this step were separated into evaporative emissions, gasoline
exhaust, and diesel exhaust to account for changes in VOC speciation as the mix of emissions from
these categories changes with temperature. Also, I/M control efficiency assumptions used in the NAPAP
inventory were removed in Step 1.
At the end of Step 1, mobile source VOC, NOX, and CO emissions were aggregated to county totals,
retaining separate totals for the evaporative emissions, gasoline exhaust, and diesel exhaust. The
resulting interim mobile inventory was then processed through FREDS to produce gridded, speciated,
hourly inventories for the three summer day-types (weekday, Saturday, and Sunday).
In Step 2, the mobile emission estimates were altered at the grid level to reflect observed temperatures
on each episode day modeled. This step was performed in the ROM emissions processors using a table
of mobile source emission factors for different daily average temperatures and diurnal ranges. The table
(given in Appendix G for MOBILE4) covers average daily temperatures ranging from 40°F to 90°F and
diurnal variations of 0°F to 40°F. Because vehicle and road classes in the original inventory are
4-10
-------
grouped together at Step 2, the final adjustment table contains "composite" emission factors, or
weighted averages of factors for the various classes which represent typical distributions in vehicle and
road types over the ROMNET region. However, VOC emission factors are divided into evaporative,
gasoline exhaust, and diesel exhaust components to allow recalculation of VOC species breakdowns
based on final grid- and day-specific temperatures. In this step, emissions control efficiencies are also
revised to reflect grid-specific temperatures and are then applied to the "uncontrolled" inventory.
After the final adjustment of VOC emissions, VOC species profiles were recalculated for each grid and
day. This was done by using three separate sets of VOC speciation factors to calculate molar emissions
by reactivity class for the evaporative, gasoline exhaust, and diesel exhaust components of VOC. Total
moles of each reactivity class were then calculated by combining the results for the three VOC compo-
nents. With this recalculation, the final speciation was representative of grid- and day-specific tempera-
tures.
The result of Step 2 was a gridded hourly inventory for each day modeled, in which the emissions of
VOC, NOX, and CO, as well as VOC speciation in each grid, reflect the day-specific temperatures in that
grid. These mobile emissions were merged with point and area emissions categories in the ROM emis-
sions processors.
4.3.3 Characteristics of the 1985 Base Case Emissions
Source Category Distribution
The distribution of emissions between major source categories (point, area, and mobile) and within each
of these categories is an important component of an emissions inventory. While the distribution varies
from State to State, the discussion in this subsection will focus on the distribution of emissions within the
US portion of the ROMNET domain. For mobile sources, a daily average temperature of 85°F with a
diurnal range of 75°F to 95°F was used in calculating emissions. A detailed listing of the ROMNET
typical weekday VOC, NOX and CO emissions by category for all States in the domain is contained in
Appendix H.
Figure 4-4a illustrates the relative contributions of VOC emissions from point, area and mobile sources.
The portion of the VOC inventory due to area and mobile sources are nearly equal, while point-source
emissions contribute only 7.5 percent of the total VOC emissions. Point source emissions have been
broken into seven categories: utilities, chemical processes, industrial/institutional boilers, surface coat-
4-11
-------
ings, storage tanks, solvent use, and other. The relative contributions from these categories are also
shown in Figure 4-4b. VOC emissions from chemical processes, surface coating and solvent use
account for more than half of all VOC emissions from point sources.
The distribution of VOC area-source emissions is shown in Figure 4-4c. Major categories include: sol-
vent use; treatment, storage, and disposal facilities (TSDFs); combustion; fires; gasoline marketing; off-
highway vehicles; and other. Over half of all VOC area source emissions come from solvent use.
Mobile-source VOC emissions are split into three categories: evaporative, exhaust, and diesel. As
shown in Figure 4-4d, evaporative emissions comprise nearly three-quarters of the VOC emissions from
mobile sources. This is based upon a Reid Vapor Pressure (RVP) of 11.5 pounds per square inch for
gasoline. As the RVP is lowered in future years or in control strategy simulations, the relative contrib-
utions of evaporative and exhaust emissions can change significantly.
Over half of the NOX emissions in the ROMNET domain are released by point sources, one third from
mobile sources, and the remainder from area sources, as illustrated in Figure 4-5a. Major categories of
point source NOX emissions include: utilities, chemical processes, industrial/institutional boilers, and
other. The majority of the point-source NOX emissions, over eighty percent, are from utilities, as shown
in Figure 4-5b.
Major NOX area-source categories include: off-highway vehicles, combustion, and fires. As shown in
Figure 4-5c, off-highway vehicles and combustion comprise over 99 percent of the area-source NOX
emissions.
Spatial Patterns
County-level emissions were allocated to grid cells based upon land use patterns, population distrib-
utions, and spatial distributions of other emission surrogates. Figure 4-6 displays the spatial distribu-
tion of VOC and NOX emissions across the domain. High emission densities for VOC are noted near
several cities: Detroit, Cleveland, Cincinnati, Charleston, Richmond, Washington DC, Baltimore,
Philadelphia, Pittsburgh, New York, and Boston. Several grids in the core of New York City have the
highest level of VOC emissions in the region. As shown in Figure 4-4, most VOC emissions are derived
from mobile or area sources. Both of these source categories are clustered near large cities.
However, the high NOX emission densities seen in Figure 4-6 are not related to large cities, but are
associated with large combustion sources. This is further illustrated in Figure 4-7, which shows the
4-12
-------
locations of industrial boilers and utilities with greater than 500 tons per year of NOX. While large
combustion sources are located in the Northeast Corridor, the majority of the large utilities are located in
the western and southern portion of the domain, in Ohio, Pennsylvania, West Virginia, and Virginia.
Temporal Patterns
As indicated in Section 4.2.1, emission rates for individual days are treated as one of the following: a
weekday, a Saturday, or a Sunday. Due to varying operating schedules, emission rates for these three
categories differ. Emission totals for VOC, NOX and CO for each day type are shown in Figure 4-8. The
pattern of highest emissions on weekdays, then Saturdays, and finally Sundays are shown for all three
pollutants.
Daily emission rates are converted to hourly values by applying operating schedules for point sources or
temporal profiles for area and mobile sources. Figure 4-9 and Figure 4-10 show the results of applying
these profiles for VOC, NOX, and CO emissions for a typical weekday. VOC hourly emission rates for
area and mobile sources show considerable diurnal variation. Point-source VOC emissions are rela-
tively constant throughout the day.
Diurnal NOX emission rates for mobile sources contain large variations due to the assumed traffic pat-
tern. Both area- and point-source NOX emissions show higher emission rates during daylight hours, but
little variation overall. Hourly CO mobile-source emissions also show considerable hourly variation due
to the assumed traffic pattern. Area-source CO emissions increase during daylight hours, while point-
source CO emissions do not change significantly throughout the day. On an overall basis, approxi-
mately 60 percent of the VOC, NOX, and CO emissions are emitted between 0600 EST and 1500 EST.
Effects of Temperature on Mobile Emissions
Mobile-source emissions, especially VOC and CO, are sensitive to the average daily temperature.
Hourly emission rates for mobile sources are computed by using a daily average temperature value for
each individual grid cell. Appendix G lists emission rates for selected temperatures and temperature
ranges. Figure 4-11 illustrates the spatial distribution of mobile-source VOC emissions at 1500 EST on a
"cool" and a "warm" simulated as part of ROMNET. Domain average daily temperatures for these days
were71.7°F and 78.7°F, respectively.
4-13
-------
Due to the relatively small difference in temperature between the two days, there is little variation in the
emissions pattern. On an overall basis, VOC emissions are approximately 20 percent higher on the
warm day than on the cool day.
Figure 4-12 and Figure 4-13 illustrate the domain total mobile-source VOC, NOX and CO emissions for
each day of the July 1988 episode (refer to Appendices D, E, and F for a description of the daily meteo-
rology during ROMNET episodes). The values reflect both temperature effects and temporal profiles for
weekdays versus Saturday and Sunday. VOC mobile source emissions vary between approximately
7,000 and 12,000 tons per day, NOX between 4,800 and 6,200 tons per day and CO, emissions vary
between 32,000 and 48,000 tons per day.
4-14
-------
4.4 PROJECTION YEAR AND CONTROL STRATEGY SCENARIOS
The 2005 future baseline was derived from the 1985 ROMNET base case inventories. The year 2005
was selected because it was considered to be far enough into the future to provide the time needed for
implementation of stringent control measures and yet not beyond the outer bound for developing cred-
ible projections of population and industrial growth.
In essence, the future baseline includes the net effect on emissions of expected growth as well as
reductions associated with controls from existing State measures and Federal programs. As indicated
earlier, 24 control strategies were developed and simulated during ROMNET. These strategies were
derived by applying controls to the 2005 baseline, and in general, they fall into the following categories:
technology-based controls on sources of VOC and/or NOX;
"across-the-board" VOC reductions beyond that achieved by technology; and
reactivity-based strategies.
In addition, simulations were conducted to address the efficacy of controls relative to uncertainty in
biogenic emissions and rule effectiveness assumptions. The ROM simulations were conducted in two
Phases as indicated in Table 4-1. In Phase I, MOBILE3.9 was used for all scenarios, whereas MOBILE4
was used in Phase II. Also, as described below, there were some differences between Phase I and
Phase II future baseline scenarios in the treatment of existing VOC and NOX controls. The remainder of
this Section provides an overview description of the future baseline and control strategy scenarios. The
impacts on regional VOC, NOX, and CO emissions of projection and control measures applied in the
Phase II scenarios are shown in Figure 4-14 through Figure 4-16. The change in regional emissions
resulting from the Phase I scenarios are given in Appendix I, which also contains the changes to emis-
sions for individual urban areas resulting from both Phase I and Phase II scenarios. Section 4.5 contains
specific control levels used in generating the baseline and strategy emissions inventories.
4.4.1 Overview of Future Baseline Scenarios
This section provides a description of the Phase I and Phase II baseline scenarios. Control measures in
these scenarios are summarized in Table 4-3.
BS05 - Phase I 2005 Baseline: Constant NOX
The Phase 12005 baseline takes into account New Source Performance Standards (NSPS), the Federal
Motor Vehicle Control Program (FMVCP), a proposed gasoline Reid vapor pressure (RVP) standard of
4-15
-------
9.0 psi, and existing State regulations for VOC. Existing State rules include inspection and maintenance
(I/M) programs for mobile sources, and State Implementation Plan (SIP) regulations for point and area
sources. Throughout Phase I, mobile source emissions were evaluated using MOBILE3.9, with some
adjustments to VOC factors to account for running losses and excess evaporative emissions. The
Phase I baseline also incorporates the concepts behind the protocol of an international convention on
long-range transboundary air pollution, which caps national annual NOX emissions at 1987 levels.1
Because 1987 NOX emissions estimates were preliminary at the time this inventory was created, and
also to simplify the analysis of modeling results, NOX emissions were actually capped at the levels in the
1985 ROMNET inventory. Caps were applied at the State level, meaning that total point, area and
mobile source NOX emissions in each State were held to the 1985 level. This was done by controlling
growth in utility and industrial boiler emissions at the levels necessary to offset overall NOX emissions
growth, taking into account reductions already achieved by NSPS and FMVCP rules. The NOX protocol
caps were kept for all of the Phase I strategy inventories, and no further NOX controls were applied.
BS05 - Phase II 2005 Baseline: NOX Growth
Two important changes were made between Phase I and Phase II of the ROMNET strategy analysis.
First, MOBILE4 replaced MOBILE3.9 for evaluations of highway vehicle emissions. Second, the NOX
emission caps in Phase I were replaced by growth and controls applied to specific NOX emissions
categories. Refinements for the projection algorithms were also made between Phase I and Phase II,
resulting in minor changes in emissions for some categories.
The Phase II baseline projection, like its Phase I counterpart, takes into account existing control mea-
sures: NSPS, FMVCP, the proposed RVP standard of 9.0 psi, and existing State regulations. In addition,
emissions in the utility sector were reduced by a five percent energy-efficiency conservation factor,
applied to initial growth estimates for all utility sources. This factor accounts for efficiency improvements
expected to be realized by 2005, both in the generation and use of electricity. Such improvements
would reduce utility fuel consumption, thereby reducing emissions of all pollutants.
Differences between MOBILE3.9 and MOBILE4 resulted in a 32 percent reduction in highway vehicle
emissions of VOC between the Phase I and Phase II2005 baseline inventories. This translates to a five
percent reduction in overall VOC emissions. A similar reduction, about 29 percent or 1100 tons per day
regionwide, occurred in mobile-source NOX emissions. This was offset by a 1200 ton per day (14 per-
1. Revised Draft Protocol to the 1979 Convention on Long-Range Transboundary Air Pollution Concerning the Control of Emis-
sions of Nitrogen Oxides or Their Transboundary Fluxes. United Nations Economic and Social Council. May 10,1988.
4-16
-------
cent) increase in point-source emissions, due to the removal of NOX emissions caps. Overall, NOX
emissions from all sources increased by less than one percent between the Phase I and Phase II
baseline projections.
2005 Baseline Canadian Emissions
A single set of growth and emissions control assumptions was used for Canadian emissions in the 2005
baseline and control strategy scenarios. There were some differences between the Phase I and Phase
II Canadian inventories, however, in that MOBILE3.9 was used to evaluate highway vehicle emissions in
Phase I and MOBILE4 was used in Phase II.
Environment Canada provided information on the control measures anticipated for Ontario for the 2005
baseline. For VOC, current US FMVCP standards were applied to Ontario mobile sources, and the
emissions controls in Strategy 1 (see below) were applied to point and area sources. Current US
FMVCP standards were also applied to mobile source NOX emissions, but no point or area source NOX
controls were applied. The changes in Canadian emissions between the 1985 and 2005 Phase II sce-
narios are a 42 percent reduction in VOC, a 10 percent reduction in NOX, and a 33 percent reduction in
CO.
4.4.2 Overview of the Control Strategy Scenarios
The ROMNET control strategies were designed to address several issues that were determined by the
Management Review and Technical Committees to be of greatest importance in achieving the goals of
ROMNET. These issues - posed as questions - are listed below along with the control strategies (CS) to
address them:
1. What are the relative benefits of VOC versus NOX controls in reducing ozone levels?
[CS10-CS14]
2. What is the impact of reducing regional transport on ozone concentrations in the Northeast
Corridor?
[CS01, CS02, CS03, CS24, CS25]
3. What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted
ozone levels in the Northeast to below 125 ppb?
[2005 baseline, CS01, CS05, CS10 - CS16, CS18, CS19, CS23]
4. What are the effects of reactivity-based strategies in reducing regional ozone levels?
[CS15, CS20]
4-17
-------
5. How does uncertainty in biogenic emissions affect conclusions drawn about the benefits of
controlling anthropogenic sources?
[1985H, 1985L, CS06 - CS09, CS17, CS21, CS22]
An overview of each strategy scenario is provided below. Details on specific control measures are
provided in Section 4.5. Strategies are numbered in the chronological order of simulation. The rationale
for this sequence of strategies was determined in part by the following considerations:
a technology-based approach for control (i.e., iteratively adding source-specific controls) was
adopted for ROMNET rather than starting with large across-the-board reductions to seek tar-
get combinations likely to reduce ozone to < 125 ppb;
. the initial focus was on VOC controls with subsequent incorporation of NOX controls, then
measures that affect reactivity;
across-the-board reductions were applied on top of technology-based controls on an urban
area specific basis when maximum technology was insufficient to reduce ozone to
< 125 ppb;
the extent of NOX controls was reduced in selected urban areas where NOX control appeared
to be counterproductive as determined from earlier simulations;
the impact of considering control (rule) effectiveness was simulated as one of the last strate-
gies in order to use, as a starting point, the strategy that reduced ozone to < 125 ppb
assuming all controls are 100 percent effective; and
biogenic sensitivity scenarios, which are easy to handle because they combine previous
strategies with across-the-board changes in biogenics, were simulated during various "win-
dows of opportunity while inventories for other strategies were being developed.
Several strategies were applied to the Northeast Corridor only and/or to nonattainment urban areas in
the region. The Corridor and nonattainment areas are shown in Figure 4-17.
CS01: Phase I Maximum Technology VOC
Strategy I is the Phase I maximum technology scenario which includes the same FMVCP, RVP and NOX
cap as the Phase I baseline. In addition, enhanced I/M was included for the entire ROMNET domain. A
number of control measures were also applied to point- and area-source VOC emissions. NSPS
controls, which currently cover only new sources, were applied to all applicable categories in the
maximum technology inventory. EPA Control Techniques Guidelines (CTG) were also applied to area-
and point-source categories, regardless of size. Controls were also extended to non-CTG and
non-NSPS surface-coating categories by assuming technology transfer. Stage I and Stage II evapora-
tive emissions controls were applied to gasoline marketing; prescribed forest and agricultural burning
were eliminated during the ozone season; and architectural-coating emissions were reduced by 52
4-18
-------
percent through substitution for oil-based coatings with water-based coatings. Finally, miscellaneous
nonindustrial (commercial/consumer) solvent emissions were reduced by 20 percent. Control mea-
sures in this scenario are summarized in Table 4-3. Overall, VOC emissions in CS01 were reduced by
about 45 percent from the Phase I 2005 baseline inventory. Emissions of NOX are unchanged and CO
was reduced by 13 percent.
CS02 and CS03: Spatial Analyses of VOC Controls - Transport Sensitivity Scenarios
Strategies 2 and 3 (CS02 and CS03) were permutations of the Phase I 2005 baseline and CS01 maxi-
mum technology inventories. These scenarios were developed to study the impact of applying maxi-
mum technology VOC controls in different portions of the ROMNET domain. In CS02, maximum
technology controls were applied to the Northeast Corridor, while emissions in the remainder of the
ROMNET domain were left at the 2005 baseline levels. In the Corridor, the effect was a reduction of 45
percent in VOC and 10 percent in CO with NOX emissions unchanged compared to the 2005 Phase I
baseline. Outside the Corridor emission in CS02 are the same as the 2005 baseline.
CS03 applied maximum technology VOC controls to the Corridor and nonattainment areas outside the
Corridor, again maintaining the rest of the domain at the 2005 baseline. In CS03, emissions in the US
portion of the region outside the Corridor were reduced by 24 percent for VOC and 7 percent for CO with
NOX unchanged.
CS04: Rule Effectiveness Sensitivity Scenario
In the technology-based, reactivity, and across-the-board strategies, emissions controls are assumed to
be 100 percent effective. For the reasons listed in Section 4.5, this is not the case in 'real world"
operations. The intent of this scenario was to apply more realistic estimates of control effectiveness to
the VOC controls in CS01. However, the results of CS01 indicated that considerably more VOC control
and, most likely, NOX controls would be needed to reduce predicted ozone to below 125 ppb. Thus, it
was decided to apply the concept of control effectiveness to a more stringent strategy. This was done
in CS23, and thus CS04 was not simulated.
CS05: 1989 Proposed Clean Air Act Legislation
In September 1989, the then-pending draft Clean Air Act legislation was simulated in ROMNET Strategy
5. The legislation analyzed was House Resolution 3030 with the October 1989 Waxman-Dingell tailpipe
4-19
-------
standards for highway vehicles. As with other ROMNET projection and strategy inventories, emissions
were estimated for the year 2005. The draft legislation substantially reduced tailpipe emissions from the
current FMVCP standards in a phased program, to 0.4 g/mile for NOX and 0.25 g/mile for VOC. Gasoline
RVP was reduced to 9.0 psi (as was assumed in the ROMNET baseline). Requirements for mobile-
source I/M programs varied in different areas of the ROMNET domain, depending on the severity of the
ozone problem. These programs were applied at the Metropolitan Statistical Area (MSA) level. For
moderate nonattainment areas, basic I/M programs were upgraded to EPA standards for waiver and
compliance rates. Enhanced I/M programs were applied in areas with serious and severe ozone
nonattainment problems.
Low-NOx burners were required for all coal-fired utility boilers with generating capacity greater than 75
Megawatts. Point- and area-source VOC controls were also applied at the MSA level depending on the
severity of the ozone problem. In addition, across-the-board reductions were applied to point-, area-,
and mobile-source emissions to meet various target reductions proposed in the draft legislation. Non-
attainment areas with moderate ozone concentrations were required to reduce overall VOC emissions
by 15 percent, in addition to any reductions that would result from Federal control programs. The
legislation required areas with serious and severe ozone problems to reduce emissions by 3 percent per
year (in addition to reductions from Federal programs) from the year of enactment (the assumption was
1990) until the ozone NAAQS is met. In the ROMNET analysis, this requirement was simulated by
applying a 30 percent reduction in serious areas (estimated attainment date 2000) and a 45 percent
reduction in severe areas (estimated attainment date 2005). (See Section 4.5 for a definition of ozone
severity categories.)
In addition to the above VOC controls, phased clean fuels programs were implemented in four metro-
politan areas in the ROMNET domain - Baltimore, New York City, Philadelphia, and Greater Hartford.
The phased fuel program had the effect of replacing about 4.5 percent of the total VOC in these areas
with methanol, reducing the photochemical reactivity of the VOC emissions in these areas.
The control measures applied in the Clean Air Act legislation strategy are listed in Table 4-4. In the US
portion of the region, CS05 achieved a 32 percent reduction in VOC emissions and a 32 percent reduc-
tion in NOX emissions from the Phase II 2005 baseline scenario. Emissions of CO were reduced by
4 percent.
4-20
-------
CS06 - CS09,1985L and 1985H: Blogenlc SensHivHy Scenarios for CS01 and CS05 and the 1985
Base Case
These scenarios were designed to test the sensitivity of ozone predictions to uncertainty in biogenic
emissions for base case and selected post-control scenarios. Included are ROM applications for the
1985 base case, CS01, and CS05 scenarios with biogenic emissions altered to reflect the possible
range of uncertainty ( ± a factor of 3). Each of the three anthropogenic scenarios are run with
(1) biogenic emissions reduced by a factor of 3 ("Low" biogenics) and (2) biogenic emissions increased
by a factor of 3 ("High" biogenics). In all scenarios, biogenic emissions are altered in an across-the-
board manner from the "best available science" estimates generated by the Biogenic Emissions Inven-
tory System (BEIS). The following are the specific scenarios modeled:
1985L: 1985 anthropogenics - Low biogenics
1985H: 1985 anthropogenics - High biogenics
CS06: CS05 anthropogenics - Low biogenics
CS07: CS01 anthropogenics - Low biogenics
CS08: CS05 anthropogenics - High biogenics
CS09: CS01 anthropogenics - High biogenics
CS10: Phase II Maximum Technology NOX and Enhanced Maximum Technology VOC
Strategy 10 (CS10) incorporates maximum technology VOC and NOX controls. This inventory retains all
of the VOC controls from the Phase I maximum technology case (CS01), with some added control
measures. First, highway vehicle tailpipe standards were reduced to levels substantially below the
current FMVCP standards, and even below the levels in the Clean Air legislation analysis (CS05). The
hydrocarbon standard for automobiles in CS10 was 0.125 g/mile. In addition, the mobile source con-
trols (including exhaust emissions standards and effects of evaporative emissions testing) were
assumed to be at full penetration (i.e., entire fleet), even though they would not reach this level until after
2005. Enhanced I/M was also applied across the entire ROMNET domain, including antitampering
provisions. Finally, gasoline RVP was reduced to 7.0 psi.
For area-source VOC, CS10 eliminated residential wood burning as well as all open burning during the
ozone season. This measure also affected NOX and CO emissions. VOC controls were applied to off-
highway diesel vehicles, locomotives, and diesel ships. Controls were added for four new point-source
categories and to generic non-CTG point sources.
4-21
-------
For utilities, the five percent emissions reduction for energy-efficiency improvements from the Phase II
2005 baseline was retained in CS10. This last reduction applied to utility NOX and CO emissions as well
as to VOC.
CS10 also included several maximum technology NOX emissions controls. The automobile tailpipe
standard was reduced to 0.2 g/mile, with reductions in other highway vehicle standards. As with the
mobile-source VOC, these lower standards were simulated at full penetration. NOX controls were also
applied to off-highway diesels, locomotives, and diesel ships in the area source category. Selective
catalytic reduction (SCR) was applied to all utility combustion sources. Finally, low-NOx burner or com-
parable technology was applied to industrial boilers, internal combustion engines, and process heaters,
as well as to commercial and institutional boilers in the point-source inventory. The control measures in
CS10 (Phase II Maximum Technology NOX and Enhanced Maximum Technology VOC) are summarized
in Table 4-3.
Overall, CS10 achieved a 57 percent reduction in NOX emissions, a 63 percent reduction in VOC emis-
sions, and a 21 percent reduction in CO from the Phase II2005 baseline scenario.
CS11 - CS14: Analyses of VOC Versus NOX Controls
CS11 and CS12 were designed to analyze the relative impact of VOC and NOX control measures. CS11
matches 2005 baseline VOC controls with maximum technology NOX (CS10) controls, while CS12
matches maximum VOC controls from CS10 with NOX emissions from the 2005 baseline.
Strategies 13 and 14 analyze the relative impacts of point- and mobile-source NOX controls. CS13
applies maximum technology controls to VOC emissions and to NOX point-source emissions, while
keeping mobile- and area-source NOX emissions at their 2005 baseline levels. CS14 applies maximum
technology controls to VOC emissions and mobile-source NOX emissions, keeping point- and area-
source NOX emissions at 2005 baseline levels. The point-source controls in CS13 reduced NOX emis-
sions in the US portion of the region by 45 percent. In contrast, mobile-source controls in CS14 reduced
NOX emissions by 9 percent.
CS15: Maximum VOC and NOX Technology with Reduced VOC Reactivity
CS15 includes all of the VOC and NOX control measures in CS10, as well as two additional measures
designed to reduce VOC reactivity. The first reactivity reduction measure was applied to solvent emis-
sions over the entire FtOMNET region. A cap was placed on the reactivity of all solvent emissions, at a
4-22
-------
level midway between the reactivities of isopropanol and ethyl acetate. Any solvents with higher
reactivities, in either the point- or area-source portions of the inventory, were assumed to be replaced
with the low-reactivity solvent. Maximum technology controls were also retained for solvent emissions
so that overall solvent emissions remained the same as in CS10.
The second reactivity reduction measure was the conversion of the entire Northeast Corridor highway
fleet to 100 percent methanol fuel vehicles. This produced a 33 percent increase in mobile VOC emis-
sions over those attained for gasoline under the tight maximum technology tailpipe standards. How-
ever, since most of this VOC was methanol, the overall reactivity of mobile source emissions was
reduced. Outside the Corridor, mobile source emissions were kept the same as in CS10.
To quantify the impact of reactivity reduction measures on emissions, an average reactivity term was
calculated. This average reactivity was essentially the weighted average reaction rate of VOC emissions
with hydroxyl radicals. By this criterion, the solvent substitution measure reduced the overall reactivity of
point- and area-source emissions by about 28 percent for the US portion of the region. Methanol con-
version reduced the overall reactivity of mobile source emissions in the Corridor by about 57 percent.
Total reactivity-weighted VOC emissions in the Corridor were reduced by 29 percent compared to the
2005 baseline as a result of combining the mobile- and stationary-source components of this strategy.
CS16: Maximum Technology with Additional Across-the-Board Reductions
CS16 includes the maximum VOC technologies of CS10, with additional VOC reductions in the Balti-
more/Washington, DC, Consolidated Metropolitan Statistical Areas (CMSAs) and the New York City
CMSA.
VOC emissions in CS10 were reduced by an additional 64 percent in New York City and by an additional
54 percent in Baltimore/Washington, DC. These reductions were taken across-the-board on point, area,
and mobile sources, and applied in addition to the reductions already achieved by maximum technology
VOC controls. For New York City, CS16 represents an 85 percent reduction in VOC from the Phase II
2005 baseline, and a 91 percent reduction from the 1985 base case. For Baltimore/Washington, DC,
VOC was reduced by 80 percent from the 2005 baseline, and 88.5 percent from 1985 levels.
For NOX> maximum technology controls are applied as in CS10, except in New York City, where mobile
sources are controlled but point-source emissions are at the 2005 baseline level. The NOX emissions
reduction in this area for CS16 was 32 percent rather than 58 percent as in CS10.
4-23
-------
CS17: Maximum Technology NOX with "Low" Biogenics
Strategy 17 was designed to examine the effectiveness of NOX controls if biogenic emissions were at the
low end of the uncertainty range. Anthropogenic emissions are the same as in CS11 (maximum
technology controls in NOX with VOC and CO at the 2005 baseline). Biogenic emissions are reduced by
a factor of 3 from base case levels as in scenarios 1985L, CS06, and CS07.
CS18: Maximum Technology, Reduced Reactivity and Additional Across-the-Board Reductions
In this strategy, VOC emissions from CS15 (maximum technology with reduced reactivity) were lowered
further by the across-the-board reductions used in CS16 (64 percent in New York City and 54 percent in
Baltimore/Washington, DC). Emissions of NOX and CO were identical to that in CS16.
CS19: CS18 with Alteration of NOX Controls in Baltimore/Washington, DC
CS19 was identical to CS18 except for the treatment of NOX controls in Baltimore/Washington, DC. In
these cities, maximum technology controls were placed on NOX emissions from mobile and area
sources. However, point source NOX emissions were maintained at the 2005 baseline level. Thus, NOX
reduction in this areas was 28 percent, rather than 46 percent as in CS18.
CS20: 2005 Baseline with Reduced Reactivity
The purpose of this strategy was to isolate the impact of using low-reactivity solvents and methanol
vehicles from the effects of technology-based stationary source controls and lower tailpipe emissions
standards. (Technology plus reactivity reductions are combined in CS15). In CS20, methanol fuel
conversion was adopted in the Northeast Corridor as in CS15, again at full fleet penetration. In addition,
the solvent reactivity reduction measure was adopted as in CS15. However, instead of applying
maximum technology controls, solvent emissions were left at 2005 baseline levels (although at a
reduced reactivity). All other VOC point- and area-source emissions, as well as mobile emissions
outside the Northeast Corridor, were left at 2005 baseline levels. Emissions of NOX and CO were at the
2005 baseline level across the domain.
4-24
-------
CS21 andCS22: Biogenic Sensitivity Scenarios for CS19
These two strategies were designed to examine the effects of uncertainty in biogenic emissions on the
effectiveness of anthropogenic controls in CS19. Thus, in both CS21 and CS22, anthropogenic
emissions are identical to CS19. In CS21, biogenic VOC emissions are at the 'Low1 end of the
uncertainty range and in CS22 biogenics are at the 'High' end of this range (± a factor of 3 from base
case levels).
CS23: Control Effectiveness Strategy
The purpose of this strategy was to provide a 'realistic' consideration of the types of control measures in
CS19 (maximum technology, reactivity reduction, and across-the-board VOC reductions). Recall that
controls in CS19 were assumed to be 100 percent effective. It was also assumed that the entire
passenger car fleet in the Northeast Corridor was comprised of M100 vehicles, and that stringent tailpipe
standards were in place on all vehicles outside the Corridor.
These controls were relaxed in CS23 by adding effectiveness factors in the calculation of overall control
efficiencies and by assuming a more likely projection for fleet turnover with clean fuel and reduced
exhaust/evaporative emissions vehicles. For this, the following specific changes were imposed on CS19
control measures:
an 80 percent effectiveness factor was applied for all point and area source controls;
an 80 percent effectiveness factor was applied to vehicle inspection and maintenance pro-
grams;
gasoline RVP was increased from 7.0 psi to 7.8 psi;
• use of low-reactivity solvents was reduced by assuming that only 80 percent of the potentially
affected sources comply with a solvent substitution program;
the alternative fuel vehicle program in the Northeast Corridor was changed to reflect the
schedule and M85/M100 vehicle fleet fractions used in CS05 (Clean Air Act legislation strat-
egy); and
light-duty passenger vehicle emissions for conventional engines were changed to reflect a
•phased" approach to lowering emissions standards: 1994 - NMHC 0.25 g/mi, NOX 0.4 g/mi;
2003 - NMHC 0.125 g/mi, NOX 0.2 g/mi.
As a result of considering rule effectiveness, VOC emissions in CS19 were increased by 64 percent, NOX
by 25 percent, and CO by 7 percent in the US portion of the region. Still, considering rule effectiveness
with CS19 controls, emissions in CS24 are lower than those in the 2005 baseline by 48 percent for VOC,
40 percent for NOX, and 16 percent for CO.
4-25
-------
CS24 and CS25: Spatial Analysis of VOC and NOX Controls - Transport Sensitivity Scenarios
In CS24, the CS19 controls were applied to all counties inside the Northeast Corridor, with emissions at
2005 baseline levels elsewhere in the domain. For strategy CS25, emissions in the Northeast Corridor
are at the 2005 baseline and the controls used in CS19 are applied in areas outside the Corridor. The
purpose of these two strategies, through comparisons with the 2005 baseline scenario and CS19
results, was to examine the effects on ozone levels in the Northeast Corridor of changes to ozone and
precursor transport into the Corridor as a consequence of regional strategies.
4-26
-------
4.5 EMISSIONS CONTROL MEASURES
This section contains the specific control efficiencies, emissions standards, etc., contained in the 2005
baseline scenarios and the ROMNET control strategies described in the previous section. Projection
and control methodologies together with emissions growth factors used in developing these scenarios
are provided in Appendix J.
For point sources, controls were generally specified by emissions "pod." Each pod represents an
aggregate group of numerous point source SCCs that are treated similarly for control purposes. The
point-source inventory was divided into 80 such pods. Stationary area-source emissions were divided
into 64 categories (area-source SCCs) that emit VOC, NOX, or CO. The specific pods and area-source
categories are listed in Appendix J.
In the remainder of this Section, information on stationary- and mobile-source controls included in the
Phase I and Phase II 2005 baseline scenarios is presented first. This is followed by a description of the
control measures used in the maximum-technology-based strategies. Controls in the Clean Air Act
strategy are presented next. Details on the reactivity-based strategies are provided last.
4.5.1 VOC Stationary-Source Controls for 2005 Baseline Scenarios
The 2005 stationary-source baseline inventories include two major sets of controls: NSPS controls
applied to new sources at the national level, and State controls applied at the State or county level.
Table 4-5 lists the NSPS efficiencies used for new, modified, and reconstructed point and area sources.
These were taken from a compilation of NSPS efficiencies in an EPA cost assessment of alternative
National Ambient Air Quality Standards for ozone (Battye ef a/., 1987). For the 2005 baseline analysis,
the NSPS efficiencies were multiplied by an 80 percent rule effectiveness factor.
As indicated in Appendix J, Table J-1, Equation 3, there are two types of State- and county-level effi-
ciencies that were used in developing the 2005 baseline inventories: 1985 efficiencies (Effi985) and
2005 efficiencies (Effgoos)- The approach for obtaining values for each of these efficiencies for VOC area
and point sources is described next.
Area Sources
For solvent categories (78 through 95 - see Appendix J), 1985 efficiencies were obtained from a survey
of State ROMNET participants on the level of existing controls. Solvent categories were distinguished
4-27
-------
from other area source categories because of the methods used to generate initial 1985 NEDS
emissions estimates. In general, the 1985 NEDS area source inventory began with the development of
national emissions estimates for various area source categories. Emissions were then allocated to
subcategories, States and counties based on industrial employment and population statistics. Within
this overall framework, national emissions estimates were developed using the two basic methodologies
of mass balance and emission factors. Solvent emissions in categories 78 through 95 were ultimately
derived from a nationwide mass balance on solvent sales. All other emissions estimates in the 1985
NEDS area source inventory were from standard emission factors and national assumptions on the
average level of emissions control. For the two "emission factor* categories controlled in 1985 (103 -
bulk terminals/plants and 104 - refinery fugitives), the 1985 control efficiencies used in the 2005 baseline
projections were taken from the national efficiency assumptions used in the original NEDS inventory.2 It
would be inappropriate to use the actual control efficiencies from the State survey, because these
efficiencies were not reflected in the 1985 inventory. Conversely, 1985 emissions estimates for solvents
do reflect actual controls, at least on a gross national level, because of the mass balance approach used
for the solvent categories. Therefore, the control efficiencies from the State survey were thought to
provide the best estimates of 1985 efficiencies for these categories (78 through 95). The 2005
efficiencies for all area-source categories to which additional controls are applied in the 2005 baseline
were also obtained from the State survey.
Point Sources
The NEDS point source inventory includes an emissions control efficiency entry for each separate
emissions source. In general, the 1985 efficiencies for ROMNET projections were obtained by taking
these NEDS efficiency entries and multiplying by an 80 percent rule-effectiveness factor. However, the
Emissions Committee was concerned that many existing controls may not be listed in the "control
efficiency" field. This field was designed mainly to handle add-on controls, although process modifica-
tions can also be entered as controls. The Committee was concerned about low-solvent coatings, which
they felt might not be listed as control techniques in the inventory. In these cases, controlled emissions
would reflect the low-solvent coating, but emissions reductions already achieved by the coating would
not be noted. Therefore, in control strategy evaluations, potential emissions reductions would be
counted for controls that were already in place. To avoid this problem, the State control survey results
for area source surface coating categories were extended to the corresponding point source surface
2. 58.5 percent for category 103 and 48.2 percent for category 104.
4-28
-------
coating categories. Thus, for surface coating categories, overall efficiency values from the State survey
were substituted for the control efficiency entries in the point source inventory. For point sources, the
State- and county-level controls for the 2005 baseline were left at 1985 control levels (including the
• changes discussed above for point source solvent category efficiencies).
State Survey of Control Efficiencies
As indicated above, State Emissions Committee representatives were surveyed on VOC control
technologies in place as of 1985. This survey provided inputs both for selected 1985 control efficiencies
and for all 2005 baseline control efficiencies. The survey initially was designed to address area sources,
but results were also extended to some point-source categories.
States were asked to quantify overall reduction efficiency and indicate the geographic coverage of any
rules applying to area sources. In quantifying overall reduction efficiency, the States were asked to use
three separate parameters: control technology efficiency, rule effectiveness, and rule penetration into
the inventory. These parameters were defined as follows:
Control technology efficiency is the efficiency (in percent) of controls or work practices
required by a regulation, based on uncontrolled emissions.
Rule effectiveness reflects the assumption that regulations typically were not 100 percent
effective due to limitations of control techniques or shortcomings in the enforcement process.
Several factors were taken into account in estimating rule effectiveness, including:
ambiguity in how the regulation should be applied
differences between the long-term performance of control equipment and the perform-
ance measured in performance tests
fugitive emissions not routed to the control device
clarity of procedures used in determining compliance
frequency of inspections
potential for significant violations of the regulation
follow-up activities at sources violating the regulation.
Rule penetration is the portion (in percent) of the area source category that is covered by the
regulation, in terms of uncontrolled emissions.
The product of these three factors can be used to estimate the overall reduction efficiency for a given
area-source category:
Overall efficiency = [control technology efficiency] x [rule effectiveness/100]
x [penetration/100]
Actual emissions = [uncontrolled emissions] x [1 - (overall efficiency/100)]
4-29
-------
The initial State survey produced such a wide range of responses that there was concern that parameter
estimates might cause inconsistencies in emissions projections for neighboring States. Further, some
States expressed very low confidence in their estimates. Therefore, the survey was repeated after
States had the opportunity to review the preliminary estimates made by others in the region. In addition,
States were asked in the second round to indicate their level of confidence in the parameter estimates.
This was expressed by letter "grades" from A to E, with A representing the highest degree of confidence
and E the lowest.
State responses from across the region were then averaged for each area-source category. That is, the
mean control technology efficiency was computed for each category by taking the average of all non-
zero responses. The same calculations were made for rule effectiveness and penetration. Table 4-6
summarizes the average responses for each parameter, as well as the range of responses and the
average level of confidence. The table also gives the "average" overall efficiency, obtained by multiply-
ing the mean values for control technology efficiency, rule effectiveness, and penetration. If a State did
not express a high level of confidence in all of its control factor estimates for a given category (A or B for
control technology and A for rule effectiveness and penetration), the overall reduction efficiency for that
category was defaulted to the average overall efficiency from Table 4-6. Otherwise, the overall reduction
was computed from the State's responses for control technology efficiency, rule effectiveness, and
penetration. Values for 1985 and 2005 control efficiencies used in developing the 2005 baseline VOC
emissions reflect the overall efficiency. The State-wide and county-specific 2005 control efficiencies are
provided in Appendix K.
4.5.2 NO, Stationary-Source Controls for 2005 Baseline Scenarios
Both the Phase I and Phase II 2005 baseline scenarios incorporate NSPS NOX emissions limits. NSPS
limits on NOX emissions were reflected in the Advanced Utility Simulation Model (AUSM) utility growth
factors, and also in the methodology used to generate industrial combustion growth factors. Additional
control measures specific to Phase I or Phase II baselines are described below.
Phase I - 2005 Baseline NOX
In addition to FMVCP, I/M and NSPS controls, all Phase I inventories include a State-level cap on total
NOX emissions so that total point-, area-, and mobile-source NOX emissions in each State were held to
4-30
-------
the 1985 level. The cap was implemented by applying sufficient point source control measures in each
State to offset overall NOX emissions growth, taking into account reductions already achieved by NSPS
and FMVCP rules.
Point source controls were applied preferentially to new sources and to utilities. Low-cost controls such
as low-NOx burners (LNB), staged combustion air, reduced air-to-fuel ratio, and steam injection were
used before selective catalytic reduction (SCR). The following list details the priority system used to
apply point source measures, with controls and emissions categories listed in descending order of
preference:
A. Low NOX burners, staged combustion air, reduced air-to-fuel ratio, or steam injection used on:
1) New sources
Utilities
Coal
Oil
Natural gas
Industrial combustion
Coal
Oil
Natural gas
2) Existing sources - with the same rankings as above for source categories and fuels
B. Selective catalytic reduction - applied to new sources before existing sources and with the same
rankings as above for source categories and fuels.
Table 4-7 summarizes the point source control measures needed to meet the NOX cap in each ROMNET
State.
Phase II - 2005 Baseline NOX
The Phase II 2005 baseline inventory, like all of the Phase I inventories, includes the effects of current
FMVCP standards for mobile source NOX emissions. However, mobile NOX emissions in the Phase II
baseline were based on MOBILE4 rather than on MOBILE3.9.
In addition to FMVCP and NSPS programs, the Phase II baseline includes a 5 percent reduction for NOX
from all utilities. This factor accounts for expected improvements to energy efficiency that were
expected to reduce fuel consumption. The energy-efficiency factor is applied to pods 50 through 53
(utility external combustion of coal, oil, natural gas, and other fuels), and pods 74 through 77 (utility oil
and gas turbines and reciprocating engines).
4-31
-------
4.5.3 VOC. NO*, and CO 2005 Baseline Mobile Source Controls
To create the Phase I and Phase II mobile source baseline inventories for 2005, emission factors were
calculated using MOBILE3.9 and MOBILE4, respectively. MOBILE4 inputs for the Phase II baseline are
given in Appendix L The effects of current inspection/maintenance (I/M) programs on baseline emis-
sions in 2005 were estimated by assuming that all areas in which I/M existed in 1985 will have a typical
"basic" I/M program operating at the estimated 1985 program effectiveness level for the area in question.
The basic I/M program consists of the following characteristics:
ROMNET 'BASIC1 I/M PROGRAM CHARACTERISTICS
I/M PROGRAM:
START YEAR (JANUARY 1): 1983
PRE-1981 MYR STRINGENCY RATE: 20%
MECHANIC TRAINING PROGRAM?: NO
FIRST MODEL YEAR COVEREDt 1970
LAST MODEL YEAR COVERED: 2020
VEHICLE TYPES COVERED: LDGV, LDGT1, LDGT2
1981 + MYR TEST TYPE: IDLE ONLY
1981 + MYR TEST OUTPOINTS: 1.2% CO, 220 PPM HC
AMD-TAMPERING PROGRAM SELECTED: NONE
ADDITIONAL MOBILE4 I/M INPUTS
WAIVER RATES: 5% (ALL YEARS)
COMPLIANCE RATE: 100%
INSPECTION TYPE: CENTRALIZED
INSPECTION FREQUENCY: ANNUAL
MOBILE4 runs produce the following ideal I/M control efficiencies, weighted across the 2005 vehicle
fleet:
Percent Control
Total VOC 12.1
Gasoline exhaust 18.1
Gasoline evaporation 9.3
Diesel 0.0
Nitrogen oxides 4.0
Carbon monoxide 19.9
The following I/M program effectiveness percentages were obtained from program assessments per-
formed by the EPA Office of Mobile Sources and were developed from comprehensive reviews of the
characteristics and performance of each program.3 Program effectiveness for NOX is taken to be the
same as for VOC. These effectiveness ratings combine the effects of program deviations from the pro-
totypical 5 percent waiver, 100 percent compliance rate, and centralized annual program on which the
basic I/M reductions were based.
3. Information provided by David Brzezinski, EPA Office of Mobile Sources, Ann Arbor, Ml, to Edwin L. Meyer, EPA Office of Air
Quality Planning and Standards, Research Triangle Park, NC. Revised I/M Program Effectiveness Data. December 20, 1989.
4-32
-------
1985 Program Effectiveness (Percent)
State VOC CO
Connecticut 84 81
Delaware 80 80
District of Columbia 84 81
Indiana 53 53
Kentucky 90 88
Maryland 87 85
Massachusetts 63 56
New Jersey 84 81
New York 50 50
Pennsylvania 78 74
Rhode Island 44 33
Virginia 44 34
A Reid vapor pressure of 9.0 psi was included regionwide in the 2005 baseline in anticipation of gasoline
volatility regulations by the NESCAUM States. The Phase II 2005 baseline inventory also assumes,
based on the January 19, 1990 Notice of Proposed Rulemaking (55 FR 1914), a basic evaporative test
procedure for new vehicles. This rule is projected to result in 36 percent control of diurnal and hot-soak
emissions, and 71 percent control of running losses beginning with the 1993 model year.4
4.5.4 VOC Maximum Technology Stationary Source Controls
Maximum VOC technology efficiencies used in Phase I and in Phase II are listed in Table 4-8 for point
sources and Table 4-9 for area sources. The tables also give brief descriptions of the control techniques
that form the basis for the maximum technology efficiencies. The major references for maximum tech-
nology controls were the Ozone NAAQS Cost Analysis (Battye ef a/., 1987), and the Emissions Reduc-
tion and Cost Analysis Model (ERCAM) for VOC (Pechan, 1989). Efficiencies given in these references
were based on primary sources such as EPA Control Techniques Guidelines and NSPS Background
Documents. As the tables show, maximum technology efficiencies for a number of surface coating
categories were based on incinerators and other add-on controls. Comparable efficiencies can be
achieved for some of these categories with low-solvent coatings. All of the maximum technology effi-
ciencies match or supersede any applicable NSPS rules, and all efficiencies were assumed to be 100
percent effective.
4. Information provided by Lois Platte, EPA Office of Mobile Sources, Ann Arbor, Ml, to Mark G. Smith, Alliance Technologies
Corporation. MOBILE4 Input for EPA Evaporative Test Procedure Proposal. January 30,1990.
4-33
-------
4.5.5 NOr Maximum Technology Stationary Source Controls
Maximum NOX technology efficiencies used in Phase II for point and area sources are listed in
Table 4-10. Efficiencies given in the table are based on selective catalytic reduction (SCR) for utilities,
low-NOx burners or comparable technology for industrial combustion sources, and the best identified
controls for off-highway mobile sources (Johnson, 1988; Ellison, 1988).5 All of the maximum technology
efficiencies match or supersede any applicable NSPS rules. The location of NOX point sources affected
by maximum technology controls are shown in Figure 4-18.
4.5.6 Mobile-Source VOC. NCv. and CO Maximum Technology Controls
Mobile-source controls used in Phase I consist of the following "enhanced" I/M program applied region-
wide, with 100 percent program effectiveness instead of the State-specific measures used in the base-
line:
CS1 I/M PROGRAM
I/M PROGRAM SELECTED
START YEAR (JANUARY 1): 1983
PRE-1981 MYR STRINGENCY RATE: 20%
MECHANIC TRAINING PROGRAM?: NO
FIRST MODEL YEAR COVERED: 1970
LAST MODEL YEAR COVERED: 2020
VEHICLE TYPES COVERED: LDGV, LDGT1, LDGT2
1981 + MYR TEST TYPE: LOADED /IDLE
1981 + MYR TEST CUTPOINTS: 12% CO / 220 PPM HC
ANTI-TAMPERING PROGRAM SELECTED
START YEAR (JANUARY 1): 1990
FIRST MODEL YEAR COVERED: 1970
LAST MODEL YEAR COVERED: 2020
VEHICLE TYPES COVERED: LDGV
ANNUAL INSPECTION: AIR PUMP, CATALYST, FUEL INLET,
PLUMBTESMO LEAD TEST
For Phase II, the Phase I enhanced I/M program was modified to add heavy duty vehicles and additional
variables required by MOBILE4 (5 percent waiver rate, 100 percent compliance, centralized/annual pro-
gram type). Reid vapor pressure was reduced to 7.0 psi from 9.0 psi. Tailpipe emissions standards
simulated in CS10, as shown below, reflect the most stringent control level proposed in Senate Bill 1630,
the December 20,1989 Clean Air Act legislation. A heavy duty diesel NOX emissions standard reduction
from 5 to 4 grams/brake-horsepower-hour was also included in CS10 as an approximation of the control
limit of current diesel technology.
5. Information provided by Larry Jones, Acid Deposition Branch, US Environmental Protection Agency, Research Triangle Park,
NC. Typical efficiencies for low-cost NOX controls.
4-34
-------
HC NO* Units
Light Duty Vehicles 0.125 0.20 grams per mile
Light Duty Trucks
1) 3750-5750 Ib 0.380 0.70 grams per mile
2) 5750-8500 Ib 0.460 1.10 grams per mile
Heavy Duty Diesels None 4.00 grams per brake horsepower hour
The following set of MOBILE4 base emissions and deterioration rates was used to simulate these emis-
sions standards:
HC NOX
ZM DR ZM DR Units
Light Duty Vehicles 0.085 0.028 0,127 0.017 grams per mile
Light Duty Trucks 1 0.14 0.05 0.35 0.03 grams per mile
Light Duty Trucks 2 0.22 0.08 0.64 0.04 grams per mile
Heavy Duty Diesels 0.0 0.0 3.11 0.0 grams per brake horsepower hour
ZM = Zero mile rate, OR = Deterioration rate per 10,000 miles
To approximate full penetration of these standards into the fleet, this MOBILE4 run included introduction
of these standards for light duty vehicles and trucks in 1995, and resulting emission factors for 2020
were used, since the modeled fleet in that year is anticipated to reach a steady state consisting of only
such vehicles. The combined effect of the resulting maximum technology for mobile sources is the fol-
lowing reductions over the 2005 baseline uncontrolled and 'basic' I/M composite emission factors:
Percent Reduction Over:
Uncontrolled Basic I/M
Total VOC 62.0 56.2
Gasoline exhaust 66.3 65.5
Gasoline evaporation 56.7 52.2
Diesel 0.0 .0.0
Nitrogen oxides 49.9 47.8
Carbon monoxide 37.8 22.2
Inputs to MOBILE4 used for the Phase II maximum technology VOC strategy are given in Appendix L
4.5.7 Controls Applied In the CS05 Clean Air Act Strategy
In this strategy, various VOC stationary source controls were applied at the urban level by MSA or
CMSA, depending on the severity of the ozone problem as indicated by the ozone design value. The
attainment categories, based upon ambient monitoring data from 1986 through 1988, are listed below.
4-35
-------
Attainment Category Design Value* Attainment Date
1-attainment < 0.12 ppm —
2 - marginal nonattainment 0.13 ppm December 31,1995
3 - moderate nonattainment 0.14-0.15 ppm December 31,1995
4 - serious nonattainment 0.16 - 0.15 ppm December 31, 2000
5 - severe nonattainment 0.19 ppm December 31,2010
* based on 4th highest daily maximum concentration over 3 years.
The VOC point- and area-source control measures applied by attainment category are provided in
Table 4-11 and Table 4-12, respectively. In the areas listed below, controls were also applied to dry
cleaners (70 percent control), bakeries (80 percent control) and miscellaneous industrial solvents (50
percent control).
Baltimore/Washington, DC Massachusetts
Philadelphia Manchester - Nashua, NH
Atlantic City Portland, ME
New York City Harrisburg, PA
Poughkeepsie, NY Huntingdon - Ashland, WV-KY-OH
Hartford, CT Toledo, OH
Providence, RI
Vehicle tailpipe standards included in this strategy are given below:
Start
HC NOX Units Year
Light Duty Vehicles 0.25 0.40 grams per mile 1994
Light Duty Trucks
1) 3750-5750 Ib 0.25 1.00 grams per mile 1995
2) 5750-8500 Ib 0.50 1.00 grams per mile 1995
Inputs to MOBILE4 used to generate mobile source emission factors for this strategy are given in
Appendix L
In addition to these category-specific controls, across-the-board reductions were applied to point, area,
and mobile sources in order to meet various VOC target reductions enumerated in the draft legislations.
These reductions are listed by urban area in Table 4-13.
4-36
-------
4.5.8 Development of Reactivity-Reduction Scenarios
For VOC, CS15 and CS20 were based on CS10 and the 2005 baseline, with two additional measures to
reduce VOC reactivity. The first measure is a cap on the reactivity of solvent emissions. For the pur-
poses of this cap, reactivity is defined as the overall reaction rate with hydroxyl radical per unit weight of
the solvent:
React! vity= ^RRt.xSF,.
i
where:
/ refers to the different Carbon Bond-IV classes;
RR/ is the reaction rate of CB-IV class / with the hydroxyl radical at 85°F:
43,187 reactions-kg-1-min-1 for olefin;
1,203 reactions-kg-1-min-1 for paraffin;
9,315 reactions-kg-1-min-1 for toluene;
36,433 reactions-kg-1-min-1 forxylene;
15,000 reactions-kg-1-min-1 for formaldehyde;
24,335 reactions-kg-1-min-1 for other aldehydes;
12,194 reactions-kg-1-min-1 for ethene;
142,000 reactions-kg-1-min-1 for isoprene;
1 reaction-kg-1-min-1 for nonreactive hydrocarbons;
1,600 reactions-kg-1-min-1 for methanol);
and
SF/ is the speciation factor for CB-IV class / (moles-kg-1 solvent).
The solvent reactivity cap was set at 35,000 •kg-1-min-1, which is midway between the reactivities of
isopropanol (30,000 -kg-1-min-i) and ethyl acetate (40,000 -kg^-min-1). Both of these solvents are
commonly used in surface coatings. Any solvents with higher reactivities in either the point or area
source portions of the inventory were assumed to be replaced. The replacement solvent is a mixture of
the above two solvents, with a reactivity at the level of the cap and a speciation of 29 moles paraffin per
kilogram solvent and 14 moles nonreactive hydrocarbons per kilogram.
The second reactivity reduction measure involves conversion of the entire Northeast Corridor highway
vehicle fleet to 100 percent methanol fuel (M100) vehicles.6 It is estimated that complete conversion of
the vehicle fleet to M100 would result in a 71 percent reduction in fleet-wide evaporative emissions, and
an increase in exhaust emissions of 325 percent relative to the CS10 maximum control scenario. This is
based on M100 evaporative and exhaust emission factors of 0.055 and 0.565 grams/mile, respectively.
6. Special Report: Analysis of the Economic and Environmental Effects of Methanol as an Automotive Fuel. EPA Office of Mobile
Sources, Ann Arbor, Ml. September 1989.
4-37
-------
Changes in speciation also affect these categories, resulting in a net reduction in reactivity even for
exhaust emissions. Table 4-14 lists speciation factors for VOC emissions from methanol-fueled vehicles.
These speciation factors were based on primary factors which included a breakdown of the exhaust
emissions into methanol, formaldehyde and other nonmethane hydrocarbons as 88.5, 2.7 and 8.8 per-
cent by weight, respectively. The other nonmethane hydrocarbons were further speciated using
previously-existing data for exhaust hydrocarbon speciation.7
7. Memorandum from Charles Gray, EPA Office of Mobile Sources, Ann Arbor, Ml, to John Calcagni, EPA Office of Air Quality
Planning and Standards, Research Triangle Park, NC. Motor Vehicle VOC Speciation for SIP Development. March 23, 1989.
4-38
-------
4.6 DEVELOPMENT OF A BIOGENIC-SOURCE EMISSIONS INVENTORY
4.6.1 Introduction
The abundance of naturally-occurring hydrocarbons in the atmosphere has been recognized for at least
20 years (Arnts and Meeks, 1981; Peterson and Tingey, 1980; Rasmussen, 1972). It was not until
recently, however, that modeling studies suggested the need to consider biogenic hydrocarbon emis-
sions for estimating the production of photochemical oxidant smog (Chameides ef a/., 1988; Trainer ef
a/., 1987). Even before the publication of these studies, the EPA reported on the development of a
computer algorithm for estimating biogenic emissions (Novak and Reagan, 1986). This system, called
the Biogenic Emissions Software System (BESS), was designed to produce hourly gridded hydrocarbon
emissions for an early version of the ROM. More or less parallel to the efforts at the EPA, researchers at
Washington State University devised a method for estimating seasonal, county-wide hydrocarbon
emissions across the United States (Lamb ef a/., 1987). These estimates were used for early parts of the
NAPAP. In preparing for the 1985 NAPAP emissions inventory and while testing later versions of the
ROM, researchers at Washington State University and EPA collaborated on combining features of their
two biogenic emission systems (Young ef a/., 1989). An outcome of this collaboration was the develop-
ment of a generalized scheme that could estimate hourly gridded biogenic hydrocarbon emissions for
use with either the ROM or the Regional Acid Deposition Model (RADM). This Section discusses the
formulation of this system, called the Biogenic Emissions Inventory System (BEIS). More-detailed infor-
mation on the BEIS can be found in Milich ef a/. (1990); and discussions of BEIS estimates are in Pierce
etal. (1990).
4.6.2 Description of the System
Calculations with the Biogenic Emissions Inventory System (BEIS) require consideration of biomass,
emission factors, and environmental factors. The basic equation for these calculations can be
expressed as follows:
ER<= £[BF,.EF17F(S,T)]
;
where ER is the emission rate (g/s/model grid cell), / is the chemical species (such as isoprene or
monoterpene),y is the vegetation type, BF is the leaf biomass factor (g/m2), EF is the emission factor
(/^g/g-leaf biomass/h),and F(S,T) is an environmental factor that accounts for solar radiation S and leaf
temperature T.
4-39
-------
Biomass Factors
Leaf biomass for forested areas and surface areas for nonforested areas were derived from land use
data in the Oak Ridge National Laboratory's Geoecology Data Base (Olson etal., 1980). The land use
data, some of it dating back to the early 1970's, consist of county-level acreages for forest types,
agricultural crops, and other areas such as urban, grassland, and water.
In the Geoecology Data Base, forested areas were classified as one of about 100 forest types. Lacking
detailed information on forest biomass and emission factors, each forest type was simply aggregated
into three groups: oak, other deciduous, or coniferous. Relationships suggested by Lamb et al. (1987)
were used to partition the leaf biomass for each forest group into four emission categories: (1) high
isoprene deciduous, (2) low isoprene deciduous, (3) no isoprene deciduous, and (4) coniferous. These
biomass factors (g/m2) are listed below:
Emission Category
Forest High isoprene Low isoprene Non-isoprene Non-isoprene
group deciduous deciduous deciduous coniferous
Oak 185 60 60 70
Other deciduous 60 185 90 135
Coniferous 39 26 26 559
Non-forest vegetation types were assigned to one of twenty categories. Agricultural crops not having an
explicit emission factor were categorized as a miscellaneous crop. Based on a review of land use data,
urban areas were assumed to contain 20 percent forests and 20 percent grassland. Forested regions in
urban areas were divided equally among oak, other deciduous, and coniferous groups.
Seasonal adjustments of biomass were based on frost dates for each county using a simple step func-
tion. For each month, deciduous vegetation (any non-coniferous vegetation class) within a county is
assumed to have either full biomass or no biomass. For oxidant modeling, this is not a critical
assumption because most high ozone episodes occur during the summer months.
Emission Factors
The emission factors in BEIS were largely based on Zimmerman's Tampa Bay Study (1979), primarily
because of the lack of standardized measurements from other biogenic emissions field programs.
Emission factors (/^g/g/h),standardized for full sunlight and 30°C, are listed below for the four forest
emission categories. By multiplying these emission factors by the corresponding biomass factors,
emission fluxes (/jg/m2/h)can be compared for vegetation types, as shown in Table 4-15.
4-40
-------
Emission Category
Chemical Highisoprene Lowisoprene Non-isoprene Non-isoprene
species
Isoprene
a-pinene
Monoterpene
Unidentified
deciduous
14.69
0.13
0.11
3.24
deciduous
6.60
0.05
0.05
1.76
deciduous
0.0
0.07
0.07
1.91
coniferous
0.0
1.13
1.29
1.38
Emission fluxes in Table 4-15 are expressed in terms of total non-methane hydrocarbons and were
partitioned as isoprene, monoterpene, alpha-pinene, and unidentified. In most instances, the emission
fluxes from forested areas were significantly higher than the fluxes for agricultural areas. An exception is
corn, which has an overall hydrocarbon emission flux of 3542 /zg/m2/h. Most of the emissions flux from
corn is assigned to the unidentified category because the gas chromatography analysis did not identify
specific hydrocarbon compounds. Rather than ignore the unidentified portion of the flux estimate, it is
tentatively assumed that unidentified hydrocarbons can be treated as 50 percent terpene, 45 percent
paraffin, and 5 percent nonreactive.
Sources of naturally-emitted NO include biomass burning, lightning, microbial activity in soils, and
ammonia oxidation. Although these natural sources were reportedly much smaller than anthropogenic
sources (Logan, 1983), concerns about air quality in rural areas (where anthropogenic emissions tend to
be small) suggest the need to consider natural NO emissions. Lacking sufficiently detailed emission
factors for other sources, only NO emissions from grasslands were considered in this version of BEIS.
The equation used is (Fehsenfeld, 1990):
0 = 0. 74 exp(0. 079 Ts)
where 0 is the nitrogen flux (ng of N m-2s-1), and Ts is the soil temperature (°C) as estimated from
7S = 0. 70 Ta + 3. 6, where Ta is the ambient air temperature (°C). It is assumed that the nitrogen
emitted into the air from grasslands is emitted in the form of NO.
Environmental Factors
Laboratory studies have shown that biogenic emissions from most plant species react strongly to
changes in temperature and isoprene emissions are also sunlight dependent (Tingey, 1981; Tingey ef
a/., 1980, 1981). BEIS includes adjustments for temperature and sunlight using these relationships. It
also attempts to simulate the vertical variation of leaf temperature and sunlight within a forest canopy.
4-41
-------
BEIS incorporates an adjustment factor for isoprene emissions (Tingey et a/., 1981) that is given by the
following:
l»exp(-6(T-c))
where a, b, c, d, and e are empirical coefficients that vary as a function of solar intensity S as given
below. The leaf temperature (7) is expressed in °C. The above relationship is quite sensitive to tem-
perature and sunlight as shown in Figure 4-19. For example, only a 2°C change of temperature during
conditions of bright sunlight and 30°C can result in a 50 percent change in the isoprene emission rate.
At night, isoprene is usually not emitted and is assumed to be zero in BEIS.
Light intensity Empirical Coefficients
a
1.200
0.916
0.615
0.437
b
0.400
0.239
0.696
0.312
c
28.30
29.93
32.79
31/75
d
0.796
0.462
0.077
0.160
e
1.00
1.95
4.75
10.73
800
400
200
100
Existing laboratory data for non-isoprene emitting plants have thus far identified only leaf temperature as
an important variable (Tingey et a/., 1980; Tingey, 1981). The environmental adjustment factor used in
BEIS for non-isoprene emitting plants is given by:
F(T) = exp(a[T-30])
where values for the coefficient a are given below.
Empirical Coefficient
Compound a
a-pinene 0.067
Monoterpene 0.0739
Unidentified 0.0739
A major refinement for BEIS was to include a canopy model for estimating profiles of leaf temperature
and sunlight within forest canopies. The model was adapted from research initially performed at
Washington State University (Gay, 1987). Sunlight is assumed to decrease exponentially through
hypothetical forest canopies with the rate of attenuation depending on the assumed biomass distribu-
tion. Both visible and total radiation were calculated for eight levels in the canopy. The visible portion of
the solar spectrum is assumed to decrease more quickly than the total spectrum, because leaves are
4-42
-------
more effective at absorbing visible light than other portions of the spectrum (Baldocchi ef a/., 1984). The
calculated total solar radiation is used to compute the leaf temperature at each level using a radiational
balance equation described by Gates and Papian (1971). The estimated leaf temperatures and visible
radiation values are then used with the Tingey adjustment factors which were discussed above.
Adjustments for temperature are also important for estimates of NO emission rates from grasslands.
The adjustment factor for NO is noted explicitly in the emission factor equation.
4.6.3 Computational Processes
As in ROMNET, ROM applications typically involve a simulation for about a two-week period. BEIS is
used to provide biogenic emissions for these simulations through the steps indicated in the system flow
diagram (Figure 4-20).
First, the biomass for the simulation period is computed with BEIS using the monthly adjustment flags.
The outputs of this process are data files of gridded canopy and noncanopy biomass. The next step is
to compute standardized (30°C and full sunlight conditions) emission fluxes at each grid cell. This
calculation is performed once per model simulation. The computer intensive portion of BEIS involves
hourly corrections for sunlight and temperature. For noncanopy vegetation, Tingey adjustment factors
were computed using surface temperature and solar radiation data that were provided by the ROM
meteorology-input processors. For canopy vegetation, values for surface temperature, wind speed,
relative humidity, and solar radiation were fed into the canopy model which computes profiles of leaf
temperature and sunlight. These profiles were used to compute Tingey adjustment factors as a function
of height using the assumed biomass distribution. The adjustment factors were then multiplied by the
standardized emission fluxes, and the noncanopy and canopy emission rates were merged. Emission
rates were finally converted into the appropriate chemical species. For example, the ROM uses the
following conversions (Gery etal., 1988):
1 mol isoprene is treated as 1 mol ISOP,
1 mol monoterpene or alpha-pinene is treated as 0.5 mol OLE, 6 mol PAR, and 1.5 mol ALD2,
1 mol unidentified (or unknown) is treated as 0.5 mol OLE, 8.5 mol PAR, and 0.5 mol NONR,
where ISOP, OLE, PAR, ALD2, and NONR were surrogate species used in the Carbon Bond IV mecha-
nism and refer respectively to isoprene, olefin, paraffin, higher aldehydes, and nonreactives.
Estimates of NO emissions require the grassland area in a grid cell (available from the Geoecology Data
Base, Olson ef a/., 1980) and the surface temperature (provided by the ROM meteorological input pro-
4-43
-------
cessors). Emissions were generated for each grid cell in units of moles of NO. The final grid- and day-
specific emissions of VOC and NOX derived from BEIS are merged with the anthropogenic emissions
derived from the ROM emission-processor system (refer to Section 2).
In ROMNET, biogenic emissions rates are assumed to be unaffected by growth in anthropogenic emis-
sions. For example, it was not possible to consider the potential impact on biogenic emissions of future
urbanization of current agricultural or forest lands. Thus, the base case, 2005 baseline, and control
strategy scenarios all used a single set of day-specific biogenic emissions.
As indicated in Section 4.4.2, several scenarios were designed to examine the relative effectiveness of
various anthropogenic emissions strategies considering the range of uncertainty in biogenic emissions.
Uncertainty is believed to be large for biogenic emissions due to a lack of extensive data sets for speci-
fying critical factors such as biomass emissions rates for various vegetation species and the effects on
emissions of environmental stress (e.g., insects, drought, etc.), as well as changes in land-use patterns
over the past 10 to 20 years. Although it was not possible to estimate the uncertainty in each biogenic
emissions component, an uncertainty of ± a factor of 3 is believed to be representative of the overall
range of uncertainty in biogenic emissions calculated for ROMNET. This factor was used for scaling
biogenic emissions calculated by BEIS in the biogenic sensitivity scenarios 1985H, 1985L, CS06 - CS09,
CS17, CS21,andCS22.
4.6.4 Characteristics of the Blogenlc VOC Emissions Inventory
As discussed Section 4.6.3, biogenic VOC emissions are very sensitive to ambient temperature levels. A
principal biogenic species is isoprene. Figure 4-21 and Figure 4-22 present spatial distributions of iso-
prene for a cool and a warm day for 1000 EST and 1500 EST. These biogenic emission estimates are
based on hourly temperatures for each grid cell for the hours shown.
Two comparisons are described. The first relates isoprene emissions on the same day (morning versus
afternoon). From the 'cool" day plots (Figure 4-21), the most noticeable effect is the increase in emis-
sions between the morning and afternoon across southeastern Ohio, central Pennsylvania and upstate
New York. On an overall basis, isoprene emissions increased approximately 25 percent from 1000 EST
and 1500 EST on the 'cool' day. The "warm1 day plots (Figure 4-22) show that isoprene emissions
increase in eastern Kentucky, and from western West Virginia across central Virginia and northeastward
into upstate New York, New Hampshire, and Vermont. On an overall basis, isoprene emissions
increased approximately 20 percent from 1000 EST and 1500 EST on the 'warm' day.
4-44
-------
The second comparison is for isoprene emissions on different days (cool versus warm days). Figures
4-21 (top) and 4-22 (top) indicate that isoprene emission levels are significantly higher in the morning for
the warm day. This is especially evident for much of the inland areas (West Virginia, Virginia, Pennsyl-
vania, and New York). On an overall basis, isoprene emissions were approximately 70 percent higher
during the morning of the warm day than on the cool day. Figures 4-21 (bottom) and 4-22 (bottom)
show that a very similar pattern can be observed in the afternoon isoprene emission levels. The overall
increase in afternoon isoprene emissions between the "cool" and the "warm" day was approximately 60
percent.
Figure 4-23 displays daily total biogenic emission rates for each day of the July 1988 episode. These
totals include not only isoprene, but alpha-pinene, monoterpene, and unidentified biogenic VOC emis-
sion estimates as well. Total regionwide biogenic VOC emission rates vary between 18,000 and 32,000
tons per day compared with an average regionwide anthropogenic VOC emission rate of 24,000 tons
per day.
4-45
-------
Input Data
FREDS Modules
Annual
Inventory
Methane and
Aldehyde
Factors
Hydrocarbon
Preprocessing
Model Data
Extraction
Species
Profiles
*"
PSPLIT
^
^
Speciation
Spatial
Allocation
Factors
Temporal
Allocation
Factors
Spatial
Allocation
Temporal
Allocation
Model Input
Preprocessor
Figure 4-1. Summary of software modules and input data used in FREDS.
4-46
-------
CC
LU
0_
cc
o
o
u.
Z
O
CO
CO
O
O
SO
60 70 80 90
AVERAGE DAILY TEMPERATURE (F)
Figure 4-2. Effect of average daily temperature on VOC emissions. (Emissions are given for a com-
posite of vehicle types and speeds for a diurnal temperature range of 20° F.)
•5.
cc
UJ
D_
CO
cc
o
CO
CO
o
o
2 -
10 20 30
DIURNAL TEMPERATURE RANGE (F)
40
Figure 4-3. Effect of diurnal temperature range on VOC emissions. (Emissions are given for a com-
posite of vehicle and speed classes at an average daily temperature of 85°F.)
4-47
-------
MOBILE
(47.2%)
CHEMICAL PROCESSES
(9.5%)
SURFACE COATINGS
(36.4%)
STORAGE TANKS
(4.5%)
SOLVENT USE
(8.6%)
a. Total VOC
b. Point Source VOC
COMBUSTION
TSDF's (4.0%)
(6.7%) l J F'RES
GASOLINE MKT
(14.6%)
SOLVENT USE
(57.6%)
OTHER
(7.7%)
EVAPORATIVE
(72.7%)
c. Area Source VOC
d. Mobile Source VOC
Figure 4-4. Distribution of regionwide 1985 VOC emissions.
4-48
-------
AREA
(13.7%)
a. Total NOx
CHEMICAL PROCESSES
(0.8%)
b. Point Source NOx
c. Area Source NOx
Figure 4-5. Distribution of regionwide 1985 NOX emissions.
4-49
-------
1985 Base Case Anthropogenic VOC Emissions
Emissions (tons/day):
— O isssss <— 3
<- SO amm > 50
<- 15
1985 Base Case Anthropogenic NOX Emissions
Emissions (tons/day):
- O
<- 50
<- 3
> 50
HE <- 15
Figure 4-6. 1985 base case anthropogenic emissions (tons/day) of VOC and NOX.
4-50
-------
Industrial Nox Sources Greater Than or Equal to 500 TPY
Point Bouroa EmlMlon (TPDh
•*• <- 20
* <- 30
A <- 100
> 100
Utility Nox Sources Greater Than or Equal to 500 TPY
Point Souro* Emission (TPDh
<- 20
* <- 30
A <- 100
Figure 4-7. 1985 base case anthropogenic emissions (tons/day) of NOX from industrial plants and uti-
lities.
4-51
-------
Anthropogenic Emissions by Source Type
For a Typical Weekday, Saturday, and Sunday
D
a
CO
c
o
CO
c
_o
'(O
CO
UJ
70000-
63000-
56000-
49000-
42000-
35000-
28000-
21000-
14000-
7000-
0
WK SAT SUN
I— voc —I
WK SAT SUN
\— NOX —I
WK SAT SUN
Area
Mobile V/////A Point
Figure 4-8. 1985 base case emissions (tons/day) by day type (weekday, Saturday, and Sunday).
4-52
-------
Anthropogenic VOC Emissions by Source Type
For o Typical Weekday
7-
c
g
'in
en
4-
0_
1 -
0-
O-B-D D D D D D D D-B-E1
n—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—i—
0 1 2 3 4 5 6 7 8 9 10111213141516171819 20 21 22 23
Hour (EOT)
Source Type
Area
Mobile
Point
Total
Anthropogenic NOX Emissions by Source Type
For a Typical Weekday
4-
a
"o
K
x
a
Q
2-
c
0)
-------
0)
a.
8-
7-
6-
5-
4-
2-
Anthropogenic CO Emissions by Source Type
For a Typical Weekday
I I I I T I
0 1 2 3 4 5 6 7 8 9101112131415 16-17 18 19 20 21 22 23
Hour (EOT)
Source Type
Area
Mobile
Point
Total
Figure 4-10. Diurnal profiles of point, area, and mobile sources of CO.
4-54
-------
Mobile Source VOC Emissions - 15OO EST Cool Day
Emissions (tons/hour): • • •> O.OO sssss <— O.I 5 aramnm < — O.45
<- O.6O •• > O.60
Mobile Source VOC Emissions - 1500 EST Warm Day
Emissions (tons/hour):
— O.OO seas <— 0.15 samara < — 0.45
<- 0.60 mmm > 0.6O
Figure 4-11. Mobile-source emissions (tons) at 1500 EST on a "cool" day and a 'warm1 day.
4-55
-------
Varlotions in Mobile Source VOC Emissions
For the ROMNET Domain
Q
c
o
in
•j;
LJ
15000-
13500-
12000-
10500-
9000-
7500-
6000-
4500-
3000-
1500-
0
03# 04 05 06
07 08 09* 10# 11 12
'July 1988 Episode
13 14 15 16*
Variations in Mobile Source NOX Emissions
For the ROMNET Domain
Q
c
o
c
g
en
'E
UJ
10000-
9000-
8000-
7000-
6000-
5000-
4000-
3000-
2000-
1000-
0
03# 04 05 06
07 08 09* 10# 11
July 1988 Episode
* Saturdoy j Sunday
13 14 15
I 6*
Figure 4-12. Regionwide daily total mobile-source VOC and NOX emissions (tons/day), July 1988 epi-
sode.
4-56
-------
Voriotions in Mobile Source CO Emissions
For the ROMNET Domain
50000-
45000-
40000-
35000
30000-
25000-
20000-
15000-
10000-
5000-
0
03# 04 05 06 07 08 09* 10# 11 12 13 14 15 16*
July 1988 Episode
* Saturday § Sunday
Figure 4-13. Regionwide daily total mobile-source CO emissions (tons/day), July 1988 episode.
4-57
-------
VOC Emissions
Northeast Corridor
(41 )
1 1
(73) (73) (73) (74)
60 60 60 60
71 72
(68)
52
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23 8io
Emissions Scenario
Point
Area
Mobile
Biogenic
VOC Emissions
USA Portion of the Domain Excluding NE Corridor
300-
_270-
r§240-
£ 210-
O 150-1
M 120-1
I 9^
in
1 60-1
LU
30 ^
(77) (77) (77) (77) (77) (77) (77)
65 65 65 65 65 65 65
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23 Bio
Emissions Scenario
Point KgSSSI Area V/////A Mobile ••• Biogenic
Values above bars: Percent reduction from 2005 and from (1985)
Figure 4-14. VOC emissions for the Phase II scenarios and the percent reduction from the 1985 base
case and the 2005 baseline.
4-58
-------
NOx Emissions
Northeast Corridor
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23
Emissions Scenario
Point
Area V/////A Mobile
NOx Emissions
USA Portion of the Domain Excluding NE Corridor
120-1
cMOO
a
\ 90
o 80
i—
co 70-
8 60
" 50-
(0
o
30
20
10
0
( 6)
7
(34)
35
(49)
50
(60)
60
(59) (60) (59) (59)
60 60 60 60
(46)
46
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CSZ3
Emissions Scenario
R^^i Point IS8888I Area W////A Mobile
Values above bars: Percent reduction from 2005 and from (1985)
Figure 4-15. NOX emissions for the Phase II scenarios and the percent reduction from the 1985 base
case and the 2005 baseline.
4-59
-------
CO Emissions
Northeast Corridor
300-
.270
240-
210-
o 150
^120
I 90
o 200
150-
8 100
50-
(45)
(45)
- 1
(58) (SB) (58) (57) (58) (57) (57)
23 23 23 23 23 23 23
(55!
1 8
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23
Emissions Scenario
R^^Si Point B8S8gg Area V////A Mobile
Values above bars: Percent reduction from 2005 and from (1985)
Figure 4-16. CO emissions for the Phase II scenarios and the percent reduction from the 1985 base
case and the 2005 baseline.
4-60
-------
Northeast Corridor
Urban Non-Attoinment Areas
Figure 4-17. The Northeast Corridor and nonattainment areas outside the Corridor.
4-61
-------
Major Nox Point Sources Affected by Maximum Technology Controls
(Plant Emissions Greater than or equal to 500 TPY)
Controlled
Uncontrolled
Figure 4-18. Location of major NOX point sources controlled in the maximum technology strategy.
4-62
-------
2.5
O
o
CD
oo
2.O -
1.5
o
- 1.0 H
CD
£
O 0.5 -
O.O
2.5
0
2.O -
CD
1.5 H
CO
~o
o
CD
O 0.5 -
O.O
Solar intensity
1OO yuE/m2-h
2OO /jE/m2-h
— .— 4-OO /jE/m —h
8OO /uE/m — h
O 5 1O 15 2O 25 3O 35
Leaf temperature (°C)
4-O
Alpha — pinene
— — — Monoterpene/unidentif ied
O
1 O
1 5 2O
25 3O
Leaf temperature (°C)
4-0
Figure 4-19. Relationship between meteorological temperature and (1) isoprene and alpha-pinene
emissions, (2) monoterpenes/unidentified hydrocarbon emissions (Tingey etal., 1980).
4-63
-------
;OUNTY-TO-GRID
ALLOCATION
{
COMPUTE "SUMMER"
BIOMASS
COMPUTE MONTHLY
ADJUSTMENT FACTORS
FOR BIOMASS
ADJUST BIOMASS
FOR MONTH
GRIDDED \ /CRIDDED
ONCANOPY1 / CANOPY
BIOMASS / \ BIOMASS
COMPUTE STANDARDIZED
EMISSION FLUXES
COMPUTE TEMPERATURE AND SUNLIGHT
ADJUSTMENT FACTORS
MERGE EMISSION RATES
GRIDDED
HOURLY
METEOROLOGY
DATA
JL
SPECIATE ACCORDING
TO MODEL CHEMISTRY
'GRIDDED
HOURLY
SPECIES
EMISSION
RATES
Figure 4-20. Flowchart of the Biogenic Emissions Inventory System.
4-64
-------
Biogenic Isoprene Emissions - 1000 EST Cool Day
Emissions (moles/hr):
< 2OOO
>- 80OO
>- 2OOO
>- 1OOOO
>- 6OOO
Biogenic Isoprene Emissions - 15OO EST Cool Day
Emissions (moles/hr):
< 2OOO r"~•:-; >— 2000 maamt > — 6OOO
>- 800O •• >- 10OOO
Figure 4-21. Biogenic isoprene emissions (mol C/h) for 1000 EST and 1500 EST on a "cool" day.
4-65
-------
Biogenic Isoprene Emissions - 100O EST Warm Day
Emissions (moles/hr): • • < 2OOO
Haass > -= 8OOO
>— 2OOO smiraBi! >— 6OOO
>- 1OOOO
Biogenic Isoprene Emissions - 15OO EST Warm Day
Emissions (moles/hr): i i < 2OOO
>- 800O
>— 2OOO ttaaam >— 6OOO
>- 1OOOO
Figure 4-22. Biogenic isoprene emissions (mol C/h) for 1000 EST and 1500 EST on a "warm" day.
4-66
-------
Variations in Biogenic VOC Emissions
For the ROMNET Domain
40000-
36000-
32000-
j^28000-
£ 24000-
o
^20000-
co
c
•5; 16000-1
OT
'E 12000-
UJ
8000-
4000-
0
03 04 05 06 07 08 09 10 11
July 1988 Episode
12 13 14 15 16
Figure 4-23. Regionwide daily total biogenic VOC emissions (tons/day), July 1988 episode.
4-67
-------
TABLE 4-1. ROMNET EMISSIONS SCENARIOS
PHASE I SCENARIOS (MOBILE3.9)
1985 Base Case
2005 Baseline:
No Growth NOX Scenario
CS01: Phase IVOC Maximum Technology Controls
CS02 and CS03: Spatial Analyses of VOC Controls-Transport Sensitivity Scenarios
CS04: CS01 Rule Effectiveness Sensitivity Scenario (not simulated-replaced by CS23)
1985 Base Case
2005 Baseline:
CS05:
CS06 - CS09,
1985L&1985H:
CS10:
CS11:
CS12:
CS13:
CS14:
CS15:
CS16:
CS17:
CS18:
CS19:
CS20:
CS21 and CS22:
CS23:
CS24 and CS25:
PHASE II SCENARIOS (MOBILE4)
NOX Growth Scenario
October 1989 Proposed Clean Air Act Legislation
Biogenic Sensitivity Scenarios for Phase 11985 Base Case and CS01, and Phase II
CS05 Strategies
Maximum Technology NOX Controls and Enhanced Maximum Technology VOC
Controls
Maximum Technology NOX Controls; VOC at 2005 Baseline
Enhanced Maximum Technology VOC Controls; NOX at 2005 Baseline
Point Source Maximum Technology NOX Controls; Enhanced Maximum Technol-
ogy VOC Controls
Mobile Source Maximum Technology NOX Controls; Enhanced Maximum Tech-
nology VOC Controls
Maximum Technology VOC and NOX Controls With Reactivity-Based Measures
Maximum Technology Controls With Additional Across-The-Board VOC Reduction
in New York City and Baltimore/Washington, DC and Alteration of NOX Controls
in New York City
Biogenic Sensitivity Scenario for Maximum Technology NOX Controls
CS16 With Reactivity Based Measures
CS18 With Alteration of NOX Controls in Baltimore/Washington, DC
2005 Baseline With Reactivity-Based Measures
Biogenic Sensitivity Scenarios for CS19
CS19 Rule Effectiveness Sensitivity Scenario
Spatial Analysis of VOC and NOX Controls-Transport Sensitivity Scenarios
4-68
-------
TABLE 4-2. REVISED SUMMER ALLOCATION FACTORS FOR AREA-SOURCE CATEGORIES
Connecticut
Delaware
DC
Indiana
Kentucky
Maine
Maryland
Massachusetts
Michigan
New Hampshire
New Jersey
New York
North Carolina
Ohio
Pennsylvania
Rhode Island
Tennessee
Vermont
Virginia
West Virginia
Gasoline
vessels
(52)
0.425
0.425
0.375
0.425
0.425
0.500
0.425
0.425
0.500
0.500
0.425
0.425
0.325
0.425
0.425
0.500
0.325
0.500
0.325
0.425
Gasoline
marketing
(54)
0.385
0.368
0.363
0.358
0.310
0.398
0.365
0.370
0.430
0.413
0.378
0.380
0.353
0.360
0.360
0.385
0.345
0.413
0.345
0.315
Architectural
coating
(82)
0.500
0.425
0.425
0.425
0.425
0.600
0.425
0.500
0.500
0.600
0.425
0.500
0.425
0.425
0.425
0.500
0.425
0.600
0.425
0.425
Miscel-
laneous
solvent
use
(95)
0.251
0.266
0.250
0.264
0.308
0.254
0.265
0.252
0.288
0.257
0.253
0.262
0.268
0.283
0.262
0.251
0.268
0.308
0.267
0.262
Bulk gas
terminals
and bulk
plants
(103)
0.398
0.380
0.380
0.365
0.315
0.405
0.378
0.380
0.445
0.423
0.393
0.390
0.368
0.370
0.370
0.395
0.360
0.423
0.358
0.323
Hazardous
waste
TSDF
(109)
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
0.300
Note: The numbers in parentheses under the column headings are area source classification codes.
a The original allocation factor in the NAPAP inventory was 0.25 for all of these categories, corresponding to no increase in
emissions during summer.
4-69
-------
TABLE 4-3. SUMMARY OF CONTROL MEASURES IN THE BASELINE PROJECTION AND MAXI-
MUM TECHNOLOGY INVENTORIES
Inventory
VOC control measures
NOX control measures
Phase I:
2005 Baseline
Highway vehicles:
- current FMVCP
- RVP = 9 psi
- existing I/M programs
Point and area sources:
- State controls in place as of 1985
- NSPS controls
Highway vehicles:
- current FMVCP
Area sources: no controls
Point sources:
- NSPS controls
- additional controls as necessary to offset over-
all NOX emission growth at the State level
Phase I: Highway vehicles:
Maximum - current FMVCP
Technology VOC - RVP = 9 psi
- enhanced I/M region-wide
Point and area sources:
- maximum technology for industrial sources
- low solvent coatings for architectural uses
- 20 percent reduction of consumer solvent
use
- on-board or Stage II controls for gasoline
marketing emissions
Same as Phase I baseline
Phase II:
2005 Baseline
Highway vehicles and area sources:
- same as Phase I baseline but with
evaporative test procedure
Point sources:
- 5% reduction of utility emissions (after
growth) to reflect enhanced energy efficiency
Highway vehicles and area sources:
- same as Phase I baseline
Point sources:
- 5% reduction of utility emissions (after growth)
to reflect enhanced energy efficiency
Phase II:
Maximum
Technology
NOX and Enhanced
Maximum
Technology VOC
Highway vehicles:
- 0.125 g/mi tailpipe standard
- enhanced I/M with anti-tampering
- RVP = 7 psi
- full implementation for all measures
Highway vehicles:
- 0.2 g/mi tailpipe standard
- 4 g/bk-hp-hr standard for trucks
- full implementation of all measures
Point and area sources:
- same controls as Phase I, and:
- 5% energy-efficiency reduction for utilities
- maximum technology for off-highway
vehicles
- prohibition of residential wood burning
and all open burning during ozone season
Point sources:
- 5% control for energy-efficiency
enhancement for utilities
- selective catalytic reduction (SCR) for utilities
- low-NOx burners or equivalent for industrial
and institutional combustion units
Area sources:
- maximum technology for
off-highway vehicles
- prohibition of residential wood burning and all
open burning during ozone season
4-70
-------
TABLE 4-4. CONTROL MEASURES FOR THE DRAFT CLEAN AIR ACT ANALYSIS
Source category VOC controls
NOX controls
Highway vehicles Tailpipe standards (phased in 1994-96):
Passenger cars: 0.25 g/mi
Light-duty trucks 1: 0.25 g/mi
Light-duty trucks 2: 0.5 g/mi
RVP = 9 psi
Alternative fuels for 4 urban areas
Enhanced I/M for serious and severe
nonattainment areas
Tailpipe standards (phased in 1994-96):
Passenger cars: 0.4 g/mi
Light-duty trucks 1: 1.00 g/mi
Light-duty trucks 2: 1.00 g/mi
Point sources Various controls implemented depending
on the severity of the ozone problem in
each MSA
Low-NOx burners or equivalent for utility
boilers larger than 75 MWe
Area sources Various controls implemented depending
on the severity of the ozone problem in
each MSA
None
General
Across-the-board reductions at the MSA
level, to achieve an overall reduction of
3 percent per year from 1990 to the proj-
ected attainment date
None
4-71
-------
TABLE 4-5. NSPS EFFICIENCIES FOR POINT AND AREA SOURCES
Pod
Category name
NSPS efficiency
Technology
Point Sources
2 Printing and publishing 85
3 Dry cleaning 70
4 Fixed roof tanks - crude oil 98
5 Fixed roof tanks - gasoline 96
6 Ext. float, roof tanks - crude 90
7 Ext. float, roof tanks - gas 95
8 Bulk gas terminals - splash 91
9 Bulk gas terminals - subm. fill 87
10 Bulk gas terminals - not balanced 79
15 Ethylene oxide manufacturing 98
16 Phenol manufacturing 98
17 Terephthalic acid manufacturing 98
18 Acrylonitrile manufacturing 98
19 SOCMI fugitives 56
20 Petroleum refinery fugitives 93
21 Cellulose acetate manufacturing 72
30 Rubber tire manufacturing 83
31 Green tire spray 90
33 Automobile surface coating 88
34 Beverage can surface coating 57
36 Paper surface coating 90
Area Sources
79 Dry cleaning 70
102 SOCMI fugitives 56
103 Bulk gasoline terminals/plants 91
104 Petroleum refinery fugitives 93
107 Synthetic fiber mfg. 83
Carbon adsorber
Recovery dryers
Internal floating roof
Internal floating roof
Secondary seal
Secondary seal
Subm./bal./ads./truck tests
Balanced/ads./truck tests
Carbon adsorber/truck tests
Incinerator
Incinerator
Incinerator
Incinerator
Equipment/maintenance
Equipment/maintenance
Carbon adsorber
Carbon adsorber
Solvent change
Incinerator
Incinerator
Incinerator
Recovery dryer
Equipment/maintenance
Balance/adsorber/truck tests
Equipment/maintenance
Carbon adsorber
4-72
-------
TABLE 4-6. SUMMARY OF STATE ESTIMATES FOR EXISTING AREA-SOURCE CONTROLS
Control technology
efficiency
No. states
respon-
SCC
54
61
62
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
100
101
102
103
104
105
106
107
108
109
Category name
Gasoline marketing
Prescribed burning
Agricultural burning
Degreasing
Dry cleaning
Graphic arts/printing
Rubber/plastics mfg.
Architectural coating
Auto repair coating
Motor vehicle mfg.
Paper coating
Fabricated metal coat.
Machinery mfg.
Furniture mfg.
Flat wood products
Other transp. equip.
Electrical equip, mfg.
Ship bldg./repair
Misc. ind. mfg.
Misc. ind. solvents
Misc. nonind. solvents
POTWs
Cutback asphalt
SOCMI fugitives
Bulk terminals/plants
Refinery fugitives
Bakeries
Pharmaceuticals mfg.
Synthetic fibers
Oil/gas production
TSDFs
ding
11
4
3
10
9
10
5
1
1
10
10
8
8
7
4
7
8
3
4
3
0
0
10
4
12
6
1
9
1
0
1
Overall
effi-
ciency
(%)
69.0
85.6
71.4
31.1
28.2
40.9
56.5
34.2
22.5
64.7
58.5
47.9
48.0
39.4
37.3
49.5
41.5
43.7
38.4
29.6
0.0
0.0
85.2
31.9
66.5
51.0
13.6
62.9
54.4
0.0
11.9
Mean
<%)
91
100
93
64
69
79
87
90
90
86
81
80
79
72
66
78
77
69
79
78
99
56
85
68
85
87
85
95
Range
84-95
na
80-100
50-90
57-85
65-90
80-90
na
na
70-92
50-95
70-90
70-90
25-90
50-85
70-90
60-90
47-90
60-90
50-100
95-100
40-62
n-90
60-70
na
70-90
na
na
Avg.
confi-
dence
A-
A
A-
C+
B-
B
B+
B
B
B+
B
B-
B-
B-
B
B-
B-
B+
B-
B-
A
C+
B+
B-
A
B
A
A.
Rule effectiveness
Mean
(%)
83
93
85
70
67
77
89
40
50
85
82
81
81
71
74
82
77
75
75
67
94
87
87
83
80
84
80
25
Range
30-98
75-100
75-100
50-98
40-99
50-95
75-100
na
na
40-96
69-95
50-95
50-95
50-95
50-90
50-95
50-95
na
50-90
50-80
80-100
75-95
60-100
65-95
na
60-99
na
na
Avg.
confi-
dence
B
B+
B
C
C+
C+
B-
B
C
B
B
C
C
C
D+
B
C+
C+
C-
D
B+
C
B-
B-
E
B-
E
D
Penetration
in area-source
inventory
Mean
(%)
91
93
90
70
61
68
73
95
50
89
88
74
76
76
76
77
70
85
65
57
92
66
89
92
20
86
80
50
Range
75-100
75-100
75-100
40-95
30- 90
25-90
10-100
na
na
58-100
70-100
60-95
60-95
50-95
60-90
62-100
50-90
75-95
50-90
50-70
65-100
20-100
67-100
75-100
na
75-100
na
na
Avg.
confi-
dence
B+
B+
B
C
C
C
C+
B
C
B
B
C-
C-
C-
C
C+
C-
B-
C
C-
B+
D+
B-
•C+
D
B-
E
C
4-73
-------
TABLE 4-7. SUMMARY OF NO* CONTROL MEASURES FOR PHASE I
State and pod
Connecticut
52 Util. ext. comb. - gas
Kentucky
50 Util. ext. comb. - coal
84 Cogeneration - other
87 Ind. ext. comb. - oil > 100 TRY
Maryland
50 Util. ext. comb. - coal
51 Util. ext. comb. - oil
52 Util. ext. comb. - gas
74 Util. oil turbine
75 Util. oil recip.
76 Util. gas turbine
77 Util. gas recip.
84 Cogeneration - other
87 Ind. ext. comb. - oil > 100 TRY
90 Ind. ext. comb. - gas > 100 TRY
Massachusetts
50 Util. ext. comb. - coal
51 Util. ext. comb. - oil
52 Util. ext. comb. - gas
76 Util. gas turbine
Michigan
50 Util. ext. comb. - coal
New Hampshire
50 Util. ext. comb. - coal
51 Util. ext. comb. - oil
New Jersey
50 Util. ext. comb. - coal
51 Util. ext. comb. - oil
74 Util. oil turbine
75 Util. oil recip.
76 Util. gas turbine
New York
50 Util. ext. comb. - coal
Ohio
50 Util. ext. comb. - coal
52 Util. ext. comb. - gas
76 Util. gas turbine
Pennsylvania
50 Util. ext. comb. - coal
52 Util. ext. comb. - gas
76 Util. gas turbine
Virginia
50 Util. ext. comb. - coal
51 Util. ext. comb. - oil
52 Util. ext. comb. - gas
74 Util. oil turbine
76 Util. gas turbine
84 Cogeneration - other
87 Ind. ext. comb. - oil > 1 00 TRY
Level of
control
needed
- LNB
SCR
LNB
LNB
SCR
SCR
SCR
SCR
SCR
SCR
SCR
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
LNB
. SCR
SCR
SCR
SCR
SCR
LNB
LNB
Approx. % of sources
needinq control
New
17
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
100
76
100
100
100
100
100
100
100
80
29
29
29
22
22
22
78
78
78
100
100
100
100
Existing
0
24
100
100
69
58
58
100
100
100
100
100
100
100
88
88
88
88
0
8
8
18
18
20
18
20
0
0
0
0
0
0
0
63
63
63
100
100
100
100
Overall
reduction for
category
(%)
8
39
48
42
52
44
76
70
30
71
30
50
42
42
45
21
48
32
7
7
2
17
27
45
13
31
23
2
13
17
2
9
13
59
29
62
22
70
50
42
Statewide
point source
reduction
(%)
4.5
30.7
38.5
17.4
5.7
2.4
13.0
12.1
1.9
1.3
36.4
SCR = selective catalytic reduction; LNB = low-NOx burners.
4-74
-------
TABLE 4-8. MAXIMUM TECHNOLOGY FOR POINT SOURCES
Pod
Description
Maximum technol.
efficiency (%)
Phase I Phase II
Control tech. basis
1 Solvent metal cleaning 54 54
2 Printing and publishing 85 85
3 Dry cleaning 70 70
4 Fixed roof tanks - crude oil 98 98
5 Fixed roof tanks - gasoline 96 96
6 Ext. float, roof tanks - crude 90 -90
7 Ext. float, roof tanks - gasoline 95 95
8 Bulk gas terminals - splash fill 91 91
9 Bulk gas terminals - subm. fill 87 87
10 Bulk gas terminals - not balanced 79 79
11 Service stations - Stage I 95 95
15 Ethylene oxide manufacturing 98 98
16 Phenol manufacturing 98 98
17 Terephthalic acid manufacturing 98 98
18 Acrylonitrile manufacturing 98 98
19 SOCMI fugitives 56 56
20 Petroleum refinery fugitives 93 93
21 Cellulose acetate manufacturing 72 72
22 Styrene-butadiene rubber mfg. 20 20
23 Propylene manufacturing 98 98
24 Polyethylene manufacturing 98 98
25 Ethylene manufacturing 98 98
26 Refinery wastewater treatment 95 95
27 Refinery vacuum distillation 100 100
28 Vegetable oil processing 42 42
29 Paint and varnish mfg. 92 92
30 Rubber tire manufacturing 83 83
31 Green tire spray 90 90
32 Carbon black manufacturing 90 90
33 Automobi le surface coating 88 88
34 Beverage can surface coating 57 57
35 Genl. wood surface coating 70 70
36 Paper surface coating 90 90
37 Miscellaneous surface coating 90 90
38 Food/agricultural starch mfg. 90 90
39 Coke oven byproducts plants 63 63
40 Ferrosilicon production 88 88
43 Marine vessel loading 90 90
46 Charcoal manufacturing 80 80
47 Fermentation/whiskey prodn. 85 85
48 Plastics parts coating - 80
49 Wood furniture coating - 60
95 Aircraft coating - 88
96 SOCMI reactors ~ 85
97 SOCMI distillation - 97
98 Furniture manufacturing - 41
99 Misc. noncombustion sources - 81
Carbon adsorber
Carbon adsorber
Recovery dryers
Internal floating roof
Internal floating roof
Secondary seal
Secondary seal
Subm./bal./ads./truck tests
Balanced/ads./truck tests
Carbon adsorber/truck tests
Vapor balance
Incinerator
Incinerator
Incinerator
Incinerator
Equipment/maintenance
Equipment/maintenance
Carbon adsorber
Incinerator
Flare
Flare
Flare
Covers
Firebox piping
Stripper
Afterburner
Carbon adsorber
Solvent change
Flare
Incinerator
Incinerator
Process change
Incinerator
Incinerator
None specifically
Equipment/maintenance
None specifically
Incinerator
Afterburner
Carbon adsorber
Low-solvent coating
Low-solvent coating
Incinerator
Equipment/maintenance
Equipment/maintenance
Low-solvent coating
Generic rule
4-75
-------
TABLE 4-9. MAXIMUM TECHNOLOGY FOR AREA SOURCES
Maximum technol.
efficiency (%)
sec
6
24
25
26
44
45
50
54
61
62
63
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
100
101
102
103
104
105
106
107
108
109
Description
Residential wood burning
Open burning
Open burning
Open burning
Off-highway diesels
Railroad locomotives
Diesel vessels
Gasoline marketing
Prescribed forest burning
Agricultural field burning
Orchard heaters
Degreasing
Dry cleaning
Graphic arts and printing
Rubber and plastics mfg.
Architectural coating
Auto body repair
Motor vehicle manufacturing
Paper coating
Fabricated metals coating
Machinery mfg. coating
Furniture mfg. coating
Flat wood product coating
Other transp. equip, coating
Electrical equip mfg. coating
Ship coating
Misc. ind. mfg. coating
Misc. ind. solvent use
Misc. nonindustrial solvents
POTWs
Cutback asphalt paving
SOCMI fugitives
Bulk gasoline terminals/plants
Petroleum refinery fugitives
Bakeries
Pharmaceuticals manufacture
Synthetic fibers mfg.
Oil/gas production fields
Hazardous waste TSDF
Phase I
-
-
--
-
-
-
85
100.
100
—
83
70
85
83
52
88
88
90
57
90
90
90
88
90
47
85
85
50
90
100
56
91
93
90
90
83
93
95
Phase II
100
100
100
100
80
65
55
93
100
100
100
83
70
85
83
52
88
88
90
57
90
90
90
88
90
47
85
85
50
90
100
56
91
93
90
90
83
93
93
Control tech. basis
Ban
Ban
Ban
Ban
Highway diesel technology
Highway diesel technology
Highway diesel technology
RVP reduction/vapor balance
Ban
Ban
Ban
Cover/carbon adsorber
Recovery dryer
Carbon adsorber
Carbon adsorber, etc.
Low solvent coatings
Low solvent coatings
Incinerator
Incinerator
Low solvent coatings
Incinerator
Incinerator
Incinerator
Low solvent coatings
Incinerator
Low solvent coatings
Low solvent coatings
Carbon adsorber, etc.
Product reformulation
Pretreatment
Ban
Equipment/maintenance
Balance/adsorber/truck tests
Equipment/maintenance
Incinerator
Condensers
Carbon adsorber
Equipment/maintenance
Pretreatment, etc.
4-76
-------
TABLE 4-10. MAXIMUM TECHNOLOGY EFFICIENCIES FOR NOX
Pod/SCC
50
51
52
53
55
58
59
60
70
71
72
73
74
75
76
77
84
85
86
87
88
89
90
6
24
25
26
44
45
50
Category description
Point sources
Utility ext. comb. - coal
Utility ext. comb. - oil
Utility ext. comb. - gas
Utility ext. comb. - other
Industrial process heaters
Commercial/institutional coal
Commercial/institutional oil
Commercial/institutional gas
Industrial oil turbines
Ind. oil recip. engines
Industrial gas turbines
Ind. gas recip. engines
Utility oil turbines
Util. oil recip. engines
Utility gas turbines
Util. gas recip. engines
Industrial ext. comb. - coal
Industrial ext. comb. - oil
Industrial ext. comb. - oil
Industrial ext. comb. - oil
Industrial ext. comb. - gas
Industrial ext. comb. - gas
Industrial ext.comb. - gas
Area sources
Residential wood burning
Open burning
Open burning
Open burning
Off hwy. diesel
Railroad locomotives
Diesel vessels
Technology8
SCR
SCR
SCR
Energy eff.
Low-N0x
Low-N0x
Low-NOx
Low-NOx
Low-NOx
Low-NOx
Low-NOx
Low-NOx
SCR
SCR
SCR
SCR
Low-NOx
Low-NOx
Low-NOx
Low-NOx
Low-NOx
Low-N0x
Low-N0x
ban
ban
ban
ban
hwy. dies.d
hwy. dies.d
hwy. dies.rf
Efficiency (%)
81.0Qb
81 .OQb
81.00b
5.00
45.00
36.00
42.00
42.00
70.00
30.00
70.00
30.00
94.30C
81.00b
94.3QC
81.00&
36.00
42.00
42.00
42.00
42.00
42.00
42.00
100.00
100.00
100.00
100.00
40.00
60.00
55.00
a Low-NOx technology includes low-NOx burners, staged combustion air, reduced air-to-fuel ratio, and steam injection. SCR
refers to selective catalytic reduction.
b 80 percent for SCR with a 5 percent energy efficiency reduction.
c 94 percent for SCR with a 5 percent energy-efficiency reduction.
d Transfer from highway diesel technology.
4-77
-------
TABLE 4-11. STRATEGY 5 POINT-SOURCE CONTROL EFFICIENCIES BY ATTAINMENT CATE-
GORY
VOC reductions by Attainment Category (%)
POD
1
2
3
4
5
6
7
8
9
10
11
15
16
17
18
19
20
21
22
23
24
25
26
27
28
30
31
32
33
34
36
37
39
39
43
46
47
48
49
95
96
97
99
Pod description
Solvent metal cleaning
Printing/publishing
Dry cleaning
Fixed roof tanks - crude
Fixed roof tanks - gas
EFR tanks - crude
EFR tanks - gasoline
Bulk term. - splash fill
Bulk term. - subm. bal.
Bulk term. - not bal.
Stage I
Ethylene oxide
Phenol
Terephthalic acid
Acrylonitrile
SOCMI fugitives
Refinery fugitives
Cellulose acetate
Styrene butadiene
Propylene
Polyethylene manufacturing
Ethylene
Refinery WW treatment
Refinery vacuum dist.
Vegetable oil
Rubber tire manufacturing
Green tire spray
Carbon black
Automobile coating
Can coating
Paper coating
Miscellaneous surface coating
Coke oven leaks
Coke byproduct plants
Marine vessel loading
Charcoal manufacturing
Whiskey production
Plastic parts coating
Wood furniture coating
Aircraft coating
SOCMI reactors
SOCMI distillation
Miscellaneous noncombustion
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
54
85
70
98
96
90
95
91
87
79
95
98
98
98
98
37
83.4
64.8
70
98
98
98
0
100
42
83
90
90
88
57
78
90
90
63
90
80
85
0
0
84.4
0
0
90
3
54
85
70
98
96
90
95
91
87
79
95
98
98
98
98
37
'83.4
64.8
70
98
98
98
50
100
42
83
90
90
88
57
78
90
90
63
90
80
85
80
24
84.4
85
85
90
4
54
85
70
98
96
90
95
91
87
79
95
98
98
98
98
37
83.4
64.8
70
98
98
98
50
100
42
83
90
90
88
57
78
90
90
63
90
80
85
80
24
84.4
85
85
90
5
54
85
70
98
96
90
95
91
87
79
95
98
98
98
98
37
83.4
64.8
70
98
98
98
50
100
42
83
90
90
88
57
78
90
90
63
90
80
85
80
24
84.4
85
85
90
4-78
-------
TABLE 4-12. STRATEGY 5 AREA-SOURCE CONTROL EFFICIENCIES BY ATTAINMENT CATE-
GORY
VOC reductions by Attainment Category (%)
sec
54
78
80
81
82
83
85
88
95
100
101
102
103
104
106
107
108
109
SCC description
Service stations (Stage I)
Degreasing
Printing
Rubber/plastics manufacturing
Architectural coating
Auto body repair
Paper coating
Furniture manufacture
Nonindustrial solvents
POTWs
Cutback asphalt
SOCMI fugitives
Bulk terminals/plants
Refinery fugitives
Pharmaceuticals
Synthetic fibers
Oil/gas production
TSDFs
1
0
0
0
0
65
0
0
0
27
0
0
0
0
0
0
0
0
93
2
38
35
0
65
65
0
80
0
27
0
100
44.6
51
43
44.6
54
37
93
3
71
35
24
65
65
60
80
16.4
27
75
100
44.6
51
43
44.6
54
37
93
4
71
35
24
65
65
60
80
16.4
27
75
100
44.6
51
43
44.6
54
37
93
5
71
35
24
65
65
60
80
16.4
27
75
100
44.6
51
43
44.6
54
37
93
4-79
-------
TABLE 4-1 3. STRATEGY 5 ATTAINMENT CATEGORIES AND ACROSS-THE-BOARD VOC
REDUCTIONS
Area
Albany-Schenectady-Troy, NY
Allentown-Bethlehem, PA-NJ
Altoona, PA
Atlantic City, NJ
Baltimore, MD
Boston, MA
Buffalo, NY
Canton, OH
Charleston, WV
Cincinnati, OH
Cleveland, OH
Columbus, OH
Dayton-Springfield, OH
Detroit, Ml
Erie, PA
Harrisburg-Lebanon-Carlisle, PA
Hartford, CT
Huntington-Ashland, WV-KY-OH
Johnstown, PA
Lancaster, PA
Lexington-Fayette, KY
Massachusetts (except Boston, Pittsfield, Springfield)
New York, NY-NJ-CT
Norfolk-Virginia Beach-Newport, VA
Parkersburg-Marietta, WV-OH
Philadelphia, PA-NJ
Pittsburgh, PA
Pittsfield, MA
Portland, ME
Portsmouth-Dover-Rochester, NH
Poughkeepsie, NY
Providence-Pawtucket-Woonsocket, Rl
Reading, PA
Richmond-Petersburg, VA
Scranton-Wilkes-Barre, PA
Sharon, PA
Springfield, MA
Toledo, OH
Washington, DC-MD-VA
York, PA
Youngstown-Warren, OH
Attainment
Category
2
3
2
3
4
4
2
3
3
4
3
2
3
3
2
3
4
4
2
2
2
4
5
2
3
4
3
4
4
4
3
4
3
3
2
2
4
3
4
2
2
Across-the-board
reduction a
%
9
5
3
2
21
31
14
2
0
0
4
12
4
18
6
0
27
12
8
12
17
34
41
13
0
12
1
36
23
38
9
22
0
10
4
8
31
0
18
8
3
a. Applied equally to point, area and mobile sources.
4-80
-------
TABLE 4-14. VOC SPECIATION FACTORS FOR METHANOL VEHICLES
Species class
Speciation factors (mol/kg)
Exhaust Evaporative
Olefin
Paraffin
Toluene
Xylene
Formaldehyde
Other aldehydes
Ethene (ethylene)
Isoprene
Nonreactive hydrocarbons
Methane
Methanol
0.177
1.775
0.058
0.003
0.817
0.056
1.307
0.003
0.515
4.725
25.523
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
0.00
31.20
TABLE 4-15. EMISSION FLUXES FOR VEGETATION TYPES IN THE BIOGENIC EMISSIONS
INVENTORY SYSTEM (BEIS) a
Percent Contribution
Vegetation type
NMHCflux Isoprene a-pinene Monoterpene Unidentified
Oak
Corn
Other deciduous
Coniferous
Peanuts/Rice
Tobacco
Grass/Pasture
Hay/Scrub/Range
Potato
Sorghum
Alfalfa
Barley/Oats/Rye/Misc.
Wheat
Soybeans
Water
4258
3542
3353
3106
510
294
281
189
48
39
38
38
30
22
0
73
0
63
24
20
0
20
20
0
20
50
20
50
100
0
3
10
5
21
25
10
25
25
25
25
10
25
10
0
0
3
10
6
23
25
10
25
25
25
25
10
25
10
0
0
21
80
26
32
30
80
30
30
50
30
30
30
30
0
0
a Total non-methane hydrocarbons (pg/m2/h)and percent contribution from individual chemical species, standardized for full
sunlight and 30°C.
4-81
-------
This page is intentionally left blank.
-------
SECTION 5
REGIONAL MODELING RESULTS
AND
PROJECT FINDINGS
by
Norman C. Possiel*
Keith A. Baugues
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
5.1 INTRODUCTION
During the course of ROMNET, the Management Review Committee and other participants identified
several key issues relative to the effectiveness of regional control strategies which were to be addressed
through ROM simulations. As stated in Section 4.4.2, these issues are:
1. What are the relative benefits of VOC controls versus NOX controls in reducing ozone
levels across the region?
2. What is the impact of reducing regional transport on Northeast Corridor ozone levels?
3. What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted
ozone concentrations in the Northeast to below 125 ppb?
4. How effective are potential reactivity-based strategies in reducing regional ozone levels?
5. How does the large uncertainty in biogenic emissions alter conclusions regarding the
effectiveness of control measures?
This section of the report presents the results of ROM simulations conducted to address these issues.
The ozone concentrations and spatial patterns for the Phase I11985 base case and 2005 baseline sce-
narios are described first.1 This is followed by a discussion of the findings relative to each of the above
issues. A summary of the implications drawn from these findings is given at the end of the section.
5.2 1985 BASE CASE AND 2005 BASELINE PREDICTIONS
A fifteen day episode during the summer of 1988, July 2-17, was used to simulate all emissions scenar-
ios. The rationale for selecting this episode, and a description of the meteorological conditions and
measured ozone levels during the episode, are provided in Section 3.
5.2.1 1985 Base Case Ozone Predictions - July 1988 Episode
The maximum 1-hour ozone concentrations predicted for this episode are shown in Figure 5-1. These
data are the highest concentrations in each ROM grid which are predicted for any hour during the
1. Ozone predictions for the Phase I scenarios are not described in this report since all of the key strategies were run with
MOBILE4 as part of Phase II. However, spatial plots of gridded maximum 1-hour concentrations are provided for all Phase I
and Phase II scenarios in Appendix M.
5-3
-------
15-day episode. Thus, this figure represents the overall impact during the episode, not a snapshot of
the concentrations on a single day or hour. The values in Figure 5-1 are taken from layer 1 of the ROM,
which is the lowest layer of the model.
Examining Figure 5-1 indicates that high ozone concentrations > 125ppb are predicted along the
entire length of the Northeast Corridor. Levels exceeding 200 ppb are predicted over the New York City
area with concentrations above 160 ppb near or downwind of Baltimore/Washington, DC, Philadelphia,
and Boston. Ozone concentrations in the range of 140 to 160 ppb are predicted over sections of
southern New England extending into coastal Maine.
Elsewhere in the region, concentrations exceeding 140 ppb are predicted near or downwind of Pitts-
burgh, Cleveland, Detroit, and Toronto. Surrounding these plumes is a rather broad area with ozone just
below 125 ppb in the range of 100-124 ppb. Moderate ozone levels of 80 to 100 ppb were predicted in
more rural, remote portions of the region. The elevated ozone levels extending across northern West
Virginia into central Virginia appear to be associated with transport flow regimes which pass over
Cleveland and other source areas in eastern Ohio, western Pennsylvania, and northwest West Virginia.
5.2.2 2005 Baseline Ozone - July 1988 Episode
The episode maximum ozone levels for the 2005 baseline scenario are shown in Figure 5-2. This sce-
nario reflects the net effects of growth and existing control programs. In general, emissions of VOC and
NOX in the Corridor are less than the 1985 base case by 34 percent and 6 percent, respectively. Outside
the Corridor, VOC is reduced by 33 percent and NO* is increased by 1 percent. In Ontario, Canada,
VOC is 42 percent lower and NOX is 10 percent lower than the 1985 base. A detailed description of the
Phase II 2005 baseline emissions is given in Section 4.4. The results of these emission changes on
predicted ozone levels can be seen by comparing Figures 5-1 and 5-2. In the Corridor, maximum ozone
is reduced by 10 to 15 percent across much of New Jersey and from central Connecticut northeastward
into Maine. Over New York City maximum ozone levels remain > 200 ppb. In the vicinity of Baltimore/-
Washington, DC, ozone levels were reduced by < 10 percent.
In the western portion of the domain, reductions in maximum ozone of 10 to 15 percent are predicted for
areas around Pittsburgh, Cleveland, and Detroit with a 5 to 10 percent reduction near Charleston. Also,
the spatial extent of ozone > 125 ppb has been greatly reduced near Detroit and Cleveland and peak
values are now below 140 ppb.
5-4
-------
5.3 RESULTS OF CONTROL STRATEGY SIMULATIONS FOR KEY ISSUES
The relative effectiveness of control strategies is examined by comparing the ROM layer 1 ozone pre-
dictions using various measures or metrics. Included are:
(1) episode maximum 1-hour concentrations, i.e., "acute" metric (spatial plots of gridded
1-hour episode maximum concentrations for each ROMNET strategy are included in
Appendix M);
(2) episode mean 8-hour daily maximum ozone concentrations, i.e., "chronic0 metric;
(3) daily maximum 1 -hour concentrations;
(4) diurnal time series of urban area peak hourly concentrations; and
(5) population exposure to hourly ozone concentrations exceeding 100 ppb and 125 ppb.
These metrics were chosen to highlight different facets of the response of predicted ozone levels to
control strategies. Not all metrics were examined for each issue. Metrics 1 through 4 are fairly straight-
forward. For metric 5, population exposure was calculated using the equation shown in Table 5-1. As
an example, the population exposure to ozone exceeding 125 ppb in New York City is calculated as the
product of gridded population and the number of hours in each grid with ozone > 125 ppb, summed for
all 88 grids covering the New York City area. Areas in the domain referred to in the discussion of several
of the issues are shown in Figure 5-3.
5-5
-------
ISSUE #1
What are the relative benefits of VOC control versus NOX control in reducing ozone levels across
the region?
Analysis Approach
The importance of this issue stems from the highly nonlinear role of NOX and VOC in ozone formation.
Basically, NOX can act as a source or sink of ozone, depending on the amount of reactive organics
present, meteorological conditions, and proximity to sources of NOX. In areas where the ratio of VOC to
NOX is relatively high, ozone formation is determined by the availability of NOX (i.e., ozone formation is
said to be "NOX limited"). In such areas, controlling NOX may be an effective approach to reduce ozone
levels. In areas where the VOC to NOX ratio is low, VOC control may be most effective, and controlling
NOX may actually increase ozone concentrations in some areas. The following five strategies were
designed to determine where and to what extent NOX and/or VOC controls provide the most effective
pathway to reduce ozone concentrations in the Northeast.
CS10: Combined VOC and NOX maximum technology controls
[NOX + VOC control]
CS11: Maximum technology NOX controls with VOC at the 2005 baseline
[NOjf-only controls]
CS12: Maximum technology VOC controls with NOX at the 2005 baseline
[VOC only control]
CS13: Maximum technology VOC and NOX point-source controls with NOX mobile sources at the
2005 baseline
[VOC + NOX point-source control]
CS14: Maximum technology VOC and NOX mobile-source controls with NOX point sources at the
2005 baseline
[VOC + NOX mobile-source controls]
Specifications of the control measures in these strategies are given in Section 4.4.2. The total anthro-
pogenic VOC and NOX emissions for each of these strategies and the reduction from the 2005 baseline
are shown in Figure 5-4 for the U.S. portion of the region. Emissions for 1985 are included in this figure
for reference as are biogenic emissions. Emissions reductions for selected urban areas are given in
Appendix I.
5-6
-------
It is important to note the effect of biogenic VOC emissions on the amount of reduction in total VOC
(anthropogenic plus biogenic) achieved by the control measures in CS12. This is illustrated in
Figure 5-5, which shows the effects of the CS12 controls on anthropogenic emissions across the region.
In most areas of the domain anthropogenic emissions are reduced by 50 to 70 percent. However, when
biogenic emissions are also considered, the net reduction in VOC is on the order of 20 percent or less
across much of the region outside the core grids of urban areas. This reflects the high level of biogenic
emissions in rural areas, which often exceeds emissions from anthropogenic sources. In contrast, the
reductions in NOX emissions achieved by CS11 (point-, mobile-, and area-source NOX controls), as
shown in Figure 5-6, are 30 to 50 percent across the region with greater reductions of 60 to 80 percent
in grids containing large point sources.
Discussion of Results
The impacts of the VOC and NOX strategies on the episode maximum 1 -hour ozone metric are pres-
ented first, followed by a comparison of acute versus chronic ozone metrics. Differential, day-to-day
effects are presented next and the impact of controls on population exposure is presented last.
Episode Maximum Concentrations
Impact on NOX Controls Alone
The application of NOX controls (CS11) to the 2005 baseline results in a dramatic reduction in ozone
levels across much of the region, as evident in examining Figure 5-7. In fact, except for Toronto, ozone
levels outside the Northeast Corridor have been reduced to below 125ppb. (Recall that Canadian
emissions were kept at the 2005 baseline with maximum technology controls applied only in the US
portion of the region.) In most of the area outside the Corridor ozone is less than 100 ppb with only 7
ROM grids containing peak values as high as 110 to 115 ppb. Reductions in episode maximum l-hour
ozone across individual urban areas are as follows: Pittsburgh, 25 to 30 percent; Cleveland, 15 to
25 percent; Detroit, 15 to 20 percent; and Charleston, 30 to 40 percent.
Along the Northeast Corridor, episode peak values have also declined substantially as a result of NOX
controls, except within the New York City urban area. In New York, ozone levels following NOX controls
are predicted to increase on the order of 15 percent relative to the 2005 baseline scenario. Elsewhere in
the Corridor, ozone levels declined to below the NAAQS, except near Baltimore. From the center of
Baltimore southeastward along the Chesapeake Bay, ozone levels remain at 125 ppb. Also, the highest
predictions near Philadelphia were reduced to the level of the NAAQS.
5-7
-------
Impact on VOC Controls Alone
The ROM predictions for episode maximum ozone following the VOC controls in CS12 are shown in
Figure 5-8, along with the predictions for the 2005 baseline. A comparison of these two scenarios indi-
cates some reduction outside the Northeast Corridor. However, ozone levels are 15 to 25 percent
higher in this area than with the NOX control strategy (CS11). Ozone levels > 125 ppb are predicted
near Pittsburgh and Cleveland and levels between 120 and 124 ppb near Detroit, Charleston, and Buf-
falo. Also, a fairly broad area with peak values of 100 to 120 ppb remain from Detroit southeastward to
northern Virginia and in western New York State. Recall from Figure 5-5 that the overall net VOC
emissions reduction in this part of the domain was only 20 percent due to the dominance of natural VOC
emissions.
Along the Northeast Corridor the episode peak ozone values predicted for the 2005 baseline scenario
are reduced by 10 to 15 percent following the VOC controls in CS12, with reductions in ozone greater
than 50 percent in the New York City area. Compared to NOX controls in CS11, the VOC controls pro-
duce substantially lower peak values in the New York City area and also somewhat lower ozone levels
near Baltimore. However, although the peak values are lower with VOC control, the spatial coverage of
concentrations > 125 ppb in most sections of the Corridor outside of New York City is much less with
NOX controls than with VOC controls.
Impact of Combined VOC and NOX Controls
The impact of combining the VOC maximum technology controls with the NOX controls was examined in
three strategies. In CS13, the VOC controls are combined with NOX controls on point sources. Mobile-
source NOX emissions are held at the 2005 baseline levels. The complementary strategy, CS14, con-
tains mobile-sources NOX controls and VOC controls. Point-source NOX emissions are at the 2005
baseline. Finally, in CS10, the maximum technology NOX controls on point, area, and mobile sources
are combined with VOC controls. The point-only controls in CS13 reduced total NOX emissions from the
2005 baseline by 40 percent in the Corridor and 53 percent elsewhere in the region. In contrast, the
reduction in total NOX due to mobile-source controls was less, showing an 18 percent reduction in the
Corridor and 10 percent elsewhere.
The episode maximum ozone for these three strategies (CS10,13, and 14) are shown in Figure 5-9. The
most noticeable differences between the predictions for these strategies are across West Virginia, east-
ern Ohio, and western Pennsylvania, where large numbers of point sources are located (see Figure
4-12). Near Charleston, for example, point-source controls reduced peak ozone to below 100 ppb,
whereas with mobile-source controls, predicted episode peak ozone was 122 ppb. In southwestern
5-8
-------
Pennsylvania, episode peak ozone levels in the range of 120 to 125 ppb were predicted in 11 grids
following mobile-source controls, but the peak value in these grids with point-source controls was only
116 ppb. Near Cleveland and Detroit the point-source and mobile-source controls yield nearly equiva-
lent impacts on episode maximum ozone levels even though emissions are lower with point-source
controls. This suggests that predicted ozone is more sensitive to mobile-source NOX controls on a
per-ton-removed basis. This is not unexpected given that mobile-sources emit both VOC and NOX,
which are spatially concentrated, and are greatest in urban areas, where emissions from other sources
are also high.
Along the Corridor, point-source controls appear to be more effective than the mobile-source controls in
the area from Washington, DC northeastward to just outside of New York City. This is evident by com-
paring the spatial extent of ozone > 100 ppb and > 125 ppb between CS13 and CS14 in Figure 5-9. In
New York City, where NOX controls produce a "disbenefit," peak ozone is less with mobile-source
controls. This most likely reflects the fact that the mobile-source controls did not reduce NOX emissions
to the extent of the point-source controls. In the general downwind direction from New York into south-
ern and central New England, ozone appears to be more sensitive to mobile-source NOX controls than
to point-source controls. For example, in Boston the reduction in total NOX was 34 percent due to
point-source controls but only 15 percent from mobile-source controls. However, the highest and sec-
ond highest daily maximum values were, respectively, 7 and 13 ppb lower with mobile-source than with
point-source controls.
In New York City, both CS13 and CS14 produce lower ozone than the 2005 baseline. However, ozone
levels following NOX point-source controls in CS13 are higher than with the NOX mobile-source controls
in CS14 or the VOC only controls in CS12. That is, adding the mobile-source NOX controls with VOC
controls does not significantly deteriorate ozone levels in New York relative to VOC controls alone. For
example, the number of grids with ozone > 140 ppb differ by one between CS12 (VOC control) and
CS14 (VOC + mobile-source NOX controls). Also, the mean episode maximum concentration for grids
> 140 ppb is 144 ppb for CS12 and 147 ppb for CS14. Another important point is that the spatial extent
of ozone > 125 ppb downwind of New York City over Connecticut is reduced by adding the mobile-
source NOX controls.
When the NOX mobile-source and point-source controls are combined and added to the VOC controls
(CS10), peak ozone responds in several different ways across the region. The differences are evident
by comparing the ozone predictions for CS10 with those for CS11, CS13, and CS14 (see Figures 5-7
and 5-9). In general, ozone levels for CS10 are lower than either CS13 or CS14 in all areas except New
York City and Baltimore/Washington, DC.
5-9
-------
Outside the Corridor, it appears that a major portion of the large reductions achieved by the combined
VOC and NOX controls in CS10 are attributable to the controls on NOX point sources. The results also
indicate that adding maximum technology VOC controls may provide little, if any, further reductions in
peak ozone beyond that achieved by NOX controls.
Inside the Corridor, ozone levels for the combined VOC and NOX strategy (CS10) are below the NAAQS
in all areas except near and downwind of New York City and Baltimore/Washington, DC. In Baltimore/-
Washington, DC, the spatial extent of ozone > 125 ppb has been reduced compared to either CS13 or
CS14. However, in New York City, there is a fairly large increase in ozone by combining point-source
and mobile-source controls even in the presence of maximum technology VOC controls. For example,
the number of grids with peak values > 140 ppb doubles and the maximum concentration increases to
181 ppb from 165 ppb in CS13 and 158 ppb in CS14.
Comparison of Impacts for Acute and Chronic Metrics
For New York City, Greater Connecticut, and Baltimore/Washington, DC, where disbenefits of NOX
controls were indicated in the episode maximum values, the analysis was extended to examine a more
long-term, chronic metric of ozone concentration. Here, the acute measure is derived from the episode
maximum 1-hour concentrations for all grids within each of the above areas. The chronic measure is
based on the episode mean of the daily maximum 8-hour average ozone concentrations in each grid.
This metric is the highest of the running 8-hour averages on each day, averaged for the episode by grid.
The acute and chronic metrics are presented in quantife-quantile (Q-Q) plots for each of the three areas
for CS10 (VOC plus NOX control) versus CS12 (VOC controls only). The data points in these plots rep-
resent the rank ordered ozone values with VOC controls prior to NOX control (CS12) and the corre-
sponding rank ordered ozone values after adding NOX control to VOC control (CS10). For
Baltimore/Washington, DC, the Q-Q plots of acute ozone (Figure 5-10, top) indicate a sharp curvature
throughout the distribution of gridded values. Since all but one of the data values falls below the diag-
onal line, the overall impact of NOX and VOC controls is projected to be more beneficial than VOC con-
trols alone. However, the benefit varies substantially across the distribution of grid cells with largest
benefits near the 50th percentile and smallest benefits projected near the minimum and maximum of the
distribution. The corresponding plot for the chronic ozone measure (Figure 5-10, bottom) shows much
less curvature and indicates a more uniform improvement across the grid cells than was the case for
1-hour peaks.
5-10
-------
In New York City and Connecticut the Q-Q plots are contained in Figure 5-11 and Figure 5-12, respec-
tively. These figures indicate that the impact of controls varies considerably depending upon which
percentile is being examined. In particular, for New York City, there appears to be a clear benefit of NOX
control for the lower percentiles of the acute ozone metric and a clear disbenefit (increases in ozone)
above the 50th percentile. Interestingly, the crossover between benefit and disbenefit is at a concen-
tration level just below the NAAQS. That is, the disbenefit increases with concentration for ozone levels
above the NAAQS. It is important to note that for the chronic measure, adding NOX controls in New York
City is actually beneficial relative to VOC controls alone. This is seen along the entire distribution
although the benefit diminishes with increasing concentration to where the highest mean 8-hour values
are nearly equivalent for the VOC only and VOC plus NOX controls. For Connecticut, the disbenefit
begins closer to the upper end of the distribution for the acute metric. As in New York City, the addition
of NOX controls appears to be beneficial for the chronic metric.
Day-To-Day Variations in the Impact of NOX and VOC Controls
A more complex picture of the role of NOX emerges by examining the impact of controls on individual
days during the episode simulated. For this part of the analysis, hourly ozone concentrations were
extracted for each scenario from ROM grids corresponding to the areas shown in Figure 5-3. For each
individual area, and each scenario, diurnal profiles of hourly ozone concentrations for all days in the July
1988 episode were created using the highest hourly ozone concentrations from among the grids
contained within the area for the particular scenario. The diurnal profiles for CS11 and CS12 for selected
areas are shown in Figure 5-13 and Figure 5-14. The shaded portions of the plots reveal when and to
what extent ozone predictions differed between the two scenarios. The diurnal profiles in Figure 5-13 for
the New York City area indicate that the relative disbenefits of NOX controls, as evident from the episode
maximum concentrations, are replicated each day during the episode. The values following NOX
controls are approximately twice as high as levels predicted with VOC controls (CS12). In fact, on most
days the predicted daily maximum ozone with NOX controls (CS11) is close to or above that predicted for
the 2005 baseline scenario.
For Greater Connecticut, the highest ozone levels with VOC controls tend to occur during mid-afternoon,
several hours earlier than the much larger peak concentration with the NOX control scenario (see Figure
5-13). Also, this mid-afternoon peak predicted in the VOC only strategy is largely reduced by NOX con-
trols. The early peak, which is reduced by NOX controls, appears to be due to ozone formation from local
5-11
-------
emissions coupled with transported ozone and precursors from the day before. The enhanced ozone
levels that occur late in the day with NOX control are apparently due to transport of the New York City
urban plume across portions of Connecticut.
For the Baltimore/Washington, DC area, the profiles in Figure 5-14 indicate that the benefits of NOX
controls resulted in ozone levels lower than with VOC controls on six days (CS11 < CS12 in the figure),
but the opposite occurred for eight days on which VOC controls were more beneficial. Thus, it appears
that the relative benefits of VOC and NOX controls can vary from day to day. This reflects the variation in
meteorological conditions that determines such factors as local source-receptor relationships, long-
range transport, biogenic emissions, and ozone production rates.
The diurnal plots for hourly peak ozone predicted for the area around Pittsburgh (see Figure 5-14) indi-
cate that for most days ozone levels with NOX controls are less than with VOC controls. The few days on
which ozone is higher under NOX controls have relatively low to moderate ozone concentrations (75 to
100 ppb) in the precontrol 2005 baseline scenario. The behavior of diurnal patterns for other areas
including Philadelphia, Boston, Cleveland, Detroit, and Charleston are similar to that shown for Pitts-
burgh.
NOX Control: New York City Case Study
A fuller understanding of the effects of NOX controls on ozone levels in New York City is evident by
examining predictions on an individual day. For this, the July 9, 1988 ozone and NOX concentrations
predicted for the 2005 baseline were compared to predictions following NOX controls in CS11.
Figure 5-15 presents the daytime maximum ozone predictions for the 2005 baseline and the change due
to NOX controls. The 2005 values show an area of ozone > 125 ppb extending from near Philadelphia
northeastward over the central core of New York City, then eastward across much of Connecticut. The
position of this high ozone plume reflects the general boundary layer flow on this day, as indicated from
the ROM layer 1 and layer 2 wind fields. Winds were generally light and variable with near stagnation
conditions until mid-afternoon when a southwesterly flow developed. Thus, the peak concentrations
over southwestern Connecticut and southeast New York State (160 to 180 ppb) appear to be associated
with morning emissions from the New York City urban area. Concentrations exceeding 160 ppb over
northern New Jersey likely reflect the combined impact of local emissions and interurban transport from
Philadelphia.
5-12
-------
The change in ozone levels resulting from NOX controls (see Figure 5-15) indicates that the large
increase in ozone is limited to the central core of the urban area extending along the centerline of the
plume into southern Connecticut. In other more peripheral portions of the plume, ozone levels actually
decline following NOX controls, as do ozone levels downwind of Philadelphia.
Examining the levels and patterns of emissions in this area provides further insight into these results.
The spatial distributions of morning (0500 to 1100 EST) total VOC and NOX emissions are provided in
Figure 5-16. These data show the high emissions density for both pollutants over the core of New York
City including several grids surrounding Manhattan [i.e., the Central Business District (CBD)]. The
highest emissions from these grids are 132 tons of VOC and 46 tons of NOx. This is more than double
that of surrounding grids and an order of magnitude greater than in other portions of the metropolitan
area. Even in Philadelphia, the highest gridded VOC and NOX emission are only 50 tons and 21 tons,
respectively. By comparing Figures 5-15 and 5-16, it appears that the large disbenefit of NOX control is
confined to areas over and downwind of the New York CBD grids.
Examining the time history of predicted ozone and NOX concentrations during the course of the day for
the 2005 and CS11 scenarios provides an explanation for the increase in ozone following NOX controls.
In the 2005 scenario, very high NOX concentrations are predicted within the four to six high-NOx emis-
sions grids in the New York CBD. In the morning (0600 EST), NOX levels are in the range of 60 to 80 ppb.
Concentrations increase with time to 75 to 200 ppb by 1000 and remain at this level for several hours
before declining to 50 to 100 ppb at 1300. Elsewhere in the urban area, NOX concentrations are an
order of magnitude less. In Philadelphia, NOX concentrations are a factor of 3 less than in the New York
CBD. Due to the high level of NO emissions, ozone concentrations in the New York CBD are suppressed
by 40 to 50 ppb compared to surrounding areas. During the day, ozone formation in the 2005 baseline
scenario is greatest around the periphery of the CBD, particularly on the northern fringe of the city.
In the NOX control strategy, NOX concentrations in the CBD are reduced by 40 to 50 percent during the
morning, with greater reductions up to 75 percent into midday. Correspondingly, ozone levels in the
CDB are higher than in the 2005 scenario. Until 1100 EST, ozone concentrations across the urban area
(including the CBD) are higher in the 2005 scenario than in the NOX strategy. By noon, however, ozone
in the CBD has increased rapidly to 182 ppb, which is 124 ppb higher than in the 2005 scenario at this
hour. In contrast, in most areas where ozone was at a peak in the 2005 scenario (i.e., northern New
Jersey and southwest Connecticut), concentrations following NOX controls are reduced by 20 to 30 ppb.
This trend continues into the early afternoon with ozone levels in the CBD rising to well over 200 ppb.
With the onset of a southwesterly flow in mid-afternoon, the area of enhanced ozone in the CBD is
transported northeastward into southwest Connecticut.
5-13
-------
Thus, the results indicate that the high levels of NOX emissions in the 2005 baseline play a major role in
scavenging ozone in the core of the city. When NOX emissions are reduced by controls, ozone levels
increase substantially. The widespread disbenefit of NOX controls across much of the New York urban
area seen from the analysis of episode maximum concentrations reflects the composite impact of the
day to day path of the plume emanating from the New York CBD. This indicates that NOX controls may
indeed be beneficial for certain sections of the New York urban area, if targeted away from the core of
the city.
A major caveat to these results is the added uncertainty introduced by the lack of spatial resolution of
emissions in the ROM for an urban-level analysis such as this. Thus, the further investigation of this
issue is warranted using an urban scale model to better quantify the magnitude and spatial extent of the
disbenefit of NOX controls in the New York City area.
Population Exposure
An alternative approach to quantify and compare the impacts of VOC and NOX strategies is by
examining the effects on population exposure. This metric is perhaps a more robust indicator than the
episode or hourly maximum values since it considers the frequency of high ozone hours during the
episode and any shifts in spatial patterns in predicted ozone relative to the locations of highly populated
areas. For most urban areas along the Corridor, population exposure to ozone > 125 ppb is the chosen
metric. For Massachusetts and coastal New England, and areas outside the Corridor where ozone is
below 125 ppb in several strategies, exposure to ozone > 100 ppb was examined. The exposures
calculated for the 2005 baseline and strategies CS10 through CS14 are presented in Figure 5-17
through Figure 5-20 for selected urban areas in the region. The figures indicate that reductions in
exposure from the 2005 baseline on the order of 90 percent were achieved for Massachusetts and
vicinity by the NO^only controls. The VOC-only, and VOC plus NOX controls produce less of a benefit
(Figure 5-17). In Baltimore/Washington, DC and Philadelphia, NOX controls alone or with VOC controls
produce reductions of 90 to 100 percent in exposure to ozone > 125 ppb (Figure 5-18). With just VOC
controls, exposures are higher than for the other strategies. Note that the disbenefit of NOy-only
controls seen in episode peak ozone downwind of Baltimore is not evident in the population exposure
metric. As indicated earlier, this occurs because the grids associated with the NOX disbenefit are over
areas with low population on the eastern shore of the Chesapeake Bay. As shown in Figure 5-19 for
New York City, the VOC controls alone or combined with point-source or mobile-source NOX controls
produce nearly equivalent exposure levels. However, adding the two types of NOX controls without any
VOC controls has essentially no benefit relative to the 2005 baseline. Adding VOC controls with full NOX
5-14
-------
controls (CS10) does reduce exposure considerably, but not to the extent of strategies with less NOX
control (CS12, CS13, CS14). Over Connecticut and Rhode Island (downwind of New York City), a
greater reduction in exposure is obtained when VOC controls are combined with point or mobile NOX
controls than with VOC controls alone, as is evident in Figure 5-19.
Outside the Corridor, NOX controls produce the greatest reduction in population exposure to ozone
> 100 ppb. Adding VOC controls yields little additional benefit. Note from Figure 5-20 that in Cleveland
and Detroit comparable benefits were achieved with the NOX point-source and NOX mobile-source con-
trols even though emissions reductions were larger in the point-source NOX control strategy (CS13 vs.
CS14).
Findings
The following are the key findings for ISSUE #1:
What are the relative benefits of VOC controls versus NOX controls in reducing ozone levels across
the region?
Maximum technology NOX controls appear to produce larger ozone reductions than stringent
VOC controls in many areas of the Northeast. This is particularly notable in the western por-
tion of the region where biogenic VOC emissions are highest and a large component of NOX
emissions are rural point sources. In the Corridor, except for New York City, NOX controls tend
to reduce the spatial coverage of high ozone levels, whereas VOC controls are more effective
in reducing the peak values. Combining VOC and NOX controls provides both benefits by
reducing the magnitude and spatial extent of high ozone concentrations.
In the presence of stringent VOC controls, peak ozone levels appear to be more sensitive to
mobile- rather than point-source NOX emissions reductions in several cities (e.g., Philadelphia,
Boston, Detroit, and Cleveland). In areas dominated by NOX point sources (e.g., Pittsburgh
and Charleston, WV), point-source controls are more effective than mobile-source reductions.
In New York City, NOX controls alone or with VOC controls are counter-productive relative to
VOC controls for short-term peak concentrations and population exposure. This effect is
most prominent in the core of the urban plume and appears to be associated with high NOX
emissions in the center of the urban area. Outside the main plume, ozone levels are actually
reduced by NOX controls. For longer averaging time (e.g., episode mean 8-hour daily maxi-
mum averages) NOX plus VOC controls produce lower ozone levels than VOC controls alone.
5-15
-------
• The relative benefits of VOC versus NOX controls varies by day (i.e., meteorology) in some
cities. This was particularly evident in the Baltimore/Washington, DC area.
The above findings underscore the complexity of the role of NOX in both ozone formation and in reduc-
ing ozone levels by controlling emissions. In addition, the results clearly indicate the need to carefully
select and examine a breadth of emissions and meteorological conditions before drawing conclusions
on the benefits of NOX controls. Also, given the local urban structure indicated by some of the results,
further, more spatially refined analyses are warranted.
5-16
-------
ISSUE #2
What is the impact of reducing regional transport on ozone concentrations in the Northeast
Corridor?
Analysis Approach
A key issue for the Northeast States concerns the impact of emissions outside the Corridor on ozone
concentrations within the Corridor. This concern is fostered by the regional nature of ozone levels and
the transport patterns typical of many of the most severe episodes in this region (see Section 3). Of
interest is whether, and to what extent, controls outside the Corridor make a difference to in-Corridor
ozone levels. There are two parts to this issue:
a. What is the impact of reducing transport on projected future baseline ozone levels in the Corri-
dor?
b. What is the impact of reducing transport on postcontrol ozone levels in the Corridor?
Two types of strategies were designed and simulated by the ROM to address these questions:
1. Stringent emissions controls were applied outside the Corridor while maintaining emissions at
the 2005 baseline inside the Corridor; and
2. Stringent emissions controls were applied inside the Corridor while maintaining emissions at the
2005 baseline outside the Corridor.
Predictions from the first type of strategy were compared to the 2005 baseline region-wide. This com-
parison was designed to quantify the impact of transported ozone and precursors on the future baseline
ozone levels in the Corridor. The second type of strategy was compared to strategies with controls
applied across the entire region. This comparison was used to determine the extent to which controls
outside the Corridor affect postcontrol ozone levels inside the Corridor.
The specific scenarios included in this analysis are as follows:
2005 baseline scenario;
CS19: The most stringent controls on VOC, NOX, and CO applied region-wide;
CS24: CS19 controls applied inside the Corridor - 2005 baseline elsewhere; and
CS25: CS19 controls applied outside the Corridor - 2005 baseline elsewhere.
5-17
-------
The 2005 baseline predictions were compared to those of CS25 to quantify the impacts of controls
outside the Corridor on baseline levels inside the Corridor. The CS24 predictions were compared to
those from CS19 to quantify the impact of controls outside the Corridor on postcontrol ozone levels
inside the Corridor.2
The analysis of the two transport impact questions listed above is divided into four components. First,
an assessment is made of the impact of controls outside the Corridor on ozone levels transported into
the Corridor. Second, the effects of reducing transport on 2005 baseline concentrations are presented.
This is followed by a discussion of the impact on postcontrol Corridor ozone levels due to controls in
upwind areas. Finally, a case study analysis is included to examine the impact of transport into and
along the Corridor more closely.
The effect of controls on ozone levels transported into the Corridor was analyzed by examining the layer
1 and layer 2 ozone predictions for a string of ROM grids located just beyond the far western edge of the
Northeast Corridor. The 72 grids included in this string are shown in Figure 5-21. Daily mean and max-
imum concentrations were calculated by State for three segments of this boundary: Virginia (VA),
Pennsylvania (PA), and New York State (NY). Forward trajectories using ROM layer 1 and layer 2 were
used to identify those days with transport flow regimes from this boundary into the Corridor. Examples
are shown in Figure 5-22. The layer 2 mean and maximum ozone concentrations along the boundary for
these days are provided in Figure 5-23, for the 2005 baseline (precontrol) and CS19 (postcontrol) sce-
narios. The layer 2 values were selected because this layer represents the bulk of the atmospheric
boundary layer during the day and the layer containing transport of pollutants aloft overnight.
The following procedures were used to quantify impacts on the 2005 baseline ozone levels in the Corri-
dor due to the upwind controls in CS25. First, all grids with daily maximum ozone concentrations
> 125 ppb for the 2005 baseline scenario were identified. The difference in daily maximum ozone
between CS25 and the 2005 baseline was then computed for each of these grids and averaged by
urban area by day and for the episode. Similar procedures were followed to quantify the impacts on
postcontrol ozone levels in CS19 except that grids with daily maximum values > 100 ppb were
examined. In addition, a case study analysis was conducted to describe the effects of transport on
selected high ozone events in several Corridor cities.
2. Initially, CS01 (Phase I maximum technology VOC controls region-wide) and CS02 (Phase IVOC controls in the Corridor only)
and CS03 (Phase I VOC controls in nonattainment areas) were included in this analysis. However, it was found that these VOC
controls did not appreciably reduce ozone levels outside the Corridor. Thus, there was little change in both ozone transport
into the Corridor and peak ozone concentrations in the Corridor.
5-18
-------
Discussion of Results
Impacts of Upwind Controls on Ozone Transport into the Corridor
The data in Figure 5-23 indicate that there is quite a large day-to-day and spatial variation in the levels of
ozone along the Corridor boundary. The highest concentrations lie along the PA and VA segments. On
4 days during this episode, plumes with peak values > 100 ppb in the 2005 scenario apparently entered
the Corridor through these segments. Following the application of controls outside the Corridor, the
peak values are reduced to 84 ppb or less. On average for the episode, the daily mean and peak
transport values are reduced by controls on the order of 20 percent for both segments. For the NY
segment, ozone transport levels are reduced somewhat less at 16 percent. The greatest reduction
along the boundary tended to occur on days with the highest concentrations. For example, on the
highest days, mean and peak values are reduced by 30 percent and 35 percent, respectively. The data
indicate that applying maximum technology controls to areas outside the Corridor reduces predicted
mean and peak ozone transported into the Corridor to levels well below the NAAQS for this episode.
Impacts of Reducing Ozone Transport on 2005 Baseline Ozone in the Corridor
The impact of reducing transport into the Corridor was most noticeable in the Baltimore/Washington, DC
area and in sections of southeast Pennsylvania. These areas are adjacent to the portions of the
boundary having the highest concentrations and greatest reduction in ozone transport. For example, in
Baltimore/Washington, DC the space/time average and peak reduction for all grids with daily maximum
ozone > 125 ppb were 8 ppb and 13 ppb, respectively. Looking at the grid with the highest predicted
concentration on each day in the episode shows an average reduction of 6 percent for those days with
ozone > 125 ppb. The impacts tend to diminish northeastward along the Corridor as indicated below:
BAL/DC PHL NYC CT BOS
Reduction for all spatial average: 8 ppb 5 ppb 4 ppb 3 ppb 4 ppb
grids > 125 ppb peak: 13 ppb 10 ppb 8 ppb 7 ppb 5 ppb
% change in daily episode average: 6% 4% 2% 2% 2%
maximum ozone peak: 9% 9% 5% 5% 3%
BAL/DC: Baltimore/Washington, DC
PHL: Philadelphia
NYC: New York City
CT: Portions of Connecticut outside the NYC urban area
BOS: Boston
5-19
-------
Thus, the reduction in future year transport may have a relatively small impact on future baseline ozone
levels within the major Corridor cities. Even with stringent controls in all areas outside the Corridor, peak
values along much of the Corridor would still be well above the NAAQS, without additional controls in the
Corridor beyond those in the 2005 baseline. This is evident from the plots in Figure 5-24, which show
the episode maximum 1-hour ozone concentrations for the 2005 baseline and CS25. As described
above, the largest change relative to the level of the NAAQS inside the Corridor occurs in southeast
Pennsylvania and from the Baltimore/Washington, DC area eastward across Delaware. For the condi-
tions simulated, little impact relative to the magnitude of baseline episode maximum ozone is predicted
for areas in the northern sections of the Corridor.
Impacts of Reducing Transport on Postcontrol Ozone In the Corridor
The impacts of reducing transport on postcontrol ozone in the Corridor is more significant than the
impacts on 2005 baseline levels. This is evident by comparing daily maximum predictions for CS19
(region-wide application of maximum controls) versus CS24 (emissions controls in the Corridor, but
2005 baseline emissions elsewhere). In CS19, peak ozone levels along the Corridor are reduced to
below 125 ppb. Values for individual cities are listed below along with the corresponding values for
CS24.
Episode Maximum 1-Hour Ozone Concentrations
Region-wide controls No controls outside Corridor
(CS19) (CS24)
Baltimore/Washington, DC 122 ppb 139 ppb
Philadelphia 115 ppb 123 ppb
New York City 118 ppb 123 ppb
Boston 107 ppb 113 ppb
The data indicate that the difference in ozone levels between CS19 and CS24, in ppb ozone, is
comparable with that predicted for the precontrol analysis. Now, however, in the postcontrol environ-
ment (CS19), Corridor ozone levels are much lower than in the 2005 baseline, and the additional
increment due to transport becomes much more significant. For example, in both Philadelphia and New
York City, peak values without controls outside the Corridor are increased to the threshold range of the
NAAQS. As shown in Figure 5-25, episode maximum values in CS24 rise to levels near or above the
NAAQS over a large part of Baltimore/Washington, DC urban area.
5-20
-------
Transport Case-Study Analysis
A case study analysis was conducted to provide a more detailed picture of the effects of upwind controls
on ozone concentrations within the Northeast Corridor. For this, the daily maximum ozone concentra-
tions in the Corridor were examined for the 2005 baseline and the CS19, CS24, and CS25 strategies to
identify grids showing the largest impact of upwind controls. Forward and backward trajectories were
constructed for these grids from the hour of maximum ozone using the ROM layer 2 wind fields.
Predictions of ozone, NOX, and reactive organics (ROG) were extracted for points along the trajectories
in order to provide a "quasi" Lagrangian history of species concentrations before and after the
application of controls.
The case selected for presentation is shown in Figure 5-26. The starting point for the forward/backward
trajectories was a grid near Trenton, NJ. Daily maximum predicted ozone in this grid on July 8, 1988
(145 ppb in the 2005 scenario) was reduced by 10 ppb as a result of controls on sources upwind of the
Corridor. The trajectory start time was the hour of the maximum predicted concentration, 1500 EST.
The trajectory path shown in Figure 5-26 indicates transport into the Corridor from West Virginia, over
portions of the Pittsburgh urban area, then across northern Pennsylvania during the two days prior to
July 8. The air parcel enters the Corridor northwest of Philadelphia on the evening of July 7. After
passing over Philadelphia during the morning, this air parcel continues northeastward across New York
City on the evening of July 8. On the following day, July 9, the air parcel tracks through Connecticut and
on up across Boston. Concentrations of ozone, NOX, and ROG along the trajectory are provided in
Figure 5-27 for the 2005 baseline and CS25, and in Figure 5-28 for CS19 and CS24. The trajectory and
time series profiles are divided into several segments in order to explain the behavior of pollutant con-
centrations. Transport into and within the Corridor for the 2005 baseline scenario is described first,
along with the impact on baseline concentrations of upwind controls. The impact of reducing transport
following the application of stringent controls inside the Corridor is presented last.
Impacts of Transport on 2005 Baseline Ozone
a. Upwind Patterns
Examining the data in Figure 5-27 indicates that ozone levels along segment A-B for the 2005 baseline
scenario show a gradual decline from 85 ppb to 65 ppb as the air parcel approaches the Corridor. A
NOX source is encountered just upwind of the Corridor near point B as indicated from the NOX concen-
tration profile. In response, ozone levels increase to 90 ppb. When controls are applied outside of the
Corridor (CS25), ozone levels are reduced to about 50 to 55 ppb along the entire upwind portion of the
5-21
-------
trajectory path. Concentrations of ROG upwind are actually slightly higher with the application of con-
trols outside the Corridor. As indicated in the discussion of Issue #1, this part of the region is "NOX
limited" due to the relatively large amount of biogenic sources. With the application of stringent NOX
controls, less VOC is consumed in forming ozone.
b. Corridor Transport Patterns
As the air parcel approaches and crosses the Philadelphia area on the morning of July 8 (segment D-E
in Figure 5-27) precursor concentrations increase substantially in response to the injection of fresh
emissions. Note that the small changes in incoming precursor levels due to upwind controls are over-
whelmed by Corridor emissions and have little effect on precursor levels inside the Corridor.
Ozone levels rise in the Philadelphia plume with the maximum concentrations over central New Jersey
and near New York City by late afternoon. The peak concentrations along this segment are reduced by
about 10 ppb due to controls upwind of the Corridor. The remains of the Philadelphia plume continue to
track across the New York City area during the evening and into Connecticut overnight (Segment E-F).
Precursor levels increase sharply over New York City and ROG levels remain relatively high during the
night. Ozone levels decline somewhat but remain at around 100 ppb. In this segment, the effects of
upwind controls on ozone levels are still evident from the CS25 profiles, but the reduction is now only
about 5 ppb. The following day (segment F-G), ozone levels remain at 100 ppb until 1100 when NOX
levels increase, indicating an injection of fresh emissions from layer 1. Ozone then rises back to near
140 ppb. The impact of controls outside the Corridor continues for the remainder of the trajectory at
about 3 ppb. Thus, the effects of upwind controls on 2005 baseline ozone levels are small compared
with the quantity of ozone produced by Corridor emissions. The amount of reduction in ozone (due to
upwind controls), which starts out at 25 ppb just upwind of Philadelphia, declines with time as the air
parcel progresses along the Corridor and additional large precursor sources are encountered.
This particular trajectory also reveals a 'second day" effect over Connecticut. That is, ozone levels
remaining from the Philadelphia plume combine with evening precursors injected over New York City
and with fresh emissions in Connecticut to produce ozone levels comparable with those in the plume the
day before. It is interesting to note that with controls in place outside the Corridor, the second-day peak
over Connecticut is actually several ppb greater than the peak on the previous day over central New
Jersey.
5-22
-------
Impacts of Transport on Postcontrol Ozone
In this part of the analysis, ozone concentrations are reduced inside the Corridor by stringent controls
placed on Corridor sources. In CS19, controls are applied regionwide and peak ozone levels fall below
125 ppb across the US portion of the region. In CS24, controls are maintained within the Corridor but
emissions outside the Corridor are increased to the 2005 baseline level. The difference in ozone and
precursor concentrations outside the Corridor between CS19 and CS24 is identical to that found by
comparing the 2005 baseline with CS25, since in this part of the region the CS19 emissions were used
for CS25 and the 2005 baseline emissions were used for CS24.
Inside the Corridor, significant reductions in ozone and precursors are evident between the 2005 base-
line and CS19. For example, comparing ROG concentrations from Figures 5-27 and 5-28 indicates that
ROG inside the Corridor (beyond C on the trajectory path) is reduced by 50 to 80 percent to levels
comparable to those upwind of the Corridor. The impact on postcontrol Corridor precursor levels due
to upwind controls is similar to that found in the analysis of 2005 baseline scenario above.
When controls are applied regionwide (CS19), ozone concentrations are reduced, with peak levels on
the first and second day inside the Corridor at about 110 ppb (see Figure 5-28). Without controls on
upwind sources (CS24), ozone increases by 10 to 15 ppb in segment D-E and the peak concentration
rises to near 120 ppb. A lesser increase is noted in the second-day peak.
Thus, the results of this trajectory analysis suggest that without reducing transport, ozone produced by
Corridor cities that are adjacent to the upwind boundary of the Corridor, namely Philadelphia and Balti-
more/Washington, DC, may approach the level of the NAAQS even with stringent controls on Corridor
emissions. The effects are less for areas further away from the Corridor boundary.
Findings
The following are the key findings for ISSUE #2:
What is the impact of reducing regional transport on Northeast Corridor ozone levels?
In the precontrol scenario, 2005 baseline ozone levels are not substantially impacted by
reducing transport except in Baltimore/Washington, DC and around Pennsylvania. These
areas are closest to the portion of the Corridor boundary having the greatest incoming ozone
levels during this episode. It is possible that conditions in other episodes might produce a
greater impact in other portions of the Corridor. Still, even in these two areas, the impacts are
small relative to the level of predicted ozone in the 2005 baseline.
5-23
-------
In the postcontrol scenario, the results suggest that without stringent upwind controls ozone
levels in parts of the Corridor may not be reduced to below the NAAQS even with stringent
controls along the entire length of the Corridor. Again, the effects are most pronounced for
cities near the upwind boundary of the Corridor. For episodes with higher ozone transport
into the Corridor from a northwesterly direction, larger impacts due to transport might be seen
in New York City and New England.
5-24
-------
ISSUE #3
What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted ozone
concentrations in the Northeast to below 125 ppb?
Analysis Approach
This question was examined to provide a general indication of the level of emissions reductions which
may be needed to reduce peak 1-hour ozone concentrations to below 125 ppb across the Northeast.
As discussed in Section 1, urban scale modeling by State agencies will be required to develop specific
emissions targets which demonstrate attainment. This is a consequence of the regional nature of ROM.
Another important point to make with respect to this issue is that the most effective ROMNET strategy for
an area may not be unique. That is, other combinations of controls may be at least as effective. How-
ever, at least one strategy has been identified that is sufficiently stringent to reduce predicted ozone
levels to below 125 ppb in all sections of the US portion of the modeling domain throughout the episode
simulated.
A technology-based approach was adopted to identify strategies that reduce ozone to below 125 ppb.
That is, existing or envisaged control measures were applied to the 2005 baseline on a source category
basis. For areas where such controls were insufficient, additional reductions were applied in an
across-the-board manner (i.e., all source categories were reduced by an equivalent percentage) until
ozone levels dropped below 125 ppb. The ROMNET strategies simulated to address this issue are listed
below. Unless specified, sources not affected by specific controls are at the 2005 baseline level. Emis-
sions of CO were reduced by certain maximum technology controls in CS10, and in CS12 through CS19.
2005 baseline: existing State and Federal control programs
CS05: VOC, NOX, and CO controls in October 1989 Clean Air Act proposed legislation
CS10: combined VOC and NOX maximum technology controls (CS12 + CS11)
CS11: maximum technology NOX controls on point, area, and mobile sources
CS12: CS01 VOC controls plus stringent VOC tailpipe standards
CS13: maximum technology NOX point-source controls with CS12 VOC controls
CS14: maximum technology NOX mobile-source controls with CS12 VOC controls
CS15: CS10 technology controls plus alternate-fueled (M100) vehicles along the Corridor
and low-reactivity solvent substitution regionwide.
5-25
-------
The remaining strategies focused on controls in New York City and Baltimore/Washington, DC. Ozone
levels in these areas persisted above the NAAQS even with the previous technology-based strategies.
. CS16: CSIOplus
(a) in New York City a 64 percent across-the-board VOC reduction applied in addition to
maximum technology controls; no maximum technology NOX point-source controls; and
(b) in Baltimore/Washington, DC a 54 percent across-the-board VOC reduction applied in
addition to maximum technology controls.
CS18: CS15 with CS16 measures in New York City and Baltimore/Washington, DC
CS19: CS18 except without maximum technology NOX point-source controls in Baltimore/-
Washington, DC.
The rationale for applying VOC controls and removing NOX controls in the last three strategies was
based on the response to VOC, NOX, and reactivity-based controls in these cities. These findings are
discussed in this Section and as part of Issues 1 and 4.
Control measures in the above strategies were assumed to be 100 percent effective. A final strategy,
CS23, was simulated to examine to what extent ozone levels might increase above those predicted for
CS19 if rule effectiveness and other more realistic assumptions about emissions control programs were
considered.
Meteorological conditions for the July 1988 episode were simulated for all of the above scenarios. In
addition, the June 1983 episode was simulated for the 2005 baseline and CS19 to determine if CS19
controls would maintain ozone below 125 ppb under different meteorological conditions.
Discussion of Results
The overall emissions for each strategy relative to the 2005 baseline are given in Figure 5-29 for major
urban areas in the US portion of the region with 1985 base case ozone levels > 125 ppb. The highest
and second highest daily maximum 1-hour ozone concentrations for the July 1988 episode are also
included in these figures.
Several general observations are evident from the results:
Existing control programs are predicted to produce a 10 to 15 percent reduction in maximum
ozone levels in most cities by the year 2005. Lesser reductions are estimated in Boston and
Charleston, WV. Still, existing programs are not sufficient to reduce ozone to below 125 ppb
in any of the major Northeast cities.
Explicit control measures in the October 1989 Clean Air Act legislation produce a 15 to 20
percent reduction in maximum ozone levels in most areas. Peak concentrations fall below
130 ppb in Pittsburgh and Detroit, but only Charleston drops below 125 ppb.
5-26
-------
The emissions reductions that lowered predicted ozone concentrations to below the NAAQS are
described below by urban area. A summary of these emissions reductions is given in Table 5-2.
New York City and Vicinity:
Peak ozone levels in the core of the New York City urban plume are reduced by VOC controls and
increased by NOX controls. In the presence of stringent NOX tailpipe standards, VOC reductions on the
order of 90 percent may be needed to lower ozone to below 125 ppb (CS16). Even without stringent
NOX controls, VOC reductions beyond 75 percent may be necessary (CS12). Reactivity-based
strategies may provide some additional benefits in reducing ozone in the presence of NOX controls (see
results for Issue #4).
Considering more realistic control effectiveness assumptions (CS23), peak predictions rise from
120 ppb to 140 ppb, indicating that further reductions may be required.
Baltimore/Washington, DC and Vicinity:
Peak ozone levels are right at the NAAQS with the combined VOC and NOX maximum technology
controls in CS10. Adding the across-the-board VOC reductions in CS16 changes the VOC control level
from 56 percent to 80 percent and lowers peak ozone to 118 ppb.
It should be noted that one grid remained above 125 ppb downwind of Baltimore over the eastern shore
of the Chesapeake Bay in the CS16 scenario. This value (133 ppb) appears to be greatly influenced by
a NOX point source in the same ROM grid. The highest concentration adjacent to this grid is 119 ppb.
Also, CS19, which removes point-source NOX controls, reduced this value to 103 ppb. Given the limita-
tions of ROM for such localized effects the importance of this value was discounted. Also, without
point-source NOX controls, ozone levels in other parts of this urban area increase to between 120 ppb
and 122 ppb on several days.
The application of reasonable control assumptions in CS23 increased ozone levels in Baltimore/Wash-
ington, DC to 128 ppb, or just above the level of the NAAQS.
Philadelphia and Vicinity:
The peak ozone concentrations is reduced to 124 ppb with CS11, the NOX only control strategy.
Addition VOC controls reduce the peak to 117 ppb (CS10). Given the results for CS13 and CS14
5-27
-------
(mobile-source controls vs. point-source NOX controls) it appears that the full complement of NOX
controls in addition to stringent VOC controls may be needed. Considering more realistic control
effectiveness assumptions increases the peak value to 125 ppb.
Boston and Vicinity:
Peak ozone levels are near 120 ppb for both CS11 (NOX only controls) and CS14 (VOC and stringent
NOX tailpipe standards). It appears that the full complement of NOX controls plus the maximum
technology VOC measures (CS10) may be necessary in this urban area.
The combined effects of the CS10 controls, reactivity reduction measures, and potentially the additional
upwind controls in New York City, reduce the CS10 peak of 116 ppb to 113 ppb inCSl9. Considering
the realistic control effectiveness assumptions increases the CS19 peak to 125 ppb.
Pittsburgh/Cleveland/Detroit:
In all three urban areas, the peak value is in the range of 120 to 125 ppb with VOC maximum technology
controls plus mobile-source NOX controls. Lower ozone levels, well below 125 ppb, are produced by
combining the full complement of NOX controls even without the VOC measures.
Peak values in all three cities remain well below 125 ppb when realistic effectiveness considerations
were applied to the most stringent VOC and NOX strategy (CS19).
Thus, it would appear that control strategies for these areas should emphasize NOX controls more so
than VOC control measures.
Charleston, WV:
Peak ozone concentrations in this area are most sensitive to point-source NOX controls as evident by
comparing the predictions for CS11 and CS13 (point-source plus mobile-source NOX controls and
point-source NOX plus VOC controls) with those of CS12 (VOC control) and CS14 (mobile-source NOX
plus VOC controls). Without the point-source controls, ozone levels remain close to the level of the
NAAQS. Concentrations drop dramatically with the application of the point-source NOX controls. Also,
the results suggest point-source NOX controls at the maximum technology level may be "overkill0 for this
area. Even the Clean Air Act NOX point-source controls (low-NOx burners on coal fired utilities) may
provide more than enough emissions reduction.
5-28
-------
Results for the June 1983 episode
As indicated above, the most stringent control strategy (CS19) was also applied to the June 1983
episode. The purpose of this was to confirm that the CS19 controls would also reduce ozone to below
125 ppb under different meteorological conditions. The June 1983 episode, as described in Section 3,
was characterized by multiday recirculation flows within the Corridor. In contrast, the July 1988 episode,
which was used for all strategies, was dominated by general southwest to westerly flow regimes with
only some instances of recirculation. The episode maximum ozone concentrations predicted for the
June 1983 episode are shown in Figure 5-30 for the 2005 baseline and CS19 scenarios.
These data indicate that high baseline concentrations of ozone are predicted near most major cities in
the Corridor during the June 1983 episode. Areas of highest ozone are located inland in a general
southwest through northwest direction from the center of each urban area. For example, the impact of
the New York City plume is over northern New Jersey and southeast New York State rather than over
Connecticut and Long Island as on many days in the July 1988 episode. Also note the area of ozone
above 125 ppb extending into central Vermont. Peak ozone in this area for the July 1988 episode was
< 100 ppb.
In the postcontrol scenario (CS19) ozone levels in most areas are reduced to well below the NAAQS.
The highest value is 124 ppb in a single grid southwest of Washington, DC. Surrounding values, how-
ever, are at most 111 ppb. A comparison of the July 1988 and June 1983 episode maximum 1-hour
concentrations for the 2005 baseline and CS19 is shown in Table 5-3 for several major cities. The CS19
controls appear to reduce peak ozone in the June 1983 episode to levels near or below those for the
July 1988 episode.
FINDINGS
The following are key findings for Issue #3:
What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted ozone
concentrations in the northeast to below 125 ppb?
• Stringent maximum technology VOC and NOX controls may be necessary in all areas of the
Northeast Corridor.
Additional reductions of VOC on the order of 64 percent in New York City and 54 percent in
Baltimore/Washington, DC also may be needed. At present, this level of emissions reductions
is beyond known or envisaged control technologies.
5-29
-------
Application of stringent NOX controls in New York City is counterproductive. Ozone levels
approach 125 ppb as VOC controls are increased. It appears that even without NOX controls,
stringent VOC technology and reactivity-based measures may be needed in this urban area.
Even considering rule effectiveness and a more realistic representation of control programs
(e.g., fleet turnover) ozone levels may still be near or just above 125 ppb in most sections of
the Northeast Corridor with the most stringent VOC/NOx/reactivity-reduction strategy simu-
lated. The exception is New York City, where the peak ozone level is predicted to be 140 ppb.
The effectiveness of the most stringent control strategy in reducing ozone to < 125 ppb was
confirmed using an alternate meteorological episode.
5-30
-------
ISSUE #4
What are the effects of reactivity-based strategies on regional ozone level?
Analysis Approach
The use of "clean fuels' (e.g., methanol, liquified natural gas, and reformulated gasoline) along with other
reactivity-based controls have recently been given a wider consideration by the EPA, States, and
industry as part of potential strategies to reduce ozone in areas having severe ozone problems, such as
the Northeast Corridor. Thus, several ROMNET strategies were designed to examine the relative effec-
tiveness of several of these measures on Northeast ozone levels.
The effects of reactivity-based strategies on regional ozone levels were addressed in CS15 and CS20
through comparisons with the 2005 baseline, CS10 (maximum technology VOC and NOX controls), and
CS12 (maximum technology VOC controls only). As described in Section 4, CS15 and CS20 contain
two control measures that alter the reactivity of VOC emissions:
a. substituting low-reactivity compounds for all solvent emissions categories across the entire
US portion of the region; and
b. changing the passenger car fleet along the Northeast Corridor to M100 (methanol-powered)
vehicles.
The difference between CS15 and CS20 is that CS15 also contains maximum technology controls on
VOC and NOX, whereas CS20 does not. Emissions levels in CS20 reflect the 2005 baseline except for
mobile sources in the Northeast Corridor. There, mobile emissions are greater due to the higher emis-
sion rate of M100 vehicles, though the reactivity-weighted emissions are less, as described in Section 4.
To assess the impact of reactivity reductions alone, in the absence of other VOC and NOX control mea-
sures, the predictions for CS20 are compared with the 2005 baseline. The effectiveness of CS20 is also
compared with the reductions provided by the maximum technology VOC controls in CS12. Then, the
impact of combining reactivity-based and VOC plus NOX technology-based controls is examined from
the predictions for CS15. Results from CS15 are compared with predictions for CS10 to determine the
additional benefit obtained by adding reactivity-based controls on top of technology-based measures.
The ozone metrics selected for this analysis are as follows:
The average change in daily maximum ozone levels (ppb) for days > 125 ppb in the 2005
baseline;
The number of days with ozone levels > 125 ppb; and
Percent reduction in population exposure to 1 -hour concentrations > 125 ppb.
5-31
-------
In the presentation of results, these metrics were calculated for selected portions of the region. The data
were aggregated for areas showing a similar response for the first two metrics. (There was less consis-
tency among cities in the effects of controls on population exposure.) Data for Philadelphia, Baltimore/-
Washington, DC, and Boston comprised Group 1. New York City, Greater Connecticut, and Rhode
Island show a different response to the controls and are examined individually. Other, more peripheral
areas of the Corridor, are combined in Group 2. Finally, Pittsburgh, Cleveland/Canton/Akron, and
Detroit comprise Group 3.
The first metric was calculated as follows. For each area, all days with daily maximum ozone concen-
trations >125 ppb in the 2005 baseline scenarios were identified. The difference in daily maximum
ozone between two scenarios (e.g., CS20 and CS12) was calculated for each of these days, and then
averaged for the days with ozone > 125 ppb. Finally, Group averages were computed from the average
values for individual urban areas. The other two ozone metrics are fairly straightforward. A summary of
the ozone metrics for each of the above areas is provided in Table 5-4.
Discussion of Results
Impact of Reactivity Controls Alone
The data in Table 5-4 indicate that, in general, the reactivity-based strategies are by far the most effective
in the New York City area and in the two areas directly downwind, Greater Connecticut and Rhode
Island.
In the New York plume, the largest reduction in daily maximum ozone appears to occur close by, within
New York City itself. The magnitude of the reduction, although still substantial, tends to level off with
distance downwind. Also, the frequency of high concentrations and population exposure are greatly
reduced by this strategy. For example, the population exposure was reduced by 75 percent in New York
City and by 69 percent in Greater Connecticut.
The benefits of the reactivity-based strategy is less notable in other Corridor cities (Group 1), and
decreased further in peripheral areas (Group 2). For example, the average reduction in daily maximum
ozone provided by CS20 vs the 2005 baseline is only 8 ppb in Group 1, and 5 ppb in Group 2. However,
in terms of population exposure, there is a moderate reduction in Group 1 (46 percent) with a lesser
benefit for Group 2 (33 percent reduction). The difference between the effects on daily maximum ozone
and population exposure indicates that the reductions from this scenario are occurring close to the
urban areas where population is greatest and/or that the number of hours and spatial extent of excee-
dances is being reduced to a greater extent than the daily maximum concentration.
5-32
-------
In the large urban areas outside the Corridor, as represented by Group 3, there was very little reduction
(2 ppb) in daily maximum ozone levels > 125ppb for a 23 percent reduction in reactivity-weighted
emissions. (Recall that only the solvent substitution component of the reactivity strategy was applied
outside the Corridor.) Population exposure to ozone above > 125 ppb was, however, reduced by
nearly 30 percent.
Although there was consistency between areas within each group, there were some notable day-to-day
differences in individual cities. For example, the reduction in 2005 baseline daily maximum ozone (i.e.,
"delta ozone") ranged from 3 to 20 ppb on different days in Baltimore/Washington, DC; 30 to 144 ppb in
New York City, and 5 to 22 ppb in Boston. Also, on the July 10 simulation day, ozone levels actually
increased following the application of the reactivity-based measures. This occurred across a substantial
portion of the Washington, DC area, and sections of eastern Maryland and Delaware. In these areas,
ozone levels, which were already > 125 ppb in the 2005 baseline, increased by up to 9 ppb, and pop-
ulation exposure more than doubled. The event occurred on the last day of a 3-day period with moder-
ately high ozone levels (125 ppb
-------
In New York City, the lower reactivity does appear to counteract the disbenefits of NOX controls
described in Issue #1. This is evident from the average 36 ppb reduction in daily maximum ozone
between CS10 and CS15. However, this effect decreased with downwind distance in Greater Connec-
ticut and Rhode Island.
FINDINGS
The following are the key findings for Issue #4:
What are the effects of reactivity-based strategies on regional ozone levels?
Reactivity-based strategies similar to those simulated in ROMNET may provide the greatest
benefit in large urban areas which are VOC-limited and thus are more responsive to changes
in VOC emissions. In such areas, reactivity-based measures could provide reductions in
ozone levels comparable with that provided by a technology-based approach. Also, reactivity
measures may counterbalance the negative impact of NOX controls in New York City and
other areas that show a similar response.
In other Northeast Corridor cities, and by extension, cities that respond to both VOC and NOX
controls, reactivity measures may produce a notable reduction in daily maximum ozone levels
> 125 ppb. However, for the Corridor cities, the reduction provided by the reactivity mea-
sures was only half of that from the technology-based VOC controls, and four times less than
that from the VOC plus NOX controls
In other, more peripheral sections of the Corridor, the relative benefits of VOC technology
controls alone or with NOX controls far outweigh the benefits of reactivity-based controls
In cities outside of the Corridor that were found to be most sensitive to NOX controls, there was
less reduction in ozone from the reactivity measure compared with the technology-based
controls.
5-34
-------
ISSUE #5
How does the large uncertainty in biogenic emissions alter conclusions regarding the effective-
ness of control measures?
Analysis Approach
Biogenic emissions are believed to be among the most uncertain of the various input parameters that
contribute to ozone concentrations predicted by the Regional Oxidant Model (ROM). Specification of
biogenic emissions rates is critical because of the high photochemical reactivity of principal biogenic
species, particularly isoprene, and the large amount of biogenic VOC mass believed to be emitted on
typical hot summer days. In fact, on a regional basis, total daily biogenic VOC emissions are compara-
ble in magnitude with emissions of 1985 anthropogenic VOC. Domain total biogenic VOC, based upon
the meteorology for July 11,1988, is 25,800 tons per day, while the anthropogenic VOC total is 21,000
tons per day. The Northeast Corridor is not as dominated by biogenic VOC emissions. Corridor bio-
genic VOC, based upon the meteorology for July 11, 1988, is 3,700 tons per day, while the anthropo-
genic VOC total is 8,700 tons per day. Previous simulations with the ROM indicate that biogenic
emissions play a significant role in the formation of regional ozone concentrations and the development
of ozone plumes from major source areas.
Given the importance of biogenic emissions and the large uncertainty in emission estimates, it is impor-
tant to determine whether the conclusions regarding the effectiveness of control strategies are likely to
differ if biogenic emissions are substantially different from the state-of-the-science 'best" estimates
currently used for the ROMNET model simulations. A series of ROM simulations were conducted to
examine the sensitivity of ozone predictions to uncertainty in biogenic emissions. The cases which were
analyzed are listed in Table 5-5. These strategies were modeled with biogenic emissions set at the "low"
and "high" end of the range of probable uncertainty. The uncertainty range was taken as ± a factor of
three, based on recommendations from the EPA Office of Research and Development. This represents
an overall uncertainty estimate since it was not possible to adequately infer the uncertainties of individ-
ual components of these emissions, namely biomass data, emission factors, the functional relationship
to meteorological conditions, forest canopy effects, and speciation.
In the uncertainty scenarios, biogenic emissions were altered in an "across-the-board" manner. For
example, in CS06, all biogenic species, in all grids, for all hours were reduced by a factor of three and
modeled with the anthropogenic emissions from CS05.
5-35
-------
The discussion of the results of the biogenic sensitivity simulations will focus on eight urban areas, listed
below:
Baltimore/Washington, DC Greater Connecticut
Boston Greater New York City
Charleston, WV Philadelphia and vicinity
Detroit Pittsburgh
An assessment of the impact of uncertainty in biogenics is presented in terms of (1) episode maximum
1-hour concentrations, (2) grid-hours > 125 ppb, and (3) population exposure to 1-hour levels
> 125ppb.
Discussion of Results
Episode Maximum Predicted Ozone
1985 Base Case
The spatial distribution of predicted episode maximum ozone with "best" estimate biogenics is shown in
Figure 5-31. Ozone levels > 200 ppb are predicted over the New York City area and over Lake Ontario.
Levels > 140 ppb are indicated over Pittsburgh, Cleveland, Baltimore/Washington DC and much of the
Northeast Corridor. In the "low" biogenics scenario (see Figure 5-31), ozone levels > 200 ppb are pre-
dicted over New York City. Levels > 140 ppb are indicated near Pittsburgh and for a portion of the
Northeast Corridor.
In contrast, with the "high" biogenics (see Figure 5-32) ozone levels > 200 ppb are predicted over New
York City, Lake Ontario, Washington ,DC, and off the coast of Delaware. Levels > 140 ppb are indicated
over Detroit, Pittsburgh, Cleveland, portions of West Virginia, much of Maryland, and most of the North-
east Corridor.
It should be noted that the ozone levels predicted with the "best" estimate biogenic emissions most
closely replicated the observed ozone levels.
Control Scenarios
Figure 5-33 illustrates the episode maximum ozone levels predicted for each of the eight urban areas
listed above. In all cases, the impact of the "high" estimate biogenic emissions is greater than the impact
of the "low" estimate biogenic emissions (impact is defined as the change in predicted concentration
from the 'best' estimate biogenics). This result is anticipated, because the threefold increase in bio-
genic emissions contributes a greater proportional increase in VOC mass in relation to the proportional
5-36
-------
decrease in total VOC mass with a threefold reduction in biogenic emissions, since anthropogenic
emissions are constant for a given scenario. Thus, the relative impact in each area is a function of the
relative abundance of anthropogenic VOC to biogenic VOC emissions, and the availability of NOX emis-
sions to react with the VOC emissions to form ozone.
CS19, which used "best" estimate biogenic emissions, reduced episode maximum ozone levels in the
US portion of the domain to below 125 ppb. When "high" estimate biogenic emissions were simulated,
five of the areas in Figure 5-40 would exceed 125 ppb. In the Baltimore/Washington DC area, levels are
predicted to exceed 200 ppb. When "low" estimate biogenic emissions are modeled with CS19, the
episode maximum ozone levels are reduced to approximately 100 to 105 ppb.
Grid-Hours > 125 ppb
An alternative method of assessing control strategy effectiveness is to determine the number of
grid-hours exceeding 125 ppb. This particular parameter is extremely sensitive to the uncertainty in
biogenic emissions, especially in the base case. For example, in the Baltimore/Washington, DC area for
the 1985 Base Case, 285 grid-hours exceeded 125 ppb when the "best" estimate biogenics were
simulated (Figure 5-34). The number increased to 863 grid-hours in the "high" estimate scenario, and
decreased to 42 grid-hours > 125 ppb for the "low" estimate scenario. A similar sensitivity is observed in
each of the other seven areas shown in Figure 5-34.
This sensitivity to biogenic emissions is diminished as controls are applied, especially NOX controls.
Ozone levels in CS19 with "best" estimate biogenics are below 125 ppb in the US portion of the domain.
For five of the eight areas, "high" estimate biogenics results in an increase of grid-hours > 125 ppb. The
total increase for these five areas with "high" estimate biogenics is 837 grid-hours.
Population Exposure to 1-Hour Ozone Levels > 125 ppb
Population exposure of 1 -hour levels > 125 ppb show trends similar to the grid-hours > 125 ppb. The
1985 Base Case values are extremely sensitive to biogenic VOC emission rates, but the sensitivity
decreases as controls are applied (Figure 5-35). For example, the Baltimore/Washington, DC area is
predicted to have over 75 million person-hours > 125 ppb with "high" estimate biogenics, but only 7
million person-hours > 125 ppb with "low" estimate biogenics. This range reduces to 22 million
person-hours for "high" estimate biogenics and 0 person-hours for "low" estimate biogenics when CS19
is simulated.
5-37
-------
Findings
The following are the key findings for Issue #5:
How does the large uncertainty in biogenic emissions alter conclusions regarding the effective-
ness of control measures?
ROM simulations with "best estimate" biogenic emissions yield predicted ozone concentra-
tions closer to observed values than biogenic emissions at either end of the uncertainty range
(± a factor of three), which provides some added confidence to the emissions rates used in
ROM strategy simulations.
The sensitivity of ozone levels and population exposure to biogenic uncertainty varies con-
siderably from city to city. In general, there is a greater sensitivity with increasing biogenics at
the high end of the uncertainty range than at the lower end.
Grid-hours > 125 ppb are much more sensitive to an increase in biogenic emissions than the
other metrics.
If biogenics were at the "low" end of the range, predicted ozone in Pittsburgh and Detroit
would fall below 125 ppb with the controls in CS05; all other cities would still be above this
level (note that ozone in Charleston, WV was already below 125 ppb with "best estimate" bio-
genics).
With the application of stringent controls that reduced peak ozone to below 125 ppb with
"best estimate" biogenics (CS19), ozone levels rise well above this level in the "high" biogenics
scenario in all cities except Pittsburgh and Charleston. Thus, if biogenics are actually near the
high end of the uncertainty range, then additional controls offering reductions in emissions
beyond those in CS19 will be needed in many of the Northeast cities.
5-38
-------
5.4 SUMMARY OF MAJOR FINDINGS TO KEY STRATEGY ISSUES
ISSUE #1:
What are the relative benefits of VOC versus NOx controls in reducing ozone levels across the
region?
NOX controls are more effective than VOC controls (beyond existing programs) in many areas
of the Northeast.
• The efficacy of NOX versus VOC control varies from city to city and from day to day in some
cities.
Unlike VOC control, NOX control can be counterproductive in some cities.
ISSUE #2:
What is the impact of reducing regional transport on Northeast Corridor ozone levels?
The impact of controls outside the Corridor on ozone levels inside the Corridor is minor, given
current control programs and planned pre-Clean Air Act control measures.
Following the application of stringent controls in the Corridor, upwind controls may determine
whether or not the NAAQS is met throughout the Corridor.
ISSUE #3:
What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted ozone
concentrations in the Northeast to below 125 ppb?
Reducing ozone to < 125 ppb appears to be possible in the Northeast with the application of
stringent control measures.
• The magnitude of emissions reductions required varies by city.
The greatest amount of emissions reductions are needed in the New York City urban area,
with a VOC reduction of 91 percent from present (1985 base) emissions levels, maximum
technology mobile-source NOX tailpipe standards, and reactivity reduction measures.
5-39
-------
ISSUE #4:
How effective are potential reactivity-based strategies in reducing regional ozone levels?
Reactivity-based measures appear to be most effective in areas where VOC controls produce
the largest reductions in ozone levels. However, maximum technology VOC controls are more
effective than the reactivity-reduction measures.
In such areas, the most benefit appears to occur close-in to the city where population is
greatest, with diminishing benefits evident further downwind.
ISSUE #5:
How does the large uncertainty in biogenic emissions alter conclusions regarding the effective-
ness of control measures?
• The uncertainty in biogenic emissions is large, but model predictions are closest to observed
values with 'best estimate* biogenic emissions rates.
If biogenic emissions were near the high end of the uncertainty range, then most of the
Northeast Corridor would require emissions reductions beyond the stringent measures
required to reduce peak ozone to below 125 ppb with 'best estimate" biogenics.
• If biogenic emissons were actually near the low end of the uncertainty range, measures in the
1989 proposed Clean Air Act legislation would still be insufficient to reduce ozone to
< 125 ppb throughout the Corridor.
5-40
-------
Episode Maximum Ozone: BS85
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 100 : -i >= 100 mam >= 125
>= 150 BOB >= 175 ••»>=200
Figure 5-1. Predicted 1985 base case episode maximum 1-hour ozone concentrations (ppb) for the
July 1988 episode.
5-41
-------
Episode Maximum Ozone: BS05
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 100
>= 150
>= 100
>= 175
>= 125
>= 2OO
Figure 5-2. Predicted 2005 baseline episode maximum 1 -hour ozone concentrations (ppb) for the
July 1988 episode.
5-42
-------
Massachusetts/
NE Coastline
Greater Cleveland
Greater Pittsburgh
onnecticut/
hode Island
New York City
Greater Philadelphia
Baltimore/Washington
Figure 5-3. Areas in the ROMNET domain used in calculating selected metrics.
5-43
-------
VOC Emissions
USA Portion of the Domain
40000 -
35000
'30000-
'25000-
'20000-
c
•55 15000-
10000-
5000-
0--
63 63
BS05 CS11 CS12 CS13 CS14 CS10 Bio
Emissions Scenario
point
Area
Mobile
Biogenic
NOx Emissions
USA Portion of the Domain
O
Q
CO
c
o
20000-
18000 -
16000 -
14000-
12000-
10000 -
8000-
6000-
4000-
2000-
0
57
45
57
BS05 CS11 CS12 CS13 CS14 CS10
Emissions Scenario
S3 Point $$$$$£& Area Y//////A Mobile
Values above bars: Percent reduction from 2005
Figure 5-4. Anthropogenic VOC emissions and NOX emissions for the US portion of the ROMNET
region.
5-44
-------
Percent Reduction in Anthropogenic VOC Emissions: CS12 vs. BSO5
Percent Reduction:
— 0 B;B:S;;; <— 3O amwa <•« 45
<- 6O •• <- 75 ^M > 75
Percent Reduction in Total VOC Emissions: CS12 vs. BSO5
Percent Reduction:
•« 0
<-6O
<— 3O
<- 75
<•» 45
> 75
Figure 5-5. Percent reduction in VOC emissions between CS12 and the 2005 baseline for anthropo-
genic emissions only and anthropogenic plus biogenic emissions.
5-45
-------
Percent Reduction in Anthropogenic NOX Emissions: CS11 vs. BS05
Percent Reduction:
= 0 mamum <= 3O niiiiiimin <= 45
<= 60 BEBB<= 75 •• > 75
Figure 5-6. Percent reduction in NOX emissions between CS11 and the 2005 baseline.
5-46
-------
Episode Maximum Ozone: BSO5
July 2, 1988 - July 17. 1988
Concentration (ppb):
< 1OO
>- 15O
>— 1 OO junmma >«• 125
>- 175 ••>- 2OO
Episode Maximum Ozone: CS11
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 100
>- 15O
=3 >— 1OO
" 175
>- 125
>- 200
Figure 5-7. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988 epi-
sode: 2005 baseline and CS11.
5-47
-------
Episode Maximum Ozone: BSO5
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 1OO
>- 15O
>- 1OO
>- 175
>- 125
>- 2OO
Episode Maximum Ozone: CS12
July 2. 1988 - July 17, 1988
Concentration (ppb):
< 10O
>- 15O
>- 100
>• 175
>- 125
>- 2001
Figure 5-8. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988 epi-
sode: 2005 baseline and CS12.
5-48
-------
Episode Maximum Ozone: CS13
July 2, 1988 - July 17. 1988
Concentration (ppb):
< 1OO
>- 15O
>- 1OO
>- 175
>- 125
- 2OOl
Episode Maximum Ozone: CS14
July 2. 1988 - July 17. 1988
Concentration (ppb):
< 100
>- 150
>— 100 mum >— 125
>- 175 ^•>- 2OO
Figure 5-9. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988 epi-
sode: CS13, CS14, andCSlO. (Page 1 of 2)
5-49
-------
Episode Maximum Ozone: CS10
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 1 00 Ml::*'.!* > = 1 00
>= 150 mam>= 175
>= 125
>= 200
Figure 5-9 (Page 2 of 2)
5-50
-------
Quantile—Quantile Frequency Distributions
1 —Hr Daily Maximum Ozone (ppb)
BALTIMORE & WASHINGTON, DC
200 •{
160-
o 120H
o
x
O
80-
40-
50
10
I
40
I
80
120 160
VOC Controls (CS12)
Episode Mean 8—Hr Daily Maximum Ozone (ppb)
BALTIMORE &. WASHINGTON, DC
100-I
80-
_co
o
o 60
o
x
O
o
40-
20-
90
10
i
20
I
40
I
60
80
VOC Controls (CS12)
200
.100
Figure 5-10. Quantile-quantile frequency distributions of 1 -hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for Baltimore/Washington, DC and vicinity.
5-51
-------
Quantile—Quantile Frequency Distributions
1 —Hr Daily Maximum Ozone (ppb)
NEW YORK CITY
200-
160-
§120-1
o
x
O
O
80-
40-
40
i
80
120 160
VOC Controls (CS12)
Episode Mean 8—Hr Daily Maximum Ozone (ppb)
NEW YORK CITY
100-
80-
o 60 H
40-
20-
90
i
20
i ' ' ' i ' ' ' i
40 60 80
VOC Controls (CS12)
200
100
Figure 5-11. Quantile-quantile frequency distributions of 1-hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for New York City.
5-52
-------
Quantile—Quantile Frequency Distributions
1—Hr Daily Maximum Ozone (ppb)
GREATER CONNECTICUT
200 -\
8160-1
•—s
VI
o
o 120
o
X
O
O
80-
40-
i
40
i
80
120 160
VOC Controls (CS12)
Episode Mean 8—Hr Daily Maximum Ozone (ppb)
GREATER CONNECTICUT
100-j
80-
V)
o
~c
O
O
x
O
•z.
o
60-
40-
20-
10
r
20
i
40
i
60
i
80
VOC Controls (CS12)
200
100
Figure 5-12. Quantile-quantile frequency distributions of 1-hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for Greater Connecticut.
5-53
-------
Maximum 1 -Hr Ozone (ROM 2.1) for July 1988 Episode
GREATER NEW YORK
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16
Day of Month
CS12 > CS11
CS11 >CS12
Maximum 1-Hr Ozone (ROM 2.1) for July 1988 Episode
GREATER CONNECTICUT
330-
300-
270-
240-
210-
180-
150-
120--
90-
60-
30-
|1|I[1|I|--I|I|I[I|T] T p -I | 1 ( I p I [ I J I 1
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
Day of Month
CS12 > CS11
CS11 > CS12
Figure 5-13. Diurnal time series of maximum hourly ozone concentrations in New York City and Greater
Connecticut for CS11 andCSlZ
5-54
-------
Maximum 1-Hr Ozone (ROM 2.1) for July 1988 Episode
BALTIMORE <5c WASHINGTON, DC
.o
a
a.
c
o
200-
180-
160-
140-
120--
100
80-I
60
40
20-I
i ' i ' i ' i ' I ' i ' i 'i ' i ' i ' i ' i 'i ' i ' I ' i ' i
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
Day of Month
CS12 > CS11
CS11 >CS12
Maximum 1-Hr Ozone (ROM 2.1) for July 1988 Episode
PITTSBURGH
_a
a.
o.
O
200-
180-
160-
140-
120--
100-
80
60-
40-
20-
| i | i | i | | i | i | i | | i | i | r |r— | i j i | -i | i p
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
Day of Month
CS12>CS11
CS11 > CS12
Figure 5-14. Diurnal time series of maximum hourly ozone concentrations in Baltimore/Washington,
DC and vicinity and Pittsburgh and vicinity for CS11 and CS12.
5-55
-------
July 9, 1966 Predicted Daily Maximum Ozone: 20O5 Baseline
Concentration (ppb):
< 1OO
>- 14O
3SS3 >- 1OO
^» >- 16O
>- 125
>- 180
Change in Daily Maximum Ozone: CS11 vs. 2005 Baseline
Concentration (ppb):
< -10
+ 3 to -1-9
-s — 9 to —3 aramsmi —2 to -t-2
> 10
Figure 5-15. Predicted daily maximum ozone concentrations across the New York City area for July 9,
1988: 2005 baseline and the change in ozone following the application of NOX controls in
CS11.
5-56
-------
July 9, 1988 0500-1100 EST Total VOC Emissions
Emissions (tons):
— O
<- 15
<— 3
<- 4Q
<- 1O
> 40
July 9, 1988 05OO-1100 EST Total NOX Emissions
Emissions (tons):
Figure 5-16. Morning (0500 -1100 EST) total VOC emissions and NOX emissions (tons) on July 9,1988
in the vicinity of New York City.
5-57
-------
Population Exposure to Ozone > 100 PPB
MASSACHUSETTS & COASTAL NEW ENGLAND
CS12 CS13
Scenario
CS14
Values above bars: Percent reduction from 2005
CS10
Figure 5-17. Population exposure to 1 -hour ozone > 100 ppb in Massachusetts and coastal New
England for selected scenarios.
5-58
-------
50-1
o 45
E 40
x
in 35
i_
"f 30
Q.
Population Exposure to Ozone > 125 PPB
BALTIMORE, WASHINGTON, DC & VICINITY
76
99
100
2005
CS11
CS12 CS13
Scenario
CS14
CS10
Population Exposure to Ozone > 125 PPB
PHILADELPHIA & VICINITY
100
94
100
2005
CS11 CS12 CS13
Scenario
CS14
CS10
Values above bars: Percent reduction from 2005
Figure 5-18. Population exposure to 1-hour ozone > 125 ppb in Baltimore/Washington, DC and Phila-
delphia for selected scenarios.
5-59
-------
500 H
Population Exposure to Ozone > 125 PPB
NEW YORK CITY
79
92
87
92
2005 CS11 CS12 CS13 CS14
Scenario
Population Exposure to Ozone > 125 PPB
GREATER CONNECTICUT & RHODE ISLAND
CS10
95
98
99
95
2005 CS11
CS12 CS13 CS14
Scenario
CS10
Values above bars: Percent reduction from 2005
Figure 5-19. Population exposure-to 1-hour ozone > 125 ppb in New York City and Greater Connec-
ticut/Rhode Island for selected scenarios.
5-60
-------
50-
J 45
1 40-
x
vi 35
I .50
Q.
£25
Population Exposure to Ozone > 100 PPB
PITTSBURGH & CHARLESTON
100
62
100
78
80
2005
CS11 CS12 CS13
Scenario
CS14
CS10
Population Exposure to Ozone > 100 PPB
CLEVELAND & DETROIT
CS12 CS13
Scenario
CS14
CS10
Values above bars: Percent reduction from 2005
Figure 5-20. Population exposure to 1-hour ozone > 125 ppb in Cleveland and Detroit (combined) and
Pittsburgh and Charleston, WV (combined) for selected scenarios.
5-61
-------
Cleveland L, Pittsburgh
Figure 5-21. Location of the boundary to the Northeast Corridor used in quantifying incoming ozone
transport.
5-62
-------
July 9, 1988
July 10, 1988
Figure 5-22. Layer 2 forward trajectories starting at 1500 EST from locations along the boundary to the
Northeast Corridor for July 9, 1988 and July 10, 1988. Trajectory markers are at 4-h
intervals.
5-63
-------
New York Sector of Corridor Boundary
100
S; 80
60
40
7/06 7/07 7/08 7/09 7/107/117/127/137/14
Days
Virginia Sector of Corridor Boundary
120
1OO
80
60
40
20
_l i i i
7/06 7/077/087/09 7/107/11 7/127/137/14
Days
Pennsylvania Sector of Corridor Boundary
120
100
o 60
40
20 L-i
7/06 7/07 7/08 7/09 7/10 7/117/127/137/14
Days
precontrol peak
postcontrol peak
precontrol mean
postcontrol mean
Figure 5-23. Peak and mean ozone concentrations by day within the three Northeast Corridor bound-
ary segments.
5-64
-------
Episode Maximum Ozone: BSO5
July 2, 1988 - July 17, 1988
Concentration (ppb): ' ' < 1OO ssssss >— 1OO
mmm>- 1 SO ••• >- 175
>- 125
>- 200
Episode Maximum Ozone: CS25
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 100
>• 150
>— 100
>- 175
Figure 5-24. Predicted episode maximum 1-hour ozone concentrations for the July 1988 episode:
2005 baseline and CS25.
5-65
-------
Episode Maximum Ozone: CS19
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 1OO
>- 15O
>- 1OO
>- 175
>- 125
>- ZOO
Episode Maximum Ozone: CS24
July 2. 1988 - July 17, 1988
Concentration (ppb):
< 100
>- 15O
>— 1 OO
>- 175
>- 125
>- 200
Figure 5-25. Predicted episode maximum 1-hour ozone concentrations for the July 1988 episode:
CS19andCS24.
5-66
-------
* Corridor Boundary
Figure 5-26. ROM layer 2 trajectory for the transport case study (trajectory markers are at 4-h intervals).
5-67
-------
ISO -
170 -
160 -
150 -
140 -
130 -
120 -
^ 110 -
J '00-
V 90 -
I 80-
70 -
60 -
SO -
40 -
30 -
20 -
10 -
O -
OZONE CONCENTRATIONS
• CorrMor Boundary
r....|.....TT1T..| ...|........... | | .....| .....I|II«II««I|II«I».I..|...........|.....*.....|!..........|
12:OO JulOB 12JOO JurOT 12:OO JulOa 12:OO JulOS 12lOO JuMO 12:OO Jultl
SCENARIO
BSOS
CS25
UO -
120
110
1O°
•«>
ao
I
Q£
50-
4O
30
20
10
O
NOX CONCENTRATIONS
2 f
i S
- 0
• Corridor Boundary
JulOS 12:00 JuK38 I2:OO JulO7 12:OO
12:00 JulOS 12:OO JuMO 12:OO JuMI
Figure 5-27. Time history of layer 2 ozone, NOX, and ROG concentrations: 2005 baseline and CS25.
5-68
-------
180
170
160
150
140
130
120
1 10
1°°
9° •
8°
70
60
50
40
30
20
10
O
OZONE CONCENTRATIONS
• Corridor Boundary
I2IOO JulO* 12KJO JulO7 I2:OO JulOI 12:OO JuK>» tliOO Julio 12iOO Julll
SCENARIO
CS 1 9
CS24
130
120 :
110 -
100 -
-§. 90-
>3 80-
o 70
K eo -
so-
40 •
3O •
20
10 -
0
NOX CONCENTRATIONS
N
*;\
ROG CONCENTRATIONS
C»
- o
• Corridor Boundary
12:OO JuKDS 12:OO Jul07 12:OO Ju«J8 12:00 JulOS 12:OO JuMO 12:OO JuM 1
Figure 5-28. Time history of layer 2 ozone, NOX, and ROG concentrations: CS19 and CS24.
5-69
-------
to
o
o
CM
100
*° 80
$_, 60
C 40
o
'35 20
|.
UJ
MOO
> 80
!£, 60
s«
o
e °
UJ
NEW YORK CITY
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
O
N
•320 §
to
-280 _
•240 §
r>
•200 2
-160 -1
•120 J±
O
3
•c
•o
cr
BS8S BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS15 CS18 CS19 CS23
Scenario
NOx I VOC I
BALTIMORE & WASHINGTON, DC
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
O
N
180 °
fO
O
160 O
O
hi 40 -
*•)
0)
rt
120 3'
3
100?
cr
BS85 BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS15 CS18 CS19 CS23
Scenario
NOx | VOC I
Figure 5-29. Emissions for Phase II scenarios, and predicted highest and second-highest daily maxi-
mum ozone concentrations (ppb) for selected urban areas. (Page 1 of 4)
5-70
-------
PHILADELPHIA
in
o
o
80
,60
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
O
N
180
o
•160 O
n
CD
•140 5.
•120
•100 ,
B
UJ
BS8S BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
NOx | VOC I
in
o
MOO
' 80
. 60
40
BOSTON
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
O
N
180 °
CD
O
•160 O
O
CD
•140 5.
•120
•100 .
UJ
BS8S BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
NOx I VOC I
Figure 5-29 (page 2 of 4)
5-71
-------
PITTSBURGH
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
ID
O
O
CO
<_ioo
o80
2^60
C/5 40
2 20
"tf>
U) 0
— BS8S BSOS CSOS CS1 1 CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
u5 Scenario
NOx I VOC I
o
N
O
180 ro
O
•160 §
O
CD
h!40^.
"1
03
•'» |
•100O
2nd Highest Daily Max.
Ui
BS85 BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS15 CS18 CS19 CS23
Scenario
NOx I VOC
Figure 5-29 (page 3 of 4)
5-72
-------
8
o
80
"I 2°
I °
UJ
DETROIT
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
O
N
180 §
(0
O
160 O
n
eo
M40 5.
120
BS85 BSOS CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
NOx I VOC I
in
o
o
CM
100
80
. 60
40
CHARLESTON
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
o
N
180 §
n>
O
hi 60 o
n
to
140 5.
•120 o
3
hioog
cr
UJ
BS85 BSOS CSOS CSM CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
NOx I VOC I
Figure 5-29 (page 4 of 4)
5-73
-------
Episode Maximum Ozone: BSOG
June 8, 1983 - June 20, 1983
Concentration (ppb):
< 1OO
>- 15O
>- 10O
>- 175
>- 125
>- 2OO
Episode Maximum Ozone: CS19
June 8, 1983 - June 20, 1983
Concentration (ppb):
< 100
>- 150
>- 100
>• 175
>- 125
200
Figure 5-30. Predicted episode maximum 1 -hour ozone concentrations (ppb) for the June 1983 epi-
sode: 2005 baseline and CS19.
5-74
-------
Episode Maximum Ozone: BSS5 - Phase I
July 2, 1988 - July 17. 1988
Concentration (ppb):
< 1OO sssiscs >— 1OO IIDOBH >— 125
>- 150 ••>-175 ••>-200
Episode Maximum Ozone: BS85 - Low Blogenic
July 2. 1988 - July 17, 1988
Concentration (ppb):
< 100
>- 15O
>— 100 mama >— 125
>- 175 ^M >- 200
Figure 5-31. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988 epi-
sode: Phase I 1985 base case with "best estimate" biogenics and 1985 base case with
"low" biogenics.
5-75
-------
Episode Maximum Ozone: BS85 - Phase I
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 1OO
>- 15O
>•• 1 OO siaminB >— 125
>- 175 ^» >- ZOO
Episode Maximum Ozone: BS85 - High Biogenic
July 2, 1988 - July 17, 1988
Concentration (ppb):
< 10O ssssss! >— 1OO mam >— 125
>- 15O ^•>- 175 ^^ >- 20O
Figure 5-32. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988 epi-
sode: Phase I 1985 base case with "best estimate" biogenics and 1985 base case with
"high" biogenics.
5-76
-------
200
§120
1 100
270
u
0
8210
£
I
J.18Q
5
2150
x
120
90
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
BOSTON
BS85 CS01 CS05 CS11 CS19
* HIGH6IOC8ICS 13 BESTESTMBIOGENICS A LOWBKJONICS
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
GREATER CONNECTICUT
BS85 CS01 CS05 CS11 CS19
* HKHBBGWCS 0 BESTESTWEBIOGENICS A LOWBTONKS
380
340
300
z
§260
a:
J.220
"140
100
200
,180
o
N
o
£140
§120
2
!100
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
GREATER NEW YORK OTY
BS85 CS01 CS05 CS11 CS19
4 HIOHBIOGENCS Q BESTE51WOIOGOKS A KMSOCCNICS
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
PHIWDEPHIA/KENT
BS85 CS01 CS05 CS11 CS19
* KHBKKENICS 0 BEST ESI Wit BBCOCS A LOWBIOGOCS
Figure 5-33. Predicted episode maximum 1 -hour ozone concentrations for biogenic sensitivity scenar-
ios for selected urban areas. (Page 1 of 2)
5-77
-------
300'
270
240-
n
M80
2
2150
x
<
Z 120
200
-180
£160
o
N
0
? 140
1 120
2
5 100
80
IMPACT OF BIOGOC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
BALTIMORE/WASHINGTON, DC
BS85 CS01 CS05 CS11 CS19
* HKHBIOCOICS D BEST ESTM BKXM3 A IOWHOCEMCS
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1-HR OZONE
PITTSBURGH/JOHNSTOWN/AITOONA
BS85 CS01 CS05 CS11 CS19
* HBHBKKOICS Q BESTESTDMTEBIOGMCS A LOWSOCOCS
200
J80
160
1140
I
§120
^100
200
Q.
Q.
1 160
o
N
0
£140
I
§120
2
X
2 100
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
CHARLESTON/HUNTINGTON
BS85 CS01 CS05 CS11 CS19
ft HIGHBOCENCS ED BEST ESTIMATE BIOGOCS A LOWBOGENCS
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1-HR OZONE
DETROIT
BS85 CS01 CS05 CS1I CS19
* HIGHBIOGOCS B BESTESMTEBIOGENKS A LOWHOGCNICS
Figure 5-33 (Page 2 of 2)
5-78
-------
500-
450-
* 350
ijJOO
o
§250
o
u.200
0
£150
ID
§100
z
50
2000'
1800
§ 1600
o
* 1400
u 1200
9 1000J
o
,800
£ 600
0
1 400
z
200-
0
GRID-HOUR OZONE EXCEEDENCES > 125 PPB
BOSTON
8S85 CS01 CS05 CS11 CS19
* HIGH BOGOIICS IH BEST ESTIMATE BKJGENICS A UOWBKJGfflKS
GRID-HOUR OZONE EXCEEDENCES > 125 PPB
GREATER CONNECTICUT
BS85 CS01 CS05 CS11 CS19
* MGHBOGQIICS El XSTESTIUATEBICaNICS A LOWBICCGCS
5000
4500
1/1
§4000
o
] J500
_i
u3000
o
92500
o
k2000
0
5 1500
CD
3 1000
z
500
1000
900
1/1
§ 800
o
j 700
u 600
u
9500
o
u. 400
o
5 300
m
§ 200
2
100
0
GRID-HOUR OZONE EXCEEDENCES > 125 PPB
GREATER NEW YORK CTTY
a
BS85 CS01 CS05 CS11 CS19
A HICHBBOMCS IE) BEST ESTOWI8BGEMCS A LOWBBCDCS
GRID-HOUR OZONE EXCEDENCES > 125 PPB
PHIUDEPHW/KENT
fl
BS85 CS01 CS05 CS11 CS19
ft HICHBOCENICS ED SESIESTimTEBKXffllCS A IOWBKXENICS
Figure 5-34. Predicted grid-hours exceeding 125 ppb for biogenic sensitivity scenarios for selected
urban areas. (Page 1 of 2)
5-79
-------
GRID-HOUR OZONE EXCEDENCES > 125 PPB
BALTINORE/V/ASHINGTON, DC
1200
1100
Jj 1000
0 900
i
1 onfi
_Jj 80U
jj 700
i 600
(K
" 500
° 400
u
3j 300
i 200
100
0
i
(
.
!
1
\
\
\
>
A
c
!
1 -^,
I
i i
;
1 ft «
GRID-HOUR OZONE EXCEDENCES > 125 PPB
CHARLESTON/HUNTINGTON
1
100
90
in
Ot on
D OU
0
i 7°
u 60
0
9 50-
a
0
b. 40
0
u 30
m
§ 20
z
10
0
A
|
(
.1
y
1
,^^~~~~-f
5w m n
U u H
BS85 CS01 CS05 CS11 CS19
i HIGH BOGMCS i BEST ESHMATE BKXENICS A LDW8KXIKICS
GRID-HOUR OZONE EXCEDENCES > 125 PPB
PTTTSBURGH/JOHNSTOWN/ALTOONA
BS85 CS01 CS05 CS11 CS19
ft HIGHBCGENKS ED BEST EST1MMI B«X»ICS A LOWBIOCENICS
GRID-HOUR OZONE EXCEDENCES > 125 PPB
DETROIT
500
450
V)
S;400
0
^350
y300
o
9250
a:
o
u.200
0
u 150
m
i 100
z
50
X
[
/
i
\.
\
i \
1
^^L
A O Q H
i i i i i
200
180'
in
^ 160
o
f 140-
dl20
o
9 100
n;
o
L 80
0
S 60
m
§40
z
20
i
[
/
.
i
1
\
\
i \
\
.
1 ~- — «L_^ » A
A u a n
i i i i i
BS85 CS01 CS05 CS11 CS19 BS85 CS01 CS05 CS11 CS19
ft HIGH BOGENCS i BCT ESTIMATE BKXZNICS A IDWBIOCfMCS ft KHBIOCENICS 01 BEST ESTIMATE BKXMICS A LOW BIOGDCS
Figure 5-34 (Page 2 of 2)
5-80
-------
t too
; 90
£
i 70
u
9j 60
8 50
Q.
S 40
o 30
§20
i 10
o
c
0
= 180
I60
140
u
K 100
in
? 80
60
POPULATION EXPOSURE TO I -HR MAXIMUM OZONE > 125 PPB
BOSTON
-fl-
BS85 CSOI CS05 CS11 CS19
* HKHBOGEMCS D BEST ESTIMATE BKXENICS A IWBOCENCS
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
GREATER CONNECTICUT
A
BS85
*
CSOI CS05 CS11 CS19
0 BESTESTMMtBIOCmCS A LOW90GENICS
= 1000
I 875
I 750
o
£ 625
IT
w 500
o
Q.
2 375
2 250
I 125
o
"• 0
^200
c
0
1180
E
x160
140
£120
U
5:100
in
x
z60
§«
220
o
1 0
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
GREATER NEW YORK CITY
BS85 CSOI CS05 CS11 CS19
4 HICHBOCOCS 0 BESTESTHWEBIOCENICS A LCWBIOGDICS
POPULATION EXPOSURE TO 1 -HR MAXIMUM OZONE > 125 PPB
PHILADEPHIA/KENT
Q
BS85 CS01 CS05 CS11 CSI9
4 HiGHencacs Q BESTESIMEBKJCEMCS A UMBIOGOCS
Figure 5-35. Predicted population exposure to ozone exceeding 125ppb for biogenic sensitivity
scenarios for selected urban areas. (Page 1 of 2)
5-81
-------
POPULATION EXPOSURE TO 1 -HR MAXIMUM OZONE > 1 25 PPB
BALTIMORE/WASHINGTON, DC
"?
•| 110
£ 100
X
« 90
r
1 fifl
' OU
a. 70
U)
K 60
8 50
0.
5 40
Z 7/1
o
^ ««
5 20
3
a. 10
o
n A
0
1
1
A
k
\
\
\
\
1
A
POPULATION EXPOSURE TO 1 -HR MAXIMUM OZONE > 125 PPB
CHARLESTON/HUNT1NGTON
t *
I
l-~>_
^^
A n a a
c
0
H 9
E
x 8
L.
r 7
I '
a
a 6
Jj
z 5
O 4
Q. *
X
3
z
0
P 2
D i
Q. '
0
n «
0
A
*
f
8 s 3 a s
BS85 CS01 CS05 CSI1 CSI9
i HCHBKXXHCS D BESTE51WTEBIOGOOCS A LOWBOGQCS
POPULATION EXPOSURE TO 1 -HR MAXIMUM OZONE > 125 PPB
PITTSBURGH/JOHNSTOWN/ALTOONA
BS85 CS01 CS05 CS11 CS19
ft HIGHBIOGENICS 0 BCTESIHW£BIOC£NCS A LOWBIOGCNICS
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
PITTSBURGH/JOHNSTOWN/ALTOONA
~50
"?
0
= 45
1
x40
a
f35
a
£30
y
^25
3
(/I
020
X
z
0
P 10
5
3 5
0
Q. n
U
1
1
.
*
\
\
\
A N.
I |
^^4— __ .
A u — u s
BS85 CS01 CS05 CS11 CS19
~50
c
0
= 45
£
x40
n
f35
a
g.30
UJ
^25
3
020
X
z
0
P10
3 5
0
Q. A
0
;
(
•
^
X.
\
"V
1^ t
^ __
A a a a
BS85 CS01 CS05 CS11 CS19
ft HCTBIOGOICS 0 BESTE5WE8JOGCNICS A LOWBOGENICS ft HIGHBIOGENICS 0 BESTESIIUATEBIOGENICS A LOWBOGENICS
Figure 5-35 (Page 2 of 2)
5-82
-------
TABLE 5-1. PROCEDURES USED FOR CALCULATING POPULATION EXPOSURE
T 71
where
Exp,* population exposure to ozone exceeding concentration x within area A for the entire epi-
sode; units are in population-hours
Pop/ population in grid / in area A
n n total grids in area A
t hour
T total hours in episode = (24 hours x number of days)
Oti = 1 if Xti > c, = 0 otherwise; occurrence of ozone exceeding concentration x in grid / for
hourr
c ozone concentration cutoff (i.e., 125 ppb)
5-83
-------
TABLE 5-2. SUMMARY OF EMISSIONS REDUCTIONS THAT REDUCED PREDICTED OZONE TO
< 125PPB
PEAK OZONE EMISSIONS REDUCTIONS
2005 POSTCONTROL BEYOND 2005 BASELINE
NORTHEAST CORRIDOR CITIES
Philadelphia
Boston
Bait/Wash., DC
New York City
Pittsburgh
Cleveland
Detroit
Charleston, WV
148ppb
1 58 ppb
149 ppb
268 ppb
CITIES
138 ppb
139 ppb
140 ppb
128 ppb
117 ppb
116 ppb
118 ppb
119 ppb
OUTSIDE THE
116 ppb
115 ppb
117 ppb
1 09 ppb
VOC
63%
60%
80%
85%
NORTHEAST CORRIDOR
46%
NC
NC
77%
NOX
48%
52%
15%
22%
30%
47%
60%
38%
CO
16%
18%
17%
21%
NC
NC
NC
7%
NC = no change
5-84
-------
TABLE 5-3. COMPARISON OF THE JULY 1988 AND JUNE 1983 EPISODE MAXIMUM 1-HOUR
CONCENTRATIONS (ppb) FOR THE 2005 BASELINE AND CS19
June 83 July 88
2005 CS19 2005 CS19
New York City 242 110 268 120
Baltimore/Washington, DC 160 124 149 122
Philadelphia 162 118 148 115
Boston 143 108 158 113
Pittsburgh 117 95 138 105
Cleveland 148 104 139 112
Detroit 135 99 140 109
Charleston, WV 95 69 128 95
5-85
-------
TABLE 5-4. SUMMARY OF OZONE METRICS USED TO ASSESS THE IMPACTS OF REACTIVITY-
BASED STRATEGIES
Average change in daily maximum ozone (ppb) for days > 125 ppb
2005 vs. CS20 2005 vs. CS1 2
New York City 69 80
Greater Connecticut 37 46
Rhode Island 22 28
Group 1 * 8 16
Group 2* 5 12
Group 3* 2 10
Number of days with ozone > 125 ppb
2005 CS20 CS12
New York City 12 8 6
Greater Connecticut 963
Rhode Island 7 1 0
Group 1 * 643
Group 2* 220
Groups* 220
Percent reduction in population exposure to ozone >
2005 vs. CS20 2005 vs. CS12 2005
New York City 75 92
Greater Connecticut 69 95
Rhode Island 90 100
Group 1 * 46 79
Group 2 * 33 82
Group 3 * 28 92
CSIOvs. CS15
CS10
10
4
0
0
0
0
125 ppb
vs. CS10
79
95
100
100
100
100
36
13
5
1
<1
<1
CS15
8
1
0
0
0
0
2005 vs. CS 15
93
99
100
100
100
100
1 Average for group
5-86
-------
Case
TABLE 5-5. CASES TESTED WITH VARYING BIOGENIC EMISSIONS
Description
Base 1985: Base Case 1985 emissions; "best" estimate biogenics
Base 1985L: Base Case 1985 emissions; "low" estimate biogenics
Base 1985H: Base Case 1985 emissions; "high" estimate biogenics
CS01: Strategy 1 - maximum technology VOC controls applied regionwide; "best" estimate
biogenics
CS07: Strategy 1 with "low" estimate biogenics
CS09: Strategy 1 with "high" estimate biogenics
CS05: Strategy 5 - Clean Air Act legislation controls; 'best' estimate biogenics
CS06: Strategy 5 with "low" estimate biogenics
CS08: Strategy 5 with "high" estimate biogenics
CS11: Maximum NOX controls from Strategy 10, VOC at 2005 Base Levels; "best" estimate
biogenics
CS17: Strategy 11 with "low" estimate biogenics
CS19: Maximum technology, reduced reactivity, across the board reductions and modified
NOX controls in Baltimore/Washington DC; "best" estimate biogenics
CS21: Strategy 19 with "low" estimate biogenics
CS20: Strategy 19 with "high" estimate biogenics
5-87
-------
This page is intentionally left blank.
-------
SECTION 6
STATE ACCESS AND USE
OF
ROM DATABASES
by
Ruen-Tai Tang
Computer Sciences Corporation
P.O. Box12767
Research Triangle Park, NC 27709
Ellen Baldridge
Technical Support Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
and
James A. Godowitch*
Atmospheric Sciences Modeling Division
Atmospheric Research and Exposure Assessment Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
* On assignment from the National Oceanic and Atmospheric Administration
U.S. Department of Commerce
-------
This page is intentionally left blank.
-------
6.1 INTRODUCTION
As noted in Section 2, performance of the ROM indicated that the model predicts observed regional
ozone levels reasonably well. Use of the model to estimate effects of regional control strategies on
ozone predictions was demonstrated in Section 5. There are a number of reasons why the ROM would
be advantageous in the development of required NAAQS attainment demonstrations. First, it provides a
means for ensuring consistency in the way States derive and use transport assumptions in their design
of urban strategies. Second, it provides a means of simulating the combined effects of many urban
strategies. Third, it provides the most technically-defensible way to estimate transport to urban areas of
future-year ozone/precursor concentrations. Finally, it can provide data for use in State Implementation
Plans (SIP's) that may be difficult and/or expensive to otherwise obtain.
For reasons discussed in Sections 2 and 5, it is inappropriate to use the ROM by itself to demonstrate
attainment of the ozone NAAQS in the vicinity of cities. The Urban Airshed Model (UAM) has been des-
ignated as the preferred approach for attainment demonstrations. Thus, the ROM is best used by "nes-
ting" the UAM within a ROM domain, and extracting data generated by the ROM for use as inputs to the
UAM.
Although the concept of nesting is simple, its implementation is not. For example, in using results from
the ROM to provide inputs for urban attainment demonstrations, States are likely to be interested in only
a small subset of these data. Episodes considered in attainment demonstrations will typically focus on
a one- to two-day period and concentrate on only a small portion of the ROM domain. Thus, a man-
ageable way is needed for States to retrieve data from applications of the ROM to support SIP develop-
ment.
A second problem arises from the fact that, like the ROM, the UAM is also a three-dimensional photo-
chemical grid model - and it is just as complex as the ROM. In addition, the UAM's grids are a different
size than those in the ROM, and there are more vertical layers in the UAM. Further, the UAM and its
numerous preprocessor programs require input data that is somewhat differently structured than the
data generated by the ROM.
Therefore, one of the goals of ROMNET was to provide the ROM predictions and ancillary input data sets
to States for use in urban modeling, along with methods and guidance on how to use these data for
incorporating pollutant transport into the development of ozone SIPs. Two computer systems were
developed as part of ROMNET to meet these objectives: the Gridded Model Information Support Sys-
tem (GMISS) and the ROM/UAM Interface Program System.
6-3
-------
GMISS is a database management system that provides an archive for ROM system databases and
access to these data by EPA and States. Databases associated with the ROMNET base case, future
baseline, and control strategy scenarios have been archived in GMISS. This system resides on the EPA
National Computer Center IBM 3090.
The ROM/UAM Interface Program System is a series of integrated Fortran programs that contain the
methodologies for using ROM system data to prescribe various inputs for the Urban Airshed Model
(UAM). Among the (JAM inputs provided are boundary conditions of pollutant concentrations for the
urban domain. Using ROM predictions to specify urban boundary conditions via the Interface System
will enable States to explicitly consider the effects of regional pollutant transport in urban scale model
applications.
This section of the report describes GMISS (Section 6.2) and the ROM/UAM Interface Program System
(Section 6.3). More information can be found in the user's guides for these systems as provided by
Dessent (1990) and Tang era/. (1990), respectively.
6-4
-------
6.2 THE GRIDDED MODEL INFORMATION SUPPORT SYSTEM (GMISS)
6.2.1 Overview of GMISS
As indicated above, the GMISS was designed to store and provide access to ROM system databases
needed to derive UAM inputs using the ROM/UAM Interface. The main functional components of GMISS
are contained in the UAM Subsystem. This subsystem includes loading ROM data into GMISS and
retrieval of these data by the user. Data loading functions are performed by an EPA designated Data-
base Administrator (DBA). Retrievals are made by the user through an interactive series of menus.
Access to these menus is documented in the GMISS User's Guide, (Dessent, 1990).
The next three sections provide additional information on the key user-oriented aspects of GMISS. The
contents and structure of the GMISS databases are described first, followed by the functional processes
of the UAM Subsystem. Finally, the user protocol for retrieving data is provided.
6.2.2 GMISS Databases
Concentration Data
As described in Section 2, the ROM generates concentrations of 35 chemical species at half-hour
intervals in three vertical layers for each grid cell of the modeling domain. GMISS reads ROM
Concentration Files created on the IBM-3090 mainframe computer and stores the data in GMISS
database files. In this process the half-hourly values are averaged to obtain hourly values and the data
are time-shifted from ROM's end-of-hour convention to start-of-hour ( i.e., the first hour of a day,
beginning at midnight, is hour 00; the last hour of a day, beginning at 11:00 pm, is hour 23). Each
database file holds up to one month's data for a particular species, layer, and scenario (e.g., base case,
strategy, etc). Each record in a database file has 24 hourly values for a particular date and grid cell (row
and column of the ROM grid). A separate Model Data Index File keeps track of the species, layer, model
scenario, episode dates, and names of each database file.
The Concentration Output File, which is produced by a user-initiated retrieval, always contains data in
whole-day units. There may be one or more (successive) days in the file, and data for all 24 hours of
each day and all three ROM layers are included. As described in Section 6.3, only 17 of the 35 chemical
species computed by ROM are required by the ROM/UAM Interface System. Thus, only the species
listed below are contained in the Concentration Output File.
6-5
-------
ALD2 Aldehydes (high molecular weight)
CO Carbon Monoxide
ETH Ethene
FORM Formaldehyde
H2O2 Hydrogen Peroxide
HNO2 Nitrous ACid (MONO)
HNO3 Nitric Acid
tSOP tsoprene
MTHL Methanol (MEOH)
NO Nitric Oxide
NO2 Nitrogen Dioxide
O3 Ozone
OLE Ofefins
PAN Peroxyacetyl Nitrate
PAR Paraffins
TOL Toluene
XYL Xylene ,
In the Concentration Output File the hourly values are temporally "smoothed" as required by the
ROM/UAM Interface System. Specifically, 3-hour moving averages are created for each hour by com-
puting the arithmetic means of the values for that hour, the preceding hour, and the following hour.
(There is no spatial smoothing; averages are computed separately for each grid cell in each layer.)
ROM Processor Data
As described in Section 6.3, various ROM input databases are also handled by the ROM/UAM Interface.
These inputs, as listed below, are contained in the ROM processor files described in Section 2.2:
Observed upper-air vertical profiles
Surface effective roughness
Land use (ten categories)
Terrain elevation
Surface meteorological data
Layer 2 wind fields
Layer 1 wind fields
Sky coverage (cloudiness)
Biogenic emissions
Elevation Of the boundary between layer 1 and layer 2
Elevation of the boundary between layer 2 and layer 3
Layer 1 water vapor concentration.
As with the ROM concentration file, the (JAM Subsystem reads these files and stores the data in the
database. User-requested processor data are provided in up to twelve retrieval files structured to be
compatible with the ROM/UAM Interface System. No time-smoothing is required for the ROM processor
data.
The UAM Subsystem keeps track of data in the database by using two index files, a model data index file
and a retrieval index file. The model data index file is used to establish the correspondence between the
ROM concentration data and processor data This index is updated when model concentration data or
ROM processor data are loaded into the database. The retrieval index file keeps track of data available
for retrieval in the UAM Subsystem.
6-6
-------
6.2.3 System Functional Description
There are three main processes in the UAM Subsystem: Data Load, Approval, and Retrieval. The Load
process has an interactive component and a batch component. The GMISS Database Administrator
(DBA) uses interactive menus to initiate the Load process and specify the data to be entered into the
database. In the batch environment, the Load program reads each file of ROM processor or concen-
tration data specified by the DBA, loads the data into database files, and records the presence of the
data in Index files. Each file loaded is described in a Log file, and a report is printed confirming and
summarizing the transaction. The Index file serves as a directory to keep track of all data loaded into
GMISS. The Log file is designed to provide a traceable path of the transactions in each Load process.
Before the ROM concentration data can be released for public use, they must be reviewed by data
quality analysts. In the approval process the concentration data are assigned a "restricted" status when
they are initially loaded into GMISS. After the quality assurance review, the DBA changes the status to
"public," and the data become available for UAM Subsystem users. ROM processor data, on the other
hand, are reviewed extensively before they are used in ROM simulations. Therefore, the ROM processor
data are assigned a "public* status when they are loaded, automatically making them available for
retrieval.
Retrieval is the last of the three main processes of the UAM Subsystem. Through interactive menus,
users initiate the process and specify the data to retrieve. A batch job submitted during the interactive
session performs most of the "work" of retrieving data The batch job determines which database files
contain the requested ROM System data (concentration data and/or ROM processor data), extracts the
data, computes 3-hour moving averages of concentration data, and writes one or more retrieval files in
which the data are properly organized and formatted for input to the Interface. In addition, the batch job
prints a report summarizing the data retrieved, and creates an entry in the Retrieval Log file. This log
provides a record of users accessing the system, the types of data retrieved, names of output files, and
the date and time of the retrieval. This information is used to assist in resolving any user-access prob-
lems.
6.2.4 User Protocol for Data Retrieval
For UAM Subsystem data retrievals, users specify the selection criteria for ROM concentration data, and
GMISS automatically selects the associated ROM processor data
6-7
-------
Users choose the data to be retrieved by specifying the ROM domain (e.g., ROMNET), study (e.g.,
ROMNET), scenario (e.g., 1985 Base Case), and begin date and end date (e.g., Jul 10-Jul 15, 1988).
Users also specify the subregion, or spatial "window" of the ROM domain, that corresponds to the UAM
domain of interest. If a user enters valid and appropriate values for all of these fields, the UAM Subsys-
tem has sufficient information to retrieve the data from the database. However, a user may not know
exactly what values to specify for each field. Two kinds of menus allow users to inquire what values are
possible and choose among them. They are the List-of-Values menu and the Final Data Selection menu.
QMISS displays a List-of-Values menu when a user enters a question mark ("?") in the domain, study, or
scenario field. The question mark tells GMISS to display another menu that lists valid values for the field.
When a List-of-Values menu is displayed, a user may select one of the values on the menu. If the user
enters question marks in more than one field on the initial data selection menu, GMISS displays a List-
of-Values menu for each field.
The Final Data Selection menu finishes the data selection process. It is analogous to a List-of-Values
menu; however, it presents a scrollable list of the types of data in the database that match the specifi-
cations for the domain, study, scenario, and date, and which are available for retrieval. Users select the
particular data desired from the list of available data on the Final Data Selection menu. The data
retrieved through the UAM Subsystem must represent a single domain, study, and scenario of ROM
data, and must be for successive days. In any one retrieval file, multiple scenarios, studies, domains, or
gaps in the time sequence of the data are not allowed.
By default, the UAM Subsystem generates a complete set of data for input to the UAM: a retrieval file of
ROM concentration data and 12 retrieval files of ROM processor data. Since a single set (scenario) of
ROM processor data may be used to compute species concentrations for many different emissions
scenarios, users may not need to retrieve ROM processor data every time ROM concentration data are
retrieved. To facilitate this selective retrieval, the UAM Subsystem defines the five data categories below.
ROM concentration data is one such category; groups of the twelve ROM processor files comprise the
other four categories. The files in each category are likely to be used together by the UAM.
CONC Hourly gridded ROM concentration data for the 17 chemical species UAM requires, in each
of the three ROM layers;
RAWIN Hourly non-gridded upper-air vertical profiles (observed data);
GEO Time-invariant gridded geographic data (surface roughness, land use; and terrain eleva-
tion);
BIOGEN Hourly gridded biogenic emissions;
MET Hourly gridded meteorology data (surface meteorology, layer 2 wind fields, layer 1 wind
fields, cloudiness,.height of layer 1,. height of layer 2, and layer 1 water vapor concentration).
6-8
-------
The data type selection menu allows users to choose the categories of data for which the UAM Subsys-
tem will create retrieval files. The default is that all five categories are selected. Users must choose at
least one category.
The geographic extent of the ROM data available for retrieval is much larger than the typical domain
used for urban modeling with the UAM. As indicated previously, GMISS users retrieving any category of
gridded data for use with UAM can choose a smaller rectangular array of ROM grid cells that covers only
the geographic area to be modeled by the UAM. This geographic "windowing" is based on user-entered
longitude and latitude coordinates of opposite corners of the desired UAM domain.
For selecting observed upper-air data, the UAM Subsystem provides a menu list of available stations
based upon the selection criteria specified in the initial data selection menus. The user selects the sta-
tion or stations from this menu. At least one station must be selected.
Execution of the UAM Subsystem results in the creation of up to 13 ROM System files (1 ROM output
concentration file and 12 ROM processor input data files). These are the files required to drive the
ROM/UAM Interface Program System. At this point, users may move these files to the desired computer
and set up the ROM/UAM Interface Program System for execution.
6-9
-------
6.3 THE REGIONAL OXIDANT MODEL - URBAN AIRSHED MODEL
INTERFACE PROGRAM SYSTEM
6.3.1 Introduction
One of the major objectives of ROMNET was to provide procedures for using ROM predictions to specify
boundary conditions for urban-scale models. In the process of developing such procedures, it became
evident that additional urban model inputs could be derived from ROM system data sets. Thus, the
scope of the task was expanded accordingly. It was felt that the additional information would reduce the
burden on States of gathering input data from other sources. Also, it would ensure greater consistency
between urban and regional modeling as well as between State Implementation Plans (SIPs) from States
that rely on the ROM data sets.
The Urban Airshed Model (UAM) was selected as the urban-scale model to interface with ROM, because
the UAM is the preferred modeling approach for urban ozone modeling (EPA, 1986). Although both the
ROM and the UAM are grid models using a common chemical mechanism, interfacing these two models
is not straightforward. A detailed discussion of the methodologies and assumptions used in the inter-
face is contained in Appendix N.
A set of computer programs was developed to enable the UAM system to be driven by ROM system
input and output data sets. The computer program package has been designated the ROM-UAM Inter-
face Program System. Since both the ROM and UAM are independent models with their own separate
processor systems, the interface programs serve as external links between particular ROM files and
components of the UAM system. The principal functions of the interface are to interpolate specific
gridded parameters from the ROM system, and to generate data files in compatible formats for input to
the UAM preprocessors or the UAM program. In particular, the interface programs have been designed
for use with the ROM 2.1 system files and the UAM Carbon Bond IV [UAM (CB-IV)] system.
Like the ROM, the UAM is a grid-based photochemical oxidant model that mathematically treats the
relevant physical and chemical processes important to ozone production, destruction, and removal,
albeit on a smaller spatial scale. Both models require extensive input data including concentrations for
initial and boundary conditions, and meteorological and emissions data. In all, 13 different ROM data
files are applied in the interface programs, which include gridded fields of pollutant concentrations, var-
ious meteorological parameters (e.g., winds, temperature, water vapor, etc.), land use information, and
biogenic emissions.
6-10
-------
The ROM-UAM interface system actually consists of seven main programs. Each interface program
provides a link between certain ROM data files and a particular UAM preprocessor or the model pro-
gram. Specifically, an output file generated by an interface is in a compatible format for direct input to
either a UAM preprocessor or the model.
The remainder of this section provides an overview of the interface package, including a summary of
some of the salient features and important limitations when applying the interface programs, and an
overview of the interface package and its programs. A detailed technical description of the methods and
procedures employed to derive the time dependent and spatially-varying gridded data fields for the
UAM system from the ROM data sets is given in Appendix N. The detailed format specifications of the
input/output files for each interface and instructions for their execution are given elsewhere (Tang et a/.,
1990).
6.3.2 Features and Limitations of the Interface
The principal features of the ROM/UAM Interface are enumerated below:
Interfaces link only components of the UAM (CB-IV) model system and certain outputs of the
ROM 2.1 system, and not earlier versions of these two models.
Four interfaces provide formatted input files for the UAM preprocessors and three interfaces
generate binary files directly for use in executing the UAM model.
Thirteen different ROM output files are applied in executing the various interface programs.
Users must create these spatially "windowed" data files for their particular UAM domain and
simulation period by accessing the ROM database through a data retrieval program devel-
oped for the Gridded Model Information Support System (GMISS) on the EPA NCC-IBM 3090
computer system.
Interface programs have been generalized to allow the user to input specifications about the
particular UAM domain and grid (i.e., origin, number of grid cells, and grid cell size) and the
vertical configuration of the UAM (i.e., number of lower and upper levels).
Interfacing for initial, lateral and top boundary conditions is performed for 17 of the 23 pollut-
ant species that must be specified in the UAM (CB-IV) model. Default values are defined for
the remaining six species.
6-11
-------
Lateral and top boundary concentrations are resolved hourly and spatially at each UAM grid
cell.
Options have been built into two interface programs to allow the user to apply non-ROM input
data files (i.e., concentrations, winds).
Quality-assured data may be incorporated into formatted packet files through an on-line edi-
tor at a terminal.
. A utility program included with the interface package converts any binary file into an equiva-
lent ASCII formatted data file so that its contents may be examined prior to use in model
execution.
Horizontal wind fields generated by the wind interface are determined by matching the ROM
gridded wind components from layers 1 and 2 into the vertical levels of the UAM. The meth-
odologies employed in the wind interface are those applied in the UAM Diagnostic Wind
Model system described in Volume III of the UAM User's Guide (Douglas et al. 1990).
Certain limitations also exist with this version of the interface:
No interfacing is performed for the diffusion break height needed by the UAM. The user must
apply a method recommended for deriving hourly values of this parameter as described in
Volume II of the UAM User's Guide (Morris era/. 1990b). As part of the interface package, a
processor program (PDFSNBK) produces a formatted packet file for the UAM diffusion break
preprocessor program (DFSNBK) using a user-supplied data file of diffusion break values.
The interface programs handle only a spatially-invariant diffusion break for each hour (note
that diffusion breaks can vary from hour to hour).
No interfacing of anthropogenic area, mobile, or point source emissions is performed. The
UAM point source preprocessor (PTSRCE) and the Emissions Processor System (EPS)
already exist for creating emissions for these source types (Volume IV, Causley ef. al. 1990). A
biogenics interface is limited to combining an existing binary anthropogenic area emissions
inventory file and the ROM gridded biogenic emissions of certain hydrocarbon and NOX spe-
cies.
No interfacing is performed for the UAM preprocessors, CPREP or SPREP, the chemical
parameters and the simulation control programs, respectively.
6-12
-------
6.3.3 Description of the Interface System
A general framework showing how the interface program package fits into the overall UAM model sys-
tem is depicted in Figure 6-1. The interface programs are executed before exercising any UAM prepro-
cessor program. The important first step to be performed in the interfacing process is the creation of a
set of formatted data files that are needed as inputs to the interface programs. This step is performed
using GMISS, as described in Section 6.2.
Since the retrieved data files will be formatted, the user may decide to transfer them to another computer
system where the actual execution of the interface programs, and the UAM preprocessor and model
simulations will be conducted.
The set of interface programs produce either formatted 'packet' files that are in a compatible format for
direct input to particular UAM preprocessors or binary files that are ready for direct input to the UAM
model. For the latter, certain UAM preprocessors are bypassed with no execution required. Therefore,
although the interface programs reside on the IBM 3090 computer, these codes are easily adaptable to
other computer systems after minor revisions Tang et a/., 1990). The user has the ability to examine the
contents of "packet" files generated by most interface programs with an on-line editor. Furthermore, at
this stage the user may wish to supplement these files with additional quality-assured observed data
before proceeding with the execution of a UAM preprocessor. For the interface programs that generate
binary files ready for the model, an optional feature has been built into the codes that allow the user the
flexibility of supplying an alternate data set.
In this instance, it must be noted that the user would be undertaking an additional effort to generate an
alternate data file for an interface program. A utility program is also part of the interface package for
converting a binary file into a formatted ASCII file version so that it can be examined before model
execution.
The UAM model code requires 13 binary input files in any modeling application. Table 6-1 gives a com-
plete list of the UAM preprocessor names and the binary output file generated by each program. The
UAM preprocessor programs have been grouped into categories to indicate which binary data files are
required for initial and boundary (lateral and top) concentrations, various meteorological parameters
(e.g., winds, etc) and surface features, emissions, and finally, control information about the chemical
species and reactions, and the model run parameters.
6-13
-------
A comprehensive approach was taken in interfacing the two model systems in order to take advantage
of the extensive variety of data sets available from the ROM model and processor network for use as
inputs to as many UAM preprocessors as possible. A complete list of the data files and their contents
that are used in the interface is provided in Table 6-2. The file names listed are those designated within
the ROM system and these file names will be imbedded within the extended file name of the retrieved
files generated by the user. One or more of these files are needed to exercise each interface program.
With the exception of the temperature profile data file, all of the retrieved files contain gridded fields of
the ROM processor or model results.
An overview of the complete ROM-UAM interface program package is provided in Table 6-3. It shows
the retrieved ROM files needed for each interface program. There are seven actual interface programs.
The designated name for an interface program begins with the letter "I" and the remainder of its name is
given by the UAM preprocessor program name. In addition, a processor program (PDFSNBK) that
reformats hourly diffusion break data (DBDATA), supplied by the user, is part of the interface program
package. The outputs from the other interface programs are formatted "packet" files (e.g., RTPACK, etc.)
for use as inputs to the appropriate UAM preprocessor. Since these files are formatted, the user has the
capability to examine their contents and also can insert additional data before exercising the UAM pre-
processors. The interface programs for generating winds (IWIND) and for initial and boundary concen-
trations (ICONC) produce binary files for use in UAM execution. Consequently, the diagnostic wind
model (DWM) system and the concentration preprocessors (i.e., AIRQUL, BNDARY, TPCONC) are not
exercised when these interface programs are applied. A combined area emissions file is created by the
IBIOG interface program from a user-supplied anthropogenic file and a ROM biogenic emissions file.
The output emissions file of IBIOG is also binary and ready for input in UAM simulations. However,
interfacing of anthropogenic point or area emissions is not performed. A detailed description of the
interface method for each UAM processor is given in Appendix N.
6-14
-------
ROM CONC/PF/MF DATABASE
i £
User Inputs
GMISS DATA RETRIEVAL PROGRAM
(EPA-NCC IBM 3090)
"RETRIEVED" DATA FILES
(Formatted/Transferable)
INTERFACE PROGRAMS
FORMATTED "PACKET" FILES
UAM PREPROCESSOR PROGRAMS
BINARY FILES
UAM SIMULATION PROGRAM
Figure 6-1. Flow diagram showing the data retrieval and interface processing steps performed to
generate data files for the UAM preprocessors and model.
6-15
-------
TABLE 6-1 . SUMMARY OF THE (JAM PREPROCESSOR PROGRAMS
Name
AIRQUL
BNDARY
TPCONC
DFSNBK
REGNTP
METSCL
TMPRTR
DWM1
CRETER
EPS2
PTSRCE
CPREP
SPREP
Internal name in
binary file
AIRQUALITY
BOUNDARY
TOPCONC
METEOROLOGY
DIFFBREAK
REGIONTOP
METSCALARS
TEMPERATURE
WIND
TERRAIN
EMISSIONS
PTSOURCE
CHEMPARAM
SIMCONTROL
Contents
CONCENTRATIONS
Initial concentration fields
Lateral boundary concentrations
Top boundary concentrations
AND SURFACE CHARACTERISTICS
Diffusion break heights
Model region top heights
Five meteorological parameters and photolysis rate
Surface temperature field
Horizontal wind component fields
Surface roughness and vegetation fraction factor
EMISSIONS
Anthropogenic area emissions
Major point source emissions
CONTROL DATA
Species reaction rate information
Model simulation input information
1. DWM = Diagnostic Wind Model
2. EPS = Emissions Preprocessor System
TABLE 6-2. RETRIEVED ROM FILES USED BY INTERFACE PROGRAMS
ROM system file
name
Description of file contents
ROM21 Hourly predicted concentrations of 17 species from layer 1, 2, and 3
MF165 Hourly gridded heights of layer 1
MF166 Hourly gridded heights of layer 2
MF174 Hourly gridded water vapor concentration in layer 1
PF102 Hourly vertical-interpolated temperature profiles
PF103 Hourly gridded surface air temperature
PF108 Gridded surface roughness length
PF114 Hourly gridded layer 2 horizontal wind components
PF115 Hourly gridded layer 1 horizontal wind components
PF117 Hourly gridded total sky cover fraction
PF118 Gridded fractions of eleven land use categories
PF119 Gridded terrain elevation
PF144 Hourly gridded biogenic emissions of six species
6-16
-------
TABLE 6-3. OVERVIEW OF THE ROM-UAM INTERFACE PROGRAMS
AND INPUT/OUTPUT FILES
ROM or other Interface 'PACKET1 UAM
input f i les programs file preprocessor
DBDATA
PF119 I > RTDATA
PF102 User data
PF117 (optional)
PF119
RTDATA
OBDATA
User data
(optional)
MF165
PF115
PF119
DBDATA
RTDATA
PF118
ROM21
PF119
DBDATA
User EMISSIONS*
I — > BIOASC (optional)
Binary
file
ftRRTM
L/DD 1 n
DTD TU
i\ i D * n
MCRTU
noD L n
TPRTM
1 r O i PI
| AQBIN
>I BCBIN
| TCBIN
* Anthropogenic area emissions file
6-17
-------
This page is intentionally left blank.
-------
REFERENCES AND BIBLIOGRAPHY
-------
This page is intentionally left blank.
-------
Alliance Technologies Corp., 1989. Regional Ozone Modeling For Northeast Transport - Development of
a Base Year Anthropogenic Emissions Inventory. EPA-450/4-89-008. Office of Air Quality Planning
and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711.
Alliance Technologies Corp., 1990. Regional Ozone Modeling for Northeast Transport - Projection Year
and Control strategy Emissions Inventories. Draft Report, U.S. Environmental Protection Agency,
Technical Support Division, Office of Air Quality Planning and Standards, Research Triangle Park, NC
27711.
Arnts, A. and S. Meeks, 1981. Biogenic hydrocarbon contribution to the ambient air in selected areas.
Atmos. Environ. (15) 9:1643.
Baldocchi, D., D. Matt, B. Hutchinson era/., 1984. Solar radiation within an oak-hickory forest: an evalu-
ation of the extinction coefficients for several radiation components during fully-leafed and leafless
periods. Agricultural and Forest Meteorology (32): 307.
Battye, W., M. Smith and M. Deese, 1987. Cost Assessment of Alternative National Ambient Air Quality
Standards for Ozone (Draft Report for External Review). Alliance Technologies Corporation. Prepared
for Ambient Standards Branch, U.S. Environmental Protection Agency, Research Triangle Park, NC.
BEA, 1986. 7985 OBERS BEA Regional Projections, Volume 1: Methodology, Concept, and State Data.
U.S. Department of Commerce, Bureau of Economic Analysis.
Causley, M.C., J.L Fieber, M. Jimenez, and L Gardner, 1990. User's Guide for the Urban Airshed Model.
Volume IV: User's Guide for the Emissions Processor System. EPA-450/4-90-007D. Prepared for the
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Chameides, W., R. Lindsay, J. Richardson et a/., 1988. The role of biogenic hydrocarbons in urban
photochemical smog: Atlanta as a case study. Science (241): 1473.
Chun, K. C, 1988. Uncertainty Database for Emissions Estimation Parameters: Area Source Supplement
to Interim Report. ANL/EES-TM-353, Argonne National Laboratory, Argonne, Illinois.
Computer Sciences Corporation (CSC), 1990. Gridded Model Information Support System (GMISS):
UAM subsystem design. EPA Office of Air Quality Planning and Standards, Source Receptor Analysis
Branch, Research Triangle Park, NC.
Demerjian, K.L, K.L Schere, and J.T. Peterson, 1980. Theoretical estimates of actinic (spherically inte-
grated) flux and photolytic rate constants of atmospheric species in the lower troposphere. In:
Advances in Environmental Science and Technology, Vol. 10, J. Pitts and R. Metcalf, Eds., Wiley Publ.,
New York. pp. 369-459.
Dessent, T., 1990. The GMISS UAM Subsystem User's Guide. In preparation. Prepared for the Office of
Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research Triangle Park,
NC.
Dodge, M.C., 1989. A comparison of three photochemical oxidant mechanisms. J. Geophys. Res.
94:5121-5136.
R-3
-------
Douglas, S.G., R.C. Kessler, and E.L Carr, 1990. User's Guide for the Urban Airshed Model. Volume III:
User's Manual for the Diagnostic Wind Model (Version 1.1). EPA-450/4-90-007C. Prepared for the
Office of Air Quality Planning and Standards, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Ellison, W., 1988. Assessment ofSO2 and NOxEmission Control Technology in Europe. EPA-600/2-88-
013, U.S. Environmental Protection Agency, Research Triangle Park, NC.
EPA, 1986. Guideline on Air Quality Models (Revised). EPA-450/2-78-027R. U.S. Environmental Pro-
tection Agency. Research Triangle Park, NC.
EPA, 1989. User's Guide to MOBILE4. EPA-AA-TEB-89-01, U.S. Environmental Protection Agency,
Office of Mobile Sources, Ann Arbor, Ml.
Faucett, J. and Associates, 1988. The Faucett VMT Forecasting Model (Draft Final Report). Office of
Policy Administration, Federal Highway Administration, U.S. Department of Transportation.
Fehsenfeld, F., 1990. National Oceanic and Atmospheric Administration, Boulder, CO, personal com-
munication.
Gates, D. and L Papian, 1971. Atlas of Energy Budgets on Plant Leaves, Academic Press, New York, pp.
1-16.
Gay, D., 1987. A National Inventory of Biogenic Hydrocarbon Emissions Based Upon a Simple Forest
Canopy Model. M.S. Thesis, Washington State University, 73 p.
Gery, M.W., G.Z. Whitten, and J.P. Killus, 1988. Development and testing of the CBM-IV for urban and
regional modeling. EPA-600/3-88/012, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Gery, M.W., G.Z. Whitten, J.P. Killus, and M.C. Dodge, 1989. A photochemical kinetics mechanism for
urban and regional scale computer models. J. Geophys. Res. 94:D10,12295-12956.
Godowitch, J.M., J.K.S. Ching, and J.F. Clarke, 1987. Spatial Variation of the Evolution and Structure of
the Urban Boundary Layer. Bound. Layer-Meteorol., 38:249-272.
Johnson, K., 1988. Industrial Boiler Low-NOx Combustion Retrofit Cost Algorithm. EPA-600/8-88-091,
National Acid Precipitation Assegsment Program, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Killus, J.P, and G.Z. Whitten, 1984. Technical Discussion Relating to the Use of the Carbon Bond Mech-
anism in OZIPM/EKMA EPA-450/4-84/009, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Lamb, B., A. Guenther, D. Gay, et al., 1987. National inventory of biogenic hydrocarbon emissions.
Atmos. Environ. (21) 8:1695.
Lamb, R.G., 1983. A regional scale (1000 km) model of photochemical air pollution. Part I - Theoretical
formulation. EPA-600/3-83/035, U.S. Environmental Protection Agency, Research Triangle Park, NC.
R-4
-------
Lamb, R.G., 1984a. A regional scale (1000 km) model of photochemical air pollution. Pan II - Input
processor network design. EPA-600/3-84/085, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Lamb, R.G., 1984b. Air pollution models as descriptors of cause-effect relationships. Atmos. Environ.
18(3):591-606.
Lamb, R.G., 1985. Application of the first generation Regional Oxidant Model to an assessment of the
effects of proposed 1987 emissions reductions on episodic ozone levels in the northeastern United
States. Internal Report, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Lamb, R.G., 1986. Numerical simulations of photochemical air pollution in the northeastern United
States: ROM 1.0 Applications. EPA-600/3-86/038, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Lamb, R.G., 1987. Design and applications of the Regional Oxidant Model (ROM). In: Conference Pro-
ceedings of the North American Oxidant Symposium, February 25-27 1987, Quebec, Canada. Envi-
ronment Canada, pp. 153-189.
Lamb, R.G., 1988. Simulated effects of hydrocarbon emissions controls on seasonal ozone levels in the
northeastern United States: A preliminary study. EPA-600/3-88/017, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
Lamb, R.G., and G.F. Laniak, 1985. A regional scale (1000 km) model of photochemical air pollution.
Pan III - Tests of the numerical algorithms. EPA-600/3-85/037, U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Lamb, R.G., and S.K. Hati, 1987. The representation of atmospheric motion in models of regional-scale
air pollution. J. dim. Appl. Meteorol. 26(7):837-846.
Langstaff, J. and T. Young, 1990. User's Manual for Regional Emissions Growth and Control Strategy
Software (Draft). Alliance Technologies Corporation. Prepared for U.S. Environmental Protection
Agency, Research Triangle Park, NC.
Logan, J., 1983. Nitrogen oxides in the troposphere: global and regional budgets. Journal of Geophys-
ical Research (88): 10,785.
Milich, L, Boehm, T., Coats, C., Eichinger, J., Fudge, S., Jordan, D., Maxwell, C., Olerud, D., Tang, R.,
Van Meter, A., Wayland, R., and Young, J., 1990. The Regional Oxidant Model (ROM) User's Guide. In
preparation, U.S. Environmental Protection Agency, Research Triangle Park, NC. J. Novak, Technical
Monitor.
Modica, L, D. Dulleba, R. Walters, and J. Langstaff, 1988. Flexible Regional Emissions Data System
(FREDS) Documentation for the 1985 NAPAP Emissions Inventory. Prepared by Alliance Technologies
Corporation for the U.S. Environmental Protection Agency, Research Triangle Park, NC 27711.
Morris, R.E., T.C. Myers, and J.L Haney, 1990a. User's Guide for the Urban Airshed Model. Volume I:
User's Manual for the UAM(CB-IV). EPA-450/4-90-007A. Prepared for the Office of Air Quality Plan-
ning and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
R-5
-------
Morris, R.E., T.C. Myers, E.L Carr, M.C. Causley, S.G. Douglas, and J.L Haney, 1990b. User's Guide for
the Urban Airshed Model. Volume II: User's Manual for the UAM(CB-IV) Modeling System (Preproce-
ssors). EPA-450/4-90-007B. Prepared for the Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Novak, J.H., and J.A. Reagan, 1986. A comparison of natural and man-made hydrocarbon emission
inventories necessary for regional acid deposition and oxidant modeling, paper 86-30. In: Proceed-
ings of the 79th Annual APCA Meeting. APCA, Pittsburgh, PA.
Olson, R., C. Emerson, and M. Nunsgesser, 1980. GEOECOLOGY: A County-Level Environmental Data
Base for the Conterminous United States, ORNL/TM-7351, Oak Ridge National Laboratory, Oak Ridge,
TN, 54 p.
Pagnotti, V. 1987. A meso-meteorological feature associated with high ozone concentrations in the
northeastern United States. Journal of the Air Pollution Control Association, 37(6): 720-722.
Pechan, E.H. and Associates, 1989. ERCAM-VOC: Description and Applications. Prepared for the U.S.
Environmental Protection Agency Office of Policy, Planning and Evaluation (EPA Contract No.
68-01-7047, Work Assignment No. 126).
Peterson, E. and D. Tingey, 1980. An estimate of the possible contributions of biogenic sources to air-
borne hydrocarbon concentrations. Atmos. Environ. (14) 1: 79.
Pierce, T.E., and K.L Schere, 1988. Regional ozone modeling: An investigation of hydrocarbon emis-
sions from treatment, storage, and disposal facilities on ambient levels of ozone. Internal Report, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Pierce, T., B. Lamb, and A. Van Meter, 1990. Development of a biogenic emissions inventory system for
regional scale air pollution models. In: Proceedings of the 83rd Air Waste Management Association
Annual Meeting, Pittsburgh, PA, June 24 - 29,1990; 16 p.
Pierce, T.E., K.L Schere, D.C. Doll, and W.E. Heilman, 1990. Evaluation of the Regional Oxidant Model
(Version 2.1) using ambient and diagnostic simulations. Draft Report, U.S. Environmental Protection
Agency, Research Triangle Park, NC.
Possiel, N.C., J.A. Tikvart, J.H. Novak, K.L Schere, and E.L Meyer, 1989. Evaluation of ozone control
strategies in the northeastern region of the United States. In: Proceedings of the Third U.S.-Dutch
International Symposium on Atmospheric Ozone Research and Its Policy Implications, May 9-13
1988, Nijmegen, The Netherlands. Elsevier, The Netherlands, pp. 623-632.
Rao, ST., G. Sistla, J.Y. Ku, R. Twaddell, R. Whitby, V. Pagnotti, E. Davis, N. Possiel, J. Pearson, K.
Schere, and R. Dennis, 1987. Examination of the Urvan Airshed model performance in the New York
metropolitan area. Proceedings, 80th Annual A.P.C.A. Meeting, New York, NY, 22 - 26 June, 1987.
Paper no. 87-71.8, Air Pollution Control Association, Pittsburgh, PA.
Rasmussen, R., 1972. What do the hydrocarbons from trees contribute to air pollution? Journal Air
Pollution Control Association (22) 7: 537.
R-6
-------
Roselle S. and K. Schere, 1990. Sensitivity of the EPA Regional Oxidant Model to biogenic hydrocarbon
emissions. In: Proceedings of the 83rd Air Waste Management Association Annual Meeting, Pittsburgh,
PA, 1990, 20 p.
Saeger, M., J. Langstaff, R. Walters, L Modica, D. Zimmerman, D. Fratt, D. Dulleba, R. Ryan, J. Demmy,
W. Tax, D. Sprague, D. Mudgett, A. Werner, 1989. The NAPAP Emissions Inventory (Version 2):
Development of the Annual Data and Modeler's Tapes. EPA-600/7-89-012a, U.S. Environmental Pro-
tection Agency, Research Triangle Park, NC.
SAS Institute Inc. 1985. SAS User's Guide: Basics, Version 5 Edition. SAS Institute Inc., Gary, NC.
Schere, K.L, 1986. EPA Regional Oxidant Model: ROM 1.0 evaluation for 3-4 August 1979. EPA-600/3-
86/032, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Schere, K.L, 1986. Evaluating Og predictions from a test application of the EPA Regional Oxidant Model.
In: Proceedings of the Fifth Joint Conference on Applications of Air Pollution Meteorology, November
18-21 1986, Chapel Hill, NC. AMS, Boston, MA, pp. 61-64.
Schere, K.L., and A.J. Fabrick, 1985. EPA Regional Oxidant Model: Description and evaluation plan.
EPA-600/3-85/067, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Schere, K.L, and J.H. Novak, 1986. Regional oxidant modeling of the northeast U.S. In: C. DeWispe-
laere et a/., eds. Air Pollution Modeling and Its Application, V - Vol. 10. Plenum Press, NY, pp. 45-59.
Schere, K.L, and N.P. Possiel, 1984. U.S. EPA Regional Oxidant Model - Background and overview,
paper no. 84-47.2. In: Proceedings of the 77th Annual APCA Meeting, June 25-29 1984, San
Francisco, CA. APCA, Pittsburgh, PA.
Schere, K.L, and R.A. Wayland, 1989. Development and evaluation of the Regional Oxidant Model for
the northeastern United States. In: Proceedings of the Third U.S.-Dutch International Symposium on
Atmospheric Ozone Research and Its Policy Implications, May 9-131988, Nijmegen, The Netherlands.
Elsevier, The Netherlands, pp. 613-622.
Schere, K.L., and R.A. Wayland, 1989. EPA Regional Oxidant Model (ROM 2.0): Evaluation on 1980
NEROS databases. EPA-600/3-89/057, U.S. Environmental Protection Agency, Research Triangle
Park, NC.
Tang, R.T., S.C. Gerry, J.S. Newsom, A.R. Van Meter, R.A. Wayland, J.M. Godowitch, and K.L Schere,
1990. User's Guide for the Urban Airshed Model. Volume V: Description and Operation of the ROM-
UAM Interface Program System. EPA-450/4-90-007E. Prepared for the Office of Air Quality Planning
and Standards, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Tingey, D., M. Manning, L Grothaus et a/., 1980. Influence of light and temperature on monoterpene
emission rates from slash pine. Plant Physiol. (65): 797.
Tingey, D., 1981. Atmospheric Biogenic Hydrocarbons. J. Bufalini and R. Arnts, eds., Ann Arbor Sci-
ence Publications, Ann Arbor, Ml, pp. 53-79.
Tingey, D., R. Evans, and M. Gumpertz, 1981. Effects of environmental conditions on isoprene emis-
sions from live oak. Planta (152): 565.
R-7
-------
Trainer, M., E. William, D. Parrish et a/., 1987. Models and observations of the impact of natural hydro-
carbons on rural ozone. Nature (329): 705.
Wagner, J., R. Walters, L Maiocco, and D. Neal, 1986. Development of the 1980 NAPAP Emissions
Inventory. EPA-600/7-86-057a, U.S. Environmental Protection Agency, Research Triangle Park, NC
27711.
Walters, R., L Modica, and D. Fratt, 1988. 7985 NAPAP Emissions Inventory Allocation Factors. Pre-
pared by Alliance Technologies Corporation for the U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711
Whitten, G.Z., and M.W. Gery, 1986. Development of CBM-X mechanisms for urban and regional
AQSMs. EPA-600/3-86/012, U.S. Environmental Protection Agency, Research Triangle Park, NC.
Young, J.O., M. Aissa, T.L Boehm, C.J. Coats, J.R. Eichinger, S.J. Roselle, A.R. Van Meter, R.A. Way-
land, and T.E. Pierce, 1989. Development of the Regional Oxidant Model, Version 2.1. EPA/600/3-
89/044. (NTIS PB89-194252). Atmospheric Research and Exposure Assessment Laboratory, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Zimmerman, P., 1979. Determination of Emission Rates of Hydrocarbons from Indigenous Species of
Vegetation in the Tampa Bay/Petersburg, Florida area, EPA-904/9-77-028. U.S. Enviromental Protection
Agency, Atlanta, GA, 173 p.
R-8
------- |