EPA-450/4-91-002a
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. Miiich
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
June 1991
-------
DISCLAIMER
This report has been reviewed by the Office of Air Quality Planning and Standards, United States
Environmental 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.
Environmental 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
consequence of the management direction and technical information provided by representives of the
State/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
Implementation 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
instrumental in providing leadership and guidance in the use of this model. The hundreds of ROM
applications 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 analysis and interpretation of the ROM applications was made by Mr. William Cox, EPA. Finally,
the tireless efforts of Ms. Cynthia Baines and Ms. Ullian Faison, EPA, and Ms. Christine Bullock, CSC, in
typing this report are greatly appreciated.
-------
3*-
. v«
V:
This page is intentionally left blank.
•••*}
-------
CONTENTS
Disclaimer ii
Acknowledgements Hi
List of Figures viii
List of Tables xii
EXECUTIVE SUMMARY 1 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-14
Technology Transfer „ ES-15
Need for Technology Transfer ES-15
Gridded Model Information Support System (GMISS) ES-15
ROM-UAM Interface ES-16
Accomplishments of the ROMNET Project ES-16
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
-------
CONTENTS (continued)
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 2-10
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 10 Episodes 3-8
3.2.6 Episodes Selected for ROM Simulations 3-10
4. EMISSIONS SCENARIO DEVELOPMENT 4-1
4.1 Introduction 43
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
I
VI
-------
CONTENTS (concluded)
4.5 Emission 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
4.5.3 VOC, NOX, and CO 2005 Baseline Mobile Source Controls 4-32
4.5;4 Maximum Technology VOC Stationary-Source Controls 4-33
4.5.5 Maximum Technology NOX Technology Stationary-Source Controls 4-34
4.5.6 Maximum Technology VOC, NOX, and CO Mobile Source Controls 4-34
4.5.7 Controls Applied in the CS05 Clean Air Act Strategy 4-36
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 4^9
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 For Key Issues .„ 5-40
6. STATE ACCESS AND USE OF ROM DATABASES 6-1
6.1 Introduction 5-3
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 , e-7
6.3 The Regional Oxidant Model - Urban Airshed Model (ROM-UAM) Interface
Program System 6-10
6.3.1 Introduction 6-10
6.3.2 Features and Limitations of the Interface e-11
6.3.3 Description of the Interface System 6-12
REFERENCES AND BIBLIOGRAPHY R.-(
VII
-------
LIST OF FIGURES
ES-1 The ROMNET region ES-18
ES-2 ROMNET management structure ES-19
ES-3 Components of the ROM ES-20
1-1 The Northeast Corridor and other Metropolitan Statistical Areas/Consolidated
Metropolitan Statistical Areas in the ROMNET region 1-11
1-2 Observed daily maximum 1-hour ozone concentrations for July 5-8, 1988
across the ROMNET region 1-12
1-3 ROMNET management structure 1-14
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
northeastern 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
-------
LIST OF FIGURES (continued)
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 diurnal 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 4-51
4-7 , 1985 base case anthropogenic emissions of NOX from industrial plants and
utilities 4-53
4-8 1985 base case emissions by day type 4-55
4-9 Diurnal profiles of point, area, and mobile sources of VOC and NOX 4-56
4-10 Diurnal profiles of point, area, and mobile sources of CO 4-57
4-11 Mobile source emissions at 1500 EST on a 'cool" day and a 'warm' day 4-59
4-12 Regionwide daily total mobile source VOC and NO* emissions, July 1988
episode 4-61
4-13 Regionwide daily total mobile source CO emissions, July 1988 episode 4-62
4-14 VOC emissions for the Phase II scenarios and the percent reduction from the
1985 base case and the 2005 baseline 4^63
4-15 NOX emissions for the Phase II scenarios and the percent reduction from the
1985 base case and the 2005 baseline , 4-64
4-16 CO emissions for the Phase II scenarios and the percent reduction from the
1985 base case and the 2005 baseline 4-65
4-17 The Northeast Corridor and nonattainment areas outside the Corridor 4-66
4-18 Location of major NOX point sources controlled in the maximum technology
strategy 4-67
4-19 Relationship between meteorological temperature and (1) isoprene and
alpha-pinene emissions, (2) monoterpenes/unidentified hydrocarbon emissions
.... ...4-68
4-20 Flowchart of the Biogenic Emissions Inventory System 4-69
4-21 Biogenic isoprene emissions for 1000 EST and 1500 EST on a "cool" day , 4-71
4-22 Biogenic isoprene emissions for 1000 EST and 1500 EST on a "warm" day 4-73
4-23 Regionwide daily total biogenic VOC emissions, July 1988 episode 4-75
5-1 Predicted 1985 base case episode maximum 1-hour ozone concentrations for
the July 1988 episode 5-45
IX
-------
LIST OF FIGURES (continued)
5-2 Predicted 2005 baseline episode maximum 1 -hour ozone concentrations for the
July 1988 episode 5-47
5-3 Areas in the ROMNET domain used in calculating selected metrics 5-49
5-4 Anthropogenic VOC emissions and NOX emissions for the U.S. portion of the
ROMNET region 5-50
5-5 Percent reduction in VOC emissions between CS12 and the 2005 baseline for
anthropogenic emissions only and anthropogenic plus biogenic emissions 5-51
5-6 Percent reduction in NOX emissions between CS11 and the 2005 baseline 5-53
5-7 Predicted episode maximum 1-hour ozone concentrations for the July 1988
episode: 2005 baseline and CS11 5-55
5-8 Predicted episode maximum 1-hour ozone concentrations for the July 1988
episode: 2005 baseline and CS12 5-57
5-9 Predicted episode maximum 1-hour ozone concentrations for the July 1988
episode: CS13, CS14, andCSlO 5-59
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 , 5-63
5-11 Quantiie-quantile frequency distributions of 1-hour daily maximum ozone and
episode mean 8-hour daily maximum ozone for New York City 5-64
5-12 Quantile-quantile frequency distributions of 1-hour daily maximum ozone and
episode mean 8-hour daily maximum ozone for Greater Connecticut 5-65
5-13 Diurnal time series of maximum hourly ozone concentrations in New York City
and Greater Connecticut for CS11 and CS12 ,....". 5-66
5-14 Diurnal time series of maximum hourly ozone concentrations in Baltimore/-
Washington, DC area and the Pittsburgh area for CS11 and CS12 5-67
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 5-69
5-16 Morning total VOC and NOX emissions on July 9,1988 in the vicinity of New York
City 5-71
5-17 Population exposure to 1-hour ozone > 100 ppb in Massachusetts and coastal
New England for selected scenarios 5-73
5-18 Population exposure to 1-hour ozone > 125 ppb in Baltimore/Washington, DC
and Philadelphia for selected scenarios 5-74
5-19 Population exposure to 1-hour ozone > 125 ppb in New York City and Greater
Connecticut/Rhode Island for selected scenarios 5-75
5-20 Population exposure to 1-hour ozone > 125 ppb in Pittsburgh and Charleston,
WV (combined) and Cleveland/Detroit (combined) for selected scenarios 5-76
-------
LIST OF FIGURES (concluded)
5-21 Location of the "transport" boundary used in quantifying incoming ozone
transport 5.77
5-22 Layer 2 forward trajectories starting at 1500 EST from locations* along the
•transport" boundary for July 9,1988 and July 10,1988 5-78
5-23 ROM layer 2 peak and mean ozone concentrations by day along the three
"transport" boundary segments 5.79
5-24 Predicted episode maximum 1-hour ozone concentrations for the July 1988
episode: 2005 baseline and CS25 5-81
5-25 Predicted episode maximum 1-hour ozone concentrations for the July 1988
episode: CS19 and CS24 5.33
5-26 ROM layer 2 trajectory for the transport case study „ 5-85
5-27 Time history of layer 2 ozone, NOX, and ROG concentrations: 20Q5 baseline and
CS25 , 5.86
5-28 Time history of layer 2 ozone, NOX, and ROG concentrations: CS19 and CS24 .. 5-87
5-29 Emissions for Phase II scenarios, and predicted highest and second-highest
daily maximum ozone concentrations for selected urban areas 5-88
5-30 Predicted episode maximum 1-hour ozone concentrations for the June 1983
episode: 2005 baseline and CS19 5.93
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 5.95
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 5.97
5-33 Predicted episode maximum 1-hour ozone concentrations for biogenic
sensitivity scenarios for selected urban areas 5-99
5-34 Predicted grid-hours exceeding_125 ppb for biogenic sensitivity scenarios for
selected urban areas 5-101
5-35 Predicted population exposure to ozone exceeding 125 ppb for biogenic
sensitivity scenarios for selected urban areas 5-103
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
XI
-------
LIST OF TABLES
ES-1 Organizations participating in ROMNET ES-21
ES-2 ROMNET control strategies ES-22
1-1 Organizations participating in ROMNET 1-15
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 (JAM 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 W'rthin-Corridorflow regimes 3-15
4-1 ROMNET emissions scenarios 4-76
4-2 Revised summer allocation factors for area source categories 4-77
4-3 Summary of control measures in the baseline projection and maximum
technology inventories 4-78
4-4 Control measures for the draft Clean Air Act analysis 4-79
4-5 NSPS efficiencies for point and area sources 4-80
4-6 Summary of State estimates for existing area source controls 4-81
4-7 Summary of NOX control measures for Phase I 4-82
4-8 Maximum technology VOC controls for point sources 4-83
4-9 Maximum technology VOC controls for area sources 4-84
4-10 Maximum technology NOX'controls for point and area sources 4-85
4-11 Strategy 5 point source control efficiencies by attainment category 4-86
xii
-------
LIST OF TABLES (concluded)
4-12 Strategy 5 area source control efficiencies by attainment category 4-87
4-13 Strategy 5 attainment categories and across-the-board VOC reductions 4-88
4-14 VOC speciation factors for methanol vehicles 4-89
4-15 Emission fluxes for vegetation types in the Biogenic Emissions Inventory
System , 4.39
5-1 Procedures used for calculating population exposure : 5-105
5-2 Summary of emissions reductions that reduced predicted ozone
to < 125 ppb , 5-106
5-3 Comparison of the June 1983 and July 1988 episode maximum 1-hour
concentrations for the 2005 baseline and CS19 5-107
5-4 Summary of ozone metrics used to assess the impacts of reactivity-based
strategies „ 5-108
5-5 Cases tested with varying biogenic emissions 5-109
6-1 Summary of the UAM preprocessor programs 6-16
6-2 Retrieved ROM files used by interface programs 6^16
6-3 Overview of the ROM-UAM Interface programs and input/output files 6-17
XIII
-------
V
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
-------
•SI
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
programs 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
measures, 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:
• the project scope and organization;
• the ROMNET technical program;
« the findings from the Regional Oxidant Model (ROM) simulations, conducted to address the
"strategic" issues regarding the effectiveness of regional control scenarios;
the mechanisms and procedures developed to enable States to explicitly consider regional
and interurban transport in preparing ozone SIPs; and
• a summary of the major accomplishments of the project.
1. An oxceedance 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, on the average, at any location.
ES-3
-------
•1
Readers interested in a detailed description of these topics and other aspects of ROMNET are referred
to the numbered sections of this report. Note: all figures and tables are placed at the end of the
Section in which they are referenced.
Protect 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. The project 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 (Emissions, Strategy,
Modeling) 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
chairmen 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
responsive as feasible to directions from the Management Review Committee. The Technical Coordi-
nator assisted the Program Director by managing day-to-day activities of the project, establishing prio-
rities for a diverse set of technical tasks, making sure these priorities were observed, and ensuring that
technical activities were integrated appropriately.
ES-4
-------
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.
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
direction on the particular topic.
THE ROMNET TECHNICAL PROGRAM
The ROMNET technical program consisted of three major components:
1. simulations using the 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
transport estimates and certain other inputs required for urban-scale SIP modeling; and
3. the development of the ROM-Urban Airshed Model (UAM) Interface 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-UAM Interface.
The Regional Oxldant Model
The ROM Version 2.1 (Milich ef a/., 1991) was used for estimating ozone concentrations across the
Northeast under various emissions scenarios selected for evaluation as part of ROMNET. The ROM is a
three-dimensional photochemical grid model, with a horizontal grid resolution of Vs° latitude by 1/4°
longitude (approximately 18.5 km x 18.5 km in the Northeast). Vertically, there are three layers. Two
layers lie between the surface 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 ROM Core Model. The Core Model performs the horizontal
ES-5
-------
:*,«
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 ROM
Core Model is described by the Carbon Bond 4 (CB-1V) mechanism (Gery et a/., 1989). This
mechanism is identical to the one contained in the UAM, recommended by the EPA for use in SIP
attainment demonstrations. The ROM generates 30-minute concentrations that are aggregated to
hourly averages. These data 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; 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; future baseline' predictions
were simulated.
4. Control measures were applied to future baseline emissions to derive control strategy (CS)
scenarios; post-control ozone levels were simulated.
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 emission control strate-
gies, and the conclusions drawn from ROM simulations.
Episode Selection
Two factors were considered in selecting episodes to simulate with the ROM: (1) ambient ozone con-
centrations and (2) wind flow patterns. Because the primary focus of trie 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 Corridor. 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.
•**,
1
ES-6
-------
Next, the 10 top-ranked episodes were 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.
The period July 2-17,1988, was identified by the above process as the most severe in terms of ozone
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
contained high ozone levels but was characterized by recirculation conditions. This period was chosen
for simulating 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 severe 1985 episodes
(August 7-16, 1985, and July 7-22, 1985) were chosen for a ROM performance evaluation. A more
complete 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 (primarily VOC emissions from vegetation) were calculated and included in
ROM simulations. The EPA Biogenic Emissions Inventory System (BEIS) (Pierce etal., 1990a) was
used to derive hourly biogenic emissions estimates for the ROMNET domain.
2. Mobile source emissions were upgraded twice using MOBILE3.9 and then MOBILE4, to obtain
estimates for mobile source exhaust and evaporative emissions. An earlier version of the MOBILE
model was used in NAPAP. MOBILE4 (EPA, 1989) considers such recently-identified phenomena
as evaporative running losses; it also contains an improved treatment of mobile source evaporative
emissions as a function of temperature.
Both biogenic emissions and mobile evaporative emissions can be very temperature sensitive.
Emissions of certain biogenic species are also sensitive to sunlight intensity. Because biogenic and
mobile 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.
ES-7
-------
A more complete description of the 1985 base ROMNET inventory is contained in Section 4.
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. Observed and predicted spatial patterns for daily maximum ozone were compared through inspec-
tion of contours of observed and predicted concentrations.
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.
Within the limits of this evaluation, ROM performance appears satisfactory for estimating episodic
ozone levels for use in assessing regional-scale concentrations and patterns. Results tend to confirm
that regional-scale models may not have sufficient resolution to simulate the fine-scale spatial structure
of individual urban plumes.
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. It is also 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?8
ES-8
-------
Regionwide reductions in VOC, NOX, 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
concentrations 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
Corridor?
3. What levels of VOC and/or NOX emissions reductions are necessary to reduce predicted ozone
levels 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 (CS01 through CS25, excluding CS04) 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, NOXl 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 that led to these findings are contained in Section 5 of the report.
Urban areas discussed below are defined by Metropolitan Statistical Area/Consolidated Metropolitan
Statistical Area (MSA/CMSA) boundaries.
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. The larger reductions are particularly notable
in the western portion 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 concen-
trations.
• In the presence of stringent VOC controls, peak ozone levels appear to be more sensitive to
mobile source rather than point source NOX emissions reductions in several cities (e.g., Phil-
adelphia, 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
maximum 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 variation was particularly evident in the Baltimore/Washington, DC
area
ES-10
-------
Issue #2
What is the impact of reducing regional transport on ozone concentrations in the Northeast
Corridor?
• In the pre-control 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.
« In the post-control scenario, ozone transport was found to be very important. 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
'-'-K*-.,,,1,,' ' .' • .
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 reduc-
tions is beyond known or envisaged control technologies.
The effectiveness of the most stringent control strategy in reducing ozone to < 125 ppb was
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.
ES-11
-------
lssue#4
What are the effects of reactivity-based strategies in reducing regional ozone levels?
• Reactivity-based strategies, similar to those simulated in ROMNET, may provide the greatest
benefit in large urban areas that 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 stringent 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
measures was only half of that from the technology-based VOC controls and a factor of 4
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.
Issue #5
How does the large uncertainty in biogenic emissions alter conclusions regarding the effective-
ness of control measures?
ROM simulations with 'best estimate0 biogenic emissions yield predicted ozone concentra-
tions closer to observed values than those simulations with biogenic emissions at either end
of the uncertainty range ( ± a factor of 3). This result 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.
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).
i
ES-12
-------
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 reduc-
tions in emissions beyond those in CS19 will be needed in many of the Northeast cities.
Implications of Findings
¥ ^Vv. ft' .-: -_....•• .....
There are several broad implications that can be drawn from the above findings. These are:
1. Significant reductions in anthropogenic emissions along the entire Northeast Corridor will likely be
necessary in order to reduce ozone to less than 125 ppb throughout the Corridor.
2. Along with VOC controls, NOX control measures should be considered in strategies to reduce
ozone levels in most areas of the Northeast. However, close examination of the potential effects of
NOX controls should be made to ensure that such controls will not be counterproductive.
3. The types of control measures and the degree of emissions reductions necessary to reduce ozone
to less than 125 ppb will likely vary between cities along the Corridor. That is, a single set of
controls common to all areas may not be the most effective approach for reducing ozone through-
out the Corridor.
4. Areas outside the Northeast Corridor may have to add controls, beyond those necessary to solve
their local problem, in order to reduce transport into the Corridor sufficiently for concentrations to
be reduced below 125 ppb throughout the Corridor (even with stringent controls in the Corridor).
5. The control technologies needed to achieve the emissions reductions necessary to reduce ozone
to below 125 ppb in most of the major Northeast Corridor cities are currently not available.
Therefore, a high priority should be given to development and testing of controls for both VOC and
NOX sources along with enhancing enforcement procedures to achieve the highest degree of
control effectiveness possible.
6. Regional strategy assessments of the type performed in ROMNET are useful for examining the
regional perspective to control impacts and transport. However, urban-scale analyses will also be
needed to develop local control targets and to establish a menu of specific control measures for
individual nonattainment cities.
ES-13
-------
'. .'I
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 18.5 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)
average values for each layer. This type of emissions estimate 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, because a large
fraction of NOX emissions emanate from 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 emission controls necessary to reduce ozone to < 125 ppb.
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 unquantified level of uncertainty in base case anthropo-
genic emissions. The representativeness of growth factors for estimating future year emissions as
well as the efficiency and effectiveness of controls adds to the uncertainty in emissions scenarios.
Thus, the findings from ROMNET should be used to help establish control directions and approxi-
mate 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. Thus, the
results (particularly regarding the effects of NO* controls and regional transport) may vary under
different meteorological conditions.
ES-14
-------
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. Four reasons for using the ROM to assist in developing required ozone NAAQS attain-
ment demonstrations are listed below.
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 mean's for simulating the combined effects from numerous urban areas across
the region.
3. The ROM provides the most technically defensible way to estimate future ozone/precursor concen-
trations transported into urban areas.
4. The ROM provides data for use in SIPs that 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 UAM 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 is used to provide initial and boundary con-
ditions estimates, as well as other inputs, for UAM applications.
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 that 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 GMISS and the ROM-UAM Interface.
Gridded Model Information Support System (GMISS)
GMISS contains an archive of ROM 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 concentra-
tions, selected meteorological inputs to the ROM, biogenic emissions calculated for ROM simulations,
ES-15
-------
and topographic information. The UAM subsystem of GMISS is designed to provide all of the ROM 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
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 4 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 data Interface programs prescribe a series of algorithms that match and interpolate
ROM data to the resolution used by UAM. The structure of the ROM-UAM Interface 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 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
[n the management and technical decision-making aspects of the project. The organizational structure
and communications process adopted in ROMNET may be useful as a framework for future joint efforts
among States and the EPA.
Second, the ROM was demonstrated to be a practical tool for addressing policy-related issues con-
cerning the efficacy of various regional control strategies. The ROMNET program 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 specified in the NAAQS; (d) the
ES-16
-------
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 that will enable
State agencies to quantitatively consider regional and interurban transport in preparing ozone SIPs.
The infrastructure containing these systems gives States access to ROM databases and provides
methodologies and guidance for using these data to support urban-level attainment demonstrations.
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-17
-------
Figure ES-1. The ROMNET region.
ES-18
-------
ADVISORY
COUNCIL
EMISSIONS
COMMITTEE
MANAGEMENT REVIEW COMMITTEE
PROGRAM DIRECTOR
TECHNICAL COORDINATOR
STRATEGY
COMMITTEE
MODELING
COMMITTEE
Figure ES-2. ROMNET management structure.
ES-19
-------
RAW INPUT DA
TAIR QUALITY) ,
(METEOROLOGY)
v
f EMISSIONS ")
C LAND USE ")
(JOPOCRAPHY)
TA
ROM
CORE MODEL
1
__ TRANSPORT
ROM *•
, ,, ^. PRORFSSOR t 1
CONCENTRATIONS
Figure ES-3. Components of the ROM.
ES-20
-------
TABLE ES-1. ORGANIZATIONS PARTICIPATING IN ROMNET
State/local air pollution control agencies:
Connecticut
Delaware
Kentucky
Maine
Maryland
Massachusetts
Michigan
New Hampshire
New Jersey
New York
Ohio
Pennsylvania
Philadelphia
Rhode Island
Vermont
Virginia
Washington, DC
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 (OPAR)
Office of Policy Planning and Evaluation (OPPE)
Office of Mobile Sources (QMS)
Contractors:
Computer Sciences Corporation (CSC)
Alliance Technologies Corporation
ES-21
-------
TABLE ES-2. ROMNET CONTROL STRATEGIES
Control
Strategy Number
Controls
Issue(s)
Addressed
Rationale
•ft.
CS01 (MOBILE 3.9)
CS02 (MOBILE 3.9)
CS03 (MOBILE 3.9)
4$
CS04
CS05
CS06-CS09
CS10
CS11
CS12
CS13-CS14
Maximum technology VOC 2,3
regionwide; NOX at 2005
baseline levels
Maximum technology VOC in
NE Corridor only; NOX at 2005
baseline
Maximum technology VOC in 2
NE Corridor and in other nonat-
tainment areas; only 2005
baseline elsewhere
Not performed
VOC and NOX controls pre- 3
scribed in HR3030 with October
1989 Waxman-Dingell tailpipe
standards
CS01 and CS05 varying 5
biogenic emissions by
± a factor of-3
Enhanced maximum technol- 1,3
ogy VOC and NOX controls
applied regionwide
Maximum technology NOX; 1, 3
VOC at 2005 baseline
Maximum technology VOC; 1, 3
NOX 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
controls outside
Corridor on ozone in
Corridor by comparison
withCSOI.
Test effect of "rural" VOC
controls outside
Corridor 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-22
-------
TABLE ES-2 (continued)
Control
Strategy Number
Controls
lssue(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
Corridor 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
emissions at 2005 baseline in
New York City
Maximum technology NOX 5
controls (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
emissions 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
biogenics.
Test relative importance
of reactivity reduction
and VOC technology
controls by comparison
withCSl2andCS16.
Reduce maximum ozone
to < 125 ppb through-
out the U.S. portion of
the region.
Compare effectiveness
of reactivity reductions
in base vs. post-control
scenarios and with VOC
technology controls.
Test effect of biogenic
emissions uncertainty
on a strategy that
reduces ozone to
< 125 ppb.
continued
ES-23
-------
TABLE ES-2 (concluded)
Control
Strategy Number
Controls
Issue(s)
Addressed
Rationale
CS23
CS24
CS25
CS19, but with more realistic
'rule effectiveness" assumptions
and vehicle fleet penetration
estimates
CS19 in NE Corridor;
2005 baseline elsewhere
CS19 outside Corridor;
2005 baseline in NE Corridor
Test sensitivity of results
to limitations in control
program effectiveness.
Assess relative impor-
tance of transport from
outside the Corridor vs.
controls on emissions
within the Corridor.
ES-24
i
i
-------
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
-------
•i*
1
This page is intentionally left blank.
»;
•I
5
-------
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
programs 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
measures, 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 Episode Selection - describes the episode selection process, and the
ozone levels and meteorological conditions during episodes simulated;
1. An exoeedance 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 (in EPA-450/4-91-002b) - include 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. 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 (Northeast Corridor) 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, and Detroit) are likely to contribute to the regional
ozone burden transported 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
technical 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 that sets forth the goals,
scope, technical approach, schedule, and organizational structure agreed to by the participants. This
protocol is included as Appendix A. Sections 1.4 through 1.6 describe:, 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 (Appendix A).
1.4 RATIONALE FOR THE GOALS AND SCOPE OF ROMNET
This subsection 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
transport. Thus, it is important to identify strategies that are most effective in reducing regional ozone
concentrations.
The rationale for the second goal is based upon prior urban-scale modeling analyses, which indicate
that the effectiveness of emission 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 Glean 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 modeling.
The third goal reflects the practicality of deriving urban boundary conditions from transport concentra-
tions. Procedures and guidance are needed 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. Thus, 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
problem 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
individually small, sources of volatile organic compounds (VOC), such as dry cleaners, gasoline service
stations, and bakeries; mobile source emissions of VOC, oxides of nitrogen (NOX), and carbon
monoxide (CO) from passenger vehicles and trucks; and large point sources of NOX, VOC, and/or CO,
such as power plants, industrial facilities, and gasoline terminals. Thus, evaluating regional strategies
requires specification of both emissions and controls for multiple pollutants from numerous source
types of different sizes; some source types are clustered in urban areas but 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. A further complication is 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 considered 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 under development and are generally not yet available. As a consequence,
procedures are needed for providing 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
pollutant 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
t
I
-------
(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, it 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, the Technical
Coordinator, and the 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
schedule, 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
progress 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 (Modeling, Emissions, and Strategy) 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).
The Modeling Committee functions included episode selection, upgrades to the ROM, development of
the ROM-UAM Interface, the supplemental evaluation of the ROM, ROM simulations, and development
of the 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 Councilare 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 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 30 combinations of episodes and emissions scenarios.
A supplemental evaluation of ROM was conducted using selected ROMNET episodes.
1-8
-------
• The ROM-UAM Interface was developed to provide a means for using ROM predictions to
derive UAM boundary conditions. The ROM-UAM Interface also contains procedures for
using ROM system data sets to generate meteorological, air quality, topographical, and
biogenic emissions input for UAM.
• " The GMISS was developed to:
(1) archive ROM predictions and system data sets needed for the ROM-UAM 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.
The OAQPS was responsible for episode selection and, together with Computer Sciences Corporation
(CSC), the development of GMISS. The EPA Atmospheric Research and Exposure Assessment Labo-
ratory (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 postprocessing 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 development of
the ROM-UAM Interface.
Emissions inventory Development
• The Biogenic Emissions Inventory System (BEIS) was developed to provide estimates of
biogenic emissions for VOC and NOX. As part of this task, a canopy module was implem-
ented 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
emissions inventory was obtained for use in ROMNET and modified to:
(1) upgrade mobile source emissions to MOBILE4,
(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 tempera-
ture 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.
1-9
-------
A majority of the emissions inventory work was conducted by Alliance Technologies Corporation, under
contract to the OAQPS. The BE1S 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 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 penetra-
tion 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.
• ROMNET participants 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
presented to the Management Review Committee for approval. Translation of strategies into emissions
reductions were performed by Alliance Technologies Corporation, under contract to the OAQPS, and
by CSC, under contract to the AREAL
1-10
-------
I NORTHEAST
'CORRIDOR
|H BOSTON
H| GREATER CONNECTICUT
|||NEW YORK CITY
H|j PHILADELPHIA
^PITTSBURGH
H| BALTIMORE
m WASHINGTON, DC
fflffi CLEVELAND
II CHARLESTON
gg DETROIT
liH OTHER MSAs
Figure 1 -1. The Northeast Corridor and Metropolitan Statistical Areas/Consolidated Metropolitan
Statistical Areas (MSA/CMSA) in the ROMNET region.
1-11
-------
Rgure1-2. Observed daily maximum 1-hour ozone concentrations for July 5-8, 1988, across the
ROMNET region. (Page 1 of 2)
1-12
-------
Figure 1-2. (Page 2 of 2)
1-13
-------
MANAGEMENT REVIEW COMMITTEE
ADVISORY
COUNCIL
EMISSIONS
COMMITTEE
PROGRAM DIRECTOR
TECHNICAL COORDINATOR
STRATEGY
COMMITTEE
MODELING
COMMITTEE
Figure 1-3. ROMNET management structure.
1-14
-------
TABLE 1-1. ORGANIZATIONS PARTICIPATING IN ROMNET
State/local air pollution control agencies:
Connecticut
Delaware
Kentucky
Maine
Maryland
Massachusetts
Michigan
New Hampshire
New Jersey
New York
Ohio
Pennsylvania
Philadelphia
Rhode Island
Vermont
Virginia
Washington, DC
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 (OPAR)
Office of Policy Planning and Evaluation (OPPE)
Office of Mobile Sources (QMS)
Contractors:
Computer Sciences Corporation (CSC)
Alliance Technologies Corporation
1-15
-------
."*
This page is intentionally left blank.
-------
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. Box 12767
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 emission control strategies. This Section
describes the characteristics of the ROM, the associated input-processor system, quality assurance
procedures followed for ROM simulations, and limitations to the use of the model. An evaluation of the
ROM, conducted as part of the 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) atmo-
spheric 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 photochemical 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.Ve° latitude, or about
18.5 km x 18.5 km. The modeling domain adopted for use in the 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
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 penet-
rative 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" because 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
evolution 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, because 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
2-4
-------
reactions. The 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 may have the spatial or temporal resolution necessary to
determine with confidence the wind fields in layer 1 under nighttime inversion conditions, a submodel
within the ROM system was developed to simulate the nighttime flow regime in layer 1 only. This prog-
nostic 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 ail 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 era/., 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
03, NO, NOg, CO, and other intermediate and radical species. Organic chemistry is partitioned along
2-5
-------
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
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 (e.g., paraffins);
• ALD2, the oxygenated two-carbon structure of the higher aldehydes;
« TOL, the aromatic structure of molecules with only one functional group (e.g., toluene);
• XYL, the structure of molecules with multifunctional aromatic rings (e.g., xylene);
• ISOP, the five-carbon isoprene molecule;
• NONR, a single-carbon organic structure not significantly participating in the reaction
sequence; and
« MTHL, methanol, 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: (1) raw input data, (2) the
processor network, and (3) the Core Model.
Haw Data
The first component specifies the five types of "raw" data that are required for input to the ROM: air
quality, meteorology, emissions, land use, and topography.
Air quality data consist of hourly ozone observations obtained from the EPA's National Air Data Branch.
These hourly observations are used to specify upwind boundary ozone concentrations required by the
ROM. Initial conditions are derived from the mean tropospheric background concentrations listed in
Table 2-2 (Wllus and Whitten, 1984). These values are chemically equilibrated with each other before
use by the ROM. Background concentrations are used because 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).
"*'.
2-6
-------
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 chemically equilibrated to this ozone concentra-
tion, generating 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 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
typically 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 meteorol-
ogy data and meteorology, data preprocessing can be found in Milich etal. (1991).
Emissions input data consist of annual point source emissions, with stack parameters and seasonal
day-type temporal factors; area source emissions for typical summertime Saturdays, Sundays, and a
"generic" weekday; and mobile source emissions. Biogenic emissions data are obtained and prepro-
cessed prior to inclusion in the emissions database. Detailed descriptions of emissions data used in
the ROMNET are included in Section 4.
Land use input data consist of 11 land use categories in 1/4° longitude by V6° latitude grid cells. Data
are provided for the United States 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 to aid in
the determination of surface heat fluxes.
Topography input data consist of altitude matrices of elevations for 30" x 30" cells in a 71/2° x 7V2°
grid. The data are obtained from the GRIDS.database operated by the EPA's Office of Information
Resources Management. Topography data are used in the calculation of layer heights.
2-7
-------
Processor Network;
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 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 proces-
sors 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.
Subsequent stages transform the input data into the gridded fields of temporally and spatially varying
parameter 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
processor 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 and air densities) 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.
2-8
-------
CoreModef
f t % <\ .- .. .=-... - .'
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
equations 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).
2.2.4 Quality Assurance Procedures for ROM Data Sets
Quality assurance on inputs to the ROM begins with a review of the "raw" data elements. Meteorologi-
cal 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, and temperature)
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
summaries 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
strategy scenarios are compared against base case concentrations while considering the changes in
emissions from the base scenario.
2-9
-------
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
blogenic and anthropogenic emissions. Anthropogenic emissions processors output data for each
emissions scenario and episode, because hourly mobile source emissions are adjusted by grid cell for
day-specific temperatures.
The ROM processors are run on the EPA-NCC VAX w 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 concen-
trations 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 x1 Q* 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 \a 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 Way land (1989). Schere and
Wayland (1989) used a special field-study data set from 1980, and 1980 NAPAP emissions for the ROM
simulations. However, 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 (1989) had access to extensive field measure-
ment 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.
Section 2.3 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 UAM boundary conditions.
2.3.2 Description of Evaluation Episodes
Two ozone episodes during the summer of 1985 were selected for evaluating ROM. The 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
"severe" classification is evident from the seasonal distribution of ozone > 120ppb 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 databases is described below.
Model 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 DAM domain, 8 km.
a. The ROM grid cells were overlaid on the UAM domain for the New York metropolitan area
(OMNYMAP: Rao, etal., 1987), as shown in (Figure 2-5), and the 144 ROM grid cells in a
three-cell wide band surrounding the UAM domain were selected for analysis.
b. Three-hour running averages, centered on the hour, were calculated for each hour for
each of the 144 ROM grid cells contained 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
concentrations 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 LSI to 1900 LSI) 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 layeM of the ROM (Schere and Wayland, 1989).
Furthermore, 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 a/. (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 condi-
tions are shown in Figure 2-7. After averaging at the six locations, concentrations were spatially inter-
polated (using linear averaging) along the UAM boundary. To be consistent with the ROM-UAM
Interface methodology, the hourly concentrations were used to create three-hour running averages as
described earlier.
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
distributions. 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 in 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 percen-
tile 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, although 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
compare the frequency distributions of sorted observed and sorted estimated concentrations. The
concentrations are sorted from highest to lowest and then plotted on an x-y plot. The x-axis depicts
observed data; 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 10 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 estimates from Groups 2, 3, and 4 tended to be lower than observations in the higher quantiles;
underestimates 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. Overesti-
mates 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 southern 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 analysis 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
overestimates. As discussed for the QQ plots, Group 1 tended to experience the most overestimates
(12 out of 14 daily medians were overestimated). Group 3 tended to experience the most underesti-
mates (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, 8 exceedance days were observed, although 10 excee-
dance 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, 10 exceedance days were observed
but only 6 exceedance days 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, WV, than those found in the Northeast Corridor. This result
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 perform-
ance.
July 9-11,1985
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
V
'&
-------
Corridor and significantly reduced ozone concentrations. This feature will be discussed below in the
section on boundary conditions. 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 13-15,1985
During this period, the meteorological scenario typically associated with elevated ozone concentrations
occurred in the Northeast. 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
-------
-M
July 18-20,, 1985
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
evident along the Atlantic Coast. Under light wind conditions, high concentrations were once again
positioned in small, distinct areas. Values above the standard stretched from the Delmarva peninsula
to central Massachusetts. The highest concentration (164ppb) 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. Although the
model did not replicate the extreme peaks, it satisfactorily estimated the pattern and shape of the ozone
plume near the Northeast Corridor.
2-18
-------
Boundary Conditions
One of the major uses of the databases generated by ROM is to estimate boundary conditions for the
UAM. As part of the evaluation, ROM-generated boundary conditions were 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 UAM 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 (08001ST 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. Although 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 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-
lations. In terms of model performance, model estimates and observations agreed more closely when
2-19
-------
-t
"H
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 mode! overestimated on all
five days and the average error was +26 percent.
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 con-
ditions. 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,
ranging 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. The high was apparently induced
by a cluster of thunderstorms. A stationary front extending from a low over central Indiana stretched
eastward 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
estimated 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: (1) weather data reported at each National
2-20
-------
Weather Service surface observing site, and (2) measured hourly ozone concentrations. The circled
ozone measurements 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. 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 that 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 ROM-UAM
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 2 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 Penn-
sylvania, skies were beginning to clear; although ozone concentrations remained low. Predictions in
Group 2 seem less affected by trie storm activity and are as high as 112 ppb.
2-21
-------
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 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 percent 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
UAM boundary conditions from either observed or modeled data should include careful examination of
the effects of mesoscale meteorological conditions, which have been shown to 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 was found between observed and modeled ozone concentra-
tions. The mean concentrations of the estimated and observed daily maxima agreed to within
1 percent. Concentrations at the higher ends of the frequency distribution were slightly
underestimated. At the 95 percent level, observed concentrations were 127 ppb and estimated con-
centrations were 119 ppb. As noted previously, the tendency to underestimate peak concentrations is
2-22
J
•J
-------
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 capturing 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.
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 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-
2-23
-------
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
boundary conditions.
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
currently 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 develop-
ment, additional monitoring data are needed to examine other chemical species (such as NO*,
isoprene, formaldehyde, and HMOs) and to fully evaluate model performance in rural areas.
2-24
-------
Figure 2-1. The ROMNET modeling domain; points represent corners of each grid cell.
2-25
-------
Subsidence inversion
Layer Functions
Uy*r3
Uy«r2
DAY
/>\
Cloud layer
Mixed layer
Marine layer
*»
Subsidence inversion
.*.*.;" .•?
:*£*.„•
u»-s -ANIGHT .^.£
•vi-vrf 'yft' " -*«*•"•••
__ ___
— — — — T-: — : - .. -T ;r. .-. 7- .-. .•."••
-.;•.•: ••-..:.. •"••• ...••;•/•••'•"•• '.— -•. qtlrnix
,* uytrs • *' "•.'/."• '\'. • ' •";*• •"•/»"•. T" / «i .'.*"*
•'• is-.,'.' ' ''"'' .'"... •.'••/•-. ''.•''.' .'"' *}•.*•'.•*/•-•
'' ' '1% ' ' -
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
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
- transport of aged pollutants and reactants
3 J ' Radiation inversion/ - transport of nighttime emissions from
,v 1 ( nocfumal jet layer to" stacks and warm <***&>
" - - deposition on mountainous terrain
- downward transport during |et breakdowns
- nighttime shallow mixed layer over
urban heat islands
Figure 2-2. The ROM vertical layers and their functional features.
2-26
-------
RAW INPUT DATA
CAIR QUALITY)
(METEOROLOGY)
C EMISSIONS ")
(JOPOGRAPHY)
KUM
CORE MODEL
1
_ TRANSPORT
ROM *"
^ PRf)f!Pf?SOp 1 i
~~ 1 '
'"" — ~ CHEMISTRY PREDICTED
— 1— ^ OZONE
CONCENTRATIONS
j
Figure 2-3. Components of the ROM.
2-27
-------
*" *-J
„•"*
4
: " -
*
/J>*
.*!
I OU -
UI
1
CO 40 -
O.
0.
A 30 -
I
2
§ 20 -
u.
O
u
< 10 -
1
Ui
a.
0 -1
JULY 7-22
!,! ,
1
ll ,1
'
1,
1,1
,1 .
29-MAY 12-JUN 26-JUN 10-JUL 24-JUL
AUG 7-16
'
(,„
1
III
1 .1 1 ..
7- AUG 21 -AUG 4-SEP
v* Figure 2-4. Percentage of monitors with daily maximum ozone exceeding 120 ppb in the
northeastern U.S.
Si »
>i'<
Figure 2-5. ROM grid points overlaying the UAM domain for the New York City metropolitan area.
2-28
-------
90
86
70
BS 78
-Longttud*
Figure 2-6. AIRS ozone monitoring sites divided into five geographical groups.
7S
Figure 2-7. Monitoring sites used for developing boundary conditions for the OMNYMAP domain.
REN = Rensselaer, NY
CCP = Chocopee, MA
PRK = Kent County, Rl
SCR = Scranton, PA
ALT = Allentown, PA
TR1, TR2 = Trenton, NJ
PTS = Pittsfield, MA
AGA = Agawam, MA
GRT = Groton, CT
NJ1 = Morris County, NJ
FLM = Flemington, NJ
MCG = McGuire AFB, NJ
WAR = Ware, MA
PRV = Providence, Rl
CRB = Carbondale, PA
ETN = Easton, PA
BST = Bristol, PA
CLM = Ocean County, NJ
2-29
-------
«sio oes. DATA
7aOJ PBED. DATA PTS-
SO 100 ISO . WO 350 MO
OZONE (OBSERVED), ppb
SO 100 150 MO 250
OZONE (OBSERVED), ppb
(a) Group 1: Northern Corridor
(b) Group 2: Southern Corridor
o too
50 100 ISO TOO ISO
OZONE (OBSERVED), ppb
50 100 160 200 250
OZONE (OBSERVED), ppb
(c) Group 3: Ohio Valley-Mid Atlantic
(d) Group 4: Interior Northeast
50 100 ISO 200 250
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
290
260
240
220
200
190
160-
IM
100
90-
SO-
40-
20-
0-
1
1
1
l!!
I''1
f
j!i,
i
i
III
i'l
1' I.
- '!
1
i i;
3PO
290
260
240
220
200
'* 190
£.»
o IM'
too
90
SO
40
20
0-
1 1- I! !
i jl I1 !
(
i
\}nV \\\
IW Its IM
t» IM IM Ma Ml
IW Itl IM 14]
Ml »3 2O1
(a) Group 1: Northern Corridor
(b) Group 2: Southern Corridor
300
280
2M
240
220
200
|,»
•3110
|-«
SIM
100
to-
fO'
40
20
0
ii H
! I. !i s i i i i
300
290
2M
J«
220
200
I! ,. •! :: !! K ii ,, ,i i! ii I i,
I- if ii 'i " '' n i! {; • " r ;!
til Iffl 1U
(c) Group 3: Ohio Valley-Mid Atlantic
(d) Group 4: Interior Northeast
300
290<
2U
240
220
200
'"'
l«0
14O'
120
100
00'
to-
40-
20-
0-
iu m IM m in i« 10 m BO HI m ra x»
Ju«»doy(l995»
(e) Group 5: Great Lakes
Figure 2-9. Comparison of observed (o) and modeled (x) maximum ozone for the July 1985 episode.
2-31
-------
• 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
S;
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
-------
1
MEAN DAILY RESIDUALS
30
20
10
•10
-20
0 0.1 0.2 03 0.4 0.5 0.6 0.7 0.8 0.9 1
PERSISTENCE
Figure 2-14. Mean residuals versus wind persistence for the UAM boundary; positive values are over-
prediction, negative values are underprediction.
MEAN DAILY RESIDUALS
30 p
20
10
-to 'r
•20
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
vl
:4j
-------
I *
• » . o *V • • • * \" * " " I " " *
>^. x-^—-^ i—^_ JL_^ -.
>
/
« «
'
• * '
• rf«A
• *
• • |
4Ml£
T
I
2-
^
1
\
-!•-
i
•\
4=
•* •
»
M
^
t'
•
r
•
r\
s
•
f
^
1'
.
•s
L
V
X
—
r*
*
a
fr
V
:K
»
»^'
^
/
¥
^
\f
\
M
f
J
^
4
-
•
*
\
^
r
•
*
•
(
1
"
r;
i*
I
i1
^s
•i
^x
r
(-•
*«4
u»
<
4
t
•
•
L*
•
6
^
ir
^J
£
rr
-•
4
,
^
1
•
«l
• *
4
d
»
e
/
^>
^
/
^
"
*
»
i
^
rJ
/
^
•
r.
r*
^
^
^
4
^
/•
2:
\,
\ff
H
•
33
i
ST"T'
\
\ •
j.*
™ . *
*
1 *
•4
— 8
Figure 2A 6. Division of the OMNYMAP boundary into eight groups.
2-37
-------
CONCENTRATIONS (PPB)
100
90
SO
70
60
SO
40
30
20
20
0
MEAN
OBSERVED
MEAN
PREDICTED
1 2
3 4 5 «
UAM GROUP
7 8
Figure 2-17. Mean daytime ozone concentrations by UAM group along the UAM boundary.
MEAN OVER ALL DAYS
2345
UAM GROUP
Figure 2-18. Mean residuals for each UAM group experiencing incoming flow.
i
2-38
-------
si'S'HU »i
Figure 2-19. Ozone concentrations for 0800 EST July 10,1985; observed (top), predicted (bottom).
2-39
-------
52 31 II 71
II 72 72 14
102 101 10* l!
12 II II II
1} 13 13 13
17 Ii 77 II
IS 13 7i 19
l« It 71 97
14 74 12 i<
Figure 2-20. Ozone concentrations for 1000 EST July 10,1985; observed (top), predicted (bottom).
2-40
-------
Figure 2-21. Ozone concentrations for 1200 EST July 10,1985; observed (top), predicted (bottom).
2-41
-------
Figure 2-22. Ozone concentrations for 1400 EST July 10,1985; observed (top), predicted (bottom).
2-42
I
Ai
-------
Figure 2-23. Ozone concentrations for 1600 EST July 10,1985; observed (top), predicted (bottom).
2-43
-------
•*<«5
^^
is
25oor
2000
1500
1000
200
°0
solar flux
cloud cover
0.8
0.6
0.4
0.2
0
10
IS
HOUR
£
i
~
1100
1600
1400
1200
IfWl
1IXJO
100
600
400
200
f\
°0
™ ^ „»•, ,^^M
A. .XV top of Layer 3
^-J \ _____
._____-- — ""~«N ^\ topofLayer2
\y \ .
topofLiycrl
-
y-\ ,-~\
^^^ •»«•••••, / ^* ••'* \,,-.,,,M,,,M,,
"\ /
5 11) 15 20
HOUR
a
V
10
21)
HOUR
Layer 1 wind speed
Layer 2 wind speed
Figure 2-24. Time series of selected meteorological parameters for grid cell (43,21) 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 QSPatom TRAC
O1Datom
Ozone
Hydroxyl radical
Olefinic carbon bond
High molecular weight aromatic
oxidation ring fragment
Peroxyacetyl nitrate
Paraffinic carbon bond
Peroxynrtric acid
Secondary organic oxy radical
Toluene-hydroxyl radical adduct
Toluene
NO to NOa reaction
NO to nitrate (NO3) reaction
Xylene
Methanol
Nonreactive hydrocarbons
Tracer species
TABLE 2-2. INITIAL MEAN TROPOSPHERIC BACKGROUND CONCENTRATIONS, PRECURSOR
SPECIES
Species
Concentration (ppm)
Species
Concentration (ppm)
CO
N02
NO
Ethanol
Olefins
0.1
1.0x10-3
1.0x10-3
3.5x10-4
2.1 xlO-4
Aldehydes
Formaldehyde
Toluene
Xylene
Paraffins
All others
1.12x10-3
1.4x10-3
1.4x10-4
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
JULY
Monitoring site
Ocean Co, NJ
McGuireAFB, NJ
Trenton, NJ
Trenton, NJ
Remington, NJ
Bristol, PA
Morris Co, NJ
Allentown, PA
Easton, PA
Scranton, PA
Carbondala, PA
Renssalaer, NY
Agawam, MA
Ptttsfleld, MA
Chlcopoe, MA
Ware, MA
Qroton, CT
Kent Co, HI
Providence, RI
07
s
3
W
W
W
N
N
N
E
08
3
SW,3
W
SW
W
W
W
N
E
09
3
3
3
SW, W
SW
W
W
W
N
E
E
10
g
SW,3
SW, W
SW
W
W
N
E
11
3
3
W
W
W
W
N
N
N
E
12
s
3
SW, S
W
SW
W
W
N
E
13
3
SW
3
SW
W
W
N
E
14
3
3
W
SW
W
W
NW
NW
N
E
15 16 17
3
333
SW, 3 3
W SW.W W
SW SW
W
W W
W W
NW W
NW
N N N
N
N
N
E E E
E
S
18 19 20 21 22
3 3
3 S SW, 3 S
3 SW, 3 SW, 3
W W SW.W W SW.W
SW SW
W W
W W W W
W W W
,NW W NW W
NW NW
N N
N N N N N
N N N
N
N
E E E E E
E
E
AUGUST
Monitoring site
Ocean Co, NJ
McQuireAFB.NJ
Trenton, NJ
Trenton, NJ
Remington, NJ
Bristol, PA
Morris Co, NJ
Allerrtown, PA
Easton, PA
Scranton, PA
Carbondale, PA
Rensselaer, NY
Agawam, MA
Pfttsfield, MA
Chlcopee, MA
Ware, MA
Grolon, CT
Kent Co, RI
Providence, RI
07
3
3
SW, 3
W
SW
W
W
NW
NW
N
E
08
3
3
SW. 3
W
SW
W
W
N
E
09
3
W
W
N
N
N
E
E
10
sw.s
SW
W
W
N
E
11
3
3
3
3
W
SW
W
W
NW
NW
N
N
N
E
12
s
3
SW.W
SW
W
W
W
N
N
N
E
13
3
3
S
S
sw.w
SW
W
W
N
N
N
E
14
3
S
S
W
SW
W
W
NW
NW
N
N
E
15 16
S 3
3
W SW, W
SW
W W
W W
NW W
NW
N
N N
N
E E
1. Compass points indicate the sites that 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
Mean
Obs. Model
56.8
52.7
79.9
76.6
66.6
61.1
79.6
74.1
Std. dev.
Obs. Model
Daytime
26.0
28.0
hourly
19.3
18.7
Daily maximum
25.1 22.5
29.0
21.3
95th percentile
Obs. Model
102.0
102.0
125.0
131.0
103.2
97.0
120.9
114.1
Maximum
Obs. Model
218.0
219.0
218.0
219.0
169.2
161.3
169.2
161.3
2-47
-------
I
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
88.6
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)
JulyS
July 9
July 10
July 11
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 15
Modeled
(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
n
250
171
233
320
165
187
215
200
169
35
70
222
265
177
221
110
168
200
2
159
222
331
Persistences
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. Box 12767
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
Sections 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. Descriptions of the meteorological conditions and
observed ozone concentrations during the selected episodes are included in Appendix D.
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
relative priority of each episode for simulation;
4. meteorological conditions within the 10 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 State/local monitoring sites in the ROMNET domain were extracted from AIRS for review. The
3-3
-------
resulting 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
exceedance 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
spatial distribution of high ozone concentrations with emphasis on areas along and adjacent to the
Northeast Corridor; (2) the frequency of .ozone exceedances; and (3) the representativeness of meteo-
rological 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 ranking established a relative priority of the
episodes for model simulation. Next, for the 10 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 subre-
gions along the Northeast Corridor and adjacent areas as shown in Figure 3-2. In this scheme, each
subregion 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 X^WfaxfmuniHourfy Ozone
First, a comparison was made between episodes by subregiori (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).
;- Number of Exceedance Days ( >1 25 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 episode 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 ail 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,
subregion 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, as illustrated below:
For the episode 7/4/88 - 7/18/88
Part 1 - maximum hourly ozone:
Part 2 - number of exceedance days:
score
4
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 10 top-
ranked episodes. These analyses were performed to characterize the meteorological conditions and
transport flow regimes associated with each episode to: (1) rerank the order of the top 10 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 recirculatiqn 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 that 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 wtthin-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 Dis-
persion (ATAD) model (Heffter, 1980) 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:
(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 10 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 10
episodes. These summaries provided information on temperature, cloud cover, wind speed, and
direction. This information was used in conjunction with the flow regimes to better understand the
meteorological conditions characteristic of each episode.
The flow regimes, meteorology conditions, and ozone concentration data were summarized for each
exceedance day of each top 10 episode. This information was condensed and submitted to the
Modeling Committee for review and final selection of episodes for simulation.
3.2.5 Flow Regimes During the Top 10 Episodes
The occurrence of the key flow regimes during the 10 top-ranked episodes are summarized below.
Each episode is described in rank order as given in Table 3-2.
Epfeode. #1: July 4 - July48,1988
Episode #1 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-June23,1988
A persistent surface trough along the Corridor was also the main feature of Episode #2. 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.
Epfsode #3: July 12 - Jury 23, 1980
High pressure systems over the east coast and just offshore dominated Episode #3. 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 20f 1983
Episode #4 was characterized by fronts and a weak pressure pattern within the Corridor. Recirculation
conditions occurred on 6 of the 10 days. A west to northwest flow occurred on other days.
3-8
-------
Episode #5: August 9 * August 17,1988
vlv > % i. -. ^ —-- .,"--..-
A high pressure system off the east coast was the main meteorological feature during Episode #5 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
-f • •.?£. ' , % , , .. .-•-..
A trough along the east coast with weak high pressure over western Pennsylvania were the main
meteorological features of Episode #6. 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 Episode #7. 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 #s: July 16* July 26> 1987
Weak surface flow and fronts were the main meteorological features in the Corridor during Episode #8,
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 Episode #9. 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 that 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 1988) 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 10 episodes displayed a similar
spatial extent of high ozone levels. Also, from the meteorological analysis, the July 1988 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, the July 1988 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 1988 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 10 episodes in order of ranking
indicates that the second- and third-ranked episodes (June 1988 and July 1980) contained flow
regimes that were, to a large extent, represented in the July 1988 episode (i.e., along-Corridor and
westerly flow). Additionally, there was some concern that the dominating westerly flow in the June 1988
episode would transport the predicted Corridor ozone plumes offshore over water and, thus, compli-
cate the interpretation of model predictions.
The fourth-ranked episode (June 1983) was characterized by interurban recirculation conditions, which
was quite distinctive from that in other top 10 episodes. Thus, June 1983 was chosen to complement
the July 1988 episode for strategy simulations.
In addition to the July 1988 and June 1983 episodes, two 1985 episodes were selected for use in the
ROMNET evaluation of the ROM. Episodes from 1985 were needed, because 1985 was the year of the
ROMNET base emissions inventory. The highest and second-highest 1985 episodes were selected.
These episodes were: 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
episodes 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
-------
Daily Maximum Ozone Percent les — 1988
250 -I
Q_
Q-,200-
c
O
c.
(D
^ 100-
O
O
C 50-
O
N
O
0-
I
1
01 MAY
1
01JUN
01JUL 01AUG
Date
1
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
-------
'!
'-I
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
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 13 - June 23, 1988
July 12 -July 23, 1980
June 9 - June 20, 1 983
August 9 - August 17, 1988
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?, 1983
July 29 -August 5, 1984
July 6 -July 14, 1987
August 1 3 - August 21 , 1 983
June 1 4 - June 21 , 1 987
July 9 -July 16, 1987
September 8 - September 14, 1983
May 28 - June 1 , 1 988
July 15 -July 23, 1985
May 27 -June 2, 1987
August 1 3 - August 1 8, 1 987
September 13 -September 23, 1987
July 7 -July 14, 1985
May 27 -June 2 1986
July 3 -July 9, 1986
June 27 - July 6, 1 985
June 1 8 - June 28, 1 986
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
4
5
Synoptic meteorological pattern and regime description
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).
Transition regime - high pressure along east coast extending SE and NW:
weak surface flow.
High pressure system off the east coast:
westerly or along-Corridor flow (back side of HIGH).
Pre-frontal: along Corridor flow.
Cold front or trough along east coast:
variable or southwesterly flow.
Frontal pattern between New York and Boston:
easterly flow north of front;
along-Corridor flow south of front.
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
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
and
Thomas E. Pierce*
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.
-------
4.1 INTRODUCTION
During the ROMNET project, 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 describes each of these scenarios and the development of the corresponding anthropo-
genic and biogenic emissions inventories. Section 4.2 presents an overview of the basic emissions
inventory structure common to all scenarios, along with a description of the methodologies used for
spatial, temporal, and species allocation of the emissions data. Section 4.3 provides a discussion of
the development and characteristics of the 1985 base year scenario. Section 4.4 presents the devel-
opment of the 2005 scenarios and control strategies. Control efficiencies 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, the ROMNET inventories provide emissions data for volatile organic compounds (VOC),
nitrogen oxides (NO*), and carbon monoxide (CO), which are precursors in tropospheric ozone
formation. NOX emissions are partitioned into NO and NOg. VOC emissions are assigned into one of
ten reactivity classes: olefin (OLE), paraffin (PAR), toluene (TOL), xylene (XYL), formaldehyde (FORM),
other aldehydes (ALD2), etnene (ETH), isoprene (ISOP), nonreactive (NONR), and methanol (MTHL).
VOC classifications 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 exceedanees 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
pollutant (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
Inventory. 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 aggregates of highway vehicle emissions. The mobile source data further classify VOC
emissions into evaporative 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 (SCC), and Standard Industrial Classification (SIC).
The point, area, and mobile source emissions inventories representing the various ROMNET scenarios
were developed by Alliance Technologies Corporation and delivered to Computer Sciences Corpora-
tion (CSC) for processing 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, NO*, and CO.
Entries for these other pollutants are removed in the Model Data Extraction step, 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 PSPUT program. The Spatial Allocation Module assigns
emissions to modeling grids; 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,
because 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 distributions of other emission surrogates; thus, they are specific to counties and source cate-
gories.
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
factors can be found elsewhere (EPA, 1989; Modica et a/., 1988; and Walters et a/., 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 NAPAP1985 emissions database, which in turn is
derived from the NEDS. The NAPAP database provides a number of benefits as a starting point for
ROMNET. The primary benefit 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 particulate matter. The overall
database includes two basic inventory types: an annual inventory and a set of 12 temporal scenario
inventories representing atypical weekday, Saturday, and Sunday in each of the four seasons. Further
information on the NAPAP inventories is provided by Saeger era/. (1989).
4.3.2 Adjustments 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 subsection. Other
changes made to the NAPAP inventory, including quality control review of both major CO point sources
and VOC emissions from hazardous waste treatment, storage, and disposal facilities (TSDFs) are
described in Alliance Technologies Corporation (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
different 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). None •
of the six categories rivals mobile sources for VOC emissions; therefore, 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 proce-
4-7
-------
dures used to calculate these factors are described in Alliance Technologies Corporation (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 cate-
gories, including degreasing and other industrial solvent use categories. However, these operations
are conducted indoors in controlled environments and, hence, are not influenced by outside
temperatures 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
categories 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 the ROMNET was to produce emissions estimates that
(1) reflect the EPA's latest revisions to the MOBILE emission factor model (EPA, 1989), and (2) 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 (QMS), it calculates
emission factors on a gram per mile basis for different types of vehicles at different speeds. These
emission factors 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 the 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, MOBILE3.9, 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
:l
4
-------
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
emission 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 vol-
atility. Running losses occur while the vehicle is running; 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
variation, respectively, on VOC emission factors predicted by MOBILE4. The figures also illustrate the
estimated 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
assumptions. There are 12 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 12 mobile source categories. On the
4-9
-------
other hand, lumping the 12 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 Stepl, each mobile source category in the annual inventory was adjusted separately to reflect
emission factors from new versions of the MOBILE model. For MOBILE3.9, running loss and excess
evaporative emissions are added in this step as described by Alliance Technologies Corporation
(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, gasolirie 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
4-10
-------
inventory are grouped together at Step 2, the final adjustment table contains "composite" emission
factors, or weighted averages of factors for the various classes that 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, emission control efficien-
cies, 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. The profiles were recalculated using three separate sets of VOC speciation factors to calculate
molar emissions by reactivity class for the evaporative, gasoline exhaust, and diesel exhaust compo-
nents of VOC. Total moles of each reactivity class were then calculated by combining the results for the
three VOC components. With this recalculation, the final speciation was representative of grid- and
day-specific temperatures.
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
emissions processors.
4.3.3 Characteristics of the 1985 Base Case Emissions
Source Category Distribution
The distribution of emissions between major source categon'es (point, area, and mobile) and within
each of these categories is an important component of an emissions inventory. Although the distribu-
tion varies from State to State, the discussion in this subsection will focus on 'the distribution of
emissions within the ROMNET domain, excluding emissions from the Canadian portion of the region.
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 source 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, but 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
4-11
-------
:4
1
coatings, 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:
solvent 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. The evaporative'emissions are 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 simu-
lations, the relative contributions 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 80 percent, are from utilities, as shown in
Figure 4-5b.
Major NO* 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 ceils 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.
The highest NOX emission densities seen in Figure 4-6 are not confined to large cities, but are also
associated with large combustion sources in rural areas. Figure 4-7 shows the locations of industrial
4-12
-------
boilers and utilities with greater than 500 tons per year of NOX. Although 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, NOXl 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, NOXl 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 relatively constant throughout the day.
Diurnal NOX emission rates for mobile sources contain large variations due to the assumed traffic
pattern. 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, but 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 factors for selected temperatures and temperature
ranges. Figure 4-11 illustrates the spatial distribution of mobile source VOC emissions at 1500 EST on
a "cool" day and on a "warm" day simulated as part of ROMNET. Domain average daily temperatures
for these days were 71.7 °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
"warm1 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 emissions vary 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
credible 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 subsection 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
emissions 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 subsection provides a description of the Phase I and Phase II baseline scenarios. VOC and NOX
control measures for 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 mainte-
nance (1/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. The levels were held 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
NO* protocol caps were kept for all of the Phase I strategy inventories, and no further NOX controls were
applied.
BS05 - Phase II 2005 Baseline: NO* Growth
Two important changes were made between Phase I and Phase II of the ROMNET 2005 baseline
scenario. 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
measures: 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 5 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 II 2005 baseline inventories. This reduction trans-
lates to a 5 percent reduction in overall VOC emissions. A similar reduction, about 29 percent or
1. Revised Draft Protocol to the 1979 Convention on Long-Range Transboundary Air Pollution Concerning the Control of
Emissions of Nitrogen Oxides or Their Transboundary Fluxes. United Nations Economic and Social Council. May 10,1988.
4-16
-------
1100 tons per day regionwide, occurred in mobile source NOX emissions. This reduction was offset by
a 1200 tons per day (14 percent) increase in point source emissions, due to the removal of NOX
emissions caps. Overall, NOX emissions from all sources increased by less than 1 percent between the
Phase I and Phase II baseline projections.
2005 Baseline Canadian Emissions
A single set of growth and emission control assumptions was used for Canadian emissions in both the
2005 baseline scenarios and all control strategy scenarios. No additional controls were applied in
Canada as part of the ROMNET strategies. Thus, complicated interpretation of the impact of controls
applied in the U.S. portion of the region was avoided. However, there were some differences between
the Phase I and Phase II Canadian inventories; 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, these controls included (1) the current U.S. FMVCP standards for mobile sources,
and (2) the emission controls in Strategy 1 (see below) applied to point and area sources. Current U.S.
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
scenarios 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 rejative benefits of VOC controls 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 concentrations in the Northeast to below 125 ppb?
[2005 baseline, CS01, CS05, CS10 - CS16, CS18, CS19, CS23]
4-17
-------
4. What are the effects of reactivity-based strategies in reducing regional ozone levels?
[CS15.CS20]
5. How does the large uncertainty in biogenic emissions alter conclusions regarding the effec-
tiveness of control measures?
[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
target 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
"windows 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
Strategy I is the Phase I maximum technology scenario that 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. The 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-
4-18
-------
tive emission 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 percent through substitution for oil-based coatings with water-based coatings. Finally, miscella-
neous nonindustrial (commercial/consumer) solvent emissions were reduced by 20 percent. Control
measures in this scenario are summarized in Table 4-3. Overall, VOC emissions in CS01 were reduced
by about 45 percent from the Phase 12005 baseline inventory. Emissions of NO* were 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
maximum technology inventories. These scenarios were developed to study the impact of applying
maximum technology VOC controls in different portions of the ROMNET domain. In CS02, maximum
technology controls were applied to the Northeast Corridor, but 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, emissions in CS02 were assumed to be 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 U.S.
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, emission controls are assumed to
be 100 percent effective. For the reasons listed in Section 4.5.1, 100 percent effectiveness 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 125ppb. It was decided to apply the concept of control effectiveness to a more
stringent strategy (CS23). Thus, CS04 was not simulated.
4-19
-------
CSpSJ 1989^oplise 75 megawatts. Point and area source VOC controls were 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 U.S.
portion of the region, CS05 achieved a 32 percent reduction in VOC emissions and a 32 percent
reduction in NOX emissions from the Phase II 2005 baseline scenario. Emissions of CO were reduced
by 4 percent
4-20
-------
CS06 -,CS09, ^9851. and 1985fH: Blogenic Sensitivity Scenarios for CS01 and CS05 and the 1985
f •. V t •-
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. The scenarios include 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 were 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 were altered in an
across-the-board manner from the 'best available science" estimates generated by the Biogenic
Emissions Inventory 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 0.125g/mile. In addition, the mobile source
controls (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 vessels. Controls were added for four new point
source categories and to generic non-CTG point sources.
4-21
-------
For utilities, the 5 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 emission 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 vessels 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
emissions, and a 21 percent reduction in CO from the Phase II2005 baseline scenario.
CS11 - CS14: Analyses of VOC Versus NOX Controls
.. ,tl., •; "• ^""* f
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. 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, but
mobile and area source NOX emissions were kept 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
emissions in the U.S. portion of the region by 45 percent. In contrast, mobile source controls in CS14
reduced NOX emissions by 9 percent.
CS15: Maximum VOC and NOy 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
emissions over the entire ROMNET region. A cap was placed on the reactivity of all solvent emissions,
4-22
-------
at a level midway between the reactivities of isopropenol 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 GS10.
The second reactivity reduction measure was the conversion of the entire Northeast Corridor highway
fleet to 100 percent methanol fuel vehicles. This measure produced an 88 percent increase in mobile
VOC emissions over those attained for gasoline under the tight maximum technology tailpipe
standards. However, because 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, as described in Section 4.5.8. 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-weighted emissions of stationary sources by about 29 percent
for the U.S. portion of the region compared to maximum technology strategy. Methanol conversion
reduced the overall reactivity-weighted mobile source emissions in the Corridor by about 22 percent
compared to the maximum technology strategy and by 69 percent compared to 2005 baseline mobile
emissions. By combining maximum-technology controls, solvent substitution and methanol vehicles as
in CS15, reactivity-weighted emissions in the Corridor were reduced by 74 percent compared to the
2005 baseline, but only 33 percent compared to maximum technology controls alone.
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 addi-
tional 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 Balti-
more/Washington, DC, VOC was reduced by 80 percent from the 2005 baseline, and 88.5 percent from
1985 levels.
4-23
-------
For NOXl maximum technology controls were applied as in CS10, except in New York City, where
mobile sources were controlled but point source emissions were at the 2005 baseline level. The NOX
emissions reduction in this area for CS16 was 32 percent rather than 58 percent as in CS10.
CS17: Maximum Technology NOX with "Low" Bldgenics
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 were the same as in CS11 (maximum
technology controls for NOX with VOC and CO at the 2005 baseline). Biogenic emissions were reduced
by a factor of 3 from base case levels as in scenarios 1985L, CS06, and CS07.
•-..-" f- n v f
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.
CS19t GS18 with Alteration of NOX Controls In Baltimore/Washington, DC
CS19 was identical to CS18 except for the treatment of NO* 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 area 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 were 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. As a result of applying the lower-reactivity solvents, total
4-24
-------
reactivity-weighted emissions were reduced by 25 percent compared to the 2005 baseline in areas
outside of the Corridor. Within the Corridor, where methanol vehicles were also introduced, total
reactivity-weighted emissions were reduced by 55 percent compared to the 2005 baseline.
CS21 arid CS22: Brbgenic 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 were set at the "Low" end of the
uncertainty range; in CS22, biogenics were set 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. 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
programs;
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
strategy); 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, NO* 0.2 g/mi.
4-25
-------
>J
As a result of considering rule effectiveness in CS23, VOC emissions increased by 64 percent, NOX by
25 percent, and CO by 7 percent over those in CS19 in the U.S. portion of the region. Still, considering
rule effectiveness with CS19 controls, emissions in CS24 were lower than those in the 2005 baseline by
48 percent for VOC, 40 percent for NOX, and 16 percent for CO.
CS24 and CS25: Spatial Analysis of VOC and NOX Controls - Transport SensitivityScenarios
For CS24, the CS19 controls were applied to all counties inside the Northeast Corridor, with emissions
at 2005 baseline levels elsewhere in the domain. For CS25, the CS19 controls were applied in areas
outside the Corridor, with emissions at 2005 baseline levels elsewhere. The purpose of these two strat-
egies, through comparisons with the 2005 baseline and CS19 results, was to examine the effects of
regional strategies on ozone levels in the Northeast Corridor - specifically, changes to ozone and
precursor transport into the Corridor.
4-26
-------
4.5 EMISSION CONTROL MEASURES
Section 4.5 provides the specific control efficiencies and emissions standards contained in the 2005
baseline scenarios and the ROMNET control strategies described in Section 4.4. Projection and control
methodologies together with emissions growth factors used in developing these scenarios are
provided in Appendix J.
For point sources, controls were specified by emissions "pod." Each pod represents a group of 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 (i.e., area source
SCCs) that emit VOC, NOXl or CO. The specific pods and area source categories are listed in
Appendix J.
Sections 4.5.1 through 4.5.3 present information on: VOC stationary-source controls; NOX stationary-
source controls; and VOC, NOX, and CO mobile source controls, respectively, in the 2005 baseline
scenarios. Sections 4.5.4 through 4.5.6 provide similar descriptions of the control measures used in
the maximum-technology-based strategies. Controls in the Clean Air Act strategy are presented in
Section 4.5.7. Details on the reactivity-based strategies are provided in Section 4.5.8.
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 et al., 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 (Eff1985) and
2005 efficiencies (Effaoos)- The approach for obtaining values for each of these efficiencies for VOC
area and point sources is described next.
4-27
-------
Area Sourcesi;i
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
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 methodolo-
gies 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 emission 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.
Pofnt Sources.
The NEDS point source inventory includes an emission 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
2. 53.5 percent for category 103 and 48.2 percent for category 104.
4-28
-------
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 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 are 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 parameters can be used to estimate the overall reduction efficiency for a
given area source category:
4-29
-------
Overall efficiency = [control technology efficiency] x [rule effectiveness/100]
x [rule penetration/100]
Actual emissions = [uncontrolled emissions] x [1 - (overall efficiency/100)]
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 confidence level was expressed by letter "grades" from A to E, with A repre-
senting 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
nonzero 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 "average1 overall efficiency, obtained by
multiplying 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 effi-
ciency 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 X.
4.5.2 NOy 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.
4-30
-------
Phase 1—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
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;
AJ, LOW-NO^ burners, staged combustion air, reduced air-tp-fuef ratio, or steam injection used on:
1) New sources ... ;, ',, , , " , , ,, - '-;; " " *' <
" \Utilitfes ; „ " - "-.','"'',''
' ' * " '"
, ,
f/- /;; 'Natural gas - -, v ,, *" ^ ,< , f
\;tndustriarcombustion , ,- ',„ t "- ? ' - '" -
'.' " "•-' -'-Coal , . "" ~" ~ ,
'•>'^m'.- I " , ' ' - - """ \""
\ ;,r 'Natural gas , - ' >'-'-' ,1 ,,
*'• ' "~ "* ^' ' , ' ,, : - '
2) Existing sources-with the same rankings as above for source categories and fuels
8. Selective catalytic reduction-applied; to new sources before existing sources and with the same
f,- rankings as above for source categories and fuefs, ,- -
Table 4-7 summarizes the point source control measures needed to meet the NOX cap in each
ROMNET State.
Phase 11-2005 Baseline NOX
The Phase II 2005 baseline inventory, like all of the Phase 1 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.
4-31
-------
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).
4.5.3 VOC. NCy. 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 MOB1LE4, respectively. MOBILE4 inputs for the Phase II baseline are
given in Appendix L The effects of current inspection/maintenance (I/M) programs on baseline
emissions 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'BASIC? I/M PROGRAM CHARACTERISTICS
tfM PROGRAM;
STARTYEAR (JANUARY Ifr ' * -
PRE-1981 MYR STRINGENCY-RATE::,
MECHANIC-TRAINING PROGRAM?:
FIRSTMODEL.YEARCOVERED!:
LAST MODEL YEAR COVERED? "-
VEHICLETYPES COVERED;'
198t+ MYRTESTTYPEi
1981* MYRTEST OUTPOINTS!, ,
ANTFTAMPERING PROGRAM SELECTED!
ADDITIONAL MOBILE4 t/M INPUTS
WAIVER RATES: - , '
COMPLIANCERATE::
INSPECTION: TYPE:
INSPECTION FREQUENCY;'
1983"
20%
NO
1970 '
2020 \
LDGV,LDGT1,LDGT2
IDLE ONLY
1.2?6 CO, 220 PPM HC
'NONE
5% (ALL YEARS)'
100%
CENTRALIZED
ANNUAL
MOBILE4 runs produce the following ideal I/M control efficiencies, weighted across the 2005 vehicle
fleet:
Total VOC
Gasoline exhaust;
Gasoline evaporation
Diesel
Nitrogen oxides
Carbon monoxide
Percent Control
12.1
18.1
9.3
0.0
4.0
' 19.9
4-32
-------
The following I/M program effectiveness percentages were obtained from program assessments
performed by the EPA Office of Mobile Sources and were developed from comprehensive reviews of
the characteristics and performance of each program.s Program effectiveness for NOX is taken to be
the same as for VOC. These effectiveness ratings combine the effects of program deviations from the
prototypical 5 percent waiver, 100 percent compliance rate, and centralized annual program on which
the basic I/M reductions were based.
State''" "
Connecticut
Delaware'
District of Columbia
Indiana
Kentucky:
Maryland; ~ -
Massachusetts^
^ew Jersey
NewYprjc,',,
Pennsylvania
Rhode Island
Virginia"" "' ;
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 Maximum Technology VOC Stationary-Source Controls
Maximum technology VOC 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 tech-
niques that form the basis for the maximum technology efficiencies. The major references for maximum
technology controls were the Ozone NAAQS Cost Analysis (Battye ef a/., 1987), and the Emissions
Reduction and Cost Analysis Model (ERCAM) for VOC (Pechan, 1989). Efficiencies given in these ref-
1 985 Program Effectiveness (%)
yoc \ . - ,
84 '-.
80
84 '
53
90
87
63
84^ " "',"
50
78 •• ' /
44 , , ,„,,",
44
CO
81
80
81
53
88
85
^56
81
50
74
-33
34
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. 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
-------
erences were based on primary sources such as the 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
efficiencies match or supersede any applicable NSPS rules, and all efficiencies were assumed to be
100 percent effective.
4.5.5 Maximum Technology NCv Stationary-Source Controls
Maximum technology NOX 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,
Iow-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 technol-
ogy 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 Maximum Technology VOC, NCy. and CO Mobile Source 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
baseline:
CS1I/M PROGRAM ' ' l , '
I/M PROGRAM SELECTED
STARTYEAR (JANUARY 1); 1983
PRE^SatMYR STRINGENCY-RATE: 20%.
MECHAN1CTRA1NING PROGRAM?: NO
FIRSTMODEL.YEAR COVERED: , 1970
LAST MODEL YEAR COVERED: 2020
VEH1CLETYPES COVEREDr LDGV, LDGT1, LDGT2
19814 MYR TESTTYPE: LOADED/ IDLE
1981 *MYR TEST OUTPOINTS: 13%. CO/ 220 PPM He'
ANTtTAMPERING PROGRAM SELECTED " ',
START YEAR (JANUARY 1>: 1990
FIRSTMODEL YEAR COVERED: 1970
LAST MODEL YEAR COVERED: 2020
VEHICLETYPES COVEREDr LDGV
ANNUAUNSPECTION: AIR PUMP, CATALYST, FUEL INLET,
' PLUMBTESMO LEAD TEST
5. Information provided by Larry Jones, Acid Deposition Branch, U.S. Environmental Protection Agency, Research Triangle
Park, NC. Typical efficiencies for low-cost NOX controls.
4-34
-------
For Phase II, the Phase I enhanced I/M program was modified to add heavy duty vehicles and addi-
tional variables required by MOBILE4 (5 percent waiver rate, 100 percent compliance, centralized/an-
nual program 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 NO* emissions
standard reduction from 5 to 4 grams/brake-horsepower-hour was also included in CS10 as an approx-
imation of the control limit of current diesel technology.
., '" -
Light Duty Vehicles
Ughtbuty Trucks
1) 3750-5750 Ib"
2) 5750- 8500 Ib'
Heavy Duty Diesels
HC
0.125
'0.380
0,460
None
NOX
0.20
0.70
1,10
4.00
'Units ; - ,
grams per mile
grams per mile
grams per mile
grams per brake horsepower hour
The following set of MOBILE4 base emissions and deterioration rates was used to simulate these
emissions standards:
•- -,-'-"; HC
Light Duty Vehicles
tight Duty Trucks 1
Light Duty Tritcks 2
Heavy Duty Diesels
ZM DR Units " ^
0.085 0.028 0.127 ,0.017 grams per mile
0,14 '0.05 ,,,0.35, 0.03 grams per mile
0.22- 0.08 0,64 0.04 grams per mile
^0,0 , 0,0 3,11 0.0 grams per brake horsepower hour
2M«Zero ntile rate, OB ^Deteriprationrate'per 10,000 mi(e» " ''„ %
To approximate full penetration of these standards into the fleet, this MOBILE4 run included introduc-
tion of these standards for light duty vehicles and trucks in 1995. In addition, resulting emission factors
for 2020 were used, because 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 following reductions over the 2005 baseline uncontrolled and "basic" I/M composite
emission factors:
^TotaiVQG
T^ gasoline exhaust
-'«, Gasoline evaporation
- ;' Nitrogen oxides ' ,
, -Icarbon monoxide
- Percent Reduction Over:
-Uncontrolled Basic l/M
. .. 62.0 , , ' ^ 56.2
- 66.3 , 65.5"
56,7 - - •- , 522
p.o ; o.o
49.9 47.8
3Z8 22.2
4-35
-------
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:
Attainment Category
l^attafnment , ~ '
2-margfnal nonattafnrhent
3-moderate nonattainment
4-serious nonattainment
5-severe nonattainment
*-' f Design Value*
'< 0.12 ppm
0.13 ppm
0.14 -0.15 ppm
, 0.16-0,18 ppm
^ 0.19 ppm
Attainment Date
i * j
December 31, 1995
December 31,1995
December 31, 2000
December 31, 2010
* baaed on 4th-hlghest dally maximum concentration over 3 years.
The VOC point and area source control efficiencies 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-, '
Phifadefphia '
Atlantic Cfty
New York City s ,',/ " v~,
Poughkeepsie, NY \ * J
Hartford, CT , % "; '''
Providence, Rl -
'^Massachusetts
^MancKester-Nashya, NH ' ['
Portland, ME
'Harrisburg^ PA
Huntington -Ashland, WV-KY-9H
Toledo, OH
Vehicle tailpipe standards included in this strategy are given below:
Ught Duty Vehicles
Light Duty Trucks
1) 3750-5750 ib
2) 5750-8500 Ib
HC
0.25
0.25
b.so' ,
NO*
0.40
1.00
1.00
\
Units
grams per mile
grams per mile
grams per mile
• Start
Year
1994
1995
1995
Inputs to MOBILE4 used to generate mobile source emission factors for this strategy are given in
Appendix L
4-36
-------
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.
4.5.8 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
purposes of this cap, reactivity is defined as the overall reaction rate with hydroxyl radical per unit
weight of the solvent:
Reactivity
where:
/ refers to the different Carbon Bond-IV classes;
FtR/ is the reaction rate of CB-IV class / with the hydroxyl radical at 85 °F:
43,187 reactions • min -1 for olefin;
1,203 reactions • min -1 for paraffin;
9,315 reactions • min -1 for toluene;
36,433reactions- min-1 forxylene;
15,000 reactions • min -' for formaldehyde;
24,335 reactions • min -1 for other aldehydes;
12,194 reactions • min -1 for ethene;
142,000reactions- min-1 forisoprene;
1 reaction • min -1 for nonreactive hydrocarbons;
1,600 reactions - 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 • min -1, which is midway between the reactivities of isopro-
penol (30,000 • min-1) and ethyl acetate (40,000 • 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.
4-37
-------
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. These
percentages are based on M100 evaporative and exhaust emission factors of 0.055 and 0.565
grams/mile, respectively. 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 that included a
breakdown of the exhaust emissions into methanol, formaldehyde, and other nonmethane hydrocar-
bons as 88.5, 2.7, and 8.8 percent by weight, respectively. The other nonmethane hydrocarbons were
further speciated using previously-existing data for exhaust hydrocarbon speciation.7
i
I
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.
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
emissions for estimating the production of photochemical oxidant smog (Chameides et a/., 1988;
Trainer et a/., 1987). Even before the publication of these studies, the EPA reported on the develop-
ment 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 et a/., 1987). These estimates
were used for early parts of the NAPAP. In preparing for the 1985 NAPAP emissions inventory and
during tests of later versions of the ROM, researchers at Washington State University and the EPA col-
laborated on combining features of their two biogenic emission systems (Young et a/., 1989). An
outcome of this collaboration was the development 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). Section 4.6 discusses the formulation of this system, called the Biogenic Emissions
Inventory System (BEIS). Information on the BEIS can also be found in Milich et a/. (1991) and in Pierce
et a/. (1990a).
4.6.2 Description of the System
Calculations with the BEIS require consideration of biomass, emission factors, and environmental
factors. The basic equation for these calculations can be expressed as follows:
where ER is the emission rate (g/s/model grid cell), / is the chemical species (such as isoprene or
monoterpene), / is the vegetation type, BF is the leaf biomass factor (g/m2), EF is the emission factor
(/zg/g-leaf biomass/h),and F(S, T) is an environmental factor that accounts for solar radiation S and leaf
temperature T.
4-39
-------
Btomass 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 et a/., 1980). The land use
data, some of ft 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 a/. (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:
Forest
group
Oak
Other deciduous.
Coniferous
High isoprene
deciduous
185
, ^ eb:\
3$
Emission Category
Low isoprene Nonisoprene
deciduous deciduous
, 60
60
185
26
90
"'26
Nonisoprene
coniferous
70
'" , ,135™
'559 "'
Nonforest 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
function. For each month, deciduous vegetation (any nonconiferous vegetation class) within a county
was assumed to have either full biomass or no biomass. For oxidant modeling, this assumption is not
critical because most high ozone episodes occur during the summer months.
Emission Factors:
The emission factors in the 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 (/ig/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 (pg/m2/h) can be compared for vegetation types, as shown in Table 4-15.
4-40
I.
-------
Emission Category
,'," 'species
~ tsoprene
' a-pinene'
. >r Monpterpene ~
'Unidentified'
Highfsoprene .
deciduous
14.69
0.13
0.11
- 3.24 ,
, Low isoprene
< ' deciduous
' s 6.60
0.05
' 0.05 '
1.76
Nonisoprene
• deciduous
0.0
' '0.07
0.07
1.91
Nonisoprene
coniferous
0.0
1.13
1.29
1.38
Emission fluxes in Table 4-15 are expressed in terms of total nonmethane 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 jug/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 the
BEIS. The equation used is (Williams etai., 1990):
0 = 0. 74 exp(0. 079 7S)
where 0 is the nitrogen flux (ng of N m-2s-1), and Ts is the soil temperature (°C) as estimated from
Ts = 0. 70 Ta + 3. 6, where 7a 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.
Envfronmental 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 et
a/., 1980, 1981). The BEIS includes adjustments for temperature and sunlight using these relation-
ships. It also attempts to simulate the vertical variation of leaf temperature and sunlight within a forest
canopy.
4-41
-------
The BEIS incorporates an adjustment factor for isoprene emissions (Tingey era/., 1981) that is given by
the following:
a
dot-
exp(-6(T-c))
-d
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 (T) 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.
Lfght intensity
Empirical Coefficients
800
400
200
100
a
1.200
"6.916
1 0.615
0.437
b
0.400
6.239
0.696
0.312
C
28.30 '
29.93
32.79
31.75
d
0.796
6.462
0.077
6.160
e
J.OO
1.95
4,75
10.73
• •£
Existing laboratory data for nonisoprene emitting plants have thus far identified only leaf temperature as
an important variable CTingey etal., 1980; Tingey, 1981). The environmental adjustment factor used for
nonisoprene emitting plants is given by:
F(7~) = exp(a[T-30])
where values for the coefficient a are given below.
Compound
ct-pihene
Monoterpene
Unidentified
- Empirical Coefficient
a
0.067
0.0739
0.0739
A major refinement for the 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
distribution. 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
4-42
-------
leaves are more effective at absorbing visible light than other portions of the spectrum (Baldocchi etal.,
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 that 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. The 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 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 cal-
culation is performed once per model simulation. The computer intensive portion of the 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 et a/., 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 AL.D2.
. In addition, 1 mol unidentified (or unknown) is treated as 0.5 mol OLE, 8.5 mol PAR, and 0.5
molNONRS.
8. Memorandum from Kenneth Schere, EPA Office of Research and Development, Research Triangle Park, NC, to Thomas
Pierce, EPA Office of Research and Development, Research Triangle Park, NC. Final Guidance on Treatment of Unknown
VOC Biogenic Compounds and Emissions Inventories for ROMNET Applications. April 12,1989.
4-43
-------
Estimates of NO emissions require the grassland area in a grid cell (available from the Geoecology
Data Base, Olson etal., 1980) and the surface temperature (provided by the ROM meteorological input
processors). 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 the 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
emissions. 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 due to a lack of extensive data sets for specifying critical factors such
as biomass emissions rates for various vegetation species and the effects on emissions of environmen-
tal stress (e.g., insects, drought), 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 cal-
culated by the BEIS in the biogenic sensitivity scenarios 1985H, 1985L, CS06 - CS09, CS17, CS21, and
CS22.
4.6.4 Characteristics of the Bloqenlc VOC Emissions Inventory
As discussed in 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 distrib-
utions of isoprene for a 'cool' day and a •warm" day, respectively, 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
emissions 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 "warm" day plots (Figure 4-22) show that isoprene
4-44
-------
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.
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. These higher levels are especially evident for much of the inland areas
(West Virginia, Virginia, Pennsylvania, 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 also alpha-pinene, monoterpene, and unidentified biogenic VOC
emission estimates. 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
-------
Irroot Data
FREDS Modules
Annual
Inventory
Methane and
Aldehyde
Factors
Hydrocarbon
Preprocessing
Model Data
Extraction
Species
Profiles
>s
PSPLIT
s».
^
\t
Speciation
Spatial
Allocation
Factors
• Spatial
Allocation
Temporal
Allocation
Factors
Temporal
Allocation
Hodel Input
Preprocessor
Figure 4-1. Summary of software modules and input data used in FREDS.
4-46
-------
ee
LLJ
0_
cc
o
CC
o
o
53
w
u
o
o
SO
60 70. 80 90
AVERAGE DAILY TEMPERATURE (F)
100
Figure 4-2. Effect of average daily temperature on VOC emissions. (Emissions are given for a
composite of vehicle types and speeds for a diurnal temperature range of 20 °F.)
IX
u
a.
to
cc
O
SW*
§
o
o
C/3
i
UJ
DIURNAL TEMPERATURE RANGE (F)
Effect of diurnal temperature range on VOC emissions. (Emissions are given for a
composite of vehicle and speed classes at an average daily temperature of 85 °F.)
Figure 4-3.
4-47
-------
MOBILE
(••7,2*)
CHEMICAL PROCESSES
SURFACE COATINGS
(36.4%)
STORAGE TANKS
(4.5%)
a. Total VOC
b. Point Source VOC
SOLVENT USE
(57.6%)
COMBUSTION
TSDF's (A n%i
(6.7%)
GASOLINE MKT
(14.6%)
QTHER
(7.7%)
EVAPORATIVE
(72.7%)
c. Area Source VOC
d. Mobile Source VOC
Figure 4-4. Distribution of regionwide 1985 VOC emissions.
4-48
-------
UTILITIES
(80.7%)
a. Total NOx
CHEMICAL PROCESSES
(0.8%)
1ND/INST
b. Point Source NOx
c. Area Source NOx
Figure 4-5. Distribution of regionwide 1985 NOX emissions.
4-49
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
1985 Base Case Anthropogenic VOC Emissions
Emissions (tons/day):
= O maxm <— 1
<= 15 ^^ <=» SO
1985 Base Case Anthropogenic NOX Emissions
Emissions (tons/day): xssaa == O ssssa
15 ^^ <= 5O
Figure 4-6. 1985 base case anthropogenic emissions (tons/day) of VOC and NOX.
4-51
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Major Industrial Nox Sources
Point Source gmisaion (TPD):
<- 20
* <- 3O
A <« 100
> 100
Major Utility Nox Sources
Point Source Emission (TPD);
•*• <- 20
* <„ 30
A <. 100
> 100
Figure 4-7. 1985 base case anthropogenic emissions (tons/day) of NOX from industrial plants and
utilities.
4-53
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Anthropogenic Emissions by Source Type
For a Typical Weekday, Saturday, and Sunday
70000-
63000-
^ 56000 -
Q -49000 -
c 42000-I
en
35000-
. 28000-1
en
CO
21000-
14000-
7000-
0
WK SAT
I— voc
SUN
WK SAT SUN
NOX —I
WK SAT SUN
CO
Area
Mobile V7////A Point
Figure 4-8. 1985 base case emissions (tons/day) by day type (weekday, Saturday, and Sunday).
4-55
-------
Anthropogenic VOC Emissions by Source Type
For a Typical Weekday
01
|6H
'tn
-------
S7H
_o
"(0
.22 6-
>s
o
Q
3-
c 9-
0) ^1
1 -
0-
Anthropogenic CO Emissions by Source Type
For a Typical Weekday
auDDDDDDQDDDia
~T I I 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1—
0 1 2 3 4 5 6 7 8 9 10111213141516 171819 20 21 22 23
Hour (EOT)
Source Type *•
. « .
** " Area ~
I I Mobile B-
•e-a Point
O O O Total
Figure 4-10. Diurnal profiles of point, area, and mobile sources of CO.
4-57
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Mobile Source VOC Emissions - 15OO EST Cool Day
Emissions (tons/hour): scs
O.OO
O.4S
<= O.15
<=» O.6O
<= O.3O
> O.6O
Mobile Source VOC Emissions - 15OO EST Warm Day
Emissions (tons/hour): ESSES « O.OO sana <=« 0.1 5
.-... •• <=° O.45 ••• <- 0.6O
<- 0.30
> O.6O
Figure 4-11. Mobile source emissions (tons) at 1500 EST on a 'cool" day and a "warm" day.
4-59
-------
I
''"t
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Variations in Mobile Source VOC Emissions
For the ROMNET Domain
o
a
CO
c
o
15000 -
13500-
12000:
10500 -
9000-
7500-
6000-
4500-
3000-
1500-
0
03# 04 05 06 07 08 09* 10# 11 12 13 14 15
July 1988 Episode
* Saturday # Sunday
Variations in Mobile Source NOX Emissions
For the ROMNET Domain
16*
o
o
CO
J,
CO
o
'co
_co
LU
10000-
9000-
8000-
7000-
6000-
5000-
4000-
3000-
2000-
1000-
0
03# 04 05 06 07 08 09* 10# 11 12
July 1988 Episode
* Saturday # Sunday
.13 14 15 16*
Figure 4-12. Regionwide daily total mobile source VOC and NOX emissions (tons/day)
July 1988 episode.
4-61
-------
O
'§
UJ
50000-
45000-
40000 -
35000-
30000-
25000-
20000-
15000-
10000-
5000-
Variations in Mobile Source CO Emissions
For the ROMNET Domain
03# 04 05 06 07 08 09* 10# 11 12 13 14 15
July 1988 Episode
* Saturday $ Sunday
16*
Figure 4-13. Regionwide daily total mobile source CO emissions (tons/day), July 1988 episode.
4-62
-------
VOC Emissions
Northeast Corridor
12CH
(73) (73) (73)
60 6O 60
BS85 8S05 CS05 CS10 CS13 CSU CS15 CS16 CS18 CS19 CS20 CS23 Bio
Emissions Scenario
Point &8?ffl Area vMs/A Mobile ••§ Biogenic
Values above bars: Percent reduction from 2005 and from {1985)
VOC Emissions
USA Portion of the Domain Excluding NE Corridor
300
^.270
Q 240
8 150
to
LU
60-
30-
(48.)
22
(77) (77) (77) (77) (77) (77) (77)
65 65 65 65 65 65 65
BS85 BS05 CS05 CS10 CS13 CSU CS15 CS16 CS18 CS19 CS20 CS23 Bio
Emissions Scenario
Point KSJSSa Area V7/7//A Mobile M 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.
4-63
-------
NOx Emissions
Northeast Corridor
50
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23
Emissions Scenario
point i2SSi Area VS/MA Mobile
Values above bars: Percent reduction from 2005 and from (1985)
NOx Emissions
USA Portion'of the Domain Excluding NE Corridor
(59) (60) (59) (59)
60 60 80 60
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19
Emissions Scenario
CS20 CS23
Point B&S&3 Area W//M 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.
4-64
-------
' CO,Enriissions
.Northeast Corridor
300 -\
(50)
9_ (56) (56) (56) (56) (56) (56) (56)
19 19 19 20 19 20 20
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23
Emissions Scenario
Point Bg^i Area V////A Mobile
Values' above bars: Percent reduction from 2005 and from (1985)
CO Emissions
USA Portion of the Domain Excluding NE Corridor
400 i
(58) (58) (58) (57) (58) (57) (57)
23 23 23 23 23 23 23
BS85 BS05 CS05 CS10 CS13 CS14 CS15 CS16 CS18 CS19 CS20 CS23
Emissions Scenario
Point Area W////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.
4-65
-------
Northeast Corridor
Urban Non-Attainment Areas
Figure 4-17. The Northeast Corridor and nonattainment areas outside the Corridor.
4-66
-------
Major Nox Point Sources Affacted 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.
4-67",
-------
2.5
82.0
1 .5 -
~co
fZ3
1ET
O
E
o O.5
O.O
2.5
O
O ,
2.O -
o>
_§ 1.5
en
o
1 .O -
O O.5 -
O.O
Solar intensity
-I OO
— h
• 2OO yUE/m2—h
• 4-OO /jE/m2-h
SOO yUE/m2-h
1O 15 2O 25 3O 35
Leaf temperature (°C)
4-0
Alpha — pinene
Monoterpene/unidentif ied
~i 1 1 r
1O 15 2O 25
Leaf temperature (°C)
~1 1—
3O 35
4-0
Figure 4-19. Relationship between meteorological temperature and (1) isoprene and alpha-pinene
emissions, (2) monoterpenes/unidentified hydrocarbon emissions (Tingey et a/., 1980).
4-68
-------
I BIOMASS
IFACTORS
:OUNTY-TO-GRID
ALLOCATION
|
IOMPUTE -SUMMER-
BIOMASS
COMPUTE MONTHLY
ADJUSTMENT FACTORS
FOR BIOMASS
ADJUST BIOMASS;
FOR MONTH
GRIDDED
ONCANOPY
BIOMASS
GRIDDED
CANOPY
BIOMASS
COMPUTE STANDARDIZED
EMISSION FLUXES
COMPUTE TEMPERATURE AND SUNLIGHT
ADJUSTMENT FACTORS
T
MERGE EMISSION RATES
JL
GRIDDED
HOURLY
METEOROLOGY
DATA
SPECIATE ACCORDING
TO MODEL CHEMISTRY
^GRIDDED
HOURLY
SPECIES
EMISSION
RATES
Figure 4-20. Flowchart of the Biogenic Emissions Inventory System.
4-69
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Biogenic Isoprene Emissions - 1OOO EST Cool Day
Emissions (moles/hr):
< 2OOO
>0" 8OOO
~ 2OOO ssssss >« 6OOO I
> •» 1 0OQO a,^ >» T 2OOO i
Biogenic Isoprene Emissions - 15OO EST Cool Day
Emissions (moles/hr):
Figure 4-21. Biogenic isoprene emissions (mol C/h) for 1000 EST and 1500 EST on a "cool" day.
4-71
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Biogenic Isoprene Emissions - 1OOO EST Warm Day
Emissions (moles/hr):
< 2OOO
>= 8OOO
KS*>=- 2OOO sssss >= 6OOO
>= 1OOOO ^M >« 1 2OOO
Biogenic Isoprene Emissions - 15OO EST Warm Day
Emissions (moJes/hr):
< 2OOO sssss >=» 2OOO asses >=• 6OOO
>= SOOO «•> = 1OOOO i^mm >= 1 2OOO
Figure 4-22. Biogenic isoprene emissions (mol C/h) for 1000 EST and 1500 EST on a "warm" day.
4-73
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Variations in Biogenic VOC Emissions
For the ROMNET Domain
•40000 -
36000-
32000 -
g 28000 -I
\
£ 24000-j
o
•^ 20000 -I
CO
c •
•5 16000-1
CO
'E 12000-
uJ
8000-
4000-
03 04 05 06 '07 08 09 10 11 12 13 14 15 16
July 1988 Episode
Figure 4-23. Regionwide daily total biogenic VOC emissions (tons/day), July 1988 episode.
4-75
-------
TABLE 4-1. ROMNET EMISSIONS SCENARIOS
PHASE I SCENARIOS (MOBILE3.9)
1985 Base Case
2005 Baseline:
No Growth NOX Scenario
CS01: Phase I Maximum Technology VOC 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
4-76
-------
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.
8 The original allocation factor in the NAPAP inventory was 0.25 for all of these categories, corresponding to no increase in
emissions during summer.
4-77
-------
TABLE 4-3. SUMMARY OF CONTROL MEASURES IN THE BASELINE PROJECTION AND
MAXIMUM TECHNOLOGY INVENTORIES
Inventory
VOC control measures
NO* control measures
Phase I:
2005 Baseline
Phase II:
2005 Baseline
Phase II:
Maximum
Technology
NO* and Enhanced
Maximum
Technology VOC
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
overall NO* 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
Sam* as Phase-1 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
Highway vehicles:
- 0.125 g/mi tailpipe standard
- enhanced I/M with arrti-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-NO* 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
- «
1
4-78
-------
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-NQx 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
projected attainment date
None
4-79
-------
Pod
TABLE 4-5. NSPS EFFICIENCIES FOR POINT AND AREA SOURCES
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 - cruda 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 Acrylonrtrile 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
4-80
-------
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
Overall
effi-
ciency
Mean
Category name ding (%) (%)
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
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
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
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
4-81
-------
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 TPY
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 >1 00 TPY
90 Ind. ext. comb. - gas >100 TPY
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 > 100 TPY
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
needina 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-NQx burners.
4-82
-------
TABLE 4-8. MAXIMUM TECHNOLOGY VOC CONTROLS FOR POINT SOURCES
Pod
Description
Maximum techno).
efficiency (%)
Phase I Phase II
Control technology 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 go
32 Carbon black manufacturing 90 90
33 Automobile 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 go go
38 Food/agricultural starch mfg. go 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
4g 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
4-83
-------
TABLE 4-9. MAXIMUM TECHNOLOGY VOC CONTROLS 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 technology 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.
4-84
-------
TABLE 4-10. MAXIMUM TECHNOLOGY NOX CONTROLS FOR POINT AND AREA SOURCES
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. - coaj
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
Control
technology*
SCR
SCR
SCR
Energy eff. .
Low-NOx
Low-NOx
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-NOx
Low-NOx
ban
ban
ban
ban
hwy. diesel
hwy. diesel.^
hwy. diesel^
Maximum technol.
efficiency (%)
81 .OQ&
81.00&
81 .OQ&
5.00
45.00
36.00
42.00
42.00
70.00
30.00
70.00
30.00
94.3QC
81.00&
94.30°
81.0Qf>
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
Low-NOx technology includes low-NOx burners, staged combustion air, reduced air-to-fuel ratio, and steam injection. SCR
refers to selective catalytic reduction.
* 80 percent for SCR with a 5 percent energy efficiency reduction.
0 94 percent for SCR with a 5 percent energy-efficiency reduction.
d Transfer from highway diesel technology.
4-85
-------
;]
-.1
TABLE 4-11. STRATEGY 5 POINT SOURCE CONTROL EFFICIENCIES BY ATTAINMENT
CATEGORY
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. baf.
Bulk term, -not bal.
Stage I •
Ethylene oxide
Phenol .
Terephthalic acid
Acrylonitriie
SOCMI fugitives
Refinery fugitives
Cellulose acetate
Styrene butadiene
Propytene
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
4-86
-------
TABLE 4-12. STRATEGY 5 AREA SOURCE CONTROL EFFICIENCIES BY ATTAINMENT
CATEGORY
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
4-87
-------
,
TABLE 4-13. STRATEGY 5 ATTAINMENT CATEGORIES AND ACROSS-THE-BOARD VOC
REDUCTIONS
Area
Albany-Schenectady-Troy, NY
AHentown-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.
4-88
-------
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).«
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/M isc
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
O
Total non-methane hydrocarbons (/»g/m2/h)and percent contribution from individual chemical species, standardized for full
sunlight and 30°C.
4-89
-------
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 concerning the effectiveness of regional control strategies. These issues were to be
addressed through ROM simulations. As stated in Section 4.4.2, these issues are:
f. 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 ozone concentrations in the
Northeast Corridor?
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. What are the effects of 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 presents the results of ROM simulations conducted to address these five issues.
Section 5.2 describes the ozone concentrations and spatial patterns for the Phase I11985 base case
and 2005 baseline scenarios.1 Section 5.3 presents a discussion of the findings relative to each of the
above issues. Section 5.4 presents a summary of the implications drawn from these findings.
5.2 1985 BASE CASE AND 2005 BASELINE PREDICTIONS
A 15-day episode during the summer of 1988 (July 2-17), was used to simulate all emissions scenarios.
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 that are predicted for any hour during the 15-day
1. Ozone predictions for the Phase I scenarios are not described in this report because 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
-------
episode. Thus, this figure represents the overall impact during the episode, not a snapshot of the con-
centrations 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 ^ 125 ppb 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. From an examination of the ozone patterns on
Individual days, it appears that the elevated ozone levels extending across northern West Virginia into
central Virginia are associated with multiday transport from 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
scenario 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 NO* is 10 percent lower than the 1985 base. A detailed descrip-
tion 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 Connec-
ticut northeastward into Maine. Although ozone levels decline over New York City2, maximum concen-
trations remain >. 200 ppb. In the vicinity of Baltimore/Washington, DC, ozone levels were reduced by
^ 10 percent.
2. Urban area boundaries are defined by Metropolitan Statistical Area/Consolidated Metropolitan Statistical Area (MSA/CMSA)
boundaries.
5-4
-------
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.
5.3 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, defined below.
Metric 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).
Metric 2: episode mean 8-hour daily maximum ozone concentrations, i.e., "chronic" metric.
Metric 3: daily maximum 1 -hour concentrations.
Metric 4: diurnal time series of urban area peak hourly concentrations.
Metrics: population exposure to hourly ozone concentrations exceeding 100ppb and
125 ppb.
These metrics were chosen to highlight different facets of the response of predicted ozone levels to
control strategies. Not ail metrics were examined for each issue. Metrics 1, 3, and 4 are fairly straight-
forward.
Metric 2 was calculated by identifying the highest running 8-hour average value for each day in each
grid within a selected geographic area. On each day, the first running 8-hour period began at midnight
and the last at 1600 EST. The episode mean for each grid was computed by averaging the highest
8-hour value from each day.
For Metrics, population exposure was calculated using the equation shown in Table5-1. 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 ROMNET 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 controls versus NOX controls in reducing ozone levels across
the region?
Analysis Approach
The Importance of Issue #1 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 limited1). 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; 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.
CS1 0: Combined VOC and NOX maximum technology controls
[NOX + VOC control]
CS1 1 : Maximum technology NOX controls with VOC at the 2005 baseline
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. Biogenic emissions are included in this figure
for reference. 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. Figure 5-5 shows the effects
of the CS12 controls on anthropogenic and total VOC 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 net reduction 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 paragraphs below present: (1) the impacts of NOX and VOC controls on the episode maximum
1-hour ozone concentration (Metric 1), (2) a comparison of "acute" versus "chronic" metrics, (3) the day-
to-day variations in the impact of NOX and VOC controls, (4) a case study of NOX control in New York
City, and (5) the impact of controls on population exposure.
Episode Maximum Concentrations
Impact of 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 125 ppb. (Recall that Canadian
emissions were kept at the 2005 baseline with maximum technology controls applied only in the U.S.
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 1-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. The highest
predictions near Philadelphia were reduced to the level of the NAAQS.
5-7
-------
Impact of 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
indicates 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
Buffalo. Also, a fairly broad area with peak values of 100 ppb to 120 ppb remain from Detroit south-
eastward 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 because of 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
produce 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 controls, 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 NOX and VOC 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,
contains 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-source-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 (CS13, CS14, and CS10) are shown in
Figure 5-9. The most noticeable differences between the predictions for these strategies are across
West Virginia, eastern Ohio, and western Pennsylvania, where large numbers of point sources are
located (see Figure 4-7). 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
5-8
-------
southwestern 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 equivalent impacts on episode maximum ozone levels even though emissions are lower with
point source controls. Thus, predicted ozone seems to be more sensitive to mobile source NOX
controls on a per-ton-removed basis in these areas. This conclusion is not unexpected, because
mobile sources emit both VOC and NOXI 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 mobile source controls in the
area from Washington, DC northeastward to just outside of New York City. Figure 5-9-shows the spatial
extent of ozone > 100 ppb and > 125 ppb for CS13 and CS14. In New York City, where NOX controls
produce a "disbenefit," peak ozone is less with mobile source controls - most likely, because 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 southern 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 second-highest daily maximum values were 7 and 13-ppb
lower, respectively, with mobile source controls 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
-------
- I
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 metric is the episode maximum
1-hour concentrations for all grids within each of the above areas. The chronic metric is the episode
mean of the daily maximum 8-hour average ozone concentrations in each grid.
The acute and chronic metrics are presented in quantile-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
represent the rank ordered ozone values from the grids covering each of the three areas considered.
The number of grids in each area are as follows: New York, 83; Greater Connecticut, 31; and Balti-
more/Washington, DC, 50. Comparisons are provided between VOC control prior to NOX control
(CS12) and VOC control with NOX 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.
Because all but one of the data values falls below the diagonal line, the overall impact of NOX and VOC
controls is projected to be more beneficial than VOC controls alone. However, the benefit varies sub-
stantially 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
percentiie 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 percentiie. 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. Note-that, for the chronic measure, adding NOX controls in New York City is actually
beneficial relative to VOC controls alone. This benefit is seen along the entire distribution, but it dimi-
nishes with increasing concentration-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 NO* 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 controls. The early peak, which is reduced by NOX controls, appears to be due to ozone
formation from local 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
5-11
-------
of the New York City urban plume across portions of Connecticut. This transported ozone was greatly
reduced as a result of the large reduction in ozone in the New York City plume with the application of
VOC controls.
For the Baltimore/Washington, DC area, the profiles in Figure 5-14 indicate that the benefits of NO*
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 variation reflects the
variation in meteorological conditions that determines such factors as local source-receptor relation-
ships, 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)
Indicate 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 concentra-
tions (75 to 100 ppb) in the pre-control 2005 baseline scenario. The behavior of diurnal patterns for
other areas (including Philadelphia, Boston, Cleveland, Detroit, and Charleston) are similar to that
shown for Pittsburgh.
NOx Control: New York CKy Case study
A fuller understanding of the effects of NOX controls on ozone levels in and around New York City is
evident by examining predictions on an individual day. The July 9, 1988, ozone and NOX concentra-
tions 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 area shown encompasses the New York City CMSA and portions of the
region Immediately upwind and downwind on this day. The 2005 values show a plume 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 in
these layers 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. Elsewhere in this part of the Corridor, ozone levels actually decline
by more than 10 ppb following NOX controls.
Examining the levels and patterns of emissions in the vicinity of New York City provides further insight
into these results. The spatial distributions of morning (0500 to 1100EST) total VOC and MOX
emissions are provided in Figure 5-16. These data show that the highest emissions for both pollutants
are over the core of New York City, including several grids surrounding Manhattan [i.e., the Central
Business District (CBD)]. The peak emitting grid in the CBD emits 132 tons of VOC and 46 tons of NOx
during this morning period. This level of emissions is more than double that of surrounding grids and is
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.
Predicted ozone and NOX concentrations during the course of the day were examined for the 2005 and
CS11 scenarios. This time history provides an explanation for the increase in ozone following MOX
controls. In the 2005 scenario, very high NOX concentrations are predicted within the four to six
high-NOx emissions grids in the New York CBD. In the morning (0600 EST), NOX levels are 60 to
80 ppb. Concentrations increase with time to 75 to 200 ppb by 1000 EST and remain at this level for
several hours before declining to 50 to 100 ppb at 1300 EST. Elsewhere in the urban area, NOX con-
centrations are an order of magnitude less. In Philadelphia, NOX concentrations are a factor of 3 less
than in the New York CBD. Because of 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
CBD 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; this value 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
5-13
-------
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.
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. Thus, 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, further investigation of this issue is
warranted using an urban-scale model to better quantify the magnitude and spatial extent of the dis-
benefit 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 to examine
the effects on population exposure. This metric is perhaps a more robust indicator than the episode or
hourly maximum values, because 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 NCyonly controls. The VOC-only controls and the VOC plus NOX controls produce less
of a benefit (Figure 5-17). In Baltimore/Washington, DC and Philadelphia, NO* 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
NOx-oniy controls seen in episode peak ozone downwind of Baltimore is not evident in the population
exposure metric. As indicated earlier, 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
5-14
-------
nearly equivalent exposure levels. However, adding the two types of NOX controls without any VOC
controls has essentially no benefit relative to trie 2005 baseline. Adding VOC controls with full NOX
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. In Cleveland and Detroit (see Figure
5-20), comparable benefits were achieved with the NOX point source and NOX mobile source controls
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. These larger reductions are particularly
notable in the western portion 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 MOX
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 source rather than point source NOX emissions reductions in several cities (e.g., Phil-
adelphia, 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
5-15
-------
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
maximum averages) NOX plus VOC controls produce lower ozone levels than VOC controls
alone.
• The relative benefits of VOC versus NOX controls varies by day (i.e., meteorology) in some
cities. This variation was particularly evident in the Baltimore/Washington, DC area.
The above findings underscore the complexity of the role of NOX in ozone formation and in reducing
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 Corridor?
b. What is the impact of reducing transport on post-control ozone levels in the Corridor?
Two types of strategies were designed and simulated by the ROM to address these questions:
1. Stringent emission controls were applied outside the Corridor, while maintaining emissions at
the 2005 baseline inside the Corridor; and
2. Stringent emission 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 post-control 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 post-control ozone levels
inside the Corridor.s •
The analysis of the two transport impact questions listed above is divided into four components:
• An assessment is made of the impact of controls outside the Corridor on ozone levels trans-
ported into the Corridor.
• The effects of reducing transport on 2005 baseline concentrations are presented.
• The impacts on post-control Corridor ozone levels due to controls in upwind areas are
discussed.
• 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. Th§ grids
comprising this "transport* boundary were selected because they were (1) between the Corridor and
major upwind source areas, and (2) not significantly influenced by Corridor emissions on the days
anatyzed. Forward trajectories using ROM layer 1 and layer 2 winds were used to identify those days
with transport flow regimes across this boundary into the Corridor. Nine such days were identified. In
several cases» trajectories entering the central and southern sections of the Corridor turned northeast-
ward and passed over northern sections of the Corridor on subsequent days. Examples are shown in
Figure 5-22.
Daily mean and maximum concentrations for these days were calculated for three segments of this
boundary: Virginia (VA), Pennsylvania (PA), and New York State (NY). The layer 2 mean and maximum
ozone concentrations are provided in Figure 5-23 for the 2005 baseline (pre-control) and CS19 (post-
control) scenarios. The layer 2 values were selected for analysis because this layer represents the bulk
of the atmospheric boundary layer during the day and the layer containing transport of pollutants aloft
overnight
3. Initially, CS01 (Phase I maximum technology VOC controls region-wide), 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
-------
The following procedures were used to quantify impacts on the 2005 baseline ozone levels in the
Corridor due to the upwind controls in CS25. First, all grids in the Corridor with daily maximum ozone
concentrations > 125 ppb for the 2005 baseline scenario were identified. The difference in layer 1 daily
maximum ozone between CS25 and the 2005 baseline was then computed for each of these grids.
Both the spatially averaged reduction and the largest single-grid reduction were computed for five
areas along the Corridor: Baltimore/Washington, DC, Philadelphia, New York City, Connecticut, and
Boston. The episode mean and peak percent change in daily maximum ozone were also calculated.
Similar procedures were followed to quantify the impacts on post-control ozone levels in CS19 except
that grids with daily maximum values > 100 ppb were examined.
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 boundary. The highest concentrations lie along the PA and VA segments. On four
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 average and peak reduction for all grids with daily maximum ozone
> 125 ppb were 8 ppb and 13 ppb, respectively. On average for the episode, there was a 6 percent
reduction in the daily maximum ozone when incoming transport was reduced. Note that, except for
Baltimore/Washington, DC, there is little, if any, change in episode maximum 1-hour ozone as a result of
reducing incoming transport.
5-19
-------
BAL/DC
PHL
NYC
CT
BOS
Reduction for all
grids > 125 ppb-
-Spatial average reduction;
-Peak reduction:
% change fn daily '-Episode mean reduction:
maximum ozone -Peak reduction:
V,V ^ s
Episode maximum -2005:
ozone -CS25:.,
8 ppb
13 ppb
6%
9%
5 ppb
10 ppb
'4%
9%
4 ppb
8 ppb
2%
5%
3 ppb
7 ppb
l"1 •. "Hi
"2%
5%
4 ppb
5 ppb
<
2%
3%
149 ppb 148 ppb 268 ppb 206 ppb 158 ppb
137 ppb 148 ppb 259 ppb 202 ppb 158 ppb
T v I lEtt ) ![ < H * ' i I | i"l»
BAtVDC: Baltimore/Washington, DC
PHL Philadelphia
NYC: New York City
CT: Portions of Connecticut outside the NYC urban area
BOSi Boston
. . -.- * s. •, x i <* * f
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 conclusion 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 conditions 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 Post-Control Ozone In the Corridor
The impacts of reducing transport on post-control ozone in the Corridor are more significant than the
impacts on 2005 baseline levels. This conclusion is evident by comparing daily maximum predictions
forCS19 (region-wide application of maximum controls) versus CS24 (emission 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.
5-20
-------
*-'••"; ^Episode Maximum 1-Hour Ozone Concentrations
~ ~~ f •* f "• f
..- Regionwide controls No controls outside Corridor
Baltimore/Washington, DC
" Philadelphia
New Yorft'City, , ,
Boston
(CS19)
122ppb
115ppb'
118 ppb
107 ppb
(CS24)
139 ppb
123 ppb
123 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 pre-control analysis. Now, however, in the post-control
environment (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.
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 case study, the daily
maximum ozone concentrations 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, NO*, and reactive organics (ROG) were extracted from
grids closest to the trajectory path at hourly intervals. This process was followed for all four of the
above scenarios 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, 1988. The air parcel enters the Corridor northwest of Philadelphia on the evening of July 7,
1988. After passing over Philadelphia during the morning, this air parcel continues northeastward
5-21
-------
..'I?
across New York City on the evening of July 8,1988. On the following day (July 9,1988), 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 concentrations. 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 apparently encountered just upwind of the Corridor near point B as indicated from the
NO* concentration profile. In response, ozone levels later increase to 90 ppb further downwind
between points B and C. 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 tn 3 trajectory path. Concentrations of
ROG upwind are actually slightly higher with the application of controls 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.
ti. Corridor Transport Patterns
As the air parcel approaches and crosses the Philadelphia area on the morning of July 8, 1988
(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 overwhelmed 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
5-22
-------
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 EST, 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.
Impacts of Transport on Post-Control 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 U.S. 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 because, 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
baseline 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 post-control Corridor precursor levels due
to upwind controls is similar to that found in the analysis of 2005 baseline scenario above.
5-23
-------
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 ozone concentrations in the Northeast
Corridor?
• In the pre-control 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.
• In the post-control 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
-------
ISSUERS:
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 that
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 that demonstrate attainment. This requirement 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. However, 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 U.S. 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.
Emissions 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 NO* 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, NOXl 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 emission 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 U.S. 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.
• Proposed 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:
.. v •• -x /••• " f '•*, L _. -- - . . • -'. " -"'
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 NO* 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.
Note that one grid remained above 125 ppb downwind of Baltimore over the eastern shore of the Ches-
apeake 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 limitations
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 concentration is reduced to 124 ppb with GS11, the NOX only control strategy.
Additional 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 in CS19. Considering
the realistic control effectiveness assumptions increases the CS19 peak to 125 ppb.
Ptttsburgh/Cleveiand/bettolt:
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 (CSi9).
Thus, it would appear that control strategies for these areas should emphasize NOX controls more so
than VOC controls.
Charleston,
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 controls) 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 "overkill" for this
area Even the Clean Air Act NOX point source controls (iow-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 application was to confirm that the CS19 controls would also reduce
ozone to below 125ppb 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 (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 eastern New Hampshire. Peak ozone in this area for the July 1988
episode was generally < 100 ppb.
In the post-control 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,
however, are at most 111 ppb. A comparison of the June 1983 and July 1988 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.
5-29
-------
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 reduc-
tions is beyond known or envisaged control technologies.
Application of stringent NOX controls in New York City is counterproductive. Ozone levels
approach 125ppb 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
simulated. 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 in reducing regional ozone levels?
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
U.S. 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 NOXl 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
emission 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
measures, 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.
5-31
-------
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.
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
consistency among cities in the effects of controls on population exposure.) Data for Philadelphia, Bal-
timore/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 outside of the above areas, are combined in Group 2. Finally, Pitts-
burgh, 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 are 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 versus the 2005 baseline is only 8 ppb in Group 1, and 5 ppb in Group 2.
5-32
-------
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
exceedances is being reduced to a greater extent than the daily maximum concentration.
In the large urban areas outside the Corridor, as represented by Group 3, there was very little reduction
(2ppb) in daily maximum ozone levels > 125 ppb 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 > 125ppb was, however, reduced by
nearly 30 percent
Although there was consistency between areas within each group, there were some notable day-to-clay
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, 1988 simulation day, ozone levels
actually increased following the application of the reactivity-based measures. This increase 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 population exposure more than doubled. The event occurred on the last
day of a 3-day period with moderately high ozone levels (125ppb
-------
The benefits of combining reactivity-based measures with maximum technology VOC and NO* controls
Is examined by comparing the predictions from CS15 (reactivity plus technology) with CS20 (reactivity
only) and CS10 (technology only). The ozone metrics in Table 5-4 indicate that for the Group 1, 2, and
3 areas, the average reduction in daily maximum ozone levels provided by CS15 is 4 to 10 times greater
than with CS20 (reactivity controls alone). However, the vast majority of the reduction in CS15 appears
to be associated with the technology-based controls rather than the reactivity measures (CS15 vs
CS10). Thus, in some areas where maximum technology controls containing large emissions reduc-
tions are applied, adding reactivity-based measures may only lead to small additional benefits.
In New York City, the lower reactivity does appear to counteract the disbenefits of NOX controls
described in Issue #1. This statement 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
Connecticut and Rhode Island.
Findings
The following are the key findings for Issue #4:
What are the effects of reactivity-based strategies in reducing regional ozone levels?
• Reactivity-based strategies similar to those simulated in ROMNET may provide the greatest
benefit in large urban areas that 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 those provided by a 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
measures was only half of that from the technology-based VOC controls, and a factor of 4
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
5-34
-------
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.
5-35
-------
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 emissions -
based upon the meteorology for July 11,1988 - are 25,800 tons per day; the anthropogenic VOC total
is 21,000 tons per day. The Northeast Corridor is not as dominated by biogenic VOC emissions.
Corridor biogenic VOC emissions - based upon the meteorology for July 11,1988 - are 3,700 tons per
day; the anthropogenic 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 biogenic emission estimates, it
is important to determine whether the conclusions regarding the effectiveness of control measures 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 was
conducted to examine the sensitivity of ozone predictions to uncertainty in biogenic emissions. The
cases that 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 3, based on recommendations from the EPA Office of Research and Devel-
opment. This range represents an overall uncertainty estimate because it was not possible to ade-
quately infer the uncertainties of individual 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 ail grids, for all hours were reduced by a factor of 3 and
combined with the anthropogenic emissions from CS05.
5-36
-------
The discussion of the results of the biogenic sensitivity simulations will focus on eight urban areas,
listed below:
? ''" Baltimore/Washington, DC '
•XV •> ff>f f s
- ,/, Boston
y, Charleston, WV'
C» " Detroit ,
^Greater Connecticut,
* Greater New York City
Philadelphia and vicinity
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
> 125 ppb.
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
predicted 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 Mew
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 Northeast Corridor.
Note 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 "high1 estimate biogenic emissions is greater than the
impact of the "low" estimate biogenic emissions (impact is defined as the change in predicted concen-
tration from the "best" estimate biogenics). This result is anticipated-the.threefold increase in biogenic
emissions contributes a greater proportional increase in VOC mass in relation to the proportional
5-37
-------
decrease in total VOC mass with a threefold reduction in biogenic emissions, because 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
emissions to react with the VOC emissions to form ozone.
CS19, which used "best" estimate biogenic emissions, reduced episode maximum ozone levels in the
U.S. portion of the domain to below 125 ppb. When "high" estimate biogenic emissions were simulated,
five of the areas in Figure 5-33 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 NO* controls.
Ozone levels in CS19 with "best" estimate biogenics are below 125 ppb in the U.S. 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.
5-38
-------
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 3), 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"
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 offering reduc-
tions in emissions beyond those in CS19 will be needed in many of the Northeast cities.
5-39
-------
5.4 SUMMARY OF MAJOR FINDINGS FOR KEY ISSUES
What are the relative benefits of VOC controls 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 ozone concentrations in the Northeast
Corridor?
• 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.
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.
5-40
-------
#4:
What are the effects of 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.
ISSUERS:
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 emissions were actually near the low end of the uncertainly range, measures in
the 1989 proposed Clean Air Act legislation would still be insufficient to reduce ozone to
< 125 ppb throughout the Corridor.
IMPLICATIONS OF FINDINGS:
There are several broad implications that can be drawn from the above findings. These are:
1. Significant reductions in anthropogenic emissions along the entire Northeast Corridor will likely be
necessary to reduce ozone to less than 125 ppb throughout the Corridor.
2. Along with VOC controls, NOX control measures should be considered in strategies to reduce
ozone levels in most areas of the Northeast. However, close examination of the potential effects of
NOX controls should be made to ensure that such controls will not be counterproductive.
5-41
-------
'
3. The types of control measures and the degree of emissions reductions necessary to reduce ozone
to less than 125 ppb will likely vary between cities along the Corridor. That is, a single set of
controls common to all areas may not be the most effective approach for reducing ozone through-
out the Corridor.
4. Areas outside the Northeast Corridor may have to add controls, beyond those necessary to solve
their local problem, in order to reduce transport into the Corridor sufficiently for concentrations to
be reduced below 125 ppb throughout the Corridor (even with stringent controls in the Corridor).
5. The control technologies needed to achieve the emissions reductions necessary to reduce ozone
to below 125 ppb in most of the major Northeast Corridor cities are currently not available.
Therefore, a high priority should be given to development and testing of controls for both VOC and
NOX sources along with enhancing enforcement procedures to achieve the highest degree of
control effectiveness possible.
6. Regional strategy assessments of the type performed in ROMNET are useful for examining the
regional perspective to control impacts and transport. However, urban-scale analyses will also be
needed to develop local control targets and to establish a menu of specific control measures for
individual nonattainment cities.
CAVEATS TO FINDINGS!
There are three broad caveats to the findings 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)
average values for each layer. This type of emissions estimate 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 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 emission
controls necessary to reduce ozone to < 125 ppb.
5-42
-------
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 unquantified level of uncertainty in base case anthropo-
genic emissions. The representativeness of growth factors for estimating future year emissions as
well as the efficiency and effectiveness of controls adds to the uncertainty in emissions scenarios.
Thus, the findings from ROMNET should be used to help establish control directions and approxi-
mate 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. Thus, the
results (particularly regarding the effects of NOX controls and regional transport) may vary under
different meteorological conditions.
5-43
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report
-------
Episode Maximum Ozone: BS85
July 2, 1988 - July 17, 1988
Concentration (ppb)
Figure 5-1. Predicted 1985 base case episode maximum 1 -hour ozone concentrations (ppb) for the
July 1988 episode.
5-45
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BSOS
July 2, 1988 - July 17, 1988
Concentration (ppb)
Figure 5-2. Predicted 2005 baseline episode maximum 1-hour ozone concentrations (ppb) for the
July 1988 episode.
5-47
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Massachusetts/
NE Coastline
Greater Cleveland
onnectlcut/
node Island
New York City
Greater Pittsburgh
Greater Philadelphia
Baltimore/Washington
Figure 5-3. Areas in the ROMNET domain used in calculating selected metrics.
5-49
-------
VQC Emissions
USA Portion of the Domain
40000-1
35000
30000
"• 25000
|2
~ 20000
•g 15000-1
a1
"
UJ
10000
5000-
0-
BS05 CS11 CS12 CS13 CS14 CS10 Bio
Emissions Scenario
Point SSSSSSS Area v/s/za Mobile •
Values abov* bars: Percent reduction from 2003
NOx Emissions
USA Portion of the Domain
Biogenic
CO
.0
'(/5
CO
1
UJ
20000
18000
16000
14000
12000
10000 -
8000-
6000
4000-
2000-
57
BS05 CS11 CS12 CS13 CS14 CS10
Emissions Scenario
Point Area W///A Mobile
Values above bars: Percent reduction from 2005
Figure 5-4. Anthropogenic VOC emissions and NOX emissions for the U.S. portion of the ROMNET
region.
5-50
-------
Percent Reduction in Anthropogenic VOC Emissions: CS12 vs. 2OO5
Percent Reduction:
O SS5S83 <= 3O SSSZZi <= 45
££22 <= 6O -rrffffrfft <=s 75 ^^B > 75
Percent Reduction in Total VOC Emissions: CS12 vs. 2OO5
Percent Reduction: ^^ = O
s <=• SO
Figure 5-5. Percent reduction in VOC emissions between CS12 and the 2005 baseline for
anthropogenic emissions only and anthropogenic plus biogenic emissions.
5-51
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
Jj
-------
Percent Reduction in Anthropogenic NOX Emissions: CS11 vs. 2005
Percent Reduction: ?*%zm = O sssssa <= 3O zzzm <= 45
<= 6O <— 75 mmm > 75
Figure 5-6. Percent reduction in NOX emissions between CS11 and the 2005 baseline.
5-53
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BSO5
July 2, 1988 - July 17, 1988
Concentration (ppb):
< SO SS33S >= SO
>- 1 4O ••• >= 1 60
1OO
18O
i>- 125
> 2OO
Episode Maximum Ozone: CS11
July 2, 1988 - July 17, 1988
Concentration (ppb):
Figure 5-7. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988
episode: 2005 baseline and CS11.
5-55
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BSO5
July 2, 1988 - July 17, 1988
Concentration (ppb):
< QO
>•» 14O
>"•» SO ggga > =
1OO
18O
>=. 125
Episode Maximum Ozone: CS12
July 2, 1988 - July 17. 1988
Concentration (ppb): sss® < 8O
140
Figure5-8. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988
episode: 2005 baseline and CS12.
5-57
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: CS13
July 2, 1938 - July 17, 1988
Concentration (ppb):
Episode Maximum Ozone: CS14
July 2. 1988 - July 17, 1988
Concentration (ppb):
< SO sszszs >«* SO
= 14O MI^ >= 16O
1OO
18O
>= 125
Figure 5-9. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1988
episode: CS13, CS14, and CS10. (Page 1 of 2)
5-59
-------
• 1
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: CS1O
July 2, 1988 -.July 17, 1988
Concentration (ppb): ;
Figure 5-9 (Page 2 of 2)
5-61
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Quantile—Quantile Frequency Distributions
1—Hr Daily Maximum Ozone (ppb)
. BALTIMORE & WASHINGTON, DC
200 -I
160-
o 120-1
o
x
O
O
80-
40-
80 120 160
VOC Controls (CS12)
Quantile-Quarttile Frequency Distributions
Episode Mean 8-Hr Daily Maximum Ozone (ppb)
BALTIMORE & WASHINGTON, DC
200
100-
co
CJ
80-
o 60
o
x
O
o
40-
20-
90
50
10
20
—I . 1 , 1 r—, , ,—
40 60 80
VOC Controls (CS12)
100
Figure 5-10. Quantile-quantiie frequency distributions of 1-hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for Baltimore/Washington, DC and vicinity.
5-63
-------
Quantile—Quantile Frequency Distributions
1—Hr Daily Maximum Ozone (ppb)
NEW YORK CITY
200-[
160-
_cn
e
o 120
o
X
O
O
80-
40-
40
—|—i—i—i—|—i—i—i—|—
80 120 160
VOC Controls (CS12)
200
Quantile—Quantile Frequency Distributions
Episode Mean 8-Hr Daily Maximum Ozone (ppb)
NEW YORK CITY
100-
80-
O)
2
o 60-
o
x
O
o
40-
20-
90
50
10
20
i i i i—i—i—i—i—|—i—i—i—|—
40 60 80 100
VOC Controls (CS12)
Rgure5-11. Quantile-quantiie frequency distributions of 1-hour daily maximum ozone and episode
mean 8-hour daily maximum ozone for New York City.
5-64
-------
Quantile—Quantile Frequency Distributions
1 —Hr Daily Maximum Ozone (ppb)
GREATER CONNECTICUT
200
160-
o
o
x
O
a
5
120-
80-
40-
40
—I r > i 1 1 1 1 1 1 1 , r-
80 120 160 200
VOC Controls (CS12)
Quantile-Quantile Frequency Distributions
Episode Mean 8-Hr Daily Maximum Ozone (ppb)
GREATER CONNECTICUT
100
80
o 60-]
o
x
O
o
40-
20-
10
—i—
20
—r~
40
60 80
VOC Controls (CS12)
-i 1 , |
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.
5-65
-------
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 M 18
Day of Month
CS12> CS11
CS11 > CS12
Maximum 1-Hr Ozone (ROM 2.1) for July 1988 Episode
GREATER CONNECTICUT
.a
&
c
ai
u
o
o
330
300
270-
240
21 (H
180
150
120--
90-
60
30 H
T
T
T
—1 I I I I 1—1—I—>—I—.—I—,—I—I—I—I—r—
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16
Day of Month
17 18
CS12>CS11
CS11 >CS12
Rgure5-13. Diurnal time series of maximum hourly ozone concentrations in New York City and
Greater Connecticut for CS11 and CS12.
5-66
-------
Maximum 1-Hr Ozone (ROM 2.1) for July 1988 Episode
BALTIMORE & WASHINGTON, DC
200-
180-
160-
£140-1
o 120--
100
80-
60-
40-
20-
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
200-
180-
160
140-
-Q
a.
a.
o
-| 100-1
tu
§ 80 H
o
60-1
40-
20-
~i ' i ' i ' i !—i r~r~'—i—i—i ' i '—i—i—i—i—i—1—1—i—r—>—i—1—1—i—i-
02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18
Day of Month
CS1.2 > CS11
CS11 >CS12
Figure 5-14. Diurnal time series of maximum hourly ozone concentrations in the Baltimore/
Washington, DC area and the Pittsburgh area for CS11 and CS12.
5-67
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
July 9, 1988 Predicted Daily Maximum Ozone: 2DO5 Baseline
Concentration (ppb):
< 1OO SEES >= 1OO 3SBS53 >=- 125
1-4O m^mm >= 1 6O I[^BB >ae ISO
Change in Daily Maximum Ozone: CS11 vs. 2OO5 Baseline
Concentration (ppb): mssm < — 1O
Susies* -t-3 to
-9 to -3
> 10
-2 to
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.
5-69
-------
•r:
'1
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
July 9, 1988 O5OD-11OO EST Total VOC Emissions
Emissions (tons): usssm = O mm® <«• 1
=» 15 ^mm <*= 4O
July 9, 1988 O5OO-11OO EST Total NOX Emissions
Emissions (tons):
Figure 5-16. Morning (0500 EST -1100 EST) total VOC emissions and NOX emissions (tons)
on July 9,1988, in the vicinity of New York City.
5-71
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Population Exposure to Ozone > 100 PPB
MASSACHUSETTS & COASTAL NEW ENGLAND
2005 CS11
CS12 CS13 CS14
Scenario
CS10
Values above bars: Percent reduction from 2005
Figure 5-17. Population exposure to 1-hour ozone > 100ppb in Massachusetts and coastal New
England for selected scenarios.
5-73
-------
J 45 -i
1 40
x
to 35 •
f 3
Q.
o 0
Q.-'
^-^
2 20^
50
Population Exposure to Ozone > 125 PPB
BALTIMORE. WASHINGTON, DC & VICINITY
76
CS11
CS12 CS13
Scenario
100
CS14 CS10
Values above bars: Percent reduction from 2005
Population Exposure to Ozone > 125 PPB
PHILADELPHIA & VICINITY
2005
100
CS11
CS12 CS13
Scenario
CS14
100
CS10
Values above bars: Percent reduction from 2005
Figure 5-18. Population exposure to 1 -hour ozone > 125 ppb in Baltimore/Washington, DC
and Philadelphia for selected scenarios.
5-74
-------
^500-
o 450 H
50-f
Population Exposure to Ozone > 125 PPB"
NEW YORK CITY
79
92
87
92
2005
CS11 CS12 CS13
Scenario
CS14
CS10
Values above bars: Percent reduction from 2005
Population Exposure to Ozone > 125 PPB
GREATER CONNECTICUT & RHODE ISLAND
2005
CS11
CS12 CS13'
Scenario
CS14
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 Connecticut/Rhode Island for selected scenarios.
5-75
-------
' I
I
^50-
J 45-
'£ 40-
X
« 35-
f 30^
Q.
J.25-
£ 20-
§ 15-
Q.
uj 10-
Population Exposure to Ozone > 100 PPB
PITTSBURGH & CHARLESTON
•= o-
,£ 2005
100
62
100
78
CS11 CS12 CS13
Scenario
CS14
Values above bars: Percent reduction from 2005
Population Exposure to Ozone > 100 PPB
CLEVELAND & DETROIT
2005
CS11
CS12 CS13
Scenario
CS14
Values above bars: Percent reduction from 2005
80
CS10
CS10
Figure 5-20. Population exposure to 1-hour ozone > 125ppb in Pittsburgh and Charleston, WV
(combined) and Cleveland and Detroit (combined) for selected scenarios.
5-76
-------
Figure 5-21. Location of the "transport' boundary used in quantifying incoming ozone transport.
5-77
-------
fi
July 9, 1988
July 10, 1988
Figure 5-22. Layer 2 forward trajectories starting at 1500EST from locations along the "transport"
boundary for July 9,1988, and July 10,1988. (Trajectory markers are at 4-h intervals.)
5-78
-------
New York Sector of Transport Boundary
100
4O
7/06 7/07 7/08 7/O9 7/IQ7/11 7/127/137/14
Days
Virginia Sector of Transport Boundary
120
10O
a.
o.
a
7/06 7/07 7/08 7/09 7/10 7/11 7/127/137/14
Days
Pennsylvania Sector of Transport Boundary
12O
100
I 80
60
40
20
7/O6 7/O7 7/O8 7/09 7/1O 7/11 7/12 7/137/14
Days
2005 (precontroi) peak
2005 (precontroi) mean
CS19 (postcontrol) peak
CS19 (postcontrol) mean
Figure 5-23. ROM layer 2 peak and mean ozone concentrations by day along the three transport-
boundary segments.
• 5-79
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: B3OS
July 2, 1988 - July 17, 1988
Co
Concentration (ppb):
Episode Maximum Ozone: CS25
July 2. 1988 - July 17, 1988
Concentration (ppb): ss < SO sssss >. SO ssssrax-lOO
' i rn •... - 1 rn
Figure 5-24. Predicted episode maximum 1-hour ozone concentrations for the July 1988 episode
2005 baseline and CS25.
5-81
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: CS19
July 2, 1988 - July 17, 1988
Concentration (ppb): 535553 < SO
Episode Maximum Ozone: CS24
July 2, 1988 - July 17, 1988
Concentration (ppb): ssss < SO toast >- SO ma>«1OO
- — """*
— """*
Figure 5-25. Predicted episode maximum 1-hour ozone concentrations for the July 1988 episode
CSl9andCS24.
5-83
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
* Corridor Boundary
Figure 5-26. ROM layer 2 trajectory for the transport case study. (Trajectory, markers are at 4-h
intervals.)
5-85
-------
•1
180;
170-
16O ;
ISO ;
14O •
130;
12O ;
.—* uo'.
^T 90-
I =0:
70;
80;
SO;
4O-
30;
20;
1O ;
0-
130;
120;
80;
SO-
70;
CO;
SO;
40 ;
30;
20-
10 ;
OZONE CONCENTRATIONS
• ConrMor Boundary
I2.-OO JuJO4 12.-OO JuKJ7 I2.-OO
12:OO JuWB 12-.OO JullO 12:OO Julll
SCENAR I 0
BSOS CS25
NOX CONCENTRATIONS
ROG CONCENTRATIONS J\ E
rw
• 5
• *
»
• 0
• Corridor Boundary
jutoa 12:00
t2:OO JU1O7 12:OO JuJO« 12:00 JuiOB 12:OO JullO 12:00
Figure 5-27. Time history of layer 2 ozone, NOXl and ROG concentrations: 2005 baseline and CS25.
5-86
-------
18O
170
160
ISO
14O
130
120
_^ 110
"g- 100 •
*~Z> 9O
1 «>
° 70
eo
so
40
3O
20
10
0
OZONE CONCENTRATIONS
• ConrMor Ooundoiy
'l'"""""l"
JuM4 12tOO
SCENAR I 0
OS 1 9
JuMO 1IiOO JoHO H:OO Jul11
CS24
130
120
110
100
90
ao
70
80
50
40
30
20
to
O
NOX CONCENTRATIONS
7
• 8
•S
• 4
• 3
• 2
- 0
ROG CONCENTRATIONS
• Corridor Boundwy
iMiHiiiiiiiiiMHiiiiiiiiiUiiiiiiiiiiriiiijiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiiniiiniiiiiniM.ii,,,
121OO JulO« 12:OO JuWT <2iOO JtHOS 12:OO JolOfl 12:OO JullO 1*OO Julll
Figure 5-28. Time history of layer 2 ozone, NOX, and ROG concentrations: CS1.9 and CS24.
5-87
-------
.- 1
t.fl
[O
WlOO
<*->
o 80
SS 60
§40
I 20
E o
ui
NEW YORK CITY
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
BS8S BS05 CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS15 CS18 CS19 CS23
Scenario
NOX mm voc
O
-320 g
-280
•140
•120
BS85 BS05 CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
NOX •§ voc
180
100 S-
Figure 5-29. Emissions for Phase II scenarios, and predicted highest and second-highest daily
maximum ozone concentrations (ppb) for selected urban areas. (Page 1 of 4)
5-88
-------
.
o 80
& 60
| 40
| 20
S 0
PHILADELPHIA
Highest Daily Max.
2nd Highest Daily Max.
125 ppb Reference
-160
tu
BS85 BSOS CSOS CS11 CS12 CS13 CS14 CSIO CS16 CS1S CS18 CS19 CS23
Scenario
NOX m voc
180
3
o>
3
o
(0
-140
120 §
^
-loo "a-
Highest Daily Max.
2nd Highest Daily Max
12S ppb Reference
BS85 BSOS CSOS CS11 CS12 CS13 CS14 CSIO CS16 CS1S CS18 CS19 CS23
Figure 5-29 (page 2 of 4)
5-89
-------
Highest Daily Max.
2nd Highest Daily Max
125 ppb Reference
BS8S BSOS CSOS CS1I CS12 CS13 CS14 CSIO CS16 CS15 CS18 CS19 CS23
Scenario
NOx •• VOC
Highest Daily Max.
2nd Highest Daily Max
125 ppb Reference
BS85 BSOS CSOS CSH CS12 CS13 CS14 CSIO CS16 CS15 CS18 CSI9 CS23
Scenario
NOx Hi VOC
Figure 5-29 (page 3 of 4)
5-90
-------
0 80
&60
<2 40
U
DETROIT
Highest Daily Max.
2nd Highest Daily Max,
125 ppb Reference
180
•160
•140
O
N
O
3
(0
O
o
3
o
CO
a
!•+
3
•120
BS85 BS05 CSOS CS11 CS12 CS13 CS14 CS10 CS16 CS15 CS18 CS19 CS23
Scenario
HOx HI VOC
Highest Daily Max.
2nd Highest Daily Max
125 ppb Reference
!•
UJ
BS85 BSOS CSOS CS11 CS12 CS13 CSI4 CS10 CS16 CS1S CS18 CS19 CS23
Scenario
voc
Figure 5-29 (page 4 of 4)
5-91
-------
M
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BSO5
June 8, 1983 - June 2O, 1983
Concentration (ppb):
Episode Maximum Ozone: CS19
June 8, 1983 - June 2O, 1983
Concentration (ppb):
SO SaamUi >°* SO 88H8S3 >•» 1 OO
2^!LJ40 ••>=» 16O mmmm>** ISO
>= 1251
> 2OO I
Figure 5-30. Predicted episode maximum 1 -hour ozone concentrations (ppb) for the June 1983
episode: 2005 baseline and CS19.
5-93
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BS85 - Phase I
July 2, 1988 - July 17, 1988
Concentration (ppb): ss < SO SMSS >= SO
mam >=> 14-O m^ >== 1 6O
Episode Maximum Ozone: BS85 - Low Biogenic
July 2, 1988 - July 17, 1988
Concentration (ppb): sssss? < SO sss-s >= SO SESSBS >« 1OO
SBHBH >=» 1 4O MBIH >= 1 GO i^M >= 18C
Figure 5-31. Predicted episode maximum 1-hour ozone concentrations (ppb) for the July 1986
wrthlow" 1985 **** ^ wWl "beSt 6Stimate" bi°9enics and 1985 base case
5-95
-------
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
Episode Maximum Ozone: BSQS - Phase I
July 2, 1988 - July 17, 1983
Concentration (ppb):
< SO masa >=- SO
*"* ........ '
Episode Maximum Ozone: BS85 - High Biogenlc
July 2, 1988 - July 17, 1988
F»centration (ppb): S5SS3 < SO
aaamm >= •{ 4Q
Figure 5-32. Predicted episode maximum 1-hour
ozone concentrations (ppb) for the Juiv 1988
>best esti-te" "ooeS ind 198*5 bat cS
5-97
-------
' »"l
•f<'
This page is intentionally left blank to accommodate
reproduction needs of the color version of this report.
-------
200
Q.
0.
1 160
o
£140
I
Il2fl
2
X
<
2 100
80'
300
270
240
2
Z150
'120
90
IMPACT OF BIOGENICVOC EMISSIONS ON MAXIMUM 1-HR OZONE
BOSTON
BS85 CS01 CS05 CSI1
* HKHBKKEHCS i BESTESWEBKXfflCS A IWHOGMCS
IMPACT OF 8KJGENICVCC EMISSIONS ON MAXIMUM 1-HR OZONE
GREATER CONNECTICUT
—l—
CS19
—l—
CS01
—I—
CS05
—l—
CSI1
BS85 CS01 CS05 CSI1 CS19
* mGHBBGOKS 0 8ESr£5HM«EBIOG0CS A LOW8KKENK3
380
^340
3
|300
o
S260
x.
j.220
2
§180
I J
140
100-
200
to
0.
1 160
o
N
0
140
§120
!100
IMPACT OFHOGENICVOC EMISSIONS ON MAXIMUM 1-HR OZONE
GREATER NEW YORK CITY
BS85 CS01 ' CS05 CS11 CS19
* HKHBIOCENCS 0 BESTESMKBIOGENICS A LOWBiOGENCS
IMPACT OF HOGEMCVOC EMISSIONS ON MAXIMUM 1-HR OZONE
PHIUDEPHIA/KENT
T
~T
BS85 CS01 CS05 CS11
* HffiHBIOGEMCS 0 BESTESIWAIEBBGENICS A LWBIOGEMCS
—1—
CS19
Figure 5-33. Predicted episode maximum 1-hour ozone concentrations for biogenic sensitivity
scenarios for selected urban areas. (Page 1 of 2)
5-99
-------
•1
..-'I
300
270
240
EL
I
§210
,1180
3
§150
'120
90
200
!]$fl
§120
;100
80
IMPACTOFKOGENICWCEMSSlOfiSOHWAXlHlIM 1-HR OZONE
BMJWOREAASHINGTON,DC
—1—
BS85
—l—
CS01
—I—
CS05
—l—
CS11
—I—
CS19
* HCH90C8KS 0 KSTESIMflEBIOCOIKS A UWBOGBflCS
MWTOFBIOGEMCVDCBISS10HS ON HAMUUM 1-HR OZONE
PHTSBURGH/JOHNSrOWN/ALTOONA
—l—
CS05
—l—
CS1I
BS85 CSOI CS05 CS1I CS19
* HCH90C8KS D BESTESMttlEBOCENKS A LOW8KXMS
200
CQ
0.
1160
o
S
120
2 100
200
,180
160
§120
IMPACT OF BIOGOflC VOC EMISSIONS ON MAXIMUM 1 -HR OZONE
CKARLESTON/HUNTINGTON
—I—
CSOI
CS05
—l—
CS11
—I—
CS19
* HKH8KKENICS Q GBrESHM«ESIOC£NICS A LOWaOCEKCS
IMPACT OF BIOGENIC VOC EMISSIONS ON MAXIMUM 1-HR OZONE
DETROIT
i i 1 1 r—
BS85 CS01 CS05 CS11 CS19
* KCHBBGENICS 0 KSTESIMAIEKOGENICS A IWSOGffilCS
Figure 5-33 (Page 2 of 2)
5-100
-------
GRIO-HOUROZONEEXC£EDENCES>I25PPB GRID-HOUR OZONE EXCEEDENCES> 125 PPB
BOSTON GREATER NEW YORK CITY
500
450
§400-
o
^350-
J
o •
§250
0
u.200-
o
£150-
ID
§100-
z
50
0-
A
\
\
A \
\
1
^
A- ;
1
5000
4500
i/i
§4000
13500
1
u3000
o
§2500
Q.
0
u.2000
o
2J1500
OJ
iiooo
2
500
0
*
k
i \
\
*
•
i '
.
^-^
\J
r ] 1 | ; , 1
BS85 CS01 CS05 CS11 CS19 BS85 CSQ1 CS05 CS11 CS19
* HIGH8«K£NICS i KSIESIWAIEBKXB1ICS A OT8DOTCS * WGHBKKENICS i 8ESTEMOIOGOIICS A IOW8CGENICS
OT-HWROZWEXCEEDENC£S> 125PPB GRID-HOUR OZONE EXCEEDENCES> J25PPB
GREATER CONNECTICUT , PHUDEPHIA/KENT
2000
1800
VI
§1600
0
^1400
GJ1200
§1000
0
O
S 600-
o
§ 400-
z
200-
0-
'
c
.
k r
\
\
\
\
1
,
^^^i
A \
BS85 ' CS01 CS
,
i — 8*\^
^s
1000-
900-
in
§ 800-
f 700-
u 600-
o.
1 500-
o
u. 400
0
£ joo-
CQ'
I 200-
2
too-
0-
4
[
^
\
\
A
f
n
•
4
n ii
1 1 '
35 csn CS'S 8S85 ' CS01 CS05 CS11 CSI9
* HKHBOCffllCS B 8ESTESI1M«EBKXS«CS iLOWBfflGEMS * HKHBBGENKS 13 BESIESIIIWEfiBGENiCS A LOWBKXBK5
Figure 5-34. Predicted grid-hours exceeding 125 ppb for biogenic sensitivity scenarios for selected
urban areas. (Page 1 of 2)
5-101
-------
GRID-HOUR OZONE EXCEEDENCES > 125 PPB
BALTIMORE/WASHINGTON, DC
1200
1100
£1000
0 900
J 803
8 700
— 60C
!soo
° 400
| 300
2 200
too
0
\
\
\
i
,
-~-
i
/
]
~ , 1 _
2 • — {
\
GRID-HOUR OZONE EXCEEDENCES > 125 PPB
CHARLESTON/HUNTINGTON
.
100
90
1/1
o
f 70
-i
ui 60
u
Q rn
5 ™
O
u 40
o
S 30
CO
1 20
z
10
0-
.
•'
[ |*^
\
x
X
*
--. f
^H
9'
* HffiHBfflGEHCS Q BESIESIMWEBBGENICS A UW8KXBIICS
200-
180
in
§160
Q
f 140'
1
uil20<
u
9100'
a
o
u. 80'
0
ffl
§40-
z
20-
0-
4
. .
\
\
a \
N.
\ f
ii 1 .
u S ~~3 H
• i i 1 1 1 _, _, ,
BS85 CS01 CS05 CS11 CS19 BS85 CS01 CS05 CS11 CS19
* HCHBOCOCS D ESIESrattlEBKXHIICS A OT6IOG£MCS
ft HIGHBBGENCS Q BESTESIWreBOGENICS A LOWBHXENICS
Figure 5-34 (Page 2 of 2)
5-102
-------
1110
! 90
£
70
60
50-
z
0
§20
a.
o
«• 0
30
10
c
I'80
x 160
| 140
a
ui
§.100
M
| 80
!? 60
0
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
BOSTON
BS85 CS01 CS05 CS11 CS19
* HGHBIOGQKS 10 BESTESMEBIOGEMCS A UWBOGEHCS
POPULATION EXPOSURE T01 -HR MAXIMUM OZONE > 125 PPB
GREATER CONNECTICUT
BS85
—l—
CS01
CS05
CS11
CS19
* HBHBBGEHCS i BESTESIIkttTEBBGENICS A IOWHOGENICS
21000
» 875
U
I 750
o
§625
a:
w 500
8 375
9 250
g 125-
o
«• fl
c
1180
£
x160
f 140
a
*100
m-
o 80
20-
POPULATION EXPOSURE T01 -HR MAXIMUM OZONE > 125 PPB
GREATER NEW YORK CRY
8S85 CS01 CS05 CS11 CS19
* HGHBKKEN1CS 0 BESTESMIEBXKENICS A LOWBIOGOIICS
POPULATION EXPOSURE T01 -HR MAXIMUM OZONE > 125 PPB
PHILADEPHIA/KENT
A
BS85 CS01 CS05 CSI1 CS19
* HICHBIOGENiCS i BESTESHMAIEBIOGENKS A LMBIOGENICS
Figure 5-35. Predicted population exposure to ozone exceeding 125ppb for biogenic sensitivity
scenarios for selected urban areas. (Page 1 of 2)
5-103
-------
>'.*
I
1
POPULATION EXPOSURETO 1-W? MAXIMUM OZONE> 125 PPB
a*LTWORE/toSHINGTON,OC -
.-.120
tiltO
S
EIOO
K
£ 90
f 80
ft
£70
u
e 80
•
2 40
o 30
5 20
I to
° 0
.
[
A
\
\
\
>
A
i
*~-^^
r j
1 ' fl J
.
1
1
-slO
c
0
= 9
£
x 8
a
f 7
a
ij
in
0 4.
Q.
z
o
3 ,
O
a. n
u
I
i
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
CHARLESTON/HUNT1NGTON
f
r
! & 3 o a
1 i i1" i 1 '
BS85 CS01 CS05 CS11 CSI9 BS85 CS01 CS05 CS11 CS19
* HCHBOCQKS Q BESIESIWTEBOCEMCS A LWaOCENICS
WnXAnOHD(POSlJR£T01-HRMAXIUUMOZONe>125PPB
PmSBORGH/JOHNSTOWN/MTOOttt
* HKHKOGBIICS 0 BESTESIlMAIEBKXHdCS A LOWBOGENCS
POPULATION EXPOSURE TO 1-HR MAXIMUM OZONE > 125 PPB
DETROIT
-?50
0
5= JC
1
* T\l
1*
a
5
j
Is
>.
D
C
3 5
° 0
j.
V
\
^ ^N
1
L
-
^___^ 4
« • S n
^50
c
0
= 45
£
x40
-------
TABLE 5-1. PROCEDURES USED FOR CALCULATING POPULATION EXPOSURE
.T n
where
Pop/
n
t
T
t=l z=l
population exposure to ozone exceeding concentration x within area A for the entire
episode; units are in population-hours
population in grid / in area A
n total grids in area A
hour
total hours in episode = (24 hours x number of days)
= 1 if X« > c, = 0 otherwise; occurrence of ozone exceeding concentration x in grid / for
hour t
ozone concentration cutoff (i.e., 125ppb)
5-105
-------
•AJ
TABLE 5-2, SUMMARY OF EMISSIONS REDUCTIONS THAT REDUCED PREDICTED OZONE TO
<125PPB
Peak ozone
2005 Post-control
Emissions reductions beyond
2005 baseline
VOC
NOx
CO
Northeast Corridor cities
Philadelphia
Boston
Batt./Wash., DC
New York City
148 ppb
158 ppb
149 ppb
268 ppb
117 ppb
11.6 ppb
118 ppb
119 ppb/
63%
60%
80%
85%
48%
52%
57%
22%
16%
18%
17%
21%
Cities outside the Northeast Corridor
Pittsburgh
Cleveland
Detroit
Charleston, WV
138 ppb
139 ppb
140 ppb
128 ppb
116 ppb
115 ppb
117 ppb
109 ppb
58%
NC
NC
77%
55%
47%
60%
38%
16%
NC
NC
7%
NC « no change
5-106
-------
TABLE 5-3. COMPARISON OF THE JUNE 1983 AND JULY 1988 EPISODE MAXIMUM 1-HOUR
CONCENTRATIONS FOR THE 2005 BASELINE AND CS19
Maximum 1-hour concentrations
June 1983 July 1988
2005 CS19 2005 CS19
New York City 242
Baltimore/Washington, DC 160
Philadelphia 162
Boston 143
Pittsburgh 117
Cleveland 143
Detroit 135
Charleston, WV 95
110
124
118
108
95
104
99
69
268
149
148
158
138
139
140
128
120
122
115
113
105
112
109
95
5-107
-------
TABLE 5-4. SUMMARY OF OZONE METRICS USED TO ASSESS THE IMPACTS OF REACTIVITY-
BASED STRATEGIES
Average reduction in daily maximum
New York City
Greater Connecticut
Rhode Island
Group 1 *
Group 2 *
Group 3 *
2005 vs. CS20
69
37
22
8
5
2
ozone (ppb)
for days > 125 ppb
2005 vs. CS12 CS10VS.
80
. 46
28
16
12
10
Number of days with ozone > 125
New York City
Greater Connecticut
Rhode Island
Group 1 *'
Group 2 *
Group 3 *
Percent
2005 CS20
12 8
9 6 '
7 1
6 4
2 2
2 2
CS12
6
3
0
3
0
0
36
13
5
1
<1
<1
ppb
CS10 Cj
10
4
0
0
0
0
CS15
S15
8
1
0
0
0
0
reduction in population exposure to ozone > 125 ppb
2005 vs. CS20 2005 vs. CS12
New York City
Greater Connecticut
Rhode Island
Group 1 *
Group 2 *
Group 3 *
75
69
90
46
33
28
92
95
100
79
82
92
2005 vs. CS1Q 2005 VS. CS15
79
95
100
100
100
100
93
99
100
100
100
100
Average for group
5-108
-------
TABLE 5-5. CASES TESTED WITH VARYING BIOGENIC EMISSIONS
Case
Description
Base 1985:
Base1985L:
Base1985H:
CS01:
CS07:
CS09:
CS05:
CS06:
CS08:
CS11:
CS17:
CS19:
CS21:
CS20:
Base case 1985 emissions; 'best' estimate biogenics
Base case 1985 emissions; "low* estimate biogenics
Base case 1985 emissions; "high" estimate biogenics
Strategy 1 -maximum technology VOC controls applied regionwide;
"best" estimate biogenics
Strategy 1; "low" estimate biogenics
Strategy 1; 'high' estimate biogenics
Strategy 5-Clean Air Act legislation controls;
"best" estimate biogenics
Strategy 5; "low" estimate biogenics
Strategy 5; 'high" estimate biogenics
Maximum NOX controls from Strategy 10, VOC at 2005 base levels;
"best" estimate biogenics
Strategy 11; "low" estimate biogenics
Maximum technology, reduced reactivity, across-the-board reductions and modified
NOX controls in Baltimore/Washington DC;
•best" estimate biogenics
Strategy 19; "low" estimate biogenics
Strategy 19; "high" estimate biogenics
5-'109
-------
This page is intentionally left blank.
-------
SECTION 6
STATE ACCESS AND USE
OF
ROM DATABASES
by
Ruen-Tai Tang
Computer Sciences Corporation
P.O. Box 12767
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
-------
•if
-:,j
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 Sections. 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 strate-
gies. 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
(SIPs) 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 desig-
nated as the preferred approach for attainment demonstrations. Thus, the ROM is best used by "nesting"
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 manageable
way is needed for States to retrieve data from applications of the ROM to support SIP development.
A second problem arises from the fact that, like the ROM, the UAM is also a three-dimensional photo-
chemical grid mode! - and it is nearly as complex as the ROM. In addition, the UAM's grids are typically
smaller 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 are 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 System
(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 (NCC) 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 UAM 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 appli-
cations.
This section 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 (1991) and
Tangetal. (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 DAM 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 Database
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, 1991).
The next three subsections 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
%' '">, %, f
Concentration, Data
~ ~ f vs <•<.<• - _ . _ . "
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-hour 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). 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 ail 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 (hfgK molecular weight)
CO Carbon Monoxide
ETH Ethene
FORM Formaldehyde
H2O2 Hydrogen Peroxide
HNO2 Nitrous Acid (MONO)
HNO3 Nitric Acfd
1SOP Isoprene
MTHL Methanof (MEOH)
NO Nitric Oxide
NO2 Nitrogen Dioxide
O3 Ozone
"OL'E Oiefins'
PAN Peroxyacetyl Nitrate
PAR "Paraffins"' ' '..
TOL Toluene
XYL Xyiene
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
computing 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 Processpt Date
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-afr 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)
Biogeriic 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 UAM 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 DAM 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 concentra-
tion 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; 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 problems.
6.2.4 User Protocol for Data Retrieval
For UAM Subsystem data retrievals, users specify the selection criteria for ROM concentration data.
GMISS automatically selects the associated ROM processor data.
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
6-7
-------
=»
1
-A
•*
-i
of Interest. If a user enters valid and appropriate values for all of these fields, the DAM Subsystem 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.
GMISS 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 Ust-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 Rnal 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 (1) match the specifi-
cations for the domain, study, scenario, and date, and (2) are available for retrieval. Users select the par-
ticular 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. A single set (scenario) of ROM
processor data may be used to compute species concentrations for many different emissions scenarios;
therefore, 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 are 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 nongridded upper-air vertical profiles (observed data);
GEQ Time-invariant gridded geographic data (surface roughness, land use, and terrain elevation);
B1OGEN Hourly gridded biogenic emissions;
MET Hourly gridded meteorology data (surface meteorology; layer 2 wind fields, layer 1 wind fields,
cloudfness, height of layer 1, height of layer 2, and layer i water vapor concentration).
6-8
-------
The data type selection menu allows users to choose the categories of data for which the UAM
Subsystem 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 station
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 OX1DANT MODEL - URBAN AIRSHED MODEL
(ROM-UAM) 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 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 and use a common chemical mechanism, interfacing these two
models Is not straightforward. A detailed discussion of the methodologies and assumptions used in the
interface 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 Interface.
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 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, various mete-
orological parameters (e.g., winds, temperature, water vapor), 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 program.
Specifically, an output file generated by an interface is in a compatible format for direct input to either a
UAM preprocessor or the model.
Section 6.3.2 provides a summary of some of the salient features and important limitations when applying
the interface programs. Section 6.3.3 presents 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.
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 - not earlier versions of these two models.
• Three interfaces provide formatted input files for the UAM preprocessors and four 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 developed
for the GMISS on the 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 pollutant
species that must be specified in the UAM (CB-IV) model. Default values are defined for the
remaining six species.
Lateral and top boundary concentrations are resolved hourly and spatially at each UAM grid
cell.
6-11
-------
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 editor
at a terminal.
A utility program included with the interface package converts any binary file into an equivalent
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 method-
ologies 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 er a/., 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 etal. 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 Preprocessor System (EPS)
already exist for creating emissions for these source types (Volume IV, Causley et 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
species..
• No interfacing is performed for the UAM preprocessors, CPREP or SPREP, the chemical
parameters and the simulation control programs, respectively.
.6,3,3 Description of the Interface System
A general framework showing how the interface program package fits into the overall UAM model system
Is depicted in Figure 6-1. The interface programs are executed before exercising any UAM preprocessor
6-12
-------
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.
Because 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 ef 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 con-
verting 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
complete 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) and surface features, emissions, and finally, control information about the chemical species and
reactions, and the model run parameters.
A comprehensive approach was taken in interfacing the two model systems 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
6-13
-------
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) for use as inputs to
the appropriate UAM preprocessor. These files are formatted; thus, the user has the capability to
examine their contents and can insert additional data before exercising the UAM preprocessors. The
Interface programs for generating winds (IWIND) and for initial and boundary concentrations (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 {BIOG 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
f
User Inputs
GMISS DATA RETRIEVAL PROGRAM
(EPA-NCC IBM 3090)
i
"RETRIEVED" DATA FILES
(Formatted/Transferable)
INTERFACE PROGRAMS
FORMATTED "PACKET" FILES
UAM PREPROCESSOR PROGRAMS
^- BINARY FILES
f
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 UAM PREPROCESSOR PROGRAMS
Preprocessor
AIRQUL
BNDARY
TPCONC
DFSNBK
REGNTP
METSCL
TMPRTR
DWMi
CRETER
EPS?
PTSRCE
Internal name in
binary file
Contents
CONCENTRATIONS
AIRQUAUTY Initial concentration fields
BOUNDARY Lateral boundary concentrations
TOPCONC Top boundary concentrations
METEOROLOGY AND SURFACE CHARACTERISTICS
DIFFBREAK
REGIONTOP
METSCALARS
TEMPERATURE
WIND
TERRAIN
EMISSIONS
PTSOURCE
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
CPREP CHEMPARAM
SPREP SIMCONTROL
Species reaction rate information
Model simulation input information
1. DWM= Diagnostic Wind Model
2. EPS 3 Emissions Preprocessor System
TABLE 6-2. RETRIEVED ROM FILES USED BY INTERFACE PROGRAMS
ROM
data file
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 UAH
input files programs file preprocessor
DBDATA
PF119 I > RTDATA
PF102 User data
PF117 (optional)
PF119
RTDATA
DBOATA
User data
(optional)
MF165
PF115
PF1 19
DBDATA
RTDATA
PF118
ROM21
PF119
DBDATA
User EMISSIONS* I — > BIOASC (optional)
Binary
file
HDD IK
UDOiN
DTD f U
K 1 Dirt
MCD T U
MbolN
TDD T U
IPBIN
| AQBIN
_•* 1 Bf*DTU
•>J oUolN
j TCSIN
* Anthropogenic area emissions file
6-17
-------
This page is intentionally left blank.
i
:
-------
REFERENCES AND BIBLIOGRAPHY
-------
d
This page is intentionally left blank.
-------
Alliance Technologies Corporation. 1989. Regional Ozone Modeling For Northeast Transport - Devel-
opment of a Base Year Anthropogenic Emissions Inventory. EPA-450/4-89-008. Prepared for the EPA,
Office of Air Quality Planning and Standards, Research Triangle Park, NC.
Alliance Technologies Corporation. 1990. Regional Ozone Modeling for Northeast Transport - Projec-
tion Year and Control Strategy Emissions Inventories. Draft Report. Prepared for the EPA, Office of Air
Quality Planning and Standards, Technical Support Division, Research Triangle Park, NC.
Arnts, A., and S. Meeks. 1981. Biogenic hydrocarbon contribution to the ambient air of selected areas.
Atmos. Environ., 15(9):1643-1651.
Baidocchi, D., D. Matt, B. Hutchison, and R. McMiilen. 1984. Solar radiation within an oak-hickory forest:
an evaluation of the extinction coefficients for several radiation components during fully-leafed and
leafless periods. Agricultural and Forest Meteorology, 32(3/4):307-322.
Battye, W., M. Smith, and M. Deese. 1987. Cosf Assessment of Alternative National Ambient Air Quality
Standards for Ozone. Draft Report for External Review. Alliance Technologies Corporation. Prepared
for the EPA, Office of Air Quality Planning and Standards, Ambient Standards Branch, Research Triangle
Park, NC.
BEA. 1986. 1985 OBERSBEA 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: Emissions Preprocessor System. EPA-450/4-90-007D. Prepared for the EPA, Office of Air
Quality Planning and Standards, Research Triangle Park, NC.
Chameides, W., R. Lindsay, J. Richardson, and C. Kiang. 1988. The role of biogenic hydrocarbons in
urban photochemical smog: Atlanta as a case study. Science, 241:1473-1475.
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. Prepared for the 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. Metcaif, eds., Wiley Publ.,
New York. pp. 369-459.
Dessent, T. 1991. Gridded Model Information Support System (GMISS) User's Guide. Volume II: UAM
Subsystem. EPA-450/4-91-009. Prepared for the EPA, Office of Air Quality Planning and Standards,
Research Triangle Park, NC.
Dodge, M.C. 1989. A comparison of three photochemical oxidant mechanisms. J. Geophys. Res.,
94(D4):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
EPA, Office of Air Quality Planning and Standards, Research Triangle Park, NC.
Ellison, W., 1988. Assessment ofSOz and NOX Emission Control Technology in Europe. EPA-600/2-88-
013. EPA, Research Triangle Park, NC.
EPA. 1986. Guideline on Air Quality Models (Revised). EPA-450/2-78-D27R. U.S. Environmental Pro-
tection Agency, Office of Air Quality Planning and Standards, 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 VMTForecasting Model (Draft Final Report). U.S. Depart-
ment of Transportation, Office of Policy Administration, Federal Highway Administration.
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 pp.
Gery, M.W., G.Z. Whitten, and J.P. Killus. 1988. Development and Testing of the CBM-IVfor 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):12925-12956.
Godowttch, 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.
Heffter, J.L 1980. Air Resources Laboratories Atmospheric Transport and Dispersion Model
(ARL-ATAD). NOAA Technical Memorandum ERL ARL-81. National Oceanic and Atmospheric Admin-
istration, Environmental Research Laboratories, Silver Spring, MD.
Johnson, K. 1988. Industrial Boiler Low-NOx Combustion Retrofit Cost Algorithm. EPA-600/8-88-091.
National Acid Precipitation Assessment 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
Mechanism in OZIPM/EKMA. EPA-450/4-84/009. U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Lamb, B., A. Guenther, D. Gay, and H. Westberg. 1987. A national inventory of biogenic hydrocarbon
emissions. Atmos. Environ., 21 (8): 1695-1705.
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. Part 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, Environment Canada, Quebec, Canada,
February 25-271987. 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 Protec-
tion Agency, Research Triangle Park, NC. .
Lamb, R.G., and G.F. Laniak. 1985. A Regional Scale (1000 km) Model of Photochemical Air Pollution.
Part 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 the EPA, Research Triangle Park, NC.
Logan, J. 1983. Nitrogen oxides in the troposphere: global and regional budgets. J. Geophys. Res.,
88(C15): 10785-10807.
Milich, L, L Bender, T. Boehm, C. Coats, J. Eichinger, S. Fudge, S. Hallyburton, D. Jordan, C. Maxwell,
D. Olerud, R. Tang, A. Van Meter, R. Wayland, J. Young, O. Bullock, J. Novak, T. Pierce, S. Roselle, and
K. Schere. 1991. The Regional Oxidant Model (ROM) User's Guide. Parti: The ROM Preprocessors.
EPA/600/8-90/083a. EPA, Office of Research and Development, Atmospheric Research and Exposure
Assessment Laboratory, Research Triangle Park, NC.
Modica, L, D. Dulleba, R. Walters, and J, Langstaff. 1988. Flexible Regional Emissions Data System
(FREDS) Documentation for the 1985 NAPAP Emissions Inventory. EPA/600/9-89/047. Prepared by
Alliance Technologies Corporation for the U.S. Environmental Protection, Research Triangle Park, NC.
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. EPA, Office of Air Quality Planning and
Standards, 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. 199Gb. 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. EPA, Office of Air Quality Planning and Standards, 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 No. 86-30. In: Pro-
ceedings of the 79th Annual APCA Meeting, Air Pollution Control Association, Pittsburgh, PA, June
1986.
Olson, R., C. Emerson, and M. Nunsgesser. 1980. GBOECOLOGY: A County-Level Environmental Data
Base for the Conterminous United States. ORNL/TM-7351. Oak Ridge National Laboratory, Oak Ridge,
TN. 54pp.
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. EPA Contract No.
68-01-7047, Work Assignment No. 126. Prepared for the EPA, Office of Policy, Planning, and Evaluation,
Washington, DC.
Peterson, E. and D. Tingey. 1980. An estimate of the possible contribution of biogenic sources to
airborne hydrocarbon concentrations. Atmos. Environ., 14(1):79-81.
Pierce, I.E., and K.L Schere. 1988. Regional Ozone Modeling: An Investigation of Hydrocarbon
Emissions 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. 1990a. Development of a biogenic emissions inventory system for
regional scale air pollution models. Paper No. 90-94.3. In: Proceedings of the 83rd Annual AWMA
(formerly APCA) Meeting, June 24-29, 1990. Air and Waste Management Association, Pittsburgh, PA. 16
pp.
Pierce, T.E., K.L Schere, D.C. Doll, and W.E. Heilman. 199Gb. 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, Nijmegen, The
Netherlands, May 9-13,1988. Elsevier, The Netherlands, pp. 623-632.
Rao, ST., G. Sistla, J.Y. Ku, R. Twaddell, R. Whitby, y. Pagnotti, E. Davis, N, Possiel, J. Pearson, K.
Schere, and R. Dennis. 1987. Examination of the Urban Airshed Model performance in the New York
metropolitan area. Paper no. 87-71.8. In: Proceedings of the 80th APCA Meeting, New York, NY,
June, 22-26,1987. Air Pollution Control Association, Pittsburgh, PA.
R-6
-------
Rasmussen, R. 1972. What do the hydrocarbons from trees-contribute to air pollution? Journal of the Air
Pollution Control Association, 22(7) :537.
Roselle S. and K. Schere. 1990. Sensitivity of the EPA Regional Oxidant Model to biogenic hydrocarbon
emissions. In: Proceedings of the 83rd Annual AWMA (formerly APCA) Meeting, 1990. Air and Waste
Management Association, Pittsburgh. PA. 20 pp.
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): Devel-
opment of the Annual Data and Modeler's Tapes. EPA-600/7-89-012a. U.S. Environmental Protection
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 Oa predictions from a test application of the EPA Regional Oxidant Model.
In: Proceedings of the Fifth Joint Conference on Applications of Air Pollution Meteorology, Chapel Hill,
NC, November 18-21 1986. American Meteorological Society, Boston, MA, pp. 61-64.
Schere, K.L, and AJ. 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: Air Pollution
Modeling and Its Application, V - Vol. 10..C. DeWispelaere era/, eds., Plenum Press, NY. pp. 45-59.
Schere, K.L, and N.C. Possiel. 1984. U.S. EPA Regional Oxidant Model - Background and overview.
Paper no. 84-47.2. In: Proceedings of the 77th Annual APCA Meeting, San Francisco, CA, June 25-29,
1984. Air Pollution Control Association, 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, Nijmegen, The Netherlands, May 9-13, 1988.
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: ROM-UAM Interface Program System.
EPA-450/4-90-007E. U.S. Environmental Protection Agency, Research Triangle Park, NC.
Tingey, D., M. Manning, L Grothaus, and W. Burns. 1980. Influence of light and temperature on mono-
terpene emission rates from slash pine. Plant Physiol., 65:797.
Tingey, D. 1981. The effect of environmental factors on the emission of biogenic hydrocarbons from live
oak and slash pine. In: Atmospheric Biogenic Hydrocarbons. J. Bufalini and R. Arnts, eds., Ann Arbor
Science Publications, Ann Arbor, Ml. pp. 53-79.
R-7
-------
Tingey, D., R. Evans, and M. Gumpertz. 1981. Effects" of environmental conditions on isoprene emissions
from live oak. P/ante, 152:565.
Trainer, M., E. Williams, D. Parrish, M. Buhr, E. Allwind, H. Westberg, F. Fehsenfeld, and S. Liu. 1987.
Models and observations of the impact of natural hydrocarbons 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.
Walters, R., L Modica, and D. Pratt. 1988. 1985 NAPAP Emissions Inventory Allocation Factors. EPA-
600/7-89/01 A. Alliance Technologies Corporation. Prepared for the EPA, Research Triangle Park, NC.
Whitten, G.Z., and M.W. Gery. 1986. Development of CBM-X Mechanisms for Urban and Regional
AQSMs. EPA-600/3-S6/012. U.S. Environmental Protection Agency, Research Triangle Park, NC.
Williams, E., D. Parrish, and F. Fehsenfeld. 1990. A parametric representation of nitric oxide emissions
from soils. NAPA International Conference on Acidic Deposition: State of Science and Technology,
February 11-16,1990. Hilton Head, SC.
Young, J.O., M. Aissa, T.L Boehm, C.J. Coats, J.R. Eichinger, S.J. Roselle, A.R. Van Meter, R.A. Wayland,
and T.E. Pierce. 1989. Development of the Regional Oxidant Model, Version 2.1. EPA/600/3-89/044.
(NTIS PB89-194252). EPA, Office of Research and Development, Atmospheric Research and Exposure
Assessment Laboratory, 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. Environmental Protec-
tion Agency, Atlanta, GA. 173 pp.
R-8
-------
- w<_ w I t>^ * *-v.
(Please read Instructions on the reverse before completing)
1. REPORT NO.
£PA-450/4-91-Q02a
2.
3. RECIPIENT'S ACCESSION NO.
4. TITLE AND SUBTITLE
REGIONAL OZONE MODELING FOR NOKTHEAST TRANSPORT
- Project Final Report -
S. REPORT DATE
June 1991
6. PERFORMING ORGANIZATION CODE
Editors
Norman C. Possiel, Lenard B. Milich, and
Beverly R. Goodrich
8. PERFORMING ORGANIZATION REPORT NO
9. PERFORMING ORGANIZATION NAME AND ADDRESS
Computer Sciences Corpa
Bldg 4401, Alexander Dr
Research Triangle Pk
N.C. 27709
'Alliance Technologies Corp
100 Europa Dr, Suite 150
Chapel Hill -
N.C. 27514
10. PROGRAM ELEMENT NO.
> A24A2F
11. CONTRACT/GRANT NO.
a. 68-01-7176
b. 68-D9-0173 WA #8
12. SPONSORING AGENCY NAME AND ADDRESS
U.S. Environmental Protection Agency
Technical Support Division
Office of Air Quality Planning and Standards
MD-14 Research Triangle Park, N.C. 27711
13. TYPE OF REPORT AND PERIOD COVERED
Final
14. SPONSORING AGENCY CODE
22
15.SUPPLEMENTARY NOTES contributing authors: E.L. Meyer,
T.E. Pierce, D. Doll, L.B. Milich, J.O. Young, W.H. B
K.A. Baugues, R-T Tang. E. Baldridge, and J.A. Godowitch
N.C. Possiel, K.L. Schere,
Battye, J.E. Langstaff, M.G. Smith,
16. ABSTRACT
The Regional Ozone Modeling for Northeast Transport (ROMNET) Project was initiated
by the U.S. EPA and State and local air pollution agencies in the Northeast to address
the problem of regional transport in developing effective and equitable control program
to attain the ozone National Ambient Air Quality Standard 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 precur
sor levels transported between urban areas following application of control measures;
and; 3) to provide procedures and guidance for incorporating ozone and precursor trans-
port in future State Implementation development.
ROMNET included the application of the EPA Regional Oxidant Model (ROM) for a
number of regional emissions control strategies. These strategies were designed to
address five major issues: 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?"; and 3) "What levels
of VOC and/or NOx emissions reductions are necessary to reduce predicted ozone levels
in the Northeast to below 125 ppb?"; 4) "How effective are potential reactivity-based
strategies? ; and 5) "How does the large uncertainty in biogenic emissions alter
conclusions on the effectiveness of controls?"
7.
KEY WORDS AND DOCUMENT ANALYSIS
DESCRIPTORS
b.lOENTIFIERS/OPEN ENDED TERMS
c. COSATI Field/Group
tropospheric ozone
regional modeling
regional emissions inventories
control strategies
ozone transport
8. DISTRIBUTION STATEMENT
19. SECURITY CLASS (This Report I
21. NO. OF PAGES
321
20. SECURITY CLASS /This page)
22. PRICE
EPA Form 2220—1 (Rev. 4—77) PREVIOUS EDITION is OBSOLETE
-------
------- |