v-xEPA
United States Environmental Sciences Research EPA-600/9-84-006
Environmental Protection Laboratory February 1984
Agency Research Triangle Park NC 27711
Research and Development
Proceedings of the
EPA-OECD
International
Conference on Long-
Range Transport
Models for
Photochemical
Oxidants and Their
Precursors
p^.'t"* • •
i-.'
OF
-------
EPA-600/9-84-006
February 1984
PROCEEDINGS Of THE
EPA-OECO INTERNATIONAL CONFERENCE
ON LONG-RANGE TRANSPORT MODELS FOR
PHOTOCHEMICAL OXIDANTS AND THEIR PRECURSORS
April 12-14, 1983
Environmental Research Center
Research Triangle Park, North Carolina 27711 (USA)
Sponsored by
U. S. Environmental Protection Agency
Office of Research and Development
Environmental Sciences Research Laboratory
Research Triangle Park, North Carolina 27711 (USA)
Under the Patronage of
Organization of Economic Cooperation and Development
Environment Directorate
Paris, France
Project Officer
Basil Dimitriades
Atmospheric Chemistry and Physics Division
Environmental Sciences Research Laboratory
Research Trianqle Park, North Carolina 27711
ENVIRONMENTAL SCIENCES RESEARCH LABORATORY
OFFICE OF RESEARCH AND DEVELOPMENT
U.S. ENVIRONMENTAL PROTECTION AGENCY
RESEARCH TRIANGLE PARK, NORTH CAROLINA 27711
-------
DISCLAIMER
Peer review requirements have been fulfilled by including in the
proceedings the comments of the conference attendees on each presentation.
Presentations have received a cursory edit. Transcripts of discussions were
sent to discussers for checking, but not all responses were received in time for
use; hence, the discussions, at points, are unclear.
Views expressed by non-EPA speakers or discussers do not necessarily
reflect the views or policies of the U.S. Environmental Protection Agency.
Mention of trade names or commerical products does not constitute
endorsement or recommendation for use.
ii
-------
ABSTRACT
The U.S. Environmental Protection Agency (EPA) and the Organization for
Economic Cooperation and Development (OECD) are concerned (1) by the fact that
the photochemical oxidant pollution problem, due to large-scale formation or
long-range atmospheric transport, has international dimensions, and (2) by the
lack of ready-to-use methods for formulating optimum control strategies for
regional oxidant reduction. In reaction to these concerns, the U.S. EPA and
OECD jointly organized the international conference documented in these
proceedings.
These proceedings contain presentations made at the conference by some of
the world's foremost experts in the field of oxidant air quality modeling and
presentations made by national experts on their countries' emissions inventories
and air quality monitoring activities. Also included are discussions of the
presentations, informal presentations, panel discussions, and conference
conclusions and recommendations. Among the subjects discussed were the need for
and utility of regional oxidant models, six regional models currently under
development, and available aerometric and emissions inventory data bases in OECD
countries.
iii
-------
CONTENTS
Abstract ill
Abbreviations and Symbols xii
INTRODUCTION
A. Ellison 1
A. Galli 3
P. Lieben 5
B. Dimitriades 7
SESSION I. EXISTING REGIONAL MODELS FOR OXIDANTS
Jack Shreffler, Chairman 13
Needs and Applications of Regional Air Quality Simulation Models
for Oxidants in North America
Henry S. Cole 14
Introduction 14
The Northeast Corridor Regional Modeling Project (NECRMP) 17
Status/Schedule of NECRMP 21
Organizational Structure—EPA/State Cooperation 21
NECRMP Funding and Resources 2k
Questions Addressed in NECRMP Modeling 25
Major Difficulties and Problems 27
Conclusions 28
References 29
Needs and Applications of Regional Air Quality Simulation Models
for Oxidants in Europe
S. Zwerver and Peter Built jes 31
Introduction 31
Description of the Dutch Air Quality Management System 33
Air Quality Policy Requirements 35
Application of Photochemical Dispersion Models in
The Netherlands 38
Conclusions and Remarks 50
Acknowledgments 51
Bibliography 52
Discussion 53
Appendix A. Guidelines for Emissions Inventory
Presentations 54
Appendix B. List of Pollutant Species 57
Appendix C. Emissieregistraitie 58
-------
CONTENTS, continued
U.S. EPA Regional Oxidant Model for the Transport of Photochemical
Oxidants and Their Precursors ,
Robert G. Lamb and Joan H. Novak 67
Structure of the Model and Its Input Data Processor Network.... 67
Required Model Resolution and Current Data 7^
Model Application in Different Regions 83
References 85
Discussion 86
Appendix. Questionnaire on the Characteristics of Existing
Regional-Scale 03 Models 87
Regional Models for Oxidants: Norwegian Lagrangian Long-Range
Transport Model with Atmospheric Boundary Layer Chemistry
Oystein Hov, Anton Eliassen, Jorgen Saltbones, Ivar S.A. Isaksen,
and Frode Stordal 3^
Introduction 9^
Model Description 96
A Case Study 119
Acknowledgments 1 2k
References \2k
Discussion 1 27
Model for the Regional Transport of Photochemical Oxidants and
Their Precursors in the United Kingdom
Kenneth A. Brice 128
Introduction 128
Model Description 1 30
Results and Discussion 1 39
Summary 1^8
Acknowledgments 1 ^9
References 149
Discussion 152
Application of a Regional Oxidant Model to the Northeast
United States '
James P. Killus, Ralph E. Morris, and Mei-Kao Liu 153
Introduction 1 53
Model Equations 15^
Application of the Model to the Northeast United States 161
Model Exercises for the July 1978 Episode 171
Evaluation of Model Predictions ! 175
vi
-------
CONTENTS, continued
References [[[ 188
Discussion ................ ..................................... ]8S
Appendix . SAI Regional Oxidant Model .......................... 191
Development of. a Regional-Scale Air Quality Model
Mei-Kao Liu and Steven D. Reynolds ..................................
Introduction .............. ..................................... 19^*
Development of a Regional-Scale Air Quality Model .............. 195
Description of Model Equations ................................. 196
Simplification of the Treatment for the Surface Layer .......... 202
Chemical Kinetic Calculations .................................. 209
Applications of the Regional Transport Model ................... 213
Appendix. Incorporation of an Aerosol Module .................. 236
References ................ k ..... . ..............................
STEM Model
Gregory R. Carmichael, Toshihiro Kitada, and Leonard K. Peters ......
Introduction [[[ 2k$
Model Description .............................................. 2*t6
Results and Discussion. .... .................................... 265
Conclusions ............... , .................................... 272
Acknowledgments ........... , .................................... 272
References ................ , ...... . ............................. 273
Discussion ................ . ....................................
Acid Deposition and Oxidant Mod^l
P.K. Misra and A.D. Christie ........................................ 276
Model Description ......... , ..... . .............................. 276
Acknowledgments ........... , .................................... 279
References ................ , .................................... 279
Discussion [[[ 280
Appendix. Response to Conference Questionnaire ................ 282
The NATO/CCMS Air Pollution Model Comparison
Han van Dop [[[ 285
Introduction [[[ 285
-------
CONTENTS, continued
General Discussion Following Session I ................. . ............ 292
SESSIONS II. AVAILABLE EMISSIONS INVENTORY DATA BASE
Lars Lindau , Chairman [[[ 299
Northeast Corridor Regional Modeling Project Emissions Inventory
Joan H. Novak and James H. Southerland .............................. 300
Introduction [[[ 300
Background [[[ 30 1
Point Source Data .............................................. 307
Canadian Inventory ............................................. 312
Area Source Data ............................................... 3 1 ^
Data Quality [[[ 319
Bibliography [[[ 323
Discussion [[[ 324
Emissions Inventory Data Bases for the United States
Charles 0. Mann [[[ 326
Introduction [[[ 326
NEDS Point Source Data ......................................... 32?
NEDS Area Source Data .......................................... 329
Availability of NEDS Data ...................................... 331
Other Data Bases ............................................... 33 1
Current Developments ........................................... 33^
References [[[ 335
Discussion [[[ 336
Emissions Inventories and the National Emissions Inventory System
Arthur Sheffield [[[ 337
Introduction [[[ 337
-------
CONTENTS, continued
SESSION III. AVAILABLE AEROMETRIC DATA BASES
Dieter Jost, Chairman [[[ 361
Availability of Ozone and Ozone Precursor Data from the
SAROAD System
Jacob G . Summers [[[ 362
Introduction [[[ 362
Data Collection and Reporting Requirements ..................... 363
NAMS/SLAMS Reporting Requirements .............................. 36?
The SAROAD System .............................................. 369
Ambient Data Available for Transport Models .................... 37'
Data Availability .............................................. 377
Bibliography [[[ 381
Discussion [[[ 382
Appendix. Guidelines for Aeroroetric Data Presentations ........ 383
Northeast Corridor Regional Model Project: Data Base of Regional
Ambient Chemical and Meteorological Measurements
Norman C. Possiel and Francis S. Binkowski .......................... 385
Introduction [[[ 385
Regional Data Base Components .................................. 387
Data Base Availability ................. ......................... 407
References [[[
Discussion ............ . ........................................
Canadian Surface Air Quality Monitoring Networks
Thomas Dann and David Balsillie ..................................... Al 0
Introduction [[[ AlO
Ontario Ministry of the Environment Monitoring Network ......... 423
-------
CONTENTS, continued
Photochemical Oxidants in Northwestern Europe, 1976-1979,
a Pilot Study
Jorgen Schjoldager, Harold Dovland, Peringe Grennfelt, and
Jorgen Saltbones ^39
Introduction
Ozone Monitoring Stations
Summary of Ozone Measurements
Selected Episodes
Conclusions
Recommendations
Acknowledgments 470
Bibliography
Emissions Inventories in Europe
Lothar Kropp
Introduction
Survey of Clean Air Plans and Emissions Inventories
in the Federal Republic of Germany
Clean Air Plans
Summary 500
References 500
Appendix A. Guidelines for Emission Inventory
Presentations
Appendix B. Guidelines for Aerometricc Data
Presentations 507
SESSION IV. MODEL EVALUATIONS, PANEL DISCUSSIONS
Dieter Jost, Chairman 509
SESSION V. MODEL EVALUATIONS, PANEL PRESENTATIONS 511
STEM Model
Peter Builtjes 512
SAI Model
Han van Dop 519
Hov Model
Elidoro Runca 525
-------
CONTENTS, continued
Lamb /Novak Model
Frank Smith [[[ 531
Uk/ADOM Model
Anton Eliassen [[[ 539
SESSION VI. CONCLUSIONS AND RECOMMENDATIONS
Pierre Lieben , Chairman ..................................................
-------
LIST OF ABBREVIATIONS AND SYMBOLS
ABBREVIATIONS
AAQS — Ambient Air Quality Standard
AGL — Above ground level
AIRS — Aerometric Information Retrieval System
AQCR — Air Quality Control Region
AQMS — Air Quality Management System
AT — Air temperature
b-scat — Light-scattering coefficient
BMC — b-matrix compiler
BNL — Brookhaven National Laboratory
CAA — Clean Air Act
CCN — Cloud condensation nuclei
CDHS — Comprehensive Data Handling System
CMA — Census Metropolitan Area
CMC — Canadian Meteorological Center
CPU — Central processing unit
DOC — Department of Commerce
DOE — Department of Energy
DPT — Dew point temperature
ECAO — Environmental Criteria and Assessment Office
xii
-------
ECE — Economic Commission for Europe
EIS — Emissions Inventory System
EIS/AS — Emissions Inventory Subsystem/Area Sources
EIS/P&R — Emissions Inventory Subsystem/Permits and Registration
EIS/PS — Emissions Inventory Subsystem/Point Sources
EKMA — Empirical Kinetic Modeling Approach
EMEP — European Monitoring and Evaluation Programme
EMSL — Environmental Monitoring Systems Laboratory
EPA — Environmental Protection Agency
EPRI — Electric Power Research Institute
EPS — Environmental Protection Service
ERG — Environmental Research Center
ESRL — Environmental Sciences Research Laboratory
FRG — Federal Republic of Germany
GC — Gas chromatography
GMT — Greenwich Mean Time
GOES — Geostationary Operational Environmental Satellite
GWL — Grosswetterlagen
HATREMS — Hazardous and Trace Emissions System
HDD — Heavy-duty, diesel-powered
HDG — Heavy-duty, gasoline-powered
HDV — Heavy-duty vehicle
HERL — Health Effects Research Laboratory
HEW — Department of Health, Education, and Welfare
HPA — Heavily polluted area
IERL — Industrial Environmental Research Laboratory
Xlll
-------
IVL — Swedish Water and Air Pollution Research Institute
LDV — Light-duty vehicle
MAP3S — Multistate Atmospheric Power Production Pollution Study
MIF -- Model input field
MPO — Metropolitan Planning Organization
c
NAAQS — National Ambient Air Quality Standards
NADB — National Aerometric Data Branch or National Air Data Branch
NAMS — National Air Monitoring System
NAPAP — National Acid Precipitation Assessment Program
NAPS — National Air Pollution Surveillance
NASA — National Aeronautics and Space Administration
NATO — North Atlantic Treaty Organization
NATO/CCMS — North Atlantic Treaty Organization/Committee on the Challenges of
Modern Society
NAVF — Norwegian Research Council for Science and the Humanities
NCC — National Computer Center
NECRMP — Northeast Corridor Regional Modeling Project
NEDS — National Emissions Data System
NEROS — Northeast Regional Oxidant Study
NWS — National Weather Service
OAQPS — Office of Air Quality Planning and Standards
OECD — Organization for Economic Cooperation and Development
ORD — Office of Research and Development
PAQSM — Photochemical Air Quality Simulation Model
PEPE — Persistent elevated pollution episode
PIF — Processor input file
QSSA — Quasi-steady-state approximation
xiv
-------
RAMC — Regional Air Mass Characterization
RAPS — Regional Air Pollution Study
RMDHS — Regional Model Data Handling System
ROM — Regional Oxidant Model
RTI — Research Triangle Institute
RTM — Regional Transport Model
SAI — Systems Applications, Inc.
SAROAD — Storage and Retrieval of Aerometric Data
SCC — Source classification code
SFT — Norwegian Pollution Control Authority
SIC — Standard Industrial Classification
SIP — State Implementation Plan
SLAMS — State and Local Air Monitoring Stations
SURE — Sulfate Regional Experiment
TNO — The Netherlands Organization for Applied Research
TSP — Total suspended particulate
TSR — Total solar radiation
UK — United Kingdom
USA — United States of America
UTM — Universal Transverse Mercator
UV — Ultraviolet
UVR — Ultraviolet radiation
VOC — Volatile organic compound
WHO — World Health Organization
xv
-------
SYMBOLS
C — Carbon
HC — Hydrocarbon
NMHC — Nonmethane hydrocarbon
NO — Nitric oxide
NOX — Nitrogen oxides
N02 — Nitrogen dioxide
N03 — Nitrate
03 — Ozone
PAN — Peroxyacetyl nitrate
PEN — Peroxybenzyl nitrate
SOX — Sulfur oxides
S02 — Sulfur dioxide
S04 — Sulfate
TNMHC — Total nonmethane hydrocarbon
xvi
-------
INTRODUCTION
Alfred H. Ellison
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
The EPA Office of Research and Development (ORD) is composed of a
headquarters located in Washington, DC, and 14 laboratories located throughout
the United States. Four of these laboratories are located at the Environmental
Research Center (ERG) in the Research Triangle Park.
The major focus of ERC is air pollution research, although other
environmental research is conducted. The four laboratories at ERC are the
Health Effects Research Laboratory (HERL), the Environmental Monitoring Systems
Laboratory (EMSL), the Industrial Engineering Research Laboratory (IERL), and
the Environmental Sciences Research Laboratory (ESRL).
Most of HERL's research in health effects involves animal studies.
However, HERL also operates a clinical facility in Chapel Hill in which human
subjects are placed in a chamber, exposed to pollutants, and monitored for
exposure effects.
EMSL is responsible for monitoring activities related primarily to air
pollutants. Recently, EMSL has conducted a lot of work in hazardous waste
monitoring.
-------
lERL's focus is the development of control systems that are needed for
stationary sources of air pollution. As a laboratory, they are concerned with
assessing source pollutants and the control systems available for these
pollutants. Where there are difficiencies, IERL attempts, at least, to develop
prototype systems for controlling air pollution from stationary sources.
ESRL conducts mostly air pollution research and atmospheric sciences
research, which brings us to the subject of this conference.
In addition to the four laboratories, several other organizational units
are located at ERG. One is the Office of Administration, which provides
administrative support to ERG. Another is the National Computer Center (NCC),
which includes the two large computers used in EPA's data processing activities.
The Office of Air Quality Planning and Standards (OAQPS) is primarily
responsible for writing the regulations related to air pollution. Finally, the
Environmental Criteria and Assessment Office (ECAO) produces criteria documents.
These documents, which are primarily air pollution documents, are produced
approximately every 5 yr and review all of the available information on a
particular pollutant or toxic chemical.
-------
Alfred Galli
Office of Research and Development
U.S. Environmental Protection Agency
401 M Street, SW
Washington, DC 20460 (USA)
Good morning. Speaking as a member of the OECD AMP Group the and U.S.
EPA-ORD staff, I am delighted that the Group's efforts in the photochemical
oxidant control area have culminated in this extremely useful project and that
the research staff of the U.S. EPA will have the opportunity to be of
assistance. Reflecting ongoing concerns of the U.S. EPA, the past and present
U.S. representatives in the AMP Group have repeatedly expressed interest in the
development and use of regional-scale models for pollution control strategy
development purposes. The long-range transport of polluants is now an
established fact and has consequences of concern both within a country, such as
the U.S., and to clusters of countries, such as Europe, North America, etc. To
understand such problems and to design for each of the involved countries
effective and equitable efforts for their control, it is imperative that
regional models be used as tools. However, regional models, especially those
for oxidants, as you will hear repeatedly in this conference, are extremely
difficult to develop and use, and they require substantial commitments on the
part of the interested countries.
The United States is certainly an interested country. We are interested
because we experience both domestic and international problems associated with
long-range pollutant transport. As a result, we have expended and continue to
-------
expend substantial resources in efforts to solve these problems, and we would
certainly welcome the opportunity to pool our resources with those of other
countries for that purpose.
-------
Pierre Lieben, Secretary
Organization for Economic Cooperation and Development
AMPG/Environmental Directorate
Paris, France
This workshop has been designed to comparatively examine existing models
for long-range ozone transport with respect to their conceptual validity,
complexity, and input data as well as with respect to resource requirements for
evaluating and subsequently utilizing one or more of the models in order to
eventually devise a strategy for OECD member countries to satisfactorily control
the photochemical oxidants pollution problem on an international scale. Such a
plan should address the following points:
• Determine the state of the art of existing emissions inventories and
develop a plan to complete these as model input data.
• Refine the best available model(s) that can be used with those data
bases that appear to be easily obtainable.
• Discuss the merits of selected model(s) with respect to their serving as
bases for developing control strategies.
• Recommend a chemistry to be used in the suggested model(s) that best
fits the input data likely to be available.
• Examine whether the aerometric data base (including boundary condition
data) is sufficient and, if not, formulate a plan to get the missing
data.
• Identify dates at which the model(s) will be fully operational.
• Define the modeling domain in consideration of the emissions and
aerometric data available, the topography, and the model capability.
-------
It is hoped that during Session VI the above questions will be answered.
It is suggested that this outline form the basis for discussions during the
meeting, especially in Section IV.
-------
Basil Dimitriades
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711 (USA)
To the introductory remarks offered by the preceding speakers, I wish to
offer a few comments on the rationale behind the conference agenda, in the hope
that this will further help in focusing the conference discussions. These
comments are based on the extensive experiences of the U.S. EPA staff with
physicochemical modeling in general and with regional 03 modeling in particular.
They are intended to convey the message that in order for the conference to be
successful, it is crucial that the conferees go away with a realistic
appreciation of the complexity of comprehensive regional 03 models, and of the
penalties and merits associated with model simplification. Crucially important
also is the need for the conferees to appreciate the skill and effort
requirements of a regional 03 model evaluation effort, such as the one
contemplated by OECD.
The U.S. EPA has been involved in physicochemical modeling of air quality
for almost a decade. Model development efforts started in earnest in 1974 when
the Regional Air Pollution Study (RAPS) was initiated in St. Louis for the
purpose of providing a field data base for modeling urban air quality. The
emphasis at that time was on urban-scale modeling, and the efforts eventually
resulted in several urban-scale PAQSMs for 03, some of which were extensively
evaluated by the U.S. EPA using the RAPs data base. From these experiences with
the urban models, it became apparent that the most useful and valid
physicochemical models, that is, those with the highest degree of spatial and
-------
temporal resolution and with detailed chemistry, should also have extremely high
skill and resource requirements. To illustrate the latter problem, for each of
these models now in existence, there are currently only two to three
institutions at the most in the United States that have demonstrated their
ability to use the model with confidence.
Development of regional PAQSMs was initiated in the late 1970s after the
air pollution scientists realized that air pollutants can travel long distances
and that photochemical smog problems cannot necessarily be solved or alleviated
by controlling local emissions only. These realizations led to the conclusion
that for control strategies to be rational, it is imperative that the role and
contribution of distant upwind sources be considered, and prompted the U.S. EPA
to launch an extensive regional model development program.
Unlike the approaches taken by others, the approach taken by the U.S. EPA
modelers to regional-scale modeling was not based on simple expansion of
urban-scale models. Such an approach is not apropriate for truly regional
models since it is limited to treating only those physical processes that occur
or are important during a single solar day and within a few miles from the
source. The U.S. EPA regional model currently under development will treat, in
addition, nighttime chemistry, very slow reactions, biogenic organics chemistry,
air movements associated with the nocturnal jet, cumulus cloud effects, and the
processes and effects that have significance mainly in connection with multiday
pollutant transport. Thus, relative to other regional modeling approaches, this
-------
one approved by the U.S. EPA is conceptually more valid. However, it is also
enormously more complex, difficult, and costly to use.
These comments on the differences among the various regional modeling
approaches are offered for the purpose of pointing out two implications bearing
on the conference conclusions and followup decisions. The first implication is
that, for a given regional model application, the most detailed and conceptually
valid model is not necessarily the most appropriate and desirable choice.
Equally useful may be models that have somewhat lower validity but are much less
difficult and costly to use. It is precisely for the purpose of helping the
conferees judge the existing models from such standpoints that we included in
the agenda presentations on the needs and intended applications of regional 03
models in the United States and in Europe. I urge the conferees to consider
carefully these presentations in their deliberations regarding the relative
merits and limitations of existing models with respect to serving the OECD
needs.
The second implication is that the conferees should come away with a clear
and realistic appreciation of the differences among the various existing models
with respect not only to scientific validity but also to practicality. In the
hope of facilitating comparisons of models, we have requested that the modelers
describe their models following given, detailed guidelines (questionnaires).
Adhering to these guidelines will certainly ensure a measure of comparability,
but it is questionable whether these standardized model descriptions alone will
allow useful judgments to be made upon the practicality aspect. Such
information will have to be extracted from those with extensive experience in
-------
the evaluation and application of physicochemical models. It is for this
purpose that we have included in Sessions IV and V of the conference panel
discussions addressed to, among other things, resources needed for model
evaluation. Such discussions should deal not only with money and man-years but
also expertise requirements. Only for the purpose of illustrating this point, I
offer some very rough estimates made by the U.S. EPA modelers for testing the
U.S. EPA regional model with European field data. Such an effort, encompassing
processing of European emissions and aerometric data, revising the model to
increase its domain (to cover the European OECD-member countries) and to allow
for varying terrain roughness (several European countries are predominantly
mountainous), and running the model, is estimated to be roughly three expert
man-years or more. Since the U.S. EPA staff with working experience in the
regional modeling area consists of three persons, it follows that such a
model-testing project would require the full-time involvement of the entire U.S.
EPA regional modeling capabiliy for one year or more. Although these estimates
are very rough, they nevertheless illustrate the point that such a project would
require almost prohibitively large commitments in resources It is crucial,
therefore, that the conferees appreciate such problems and seek compromises that
entail realistic model testing efforts and also adequately serve the needs of
OECD.
My last comment is to stress that we at the Environmental Sciences Research
Laboratory of the U.S. EPA are truly delighted to have this opportunity to
discuss with our international colleagues this extremely important subject. We
look forward to the presentations and followup discussions, and we feel
committed to assisting OECD in any way we can in this worthwhile undertaking.
10
-------
We have an enormous respect for our international colleagues and we would be
anxious to continue the close contacts that we expect to make during this
conference. In this vein o£ enthusiasm, I would like to submit for
consideration by the U.S. EPA and OECD that the U.S. EPA host within the ESRL
facility one or two OECD modelers to work with us for a period of one year or so
in the regional modeling area. We would certainly gain much from such a close
exposure to our guest experts, and OECD would enhance its capability for
conducting future modeling projects, such as the one contemplated as a followup
to this conference.
My wishes to you all for a productive and enjoyable meeting.
11
-------
SESSION I
EXISTING REGIONAL MODELS FOR OXIDANTS
April 12, 1983
13
-------
NEEDS AND APPLICATIONS OF REGIONAL AIR QUALITY SIMULATION MODELS
FOR OXIDANTS IN NORTH AMERICA*
Henry S. Cole
Monitoring and Data Analysis Division
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
INTRODUCTION
The purpose of this paper is to discuss the needs and applications of
regional air quality simulation models for oxidants in North America. The paper
focuses on 03, the oxidant designated in the National Ambient Air Quality
Standard (NAAQS), and on the northeastern United States, the area that is being
modeled by the U.S. Environmental Protection Agency (EPA).
Ozone is one of the most serious and widespread air pollution problems in
the United States. About one-third of the nation's population lives in 32 urban
areas that have been designated as nonattainment regions for 03. Under the
Clean Air Act Amendments of 1977, states have primary responsibility for
developing and enforcing programs to control 03. In accord with this
responsibility, the states in 1982 issued plans that describe how they intend to
achieve the NAAQS for 03 by 1987, the statutory deadline. These State
Implementation Plans (SIPs) are currently under review by EPA (1983).
*This paper has been reviewed by the Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, and approved for publication.
Mention of trade names or commercial products does not constitute endorsement or
recommendation for use.
14
-------
The problem of interurban/regional transport of 03 and/or precursors was
first recognized in the Los Angeles Basin during the 1950s when the impact of
Los Angeles emissions on the San Bernadino and Riverside areas of the Eastern
Basin was established (HEW, 1970). During the past 10 yr, numerous studies have
demonstrated the significance of 03 and precursor transport. The maximum 03
concentrations associated with urban source regions are recorded tens of
kilometers downwind of the source regions (Martinez and Meyer, 1978).
Significant transport of 03 over hundreds of kilometers has also been
demonstrated in the Midwestern United States (White et al., 1976, Vukovich,
1977) and in the Northeastern United States (Cleveland et al., 1976).
Several studies present evidence for interstate transport in the
Northeastern United States. Figure 1 shows the isopleths of maximum 03 observed
on a day when the winds were from the southwest, i.e., along the metropolitan
corridor. The very high 03 concentrations in Connecticut (about 100 km downwind
of New York City) appear to be associated with morning rush hour emissions from
the tri-state metropolitan area of New Jersey, New York, and Connecticut.
During an EPA study based on Lagrangian aircraft measurements, the Baltimore 03
plume was tracked for 400 km into New England (Clark et al., 1982). Other
studies indicate that substantial quantities of 03 are periodically transported
from the Midwest and Gulf State cities to the Northeastern States (Clarke and
Ching, In press; Wolff and Lioy, 1980).
Unfortunately, the legal framework for 03 control is not well suited to the
regional nature of the problem. The responsibility for attainment rests with
individual states, and there is no formal mechanism for multistate control
15
-------
8t>
Figure 1. Isopleths of maximum observed 03 concentrations (ppb) for
July 16,1980.
16
-------
programs. Secondly, regional oxidant models have not been available to the
states for the 1982 SIPs. In most cases, the states have used the simple
city-specific EKMA Model (EPA, 1981), which treats transport simplistically. In
most cases, days with significant 03 transport were not included in the
analyses, the assumption being that control efforts in upwind cities would
eliminate this problem in the future. Although this approach appears to be
reasonable for areas with moderate 03 levels, the use of refined regional
oxidant models may be desirable for areas that will have difficulty attaining
the 03 NAAQS by 1987.* The experience with 03 SIPs suggests that a change from
the traditional approach to a multistate approach may be required to develop an
effective and equitable control program for the Northeast United States.
THE NORTHEAST CORRIDOR REGIONAL MODELING PROJECT
The Northeast Corridor Regional Modeling Project (NECRMP) is a joint
program of the Northeastern United States and the EPA (EPA, 1980). Initiated in
the late 1970s, NECRMP grew out of the recognition .that a serious and widespread
03 problem exists in the Northeastern United States and that violations are
strongly affected by the regional transport of 03 and its precursor pollutants.
The NECRMP region, shown in Figure 2, encompasses the region from northern
Virginia to northern New England and from the East Coast to eastern Ohio and
southern Ontario. Five major urban areas—Washington, DC, Baltimore,
Philadelphia, New York City, and Boston—are included in the area. The purpose
*For example, the SIP estimates for NMHC reductions were 60% for New York City
and 85% for Los Angeles (EPA, 1983).
17
-------
(0
B
o
•o
w
o
w
(-1
(0
T3
c
3
o
to
CM
0)
1-1
3
18
-------
of the project is to use refined photochemical models to develop effective and
equitable strategies for 03 control in the region. As Figure 3 shows, the basic
concept of the program is to integrate the use of urban- and regional-scale
models.
A key component of NECRMP is EPA's Regional Oxidant Model (ROM)
(Lamb, 1982). ROM is a three-dimensional photochemical grid model that
simulates the effects of emissions, meteorology, chemistry, and deposition on
concentrations of 03 and other pollutants (CO, NO, N02, NMHC, etc.). Its scale
enables the model to represent transport between cities and over the full extent
of the NECRMP region. Both single-day and multiday simulations are possible.
ROM will be used: (1) to ascertain the importance of transport on a scale of
100 km or more, (2) to estimate broad reductions :n emissions required to meet
the 03 standard across the Corridor region, and (3) to provide estimates of
future-year boundary concentrations that are required as inputs for the
urban-scale model applications.
The ROM, however, has too course a grid (18 km x 18 km horizontal cells) to
simulate the gradients and the peak concentrations that occur immediately
downwind (within 100 km) of urban areas. Thus, the NECRMP concept incorporates
the use of an urban-scale photochemical grid model (i.e., the Urban Airshed
Model) to develop strategies that will ensure attainment in the major urban
areas. Initial plans called for modeling in all five of the major urban areas.
However, limited resources will probably restrict the number of cities that can
be modeled to one or two. (Airshed modeling for Philadelphia is in progress and
the greater New York City metropolitan area is a second candidate.)
19
-------
I Regional
I Air Quality +
I Meteorology
1. Evaluation
2. Application
Regional
Inventory
Regional I
Model |
1. Base Case
2. Projections!
Growth
Control
I Broad Scale I
I Target Emissions I
I Reductions I
Future Boundary I
Concentrations I
Integrated
Control
Programs
I
I
Urban
Inventory
Urban
Model
1. Evaluation
2. Application
I Urban
Control
I Strategies
Urban I
Air Quality 4- I
Meteorology I
Figure 3. Basic NECRMP concept (how regional and urban
modeling interface).
20
-------
STATUS/SCHEDULE OF NECRMP
The status and schedule of NECRMP are summarized in Table 1. All of the
data bases required for regional modeling have been completed or are nearly
complete. ROM verification activities are scheduled for 1984 and 1985, and
regional control tests will be conducted in 1986. With regard to the urban
scale, extensive modeling will be conducted for Philadelphia in 1983 and 1984.
During this period, the Office of Air Quality Planning and Standards (OAQPS)
will: (1) conduct base-case simulations and model evaluation exercises for
Philadelphia, (2) conduct a series of sensitivity tests, and (3) run a limited
number of future-year projection and control-strategy simulations based on rough
estimates of boundary concentrations upwind of Philadelphia. The tests for the
latter activity can be rerun by using the boundary concentrations output from
ROM for the 1986-1987 time frame.
ORGANIZATIONAL STRUCTURE—EPA/STATE COOPERATION
Table 2 shows the organizational structure of NECRMP. The role of the
NECRMP Policy Group is to set the overall direction of the project and is
composed largely of the directors of state air pollution control agencies and
air branch chiefs of the regional EPA offices. OAQPS and the Office of Research
and Development, Environmental Sciences Research Laboratory (ORD/ESRL), are
21
-------
Status
TABLE 1. SCHEDULE OF NECRMP ACTIVITIES
Activities3
Completed
Remaining
Air quality/meteorological field programs (1979, 1980)—data
bases now available.
Annual county-based emissions inventories—complete spatial,
temporal, and species allocation
files available; some remaining problems.
%
Automated regional inventory data handling system—on-line.
First generation ROM—on-line.
Episode characterization/selection of test days (fall 1983).
Complete regional emissions inventory for ROM (late 1983)
Complete second-generation ROM (spring 1984).
Complete validation of ROM-2 (spring 1985).
Complete projection and control strategy emission tapes
(spring 1985).
Simulations for projection and control strategies 1986.
aAll dates are in calendar years.
22
-------
TABLE 2. NECRMP ORGANIZATION
Group
Composition
Function
Interagency
Policy Group
Interagency Model
Work Group
States, regional
EPA offices
Office of Air
Quality Planning
and Standards
Office of Research
and Development
Administrators from States, EPA
(regional offices, Office ot Air
Quality Planning and Standards,
and Office of Research and
Development)
Technical representatives of
States, EPA (regional offices,
Office of Air Quality Planning
and Standards, and Office of
Research and Development)
Overall direction of
project, recommenda-
tions on priorities,
policy questions.
Formulation of
modeling protocol,
resolution of tech-
nical issues, recom-
mendations to Policy
Group and to EPA.
Funding/resources for
emissions/air quality
data bases.
Overall project coor-
dination/management .
Chair Policy and
Work Groups. Final
emissions inventory.
Aerometric analyses
(episodes ) .
Development, refine-
ment, validation,
application of ROM.
represented at upper management levels. The NECRMP Work Group, which consists
of technical representatives of the states, ORD, and OAQPS, is responsible for
developing modeling objectives and procedures that are consistent with the
directives of the Policy Group. The interagency composition at both the policy
and technical levels is designed to ensure that the modeling protocol will be
responsive to the needs and limitations of the participating agencies. The EPA
23
-------
project manager serves as the chairman of the Work Group and the executive
secretary (ex officio) of the Policy Group.
EPA's responsibilities are as follows: ORD/ESRL is responsible for
developing, testing, and applying the ROM. OAQPS is responsible for
coordinating the project, developing the ROM input data bases (emissions, air
quality, and meteorological), and conducting the Philadelphia airshed model
study. State and local agencies are responsible for supplying the basic
emissions inventories and considerable air quality and meteorological data to
OAQPS and for participating in planning the program as discussed above.
NECRMP FUNDING AND RESOURCES
Approximately $8 million have been spent thus far on NECRMP (about
two-thirds by EPA and one-third by the states), and an additional $4 to
$5 million will be needed to complete the project. Major emissions and
aerometric data bases have been assembled largely by contractors. Analyses of
03 episodes and plume transport and transformation have been performed
internally and by contractors. Preparation of the final ROM model inputs
(emissions, meteorology, and air quality) is being done largely in-house, as are
the development, testing, and application of the ROM. The Philadelphia modeling
project is being carried out by a contractor under the supervision of OAQPS.
24
-------
QUESTIONS ADDRESSED IN NECRMP MODELING
The Policy Group and the Work Group are currently identifying the questions
to be tested with the ROM (and urban-scale models). Although it is not possible
at present to give a definitive list of these questions, examples are given in
Table 3.
The general procedure that will be used to answer questions related to
control strategies is shown in Table 4. The first step is to identify days that
represent the important types of high 03 episodes in the region. (The premise
is that the meteorological conditions of the test days will represent those
associated with similar episodes in future years.) The next step is to model
the same test days by using "baseline projection" emissions inventories for a
future year in which attainment is required. (The baseline projections reflect
changes in emissions associated with changes in population and economic activity
and also scheduled reductions mandated by control programs already in place.)
The results of these model simulations are then used to determine the types of
episodes that are likely to require greater control. These episodes will then
be modeled again to determine the types of additional control programs that will
be required for attainment of the 03 NAAQS. As stated previously, parallel
studies will be conducted with urban-scale models in order to determine the type
of control programs that may be necessary to assure compliance in the urban
areas and downwind environs. The urban analyses will use forecasts of
future-year boundary concentrations supplied by the ROM.
25
-------
TABLE 3. EXAMPLES OF QUESTIONS TO BE ANSWERED BY THE STUDY
1. What are the relative contributions of transported and urban (local)
emissions in different parts of the region? What are the relative
contributions of different source regions to high 03 in different parts of
the region?
2. What are the relative contributions of various types of sources to high Oa
occurrences, e.g., stationary sources vs. mobile sources?
3. How will boundary concentrations for urban areas change in the future?
(Boundary concentrations are used as inputs for urban-scale photochemical
modeling.)
4. What levels of precursor control are required to attain the NAAQS for 03 and
how effective are various types of control programs or approaches? Specific
questions related to control:
a. If the region is approached as a whole, what level of precursor control
(% NMHC reduction) is required to attain the NAAQS?
b. How is the estimate in (a) affected by regional changes in the level of
NOX emissions?
c. What percent reductions in NMHC (and/or NOX) are required for the
different urban regions in order to reduce regional 03 values to the
NAAQS?
d. Does it make sense to .reduce emissions of sources located in attainment
regions?
e. What is the relative impact (effectiveness) of reducing emissions from
mobile vs. stationary sources?
f. What would be the effect of changes in regional fuel composition or of
changes in currently mandated automobile emission control programs?
g. How effective are specific control measures on a regional basis, e.g.,
substitution of solvents, vapor recovery measures, traffic reduction?
26
-------
TABLE 4. BASIC NECRMP APPROACH
1. Identify important types of high 03 episodes.
2. Perform base-case ROM simulation/model validation.
3. Develop future-year projection emissions inventories.
4. Perform future-year projection simulations with ROM.
5. From the results of (4), determine which episodes/areas require additional
control.
6. Prepare various types of "control scenario" emissions inventories.3
7. Apply ROM for the critical episodes in (5) by using various control scenario
inventories (6).
"Initial simulations will be based on simple sensitivity tests to focus on
control requirements. This will be followed by more specific control programs.
MAJOR DIFFICULTIES AND PROBLEMS
Many problems associated with NECRMP are related to the enormous size and
complexity of the program. The large number of agencies and tasks involved
requires careful coordination on a continuing basis. In addition, circumstances
and perceptions are different in each state, and it is not surprising that
positions on key items and the degree of interest in the project vary from
agency to agency. Furthermore, large spending requirements have come at a time
of shrinking budgets. Diminished levels of funding have reduced the ability of
the states to assemble urban emissions inventories, and the number of cities for
which the Urban Airshed Model can be used has been reduced substantially.
Perhaps the greatest problem is the long lead time and uncertainty associated
with the project. The possibility remains that major technical (or funding)
problems will cause delays that will jeopardize the regulatory utility of the
27
-------
project. Another serious problem is that there is presently no firm legal basis
for converting the results of NECRMP analyses into binding multistate air
pollution control programs. Despite the problems and uncertainties, the
participants have continued to support NECRMP as a viable, and perhaps the best,
means for developing controls that are equitable and effective.
CONCLUSIONS
This paper has described EPA's major ROM program, NECRMP. This program was
initiated in the late 1970s and is scheduled for completion in 1987. Continued
support for this program stems from the need to develop control programs that
are effective and fair and that are based on sound scientific analyses. The use
of refined state-of-the-art photochemical models that incorporate the transport
of 03 and precursors provides a more credible basis for expensive control
programs than do simple models such as EKMA.
A major beneficial result is the joint EPA/state participation in regional
modeling efforts. The establishment of the Interagency Policy and Work Groups
in 1981 has increased the interest and involvement of the states in NECRMP. The
interagency committees have not only contributed to the flow of information but
have also been useful in working through complex problems and in resolving the
differences between agencies. The cooperative process in molding the study will
hopefully lead to multistate control strategies that are perceived as fair and
effective. Nevertheless, there are no provisions in the current Clean Air Act
Amendments that legally bind the States to use the results of regional modeling
efforts. The currently mandated SIP approach is not well suited to the regional
28
-------
nature o£ the 03 problem and has led to considerable confusion and conflict.
Moreover, regional efforts such as NECRMP, although beneficial, operate on a
purely voluntary basis. The answer to this problem may lie in revising the
amendments to permit multistate/EPA implementation programs for pollutants and
areas that are strongly affected by regional transport.
REFERENCES
Clark, J. F., and J. K. S. Ching. In press. Aircraft Observations of Regional
Transport of Ozone in the Northeastern United States. Meteorology and
Assessment Division, Environmental Sciences Research Laboratory, U.S.
Environmental Protection Agency.
Clark, T. L., J. F. Clarke, and N. C. Possiel. 1982. Boundary Layer Transport
of NOX and 03 from Baltimore, Maryland—A Case Study, Paper 82-24.3, Air
Pollution Control Association, Annual Meeting, New Orleans, Louisiana.
Cleveland, W. S., et al. 1976. Photochemical air pollution: Transport from the
New York City area into Connecticut and Massachusetts. Science,
191:179-181.
Lamb, R. G. 1982. A Regional Scale (1000 km) Model of Photochemical Air
Pollution, Part I: Theoretical Formulation. Meteorology and Assessment
Division, Environmental Sciences Research Laboratory, U.S. Environmental
Protection Agency.
Martinez, E. L., and E. L. Meyer. 1978. Urban-Nonurban Ozone Gradients and
Their Significance. Air Pollution Control Association, Special Conference
on Ozone/Oxidants: Interactions with the Total Environment, Dallas, Texas,
March 12, 1976. Reported in Air Quality Criteria for Ozone and Other
Photochemical Oxidants. Vol. 1, EPA-600/8-78-004, Office of Research and
Development, U.S. Environmental Protection Agency.
U.S. Department of Health, Education, and Welfare. 1970. Air Quality Criteria
for Photochemical Oxidants. Public Health Service.
U.S. Environmental Protection Agency. 1983. A Review of the Modeling Analyses
Supporting 1982 State Implementation Plans for Ozone (Draft). Office of
Air Quality Planning and Standards.
U.S. Environmental Protection Agency. 1981. Guideline for Use of City-Specific
EKMA in Preparing Ozone SIPs. EPA-450/4-80-027, Office of Air Quality
Planning and Standards.
29
-------
U.S. Environmental Protection Agency. 1980. Northeast Corridor Regional
Modeling Project Study Protocol, Office of Air Quality Planning and
Standards.
Vukovich, F. 1977. In: International Conference on Oxidants, 1976—of
Evidence and Viewpoints, Part V. The Issue of Oxidant Transport.
EPA-600/3-77-117, U.S. Environmental Protection Agency.
White, W. H., et al., 1976. Formation and transport of secondary air
pollutants: Ozone and aerosols in the St. Louis urban plume. Science,
194:187-189.
Wolff, G. T., and P. J. Lioy. 1980. Development of an ozone river associated
with synoptic scale episodes in the Eastern United States. Environmental
Science and Technology, 14(10):1257-1260.
30
-------
NEEDS AND APPLICATIONS OF REGIONAL AIR QUALITY
SIMULATION MODELS FOR OXIDANTS IN EUROPE*
S. Zwerver
Head of the Air Quality Division
Ministry of Housing, Physical Planning and Environment
Directorate Air
Dokter Reijersstraat 12
2265 BA Leidschendam, The Netherlands
P.J.H. Builtjes
Project Leader, Air Quality Management System
MT-TNO, Department of Fluid Dynamics
P.O. Box 342
7300 A Apeldoorn, The Netherlands
INTRODUCTION
Although the oxidant problem in Europe has not reached the level it has in
the United States, it has become a subject of growing concern, especially
because of its possible link with acidification. This paper emphasizes the
Dutch situation in particular, because we understand the circumstances
prevailing in that part of Europe. However, the Dutch situation can often be
viewed as representative of larger parts of Western Europe as well.
Although The Netherlands is situated at a high latitude (52° N),
substantial 03 levels can prevail in the summer season. For example, the EPA
standard of 120 ppb (240 /ig/m3) was surpassed on 13 days during the summer of
1982. During such episodes, when easterly winds prevail, the 03 levels are
usually high over all of Western Europe. Maximum values have reached
approximately 270 ppb (540 /ig/ra3).
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
31
-------
In The Netherlands, an infrastructure has been developed to provide
information for the formulation of air pollution abatement policies. This
structure consists of monitoring data, emissions inventories, research programs,
modeling, and the Air Quality Monitoring System (AQMS), which plays a central
role in integrating and directing the use of these different types of
information.
The Netherlands and surrounding countries have a rather dense air quality
measuring network. In The Netherlands, an area of 34,000 km2, hourly
concentrations are available for S02 (200 stations), NO and N02 (92 stations),
03 (30 stations), and CO (41 stations). This information is supplemented with
measurments taken at selected sites, particularly measurements on the
determination of compounds of interest to photochemistry. It includes the
analysis of up to 200 different chemical species and the physical and chemical
structure of aerosols. In addition to ground-level measurements, aircraft
measurements are conducted. Flowers and crops are also used as biological
indicators of air pollution. Meteorological information, such as wind speed,
wind direction, and temperature, is generated by over 60 stations. Higher level
information can be obtained from a meteorological mast (200 m) and several other
masts. A detailed emissions inventory on a 1 km x 1 km scale is available.
Information from this inventory has been processed for model applications,
including photochemistry.
To indicate the photochemical patterns of ambient air quality in The .
Netherlands, Figure 1 shows the 03 and N02 pattern for a 20-yr period. Figure 2
gives general trends for significant air pollutants in The Netherlands.
32
-------
Although S02 levels have decreased during the last two decades, there has been
no decrease in photochemically related pollutants. The concentration of PAN,
for example, has increased significantly.
DESCRIPTION OF THE DUTCH AIR QUALITY MANAGEMENT SYSTEM
In 1978, the Ministry of Housing, Physical Planning and Environment
instructed The Netherlands Organization for Applied Scientific Research (TNO) to
construct the Air Quality Management System (AQMS) in an attempt to quantify the
diffuse information that eventually leads to policy decisions. The system
contains a socio-economic module (based on scenarios for the future), a
transmission module, and a module containing consequences accessible to
assessment. The interrelations between the modules are of great importance.
For AQMS, transport and dispersion models are important tools, but the AQMS also
compels modelers to adapt their models to policy purposes. Thus, models should
be practical and should not aim at achieving more detail and accuracy than that
in accordance with the overall.results of the system.
The AQMS for NOX was recently completed. It contains models for annual
averages and higher percentages of N02, an episodic SAI airshed model,
semi-empirical relations to determine the consequences of NOX emissions with
respect to photochemistry, traffic emission dispersion models, and models for
determining wet and dry deposition.
33
-------
arts- is
25 JUL 1*80 13H
Figure 1. Measured concentrations of S02, N02, 03 (/jg/m3), and oxidants
N02 + 03 (ppb) for July 25, 1980, 1500 h (Automated Air Quality
Monitoring System, Dutch National Institute of Public Health). The
arrows indicate windspeed in meters per second.
34
-------
so,
ug/m3
t
300
2OD- 20
100- -10
NO, 03
uj/ms ug/m3
I960
1965
1970
1975
I960
Figure 2. Trend of yearly averaged concentration levels in The Netherlands.
S04* and N03~ measured in rain water.
Table 1 indicates how the AQMS integrates modeling results and presents
them in a manner compatible with policymaking. For a more complete description
of the AQMS, see Zwerver (1982) and Bovenkerk et al. (1982).
AIR QUALITY POLICY REQUIREMENTS
The relative position of models in the field of air pollution research may
be indicated by the amount of money spent on a model's development and
application on the one hand and its infrastructure on the other. Table 2 gives
a rough estimate of the money spent by the Ministry over the last 10 yr.
Table 2 suggests that the development and use of models is relativley
cheap. However, models require information infrastructures. Although the
development of complex models may be relatively cheap, the application costs may
be considerable, a factor arguing for relatively simple and practical models.
35
-------
TABLE 1. THE NO.-AQMS AND ITS RESULTS
Environmental Issue
Standard Violation
Minimal Required
Improvement
(7.)
Minimal Required
NO. reduction
Exposure of humans to
NO concentrations
No violation
Exposure of humans to
NO, concentrations
Close to process
industry
In cities
At street level
In cars
At home
50
5
25-35
30-35
up to 80
Optimal
Dutch traffic, 10
Dutch traffic, 50-70
Dutch traffic, 50-70
regulations for
burners, etc.
Exposure of human
HN03 concentration
Visibility
No violation
No standard
Improvement, 50-80
20-30
Europe & Netherlands
100, 50
Exposure of human
to Os concentrations
In over half
the country,
several days
year
30
Europe & Netherlands
N0y, more than 80;
HC, more than 40
Exposure of humans to
PAN and other
photochemical
products
Exposure of cultivated
plants to N02
No standard
In more than half
the country
50-80
Europe & Netherlands,
80
Exposure of cultivated
plants to 03
In more than half
the country for
several days a
year
40
Europe & Netherlands,
NO,, 80; HC, 40
Exposure of cultivated
plants to PAN
Exposure of natural
vegetation to NOj
and O3
No standard
No standard
available, but
more severe than
on cultivated
plants for
sensible locations
30-40
Europe & Netherlands,
50-80
Exposure of flora,
fauna, and eco-
systems to
acidification
75
Europe & Netherlands.
NOX + S02 +• NH3; 75
Exposure to eutrifi-
cation
No standard
Exposure of materials
to N02, 03, (SO,) in
musea, etc.
Yes
Does not apply
Exposure of materials
to N02, nitrates,
S02, sulphates
Climatological changes
No standard
No standard
To avoid anv damage
Europe & Netherlands,
NO, 70, S02 50
36
-------
TABLE 2. ESTIMATED BREAKDOWN OF MONEY SPENT BY
MINISTRY OF HOUSING, PHYSICAL PLANNING
AND ENVIRONMENT
Expendi ture
Project (million Dfl.)
Emissions inventory 50
Monitoring network 130
Research on photochemistry 15
Model development and application 15
AQMS structure 7
Requirements Related to Resolution and Degree of Discrimination
Questions regarding resolution in time and space, the degree of
discrimination, the abatement of traffic and industry emissions, and the
efficiency of NOX and/or HC abatement arise in model applications. At the
moment, policymakers need rather simple answers to these questions, although the
questions themselves probably cannot be answered easily. The problems that
arise are: What are the essential details and how sophisticated should the
model be in order to determine which compounds should be abated, what spatal
scale of abatement is required, and what are the quantitative results.
Ozone formation is a large-scale phenomenon, and 03 concentrations show a
rather flat pattern. So, model results should indicate the minimum spatial
dimensions of the areas to which abatement should be applied and the compounds
that should be abated (HC, NOX, or both). The fact that 03 formation is not a
37
-------
linear process implies that several abatement strategies should be separately
calculated in order find the most efficient one. Then, the problem of 03
variance and abatement on a smaller scale becomes apparent, and large-scale
models can be used to define the boundary conditions for smaller areas
(including individual plumes). As stated previously, the oxidant problem is
just one of many air quality problems facing The Netherlands. In Table 3, the
authors give their personal views of present environmental priorities.
APPLICATION OF PHOTOCHEMICAL DISPERSION MODELS IN THE NETHERLANDS
History
Photochemical phenomena have been investigated in The Netherlands since
approximately 1970. This field of investigation was soon extended from the
development of measuring devices. In addition to conducting smog chamber
experiments and making detailed analyses of the phenomena, researchers developed
and applied models to practical situations. First, the EKMA approach and box
models were used; later, the SAI airshed model was used. Much more detailed
field experiments were also carried out, including the use of airplanes and the
measurement of over a hundred different HCs. Table 4 gives an overview of the
historical development of photochemical dispersion models (see also Guicherit et
al., 1978).
38
-------
TABLE 3. PRIORITY AND MODEL PERFORMANCE RELATED TO ENVIRONMENTAL ASPECTS AND ESTIMATED NEEDS
FOR ABATEMENT*
Environmental Aspects,
Concentrations, Phenomena,
and Subjects
O3 concentrations
Humans
Crops
Natural vegetation
N02 concentrations
Humans
Crops
Natural vegetation
Humans in houses, cars,
and traffic
Acidification
Eutrification
Dying forests
Secondary products
(PAN, HN03, aerosols
Spatial
Europe
Europe
Europe
Mesoscale
Mesoscale
Mesoscale
Local and
microscale
Europe
National scale
(NHj)
Mesoscale
Scale and
Resolution Time
Episodic (days)
Episodic (days)
Episodic (days)
98th percentile
Hourly based
Hourly based
Hourly peaks
Long-term0
Long-term
Long-term
98th percentile
Hourly based'
Grid
~20 x 20 km2
*20 x 20 km2
~20 x 20 km2
1-400 km2
1-400 km2
1-400 km2
Population
exposure
statistics
Depends on land
use, 400 km2 up
to the size of
country
1-400 km2
Priority
1-2
3
3 (1 if 03 is
main cause
cause of dying
forests)
2-3
3
3
1-2 (1 if 03
is also
strongly
involved;
subject of
Dutch
inhalation
toxicology
research
1
1
1
1-3?
carcinogenics,
aldehydes, etc.)
Visibility
Climate
Materials
Mesoscale
From conti-
nental to
microscale
Europe/
mesoscale
Episodic
Long-term,
episodic
Long-term
Vertical 3
columns.
countries
1-3?
400 km2. 3
countries
'See Table 1 for preliminary estimates of necessary abatement.
"Outside The Netherlands (e.g., Scandinavia), the problem has a more or less episodic character
(episodic rains). In The Netherlands, dry deposition contributes most in the long term.
'For deposition, long-term and 400 km2 up to size of country.
39
-------
TABLE 4. HISTORICAL OVERVIEW OF DUTCH RESEARCH ON PHOTOCHEMISTRY
Year
Research Activity
Major Emphasis
of Activity
1970 Development of monitoring methods and
instruments (03, NOX, HC, UV, PAN)
1975 Monitoring of mainly gaseous components
Natural 03 and PAN levels
Trend analysis
Models (Box model, EKMA variant)
Smog chamber research
Rough abatement scenarios
1978 Integrated research programs:
Monitoring & physical characterization
of aerosols, gaseous components
Smog chamber research of aromatics
"Aged smog" (sampled at 200 m)
Conversion of S02 interacting with
aerosols
Monitoring method for OH radicals
(failed)
Reactive NMHC detector
Preparation for and application of
models (EKMA and SAI airshed models)
1982 Mesoscale and large-scale models
Plume modeling
Monitoring equipment
Intrusion of 03
Urban scale
Ratio of contributions
of traffic and industry
HC/NOX abatement
Contribution of natural
emissions
Relative importance of
the pollutant inflow
Traffic/industry
HC/NOX abatement
40
-------
The EKMA Approach
An EKMA-type approach, suited especially to the Dutch situation and
developed by Guicherit (1978), has been used to gain an initial insight into
photochemical processes. The EKMA approach has also been used in the framework
of a large project, aimed at the experimental determination of oxidant levels in
The Netherlands. Measurements were performed at a 200-m-high meteorological
mast, and several flights were conducted. The EKMA approach is used to
determine, in a qualitative manner, the influence of Dutch sources on 03
formation. The results indicate that this influence is relatively small.
Combining the results of aircraft measurements (see Figure 3) and EKMA
calculations led to rough estimates of downwind contributions to the 03 levels
(see Table 5) of typical emitters. At the moment, an EKMA-type box model is
used in a trajectory mode in combination with the SAI airshed model. A few runs
are being conducted with the airshed model, supplemented with box model
trajectory calculations, in which both EKMA and other chemical mechanisms are
incorporated.
The Oxidant Approach
An analysis of NO, N02, and 03 data from the Dutch National Monitoring
Network showed the limiting influence of Ox (= N02 + 03) levels on N02
formation, indicating a possible important inflow of Ox, 03, or 03 precursors in
our area. These oxidant levels were mainly determined by 03 inflow across the
Dutch borders. In favor of the Dutch Environmental Council, which is preparing
its comments on the N02 standard proposed by the Dutch Council for Public
41
-------
120.1*0ppb
WO.ttOppb
M.IMppb
60. Mppb
Figure 3. Example of measured 03 profiles at 650-m
height above Rijnmond area.
TABLE 5. DOWNWIND CONTRIBUTIONS TO 03 LEVEL
Emitter/Pollution Source
Approximate Increase of 03
Downwind from Source (ppb)
City (100,000 inhabitants)
Rijnmond (Rotterdam industrial area)
Southern part of The Netherlands
Transboundary inflow of precursors
Transboundary inflow of 03
1
20
10
20
80
Health, Van Egmond et al. (1982) developed and applied a simple oxidant model to
establish a simple, empirical relation between the probability of exceeding the
N02 standard and the NOX concentration levels (see Figure 4). This simplified
42
-------
NOxconcMtratlon
(ppb) 160
140
120
100
80
60
40
20
I0<0»
' «27ppb
20 <0,
< 37ppb
Figure 4.
0 8 162432404856647280
— N02 conetntrttion (ppb)
Limitation of the N02 level (limit depends on the
actual NOX and oxidant levels).
relationship, based on the photostationary equilibrium of NO, N02, and Oa, has
been applied in a three-layer numerical grid model (Van Egmond, 1982(b)).
The model, incorporating a very simple chemical scheme, accurately
described the Ox and N02 situation in The Netherlands, once again indicating the
large inflow of oxidants into the Dutch area.
Nonepisodic Modeling of N02 Concentrations
Although not directly related to 03 levels, the influence of the inflow of
primary and secondary pollutants on. a long-term basis is nevertheless worth
mentioning. In terms of N02 frequency distributions and estimates of yearly
nitrate depositions in Dutch territory, van den Hout et al. (1983) applied a
Gaussian-type model combined with wind-direction-dependent empirical factors for
the N02/NOX conversion. This model, combined with other approaches (the oxidant
model and a preliminary microscale model), has led to the results presented in
43
-------
Table 2. These results also demonstrate the importance of inflow from outside
The Netherlands. Similar considerations apply to the deposition of nitrates.
About 70% of this deposition originates from across the border and contributes
20% to the acidification of the Dutch soil. Vice versa, Dutch NOX emissions are
deposited in other countries.
Application of the SAI Airshed Model
SAI airshed model calculations were carried out for an episode of
relatively high oxidant levels (June 7-8, 1976). Runs to test the model and
establish its sensitivity to the assumed boundary conditions were followed by
runs for application purposes. In these runs, several abatement reference cases
involving traffic and industry NOX and HC emissions were simulated.
Figure 5 shows the region covered by the model (310 km x 230 km). In this
region, three large industrial areas are situated at distances from each other
of 100 m up to 300 km. These are the Rijnmond (Rotterdam harbor) area in the
Western part of The Netherlands, the Antwerp area in the Western part of
Belgium, and the Ruhr area in the Western part of Germany. In addition to the
industrial and urban areas, the region covered by the model includes rural areas
with low NOX and HC emission densities.
The well-known SAI airshed model (Reynolds, 1979) was originally developed
for urban-scale photochemical episodes. For this application, the model has
been "stretched out" in space to regional scale. Hourly concentrations of
several air pollutants, 03, PAN, NO, N02, and four HC classes have been
44
-------
-310km
Figure 5. The modeling area.
calculated for 10-km grid distances; five levels have been used in the vertical
direction.
Test Runs—
The amount of pollutant inflow into the area is uncertain. For this
reason, boundary conditions were changed in order to test the sensitivity of the
model and to check how realistic the assumed boundary conditions were.
The originally selected conditions gave the best overall agreement with the
monitoring data. Generally, the results of the calculations fit and correlated
quite well with the measurement data. For instance, a calculated maximum 03
value of 244 jjg/ra3, averaged over 11 monitoring stations, compares very well
with a measurement of 259 >ig/m3. The temporal correlation coefficient for
monitoring stations with 03 values higher than 40 /jg/m3 was 0.91; the spatial
correlation of the maxima was 0.62.
45
-------
Figure 6 shows observed and calculated measurements (03, NO and N02) for a
monitoring site in the industrial Rijnmond area (for further information see
Builtjes et al., 1980, 1982).
Application of the Model in the Establishment of an NOX Policy—
The main issue facing Dutch policymakers attempting to counteract the
effect of pollutants with a long lifetime is: How effective is domestic
abatement in a small country like ours in comparison with abatement measures
taken abroad? How do the main sources (traffic and industry) contribute to
environmental pollution inside and outside the country? The answers to these
questions are complicated by the nonlinear behavior of photochemical
transformations; consequently, it is not possible to express the pure and
independent contribution of one source to the concentrations of 03 and N02
without considering the emissions from other sources.
n 200
0,
uj/m3
100
NO
0
100
NOj
tig/I*3
t
0
100
90
24 4 8 \Z 16 20
— tlini (hour»)
Figure 6. Concentrations at Viaardingen. Dashed line
represents observed concentrations; solid
line represents calculated concentrations.
46
-------
In order to gain an insight into the effectiveness of emission reduction,
two sets of runs were carried out, directed to source categories and to species.
The results of the first set are shown in Figure 7 and refer to the influence of
a 100% reduction of NOX and HC emissions from traffic and industry. The 03
concentrations were nearly unaffected in all cases. Even with a 100% reduction
of all emissions in the area, 85% of the 03 still remains; the efficiency with
respect to N02 concentrations is also relatively low, but a 100% reduction of
NOX and HC emissions finally drops the N02 concentrations to a a low level.
The results of the second set of runs, shown in Table 6, show separate and
combined reductions of NOX and HC emissions in the considered area of 0%, 40%,
and 100%. It is apparent that a 40% reduction of the HC emission yields a
decrease in the maximum 03 emission by 2%. This effect is small, partly because
the flux of background HC across the boundaries into the model region is much
100
90
80
70
60
50
40
30
20
10
0
03 NO,
NO
all Mission*
no Dutch industry
no Dutch local traffic
no traffic wholo ar«a
i
!
no industry wholo aroa
i wholo aroa
Figure 7. Average reduction of maximum hourly
03, N02, and NO concentrations.
47
-------
TABLE 6. EFFECT OF EMISSION REDUCTIONS
ON THE CALCULATED MAXIMUM 03a
Run
A
B
C
D
E
F
Emission Reduction
in Model Region
(Species) (%)
None
HC 40
HC 100
NOX 40
NOX 100
HC and NOX 100
Mnximum 0-,
(,,g/m3)
275
269
250
274
205
206
"IIC emission is 105,000 kg/h; NO*
emission is 195,000 kg/h.
the flux of background HC across the boundaries into the model region is much
larger than the HC emission inside the region. For NOX, on the other hand, the
background flux across the boundaries is smaller than the emission inside the
region. Consequently, for a larger area the effect of a 40% NOX emission
reduction on 03 concentrations will be much smaller than that of a 40% HC
emission reduction, a fact that argues for HC control.
An important result of these model calculations is that a reduction in
emissions in the model region has only a limited effect on the 03 maxima. The
influence of the pollutants from sources upwind of the region is much more
important than that of local sources. The estimates of the upwind
concentrations resulting from remote sources are very uncertain, however. The
HC/NOX ratio in this background air, which is important for 03 formation
48
-------
efficiency, can be much larger than the HC/NOX emission ratio of the model the
region. For the development of efficient control strategies for situations like
the one discussed here, it is therefore necessary to perform calculations on a
spatial scale that is considerably larger than thn 300-km scale used in these
calculations (see also Van den Hout et al., 1982).
In view of the above results, it should be clear that The Netherlands, to
develop abatement strategies in their own country and to stimulate international
actions, would welcome research on and clarification of 03 and Ox (= 03 + N02)
formation on a large scale, including the use and/or development of large-scale
models.
One problem needing careful consideration is the relative efficiency of NOX
and HC abatement, in particular the emissions produced by motor vehicles. At
the moment, discussions are going on in the European Community about a further
reduction of CO and HC traffic emissions. The discussions include a detailed
consideration of matters like the European test cycle. However, the definition
of this cycle, the resulting traffic regulations, and future developments in the
construction of motor cars will determine the admissible levels of HC and NOX
car emissions and the ratio between the two in the coming 20 yr. So, a simple
or complicated model that can discriminate between abatement efficiences for NOX
and HC on a European scale would fit the European needs very well.
The tight time schedule for the above-mentioned activities favors the use
of directly applicable models instead of waiting for more complicated models
that need further development and testing or collecting detailed input data.
49
-------
In conclusion, it is still unclear whether an unambiguous solution can be
found for such phenomena as the ones discussed here. Different chemical
submodels lead to different estimates of what would be the most favorable
solution for the Dutch situation in so far as substantial NOX and/or HC emission
reduction on a European scale is considered. The margin of uncertainty seems to
be too wide for reliable interpretation. Table 7 demonstrates this uncertainty.
It gives estimates from several Dutch experts, based on their model calculations
and experience with respect to the possible influence of NOX and HC emission
reduction for 03 levels on a European scale. These are very speculative
estimates, given here only to illustrate the dilemma and the need for
large-scale models.
CONCLUSIONS AND REMARKS
The application of models in The Netherlands indicates that 03 formation is
due to large-scale phenomena caused by NOX and HC emissions on a European scale.
In order to be effective, abatement strategies should be developed with the
continental scale in mind. Large-scale models applied to a substantial part of
Western Europe, at least on a scale of a 1,000 km x 1000 km, may contribute to
the establishment of these abatement strategies. One major problem is the
effectiveness of NOX and/or HC control on the European scale. Currently used
chemical submodels give different answers to this problem. Clarification of
this point could also contribute to the discussions being held in the European
Community's working group, ERGA, with respect to the abatement of traffic
emissions. In addition to Oa, other compounds and phenomena must be considered
50
-------
TABLE 7. ESTIMATED 03 REDUCTION ON A EUROPEAN
SCALE AS A CONSEQUENCE OF HC AND/OR NOX
EMISSION REDUCTION
NOX Reduction HC Reduction 03 Reduction
30
60
90
0
0
0
30
60
60
0
0
0
15
30
60
15
30
60
0
15
50
10?
25?
50?
10-15??
20-40??
20-60??
on a mesoscale and a large scale, especially the problems related to
acidification.
ACKNOWLEDGMENTS
This paper gives an overview of photochemistry work that was conducted by
experts in the field, to whom the authors are greatly indebted.
51
-------
BIBLIOGRAPHY
Bovenkerk, M., P.J.H. Builtjes, and S. Zwerver. 1982. An Air Quality
Management System as a Tool for Establishing an S02 and NOX Policy. Report
No. 82-013644, MT-TNO, The Netherlands.
Builtjes, P.J.H., K. D. van den Hout, and S. D. Reynolds. 1982. Evaluation of
the Performance of a Photochemical Dispersion Model in Practical
Applications. Thirteenth International Technical Meeting on Air Pollution
and Its Application, lie des Embiez, France.
Builtjes, P.J.H., et al. 1980. Application of a Photochemical Dispersion Model
to The Netherlands and Its Surroundings. Eleventh International Technical
Meeting on Air Pollution and Its Application. Amsterdam, The Netherlands.
Government Publishing Office. In press. Handbook of Emission Factors. Part 1,
Industrial Sources. The Hague.
Government Publishing Office. In press. Handbook of Emission Factors. Part 2,
Nonindustrial Sources. The Hague.
Guicherit, R., editor. 1978. Photochemical Smog Formation in The Netherlands.
TNO, The Hague.
Reynolds, S. D., et al. 1979. An Introduction to the SAI Airshed Model and Its
Usage. SAI Report EF 48-53 R, EF 79-31, SAI.
Schneider, T., and L. Grant, editors. 1982. Air Pollution by Nitrogen Oxides.
Proceedings of the US-Dutch International Symposium, Maastricht, The
Netherlands, May 24-28, 1982. In: Studies in Environmental Science 21.
New York: Elsevier Scientific Publishing Company.
Van Egmond, N. D., and H. Kesseboom. 1982(a). Modeling of Mesoscale Transport
of NOX and N02; Concentration Levels and Source Contributions. In: Air
Pollution by Nitrogen Oxides. Elsevier, Amsterdam.
Van Egmond, N. D., H. Kesseboom, and R. M. van Aalst. 1982(b). Relationships
Between N02, NO and 03 Levels in the Field; the Determination of an NOX
Standard. Report 227905050, RIV, in Dutch.
Van den Hout, K. D., et al. In preparation. N02 Concentrations in The
Netherlands. IMG-TNO, The Netherlands.
Van den Hout, K. D. and P.J.H. Builtjes. 1982. Dutch Contribution to the OECD
Study on the Development of Photochemical Oxidants Control Strategies
Within an Urban Airshed. Report no. 844, TNO Research Institute for
Environmental Hygiene.
Zwerver, S. An Air Quality Management System as a Tool for Establishing a NOX
Policy. In: Air Pollution by Nitrogen Oxides. Elsevier, Amsterdam.
52
-------
DISCUSSION
J. Shreffler; There is a very strong modeling and measurement program in The
Netherlands, as seen from this presentation. One of the major isues to be
addressed at the conference should be: If we go to regional modeling, who will
do it an how will it be done to put together a package of emissions data,
aerometric data, for the large number of European countries? Among the
countries, are the measurement equipment and procedures the same? In other
words, how do we get a compatible data base for a large number of countries?
53
-------
APPENDIX A. GUIDELINES FOR EMISSIONS INVENTORY PRESENTATIONS
Data base name/source
Reasons for inventory development?
Who collects the raw data? (private
industry, national/provincial
government, etc.)
How is raw data collected?
(questionnaire, permit system,
inspection, other)
How frequent are data updated?
Are updates legally required?
List legal or confidentiality restrictions
which may prevent release of the data
Area of coverage
Coordinate system
Point source information; define a point
source
A) Raw data collected:
List stack information
List major contributing source
categories (industries)
List types of raw data collected and
temporal resolution where
appropriate
Spatial resolution
Dates of available data
B) Emission estimates:
List pollutant species
Temporal resolution of calculated
emissions
List information available for
temporal apportionment
List percentage of emissions
estimated by following methods:
Standard emission factors with
specific plant information
Nonstandard emission factors
with specific plant information
Source test
Material balance
Other, specify
What emission factors, if any, arc
used?
List publication describing emission
factor development program
General Emission Inventory in The Netherlands
Air and water quality management
Dutch Organization for Applied Research (TNO) by
Order of the Ministry of Pub. Health and Env.
Protection and the Ministry of Traffic and Public
Works
Large industries; inspection
Small industries: questionnaire exc. those of
significant env. concern; also inspection
Combustion sources >20 MWh, yearly; all other
sources, once in 3 yr.
No
Emis. data from private industry are confidential;
limited clustered data may be published
Netherlands (ca. 40,000 km2)
Dutch topographical map
Vertical stacks and chimneys
Locn. (10m). height (m), cross sect, area (mm2),
name
Food, paper, refining, chems., bldg. mat 1., prim.
Metal, metal products, thermal generation, coke.
other industry
See Appendix C.
10 m
Previous vr. with regard to vr. of registration
1500 subtances (air and water) (Appendix B)
See Appendix C.
Information obtained from plant officials
^v
15
63
18
4 J
^.process enissions onlv
See Handbook of Emission Factors. Part 2,
Industrial Sources (In press).
54
-------
Are reported emission controlled
or uncontrolled?
Are control equipment and efficiency
information available?
Describe method of estimating
volatile organic compound emissions
Area source information; define an area
source
A) Raw data collected
List major contributing source
categories
List subclasses of stationary area
mobile sources
List types of raw data collected,
spatial and temporal resolution
where appropriate
Dates of available data
B) Emission estimates:
List pollutant species
Temporal resolution of calculated
emissions
List information available for
temporal apportionment
Describe grid system or spatial
resolution
List information available for
spatial approtionment
Are published standard procedures
used for temporal and spatial
allocation and emission
calculations?
If yes, list major references
Describe method of estimating
volatile organic compound
Both
yes
See Handbook of Emission Factors, Part 1,
Nonindustrial sources.
Bldgs. with openings; open areas (e.g.. storage
of petroleum liquids; chem. plant;
Ore and coal handling
See Appendix C.
Previous yr. with regard to yr. of registration
See Appendix B.
See Appendix C.
Information obtained from plant officials
Coordinates of mid-point of terrain
No
See Appendix B.
General emissions
Comment on completeness
As complete as possible within the scope of the
project
Comment on currentness
Summarize Quality Assurance Program
First round. 1974-1981
Second round. 1982-1984
Combustion sources >20 MWh, yearly
Forms (see Appendix C) are inspected. After
discussion with relevant registrator and approval,
a decoded printout is made and again inspected.
After final approval, decoded printout is sent
to plant officials who may comment within 3 mo.
55
-------
Who is responsible for data quality? TNO
Attach detailed record formats
Are source inventory data handled By computer
manually or by computer?
56
-------
APPENDIX B. LIST OF POLLUTANT SPECIES
2 "d roun cJ (i 9
ft-
round
CMIiSI - COMLOSIELUST PER MICUk't STOF
111 k«Ttf<>w»TCH01>'P
Idl 1 PC1HA«N
101S tTH«»N
1019 PROPAAN
1023 BUTAAN,U
1027 ISOOUTAAN
1031 PENTAStN
1035
1339 HEXAUCN
1013
1017 CTCLOHEXAAK
1JS1 KWST.,ALIFATISCH,>'£KGSEL,C2-C10
10C.O
100.0
10C.O
1CO.O
100.0
100.0
10C.O
100.0
1DC.C
100.0
100.0
10C.C
irs.c
100.0
ino.o
10C.O
100. c
loo.o
10C.Q
100.0
10C.O 1
10C.O
100. C
1CC.U
100.0
10U.O
100.0
100. 0
100.0
100. C
1CC.O
100.0
100.0
100.0
100.0
100.0
100.0
10C.C
100.0
inc. a
100. 0
100. a
100.0
loc. a
10C.O
133 16 UATER
139 SO HCTHAIN
110 SO CTHAAN
<4I>1 SS KOOLUtTERSTOFrtN C2, NNB
71 SO PROPAtU
SI SS HOOLy>TERS10FrCN CJ, NNB
56 SS KOOLJATCRS10FFCN C1, NNt
2<49 52 BUTtAN, N-
1S1 52 ISOBUTA1N
1U8 52 PENT/UN, N-
571 52 1SOPENTAAN 1 2-HETHVLBUTI AMI
711 52 KOOLU* TERSIOff [N, AL1FAT1SCH, C5
150 52 HCXAAN, tt-
73<4 55 KOOLJATERSTOFFEN C5-C6, llNB
1633 52 OIHCTHVLBUTAAN, 2,2-
679 52 HEPIAAN, N-
58 53 CrCLOHEXAN
52 SS OL1E, FRACI1E5 KOOKPUNT 100-200 C
131 99 PCTROLEUHETHER-CX7RAHECIIBAAR
135 55 KOOLyATERSIOFFENr XN8
181 55 KOOLUATERS10FFCN, VERBS. HOUI
191 55 NAFTA, KOOKTRAjECT 1C-90 C
193 55 NAFTA, KOOKTRAJECT «C-1B5 C
229 5S PCTROLEUHCTHER IKOOKP.EENZ. 1 90-120
231 55 PETROLEUHETHER fKOOKP.BENZ . ) 1M 5-1 61
3x0 55 PETROLEUHETHER IKOOKP.BEN2. 1 80-110
358 55 PETROLEUHETHER IKOOKP .BEN2 , 1 100-110
377 55 PETROLEUMS 1HE» IKOOKP .BEN2. 1 55-75
S13 SI OCTAAN, M-
550 52 KOOLUATERS10FFE", ALIF., VER2., >C3
583 55 PETROLCUHETHER IKOOKP. BEN? . 1 10-60
589 96 OPLOiHIDOELEN, ORGAN1SCH, NNB
678 55 OPLOSMIOOELEh, KOOLU ATERSTOFFEN, NNB
660 52 NONAAN, N-
712 52 KOOLUATERSTOFFLN, AL1FATISCH, C9
1081 55 PETROLCUMElHEft IKOOKP. BEN2. 1 60-95
1091 96 OPL05M100ELEN,ALlF/AROH/GECHL-KyST.
1191 SS PETROLEUMETHER IKOOKP .BEN? . 1 00-135
1198 55 KOOLWATERSTOFFCN, ALIFATISCH, NNB
1200 55 PCTDOLEUMClHCR IKOOKP .BEN? . 1 NNB
1312 55 KOOLWATERSIOFFEN, BER. «LS HETHAAN
13S2 55 KOOLWMERS10FFCN Cl t/H CS, NNB
1105 55 KOOLUATERSTOFFENiKOOKPUNT 110-1BO C
1715 97 KOOLUAIERS10FFEN, AU IF . VER2 . , TOT A»L
1781 55 KOOLdATERSTOFFEN ALIFATISCH
57
-------
APPENDIX C. EMISSIEREGISTRATIE
VertrouwelijR. Alleen voor pers^neel aang';--e:t;.n
volgens ERL-regels. Geen basis voor heffir.gen.
1- 5
6
7- 12
13- 42
43- 67
68- 92
93- 95
96- 98
99-100
101-103
104-105
106
107
108-112
•113-114
115-116
117-118
119
120-121
122-123
124-125
D
EMISSIEREGISTRATIE
Bedrijvenbestand Plant tt'/e
1. Recordcode C
2. Mutatiesoort (code 1)
3. BedrijfsQummer
4. Naara jh'Snr /i*mt
5. Vestigingsadres 3c/Jrti
6. Vestigiagsplaats /ou/n FT
7. Gemeente (code 2)
Ligging: hor. km
/i 10
ver. km
10
| ( ( |
I I I I
Q-J
[TTI
m
I i I I M I M I I M I
I I
Code
t>er
9. Toezicht lucht (code 3) Fl AJS ^ do u///A air
10. Toezicht water (code 4) Q
11. Aantal werknemers I I I
12. Aantal iastallaties:
Totaal • I ] ]
Te herregistreren I I I
Aantal grote vuurhaardeul I |
13. Registratiesoort(code
14. Basisjaar
15. Opname-instantie(code
16. Opname-persoon | | |
f
58
-------
APPENDIX C. continued
volgcns tRL-regels. Oien bisis voor beif-.i
1- 5
6
7- 12
13- 42
i3- 67
53- 73
74- 98
99-113
114-125
EMISSIEREG1STRATIE
Adressenbestand
1. Recordcode
2. ^utatiesooct (code 1)
3. Bedrijfsniunmer
5. Postadres
6. Postcode
7. Plaatsnaam/UAf<
8. Contactpersooa
9. Talefoonniimmer
59
-------
APPENDIX C. continued
1- 5
6
7-12
13-15
16-17
18-21
22-25
26-55
56-59
60
61
62-64
65-68
69-70
71-72
73-75
--76 -
77-79
80-81
82-84
85-86
87-88
89-90
91-92
Vert rcu
vclgens
EMISSIEREGISTRATIE
Installatiebestand JflSn^'i^f/O
1. Recordcode 3 9\0 9\2
2. Mutatiesoort (code 1) FJ
3. Bedn j f snummer
4. Installatienummer 1 1 1 1
5. Aantal samengenomen 1 1 I
6. Installatiesoort
(code 8)
7. Bedrijfstak (code 7)
8. Kaam /AT7&/& /r«/i mrnt
9. Ontwerpcapaciteit
mantisse
exponent 1 — 1
eenheid (code 10) U c
10. Bezettingsgraad (^) 1 — 1 — LJ /•
11. Bedrijfsuren per jaar e.
12. Bouwjaar • '[ | ] y
13. Verwachte technische
levensduur (ja'ren) [ | ] re
14. Produktiewerknemers
(aantal) | /)U
1-5. Inrichtingenbesluit -_i- - r
(code 29) LJ **
16. Ligging: hor. km I I I I /
10 n | | J
ver. km | || 1
10 m | f ]
17. Basisjaar. | | |
18. Opname-instantie 1 1 J
(code 6)
19. Opname-persoon ( [_]
k-elijk. All-en vocr ;n-rs-i. :.•••! i.-ngt^'-^tn
ERL-regels. Gcen basis voor heff inyen.
E
. fa L'
/*>$/*' for/on /?&/**&£.?*
f}tSfftb'£f or J/Sry/fa/* /s?Sik //37f6t9±
/t*G*utjry CJ/e*o fy C"o c/c
\ 1 II 1 1 1 1 1 i 1 !
arft/jstcj^acS/y
'tyaft'/y unit- £oc/e
/> ' it
?*r of £ reef tor)
fnG&r or Jif^C/ucAev>- GsnJb'oyet^
j to c/o tv/'/t, /sy/j/a/xeo
y
t t
60
-------
APPENDIX C. continued
Vertrouwelijk. Alleen voor ptrscnccl a£:ig
-------
APPENDIX C. continued
V
crtrcn-.tli jk. AjU-en v;.or ; cri • ;."tl •xr.^evczen
olger.s LRL-reg'-ls. Geen basis vcor hsffingen.
EMISSIEREGISTRATIE
-Source ft/e fe/r)
1- 5
6
7-12
13-15
16-17
18
19-38
29-41
42-43
44-46
47-48
49-51
52-54
55-57
58-61
62
63-64
65-66
67-68
Bronnenbestand lucht
_j
9\0\9\i
2. Mutatiesoort (code 1) l_|
3. Bedrijfsnununer
| 1 Jsfenf At//nt>er
4. Bronnumner 1 II J Jot/fCf Aut»&er
5. Aantal emissiepunten
6. Bronsoort (code 13)
7. Kaam Joufee. rtintt
fiUtnLer 0fjb/±eet> latest em/tu'ont
T «* / ^ ££&f
T T T 1
8. Ligging: hot. km MM
10 n 1 I 1 / /.
LJ-J /oca/ion
ver. kra MM
10 m Ml
9. Gemeente (code 2) Mil /tiunifi ' bd//'/y Coc/e
10. Geod. hoogte (m) +10 o | | 1 J
11. Geom. hoogte Mil n*t*At OtJourt£
12. Bronoppervlakte (HUB )
mantisse " ' |_
| Cross- Stc#t>n*/ <3re&
exponent | |
13. Basisjaar | | |
14. Opname-instantie (code 6) | | |
15. Opnane-persoon ' 1 1 1
62
-------
APPENDIX C. continued
V..-rtro..'-ulijk. Al!<--r. --cor ;•:: ..• '.-1 dir.ievec
voljcns ERL-regels. G<-en \ asis voor lieffingen.
EMISSIEREG1STRATIE
1- 5
6
7-12
13-15
16-19
20-22
23-24
25-28
29
30-33
34
35-36
37-38
39-42
43
44-46
47-48
49
50-51
52
53-56
57-58
59-62
Emissiebestand lucbt —
1 Rcrordcode 3 9\0\9 6
2. Mutatiesoort (code 1) [_J
f>'of»J ntJfn
1 — 1 ~1
4. Brormuinmer 1 1 1 1 -fot/rft />£ of 'jj
exponent 1 — 1
14. Temperatuur (°C) [ | j J r*f»/>e/ar'vrf
15. Frequentie: waarde |q| J numerirat ^
^^"'"^eeaheid (code 16) @ ^^
16. Tijdsduur : waarde | ^| | nur»er,c»/
^ ° p.... f ?re>
17. Tijdsopgave: maand AroA/4 C t/m \_ei '*" jj*'^.
dag ef*y g t/m gj , \'.
nur Xot| J ^,/t
f
t/e.
/4ft
63
-------
APPENDIX C. continued
63-70
71-82
82-S6
87-38
69-90
91-92
92-94
95-96
18
\
19
20
21
22
23
Gasreiniging: t
type (code 18) £ftti'/>">t»4
-------
APPENDIX C. continued
V-r t rTi-el i j k. ."• 11 •_ i n '•'- jr . .' . "i j.irgt:-e7"
.-' ' ,_/.-ns tRL-rtgels. Cc-n.basis . or heffingen.
1- 5
6
7-12
13-15
16
17-36
37-39
40-41
42-44
45-46
47-49
50-58
59-62
63-64
65-68
69
70-71
72-73
74-75
EM1SSIEREGISTRATIE
Bronnenbestand water
1. Recordcode
2. Mutatiesoort (code 1)
3. Bedrij fsnummer
4. Bronnummer
5. Bronsoort (code 21)
6. Naara
7. Ligging: hor. km
10 m
ver. km
10 m
8. Gemeente (code 2)
9. Bestemraing
-S
ource.
D
| | | J
cm
m
un
m
Source
code
/o cat
on
/Y> c/ n/c/'/y A
Coe/e
coc/e.
10. Ont.oppervl.water (code 26) | | | | ) Cac/t for rccet'vinj ft,'s£t«
11. Waterkwaliteitsbeheerder
(code 22) | | J /?<9J ^ C/o A///4 K*/er ^uttSJy
12. Volumestroon (m /jaar)
mantissa [ | | | ~] £lhot/nt or
exponent f_J
13. Basisjaar | | ]
14. Opname-instantie (code 6) | | |
15. Opname-persoon | | |
' Code (not- c^ie^ory code)
- hyc/ra/oj!e*/ c/t/Jnetf areA
- Soil
. oth
65
-------
APPENDIX C. continued
'.'•-1 Lroi;.L'l ijk. All'-cn Jj'-r p-r: :i -1 .. :..j;'---.-rer
volg-.'ns ERL-rcgels. CV-rn b.^sis voor heff.ngen.
EMISSIEREGISTRATIE
///c
1- 5
6
7-12
13-15
16-19
20-22
23-24
25-28
29
30-33
34
35-37
3S-39
40-41
42-45
46
47-48
49
50-51
52
53-56
57-58
59-62
63-70
71-73
74-75
76-77
73-79
80-81
S2-83
1. Recordcode -'rl" ° °
2. Mutatiesoort (code 1) [J
3. Bcdrij fsnuiwner
4. Bronnumner 1 1 1 1
5. Volgnumraer (niet invullen)
6. Installatienummer 1 '1 1 1
7. Arparaatnummer | | [
8. Stofniuumer . | | ] | |
9. Emissievorra (code 23) | )
10. Massa- of warmtestroom
(.ng/h of Watt)
mantisse J
exponent j~[
11. Herleiding tot peiljaar(%) | | ) |
12. Emissiemodus (code 15) I 1 1
13. Capaciteit (10%) | p)
14. Volumestroom (1/h)
maatisse
exponent (J
15 . Frequentie : waarde [ j | Hum
" eenieid (code 16) Q fotft
16. Tijdsduur : waarde nun-
eenheid (code 17) 1 fat^e.
11. iijdsopgave: maand finonSt) I Y~\ t/
dag c/*y QJ t/
uur /tour \ \ \ to
18. Waterzuivering: jbo/funor) con/fto/
type (code 24) egv/J>'r>r»f f**
rendement (0, 1%) tfl/tfeoey |~[~] []
19. Oorzaak wijziging (code 19) 1 I j
20. Soort bepaling (code 20) I, I_J Cc
21. B.isisjaar
22. Opoame-instantie (code 6) | | [
23. Opname-persoon [ 1 J
]
hfehl humier
•fourct nt/mbcS"
e3/3f£ /y Jut fttSSn k eS~
CotJe of StsAf/znce cm/Jfe
Coe/e of sAi/>t oS1 &m/'rr/oi
<3moun£ of f/ntJS/or)
firoc/esc/t'ory /eve./ *
cbfYicxjnl or l^t£S/c wl/cr Con
fr.fi/
. /r^soxt/te/v//^
ra m
^ m -
_j /> . j j. t ,
><*€ /or rCG/sfr^non •fccf>n/<»es<. •
v /*
resse. of fmi'ti/on
< e/f.
//£?/ otft
66
-------
U.S. EPA REGIONAL OXIDANT MODEL FOR THE TRANSPORT OF
PHOTOCHEMICAL OXIDANTS AND THEIR PRECURSORS*
Robert G. Lambi and Joan H. Novak)
Meteorology and Assessment Division
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
STRUCTURE OF THE MODEL AND ITS INPUT DATA PROCESSOR NETWORK
The U. S. Environmental Protection Agency (EPA) Regional Oxidant Model
(ROM) is designed to simulate hourly averaged concentrations of photochemical
pollutants over periods of several days on a three-dimensional spatial grid that
is 103 km in size, with a horizontal resolution of about 18 km x 18 km. This
model is intended to assist the states in formulating emission control plans
that will bring air quality into compliance with Federal standards. To provide
reliable service in this capacity, the model was structured to consider all of
the chemical and physical processes that are known, or presently thought, to
affect air pollutant concentrations over several-day/I,000-km-scale domains.
Among the processes included are:
• Horizontal transport;
• Photochemistry, including the very slow reactions;
*This paper has been reviewed by the Environmental Sciences Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Mention of
trade names or commercial products does not constitute endorsement or
recommendation for use.
iOn assignment from the National Oceanic and Atmospheric Administration, U.S.
Department of Commerce.
67
-------
• Nighttime chemistry of the products and precursors of photochemical
reactions;
• Nighttime wind shear, stability stratification, and turbulence
"episodes" associated with the nocturnal jet;
• Cumulus cloud effects that vent pollutants from the mixed layer, perturb
photochemical reaction rates in their shadows, provide sites for liquid-
phase reactions, influence changes in the mixed-layer depth, and
perturbhorizontal flow;
• Mesoscale vertical motion induced by terrain and horizontal divergence
of the large-scale flow;
• Terrain effects on horizontal flows, removal, diffusion;
• Subgrid-scale chemistry processes resulting from emissions from sources
smaller than the model's grid can resolve;
• Natural sources of HCs, NOX, and stratospheric 03; and
• Wet and dry removal processes (e.g., washout and deposition).
Based on analyses of aircraft 03 and NOX measurements and of meteorological
variables over the Northeastern United States, it was concluded that
incorporating all of the processes listed above would require, at the very
least, a three-level model. One level would be assigned to the surface layer, a
second level to the remainder of the daytime mixed layer, and a third to the
layer atop the mixed layer where convective clouds are often present. However,
in order for only three levels to suffice, each would have to be able to expand
and contract locally in response to changes in tho meteorological phenomena that
each layer was intended to simulate.
The EPA regional model is designed in such a way. In addition to three
layers of variable thickness, the model possesses a shallow surface layer,
called Layer 0, which is adjacent to the ground. This layer handles surface
deposition and subgrid-scale chemistry phenomena, and the concentration values
68
-------
computed in this layer represent ground-level conditions. The model is shown in
Figures 1 and 2. A more detailed specification is provided in the appendix and
a complete description is available in Lamb (In press, 1982).
A model as comprehensive as this one is necessarily complicated and a
rather sizeable team of people was needed to develop, test, and operate it.
Moreover, numerous mathematical descriptions have been proposed for the physical
and chemical processes cited earlier that affect air pollutant concentrations,
and new and improved descriptions are continually being developed. In view of
all these considerations, the EPA model was structured in a highly modular form
to aid the division of labor in the model development and maintenance, to
simplify the task of trouble-shooting, and to facilitate the interchange of
existing and future methods of describing the various physical and chemical
processes that the model represents.
The overall structure of the model is illustrated in Figure 3. The box
labeled CORE represents the computer analog of the differential, equations
describing the governing processes. This analog is in a very primitive
mathematical form in the sense that its inputs are matrices and vectors whose
elements are composites of meteorological parameters, chemical rate constants,
etc. For example, the link between CORE and the output of the module labeled
CHEM, which contains the analog of the chemical kinetics scheme, consists of two
vectors, P and Q, each of length N, where N is the total number of chemical
species simulated. The n-th element of P is the net production rate of
species n due to its chemical interaction with all other species, and the n-th
element of Q is the net rate of destruction of species n due to its chemical
69
-------
I
4=
(A U
oi
co c
i-l 0)
B
r o
u c
•H a>
S x
m a,
•O B
: -H
j->
01 >>
j= «
H -a
a>
M
3
00
•H
70
-------
I
u
c
3
U.
I
I
CO
0>
Ol l-i
C
03 -a
o>
JC -O
O
Ul
HI OJ
U ^
3 N 0>
03 S
.-I O
C
r 01
a f.
•H a
s
0] 01
C B
O) OO
X -H
H C
01
M
3
00
•H
tu
71
-------
4
LJ
O
/
^ /
^
V
/^
^
LJ
I
O
to TU
0, 3
3 O
O .-I
>-l U
60
0>
X 4J
•a
O)
G -H
O 3
•H cr
4J 0)
4J 01
0) CO I-I CO
« w o 4->
0) V4J flj
u - e ts
v w a*
M ^-^ ^^ jC
O 4J
10 10
O) C (0 0)
r-l O tt) tJ
O -H O -H
l-l M M >
•rl W 3 O
U fl O M
ana.
a>
^
«
0)
U (0
0) C <
U O v^
o o
a>
AS •>
M -^
OX
S ^
u
AJ
a>
CM
CXi-l 4J AJ
cQ -H
(B y i-l -
U -r-i (Q to
CO 00 3 t-4
•o o cr a>
AJ O (-4 O
3 i-i -H a
a, o
M 3 n> o
4-1 0)
•H - •> a
***** *^v O
•O H i-l H
CO X^
>• 01 «-"
i-i jc n o
ai a. 3
Tl) CQ P^M
o M -a u
S oo a o
o « »
o> a. .-i 4-»
j; o a)
4J H - C
O •• ^^
(0
c 4J m
o m c
•H 73 O
O
u
N O
3: a>
a>
72
H O O S
o-i
0)
M
3
00
•H
tu
-------
interaction with nil other species. Thus, any chemical kinetics mechanism can
be incorporated into the model as long as it is expressed in a form that is
compatible with the vector interfaces that link CORE with the chemistry module
CHEM. The rate equations that describe the chemical reactions are handled in
their differential form in CORE; pseudo-steady-state approximations are not
used. In the current phase of the model development, the kinetics mechanism
developed by Demerjian and Schere (1979) is used, which includes some
35 reactions for 23 chemical species.
The remainder of the inputs required by CORE are prepared by the module
designated BMC (b-matrix compiler) in Figure 3, which performs essentially the
same task that language compilers perform in computers. The BMC translates the
parameters in the model input field (MIF) into the matrix and vector elements
that are required to operate the algorithms in CORE. These parameters consist
of the layer thicknesses, horizontal winds in each layer, interfacial volume
fluxes, deposition velocities, etc.
The variables in the MIF are supplied in turn by a network of
interconnected processors (labeled P7, P8, etc. in Figure 3), several of which
are rather complex models in themselves. These processors generate the wind
fields, the interfacial surfaces that separate the layers, turbulence
parameters, and source emissions. Their inputs consists of information
generated by other processes in the network and of partially processed raw data
that are transferred through the processor input file (PIF). The specific data
requirements of the processor network are described later.
73
-------
Each of the processors, including CHEM, can be replaced by any module that
performs the same functions and that is compatible with the data channel
interfaces, whose specifications are a fixed part of the network. Therefore, no
specific method of treating meteorological variables is an integral part of the
model. Only the data channels, the BMC, and the mathematical equations in CORE
that describe the volume-averaged concentrations of pollutants within each of
the model's four layers are firm parts of the system.
REQUIRED MODEL RESOLUTION AND CURRENT DATA
The ROM typically requires data in gridded form for the entire modeling
region, which currently extends from 69° to 84° west longitude and from 38° to
45° north latitude. The grid system is defined in curvilinear coordinates, with
a grid spacing of 1/4° east/west longitude and 1/6" north/south latitude,
resulting in 2,520 grid cells (60 columns and 42 rows) of slightly varying area,
approximately 18.5 km x 18.5 km. The model generally requires hourly temporal
resolution for most input parameters, even though the standard model time step
is 30 rain. In the absence of more temporally resolved raw data, the model
preprocessors use a variety of interpolation schemes to provide the final
required temporal resolution.
Different types, forms, and resolutions of raw data must be standardized
and combined into the consistent, chronologically ordered data sets required by
the preprocessors. Each raw data set is transformed into a standard format
associated with that data type and all data of similar type are merged and
74
-------
sorted. Thus, consistent quality control checks, graphical analysis, and
standardized input/output procedures can be applied.
Much of the raw data were obtained from various national data bases.
However, the complexity and scope of the ROM required more detailed and
extensive information than those standard data bases provided. Therefore, EPA
conducted several special field programs to gather the ambient air quality and
meteorological data required for ROM development, evaluation, and application,
including: the Northeast Corridor Regional Modeling Project (NECRMP), the
Northeast Regional Oxidant Study I and II (NEROS), and the Persistent Elevated
Pollution Episode (PEPE) Study. (For details, see Freas 1983; Possiel and
Freas, 1983; and Possiel et al., 1982). In further discussions, NECRMP will be
used to designate data collected during any of these field studies.
Upper Air Data
The two major U.S. sources of upper air data for ROM were the National
Weather Service (NWS) radiosonde data obtained from the National Center for
Atmospheric Research and the NECRMP radiosonde, pibal, and acoustic sounder
data. Upper air data for 24 Canadian stations were obtained from the National
Meteorological Center (NMC). Ten NWS stations fell within the model domain, and
14 additional sites surrounding the region and extending into Canada were
required to resolve boundary conditions. The NWS stations are evenly
distributed across and around the modeling region. The NECRMP upper air network
consists of six radiosonde sites, six pibal sites, and four sodar locations
aligned along the urban corridor from Virginia to Massachusetts. Typical
75
-------
measurements include vertical profiles of pressure, temperature, dew point
temperature, dew point depression, relative humidity, wind speed, and wind
direction at all mandatory and significant levels (pressure altitudes), up to
100 mbar for NWS soundings and 700 mbar for NECRMP releases. The NWS normally
releases radiosondes at 12-h intervals; however, several selected sites
increased their release frequency to 6-h intervals during the special field
study periods—August 1979 and June, July, and August 1980. NECRMP releases
occurred in the early morning, mid-morning, and early afternoon. Vertical
profile data were also available from aircraft flights, tetroons, small and
large tethered balloons, 3-D sodar, and minisondes.
In addition to hourly gridded values in the three layers for all listed
parameters, the preprocessors required vertical wind profiles (50-m resolution)
for each radiosonde release to calculate flow fields for layer-averaged winds.
Station elevations were also necessary.
Surface Meteorology
Hourly surface meteorological data within the modeling region were compiled
from three sources: (1) approximately 160 NWS and NMC-supplied Canadian sites,
(2) 41 NECRMP sites, and (3) 27 SAROAD (Storage and Retrieval of Aerometric
Data) sites. The NWS and NMC data obtained from the NOAA Techniques Development
Laboratory encompass all of North America and include hourly values for
temperature, dew point temperature, wind direction, wind speed, pressure, sky
cover, ceiling, cloud amounts, and cloud heights. The NWS sites are evenly
distributed across the region. The NECRMP and SAROAD sites report hourly wind
76
-------
speed, wind direction, ambient temperature, and solar radiation. NECRMP sites
are located along the urban northeast corridor and SAROAD sites are distributed
somewhat more evenly. Surface meteorological data were available for
August 1979 and July-August 1980. Station elevations and hourly gridded values
for all measurements mentioned above except solar radiation were required by the
preprocessors to calculate parameters such as friction velocity, Obukov length,
and heat flux. Surface parameters were gridded by using a 1/R2 weighting
function.
Emissions Data
The EPA National Emissions Data System (NEDS) data files were not current
or detailed enough for direct use in the ROM. Therefore EPA, in conjunction
with States in the modeling domain, compiled an improved 1979/1980 NECRMP
emissions inventory (EPA, 1982), specifically addressing the ROM requirements.
Canadian emissions inventories with data ranging from 1976 to 1980 were obtained
from Environment Canada to provide emission information for those portions of
Canada included in the modeling domain.
Point sources are those stationary sources typically emitting greater than
100 tons of any pollutant per year. Annual U.S. emissions for point sources are
reported for NOX, VOCs, CO, SOX, and TSP. Primary emphasis in terms of data
collection and quality assurance is placed on VOCs and NOX. Source locations
are resolved to the nearest 100 m, and information on individual stack diameter,
temperature, exhaust flow rate, and height are available. Emissions are
~v,
reported for approximately 1,400 source classification codes (SCC). For
77
-------
electric utilities, fuel- and State-specific seasonal factors are calculated
from power generation statistics, and hourly factors are derived from hourly
power plant fuel use data. Other point source categories rely on plant-specific
operating data for temporal resolution of emissions. Uniform distributions are
assumed in the case of missing operating data (EPA, 1983).
The Canadian inventory contains the same types of pollutant emission and
stack parameter information. The annual data on Canadian point source
emissions, however, are reported for 62 different standard industrial
classification (SIC) codes, and seasonal information is available for NOX and
SOX only. Therefore, U.S. temporal allocation factors will be used to
distribute most point emissions until further temporal information can be
obtained.
U.S. area sources are typically mobile sources and small stationary sources
individually emitting less than 100 tons of pollutant per year. Annual
county-level VOC and NOX emissions data are available for 54 area source
categories. Primarily based on information gathered from the literature and
previous studies, seasonal, daily, and hourly allocation factors were developed
for all area source categories. County emissions were apportioned to the model
grid system according to the known distribution of surrogate indicators such as
housing, population, urban land, agricultural land, composite forest, land area,
airport, and park locations.
Annual Canadian area source data were reported for all five pollutants
(NOX, VOCs, CO, SOX, TSP), on the Canadian polar stereographic grid system used
78
-------
by the Canadian Meteorological Center (CMC). The side length of a grid is
127 km. Currently, only population data have been obtained for finer spatial
resolution. However, additional surrogate information will be available from
Canada in the near future. Only total emissions per pollutant are reported for
each CMC grid, but percentage contributions from the 54 area source categories
have been calculated.
The current chemical mechanism in the ROM expects VOC emissions to be
disaggregated into four reactive classes: olefin, paraffin, aldehyde, and
aromatic. The speciation methodology, which makes use of species profiles
associated with process-related groups of point and area source categories, is
flexible enough to generate factors to speciate reported VOC emissions into any
chemical classification scheme required for the regional model. NOX is also
split into its NO and NOz components, based on species profile information.
Hourly gridded values for these reactive HC classes, NOX related species, CO,
and initial estimates of the remaining species treated in the current chemical
mechanism are required for model operation.
Land Use Data
A National Land Use and Land Cover Inventory (Page, 1980) was compiled by
the EPA Environmental Monitoring Systems Laboratory for the specific latitude/
longitude based grid system used in ROM. The land use and land cover data were
derived from U.S. Geological Survey maps and Landsat imagery acquired during the
periods July 23 through October 31, 1972, and January 1 through March 15, 1973.
79
-------
The percentage of total land use and land cover in each grid cell is available
for 10 categories:
• urban land • mixed forest land (including forested
wetland)
• agricultural land • water
• range land • land falling outside the study area
• deciduous forest land • nonforested wetland
• coniferous forest land • mixed agricultural land and range land
The inventory extends from 105° to 65° west longitude and 20° to 50° north
latitude. The land use and land cover percentages are required by a ROM
preprocessor to calculate deposition resistances and surface roughness. These
data are also used in the spatial allocations of emissions inventories.
Topography Data
U.S. Air Force average elevation data were obtained from the National
Center for Atmospheric Research (NCAR). Data consisted of mean elevation for
global areas of 1° latitude by 1° longitude in 30-min components, with 5-min by
5-min areas for Europe, a portion of North Africa, and North America, excluding
Alaska and parts of the Northwest Territories. Raw data for 5-min by 5-min
areas are averaged for one model grid cell (15 ft x 10 ft) and then smoothed
over a nine-cell grid area. The smoothed elevation data and the local maxima
are used to incorporate terrain effects on surface deposition, horizontal winds,
and mean vertical motion.
80
-------
Cloud Cover and Radiation Data
Cloud cover data were available from two sources: the NWS surface stations
and the Geostationary Operational Environmental Satellite (GOES) imagery. The
NWS cloud data were described in the Surface Meteorology section above.
Satellite images for 3 days in August of 1979 and 20 days during the period July
25 through August 25, 1980, were processed to obtain the following data for each
model grid cell: (1) fractional coverage of all clouds, other than cumulus;
(2) fractional coverage of only cumulus clouds and unobscured by higher clouds;
and (3) average height of the cumulus cloud tops. Data images typically cover
most of the model domain and are available four to five times per day.
Data were collected during August 1979 and July to August 1980. Total and
ultraviolet radiation measurements were recorded at 10 surface monitoring sites
and on approximately 15 aircraft flights. Spatial distribution is limited to
the urban corridors and actual flight paths. Temporal resolution can vary from
10 min to 1 h.
Both cloud cover and radiation data can be used to vary photolytic rate
constants in each grid for each model time step (30 min). The current
methodology is to parameterize the effects of clouds on the solar spectrum
rather than to require radiation measurements. Primarily, the hourly surface
cloud data are gridded by using a 1/R interpolation with the scan radius
lechnique. Cubic spline interpolation is used to derive the 30-min cloud cover
values for calculating the gridded cloud transmissivity required to vary the
photolytic rate constants.
81
-------
Aircraft Data
During the 1979 field study, three aircraft were instrumented to provide
continuous measurements of 03, NOX, S02, light-scattering coefficient (b-scat),
temperature, relative humidity, and periodic canister samples for HC analysis.
Sampling frequency ranged from 10 samples per second to one sample every 10 min,
and data were collected for approximately 140 flights. The flight patterns were
designed to provide horizontal and vertical distribution of measured parameters
within specific air parcels as they were transported across the region.
Approximately 19 aircraft were involved in the 1980 field studies,
collectively recording about 200 flight days of data during the 30-day study
period. Additional measurements include dew point temperature, total and
ultraviolet radiation, turbulence data, condensation nuclei, winds, sulfates,
nitrates, and cloud chemistry in gas-phase, aerosol, and rain/cloud water forms.
The 1980 aircraft sampling patterns were designed for several purposes:
characterization of urban plumes, examination of urban plume interaction, and
characterization of the regional air mass.
Ambient concentration data are required for testing the performance of the
ROM and for determining boundary conditions. Other specialized data sets are
being analyzed to provide improved parameterization of meteorological and
chemical mechanisms currently in the ROM.
82
-------
Surface Air Quality
The two major sources of surface air quality data were 79 NECRMP sites and
81 SAROAD sites. Hourly measurements of 03, NO, N02, NOX, nonmethane organic
compounds, methane, and CO were recorded during July through mid-September 1980.
SAROAD sites are scattered across the entire region, whereas the NECRMP sites
are aligned along the urban corridor.
Additional field study measurements, primarily of NO, NOX, and O3, were
taken at four stationary ground platforms and two mobile laboratories. Surface
air quality measurements are used in conjunction with aircraft data for model
boundary conditions, initial conditions, and evaluation.
MODEL APPLICATION IN DIFFERENT REGIONS
The design and implementation of the ROM theoretically provide maximum ease
and flexibility for updating the model and preprocessor, such as shifting the
model domain. Software modifications would be minimal if the same grid pattern
(60 x 42) or a smaller grid pattern were defined over another region of
interest. If, however, a larger number of rows and/or columns were necessary to
define the modeling region, then approximately three person-months would be
required to adequately modify and test all software components, i.e.,
approximately 40 independent programs.
The major effort required in either case is the acquisition and preparation
of raw data applicable to the chosen region. Available data of each type
83
-------
discussed in the previous section must be reviewed for spatial and temporal
adequacy. If required data are readily available, approximately three to four
person-months of programming and meteorological skills are necessary to verify
raw data and interface with existing preprocessors. If certain data sets are
inadequate or nonexistent, then the additional costs of procuring supplementary
data must be evaluated. The cost of independently obtaining data typically
measured through national networks or special field studies (i.e., surface and
upper air meteorology and air quality, and emissions) is generally very high.
Individual estimates are highly dependent on actual requirements. Preparation
of a land use and land cover inventory similar in scope and resolution to the
present U.S. inventory would cost approximately $15,000. Topography data are
currently available for all of Europe. Even if adequate annual point and area
source emissions inventories were available for the specified region, the cost
of developing specific temporal, spatial, and VOC speciation factors for the
region must be added. The cost to EPA for the current set of factors was about
$50,000. The estimated computer cost (with EPA's National Computer Center
UNIVAC 1100/82) of generating a temporally and spatially resolved emissions
inventory compatible with the ROM for one emission scenario is $15,000 (about
$200 per hour). A single execution of the preprocessor system on the UNIVAC
1100/82 requires approximately 3 to 4 CPU hours, and a single 24-h model
simulation requires approximately 10 CPU hours. Thorough testing of a modified
model and preprocessor system would require a 3- to 4-mo commitment of a person
highly knowledgeable in ROM theory and operation.
In summary, the application of the U.S. EPA ROM for a domain other than the
current Northeastern United States is, in theory, fairly straightforward.
84
-------
However, because of extensive data requirements and the complexity of the
software system, the actual accomplishment of this objective requires
significant resources, both in dollars and skilled personnel.
REFERENCES
Demerjian, K. L., and K. L. Schere. 1979. Applications of a Photochemical Box
Model for Ozone Air Quality in Houston, Texas. In: Proceedings,
Ozone/Oxidants: Interaction with the Total Environment II, Houston, Texas,
Air Pollution Control Association, Pittsburgh, Pennsylvania, 1979.
pp. 329-352.
Freas, W. P. 1983. Northeast Corridor Regional Modeling Project Data Base
Description. Office of Air Quality Planning and Standards, U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina.
Lamb, R. G. In press. Air Pollution Models as Descriptors of Cause-Effect
Relationships. Paper presented at the Joint WHO/IIASA Workshop on Ambient
Air Pollution - Health Effects and Management, Vienna, Austria, July 1982.
Lamb, R. G. In preparation. A Regional Scale (1000 km) Model of Photochemical
Air Pollution. Part 2: Input Processor Network Design.
Lamb, R. G. 1982. A Regional Scale (1000 km) Model of Photochemical Air
Pollution. Part 1: Theoretical Formulation. EPA-600/3-83-035, U.S.
Environmental Protection Agency. 226 pp.
Page, S. H. 1980. National Land Use and Land Cover Inventory. Lockheed
Engineering and Management Services Company for the Office of Research and
Development, U.S. Environmental Protection Agency, Las Vegas, Nevada.
7 pp.
Possiel, N. C., and W. P. Freas. 1983. Northeast Corridor Regional Modeling
Project Description of the 1980 Urban Corridor Field Studies. Office of
Air Quality Planning and Standards, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
Possiel, N. C., J. F. Clarke, T. L. Clark, J. K. S. Ching, and E. L. Martinez.
1982. Recent EPA Urban and Regional Oxidant Field Programs in the
Northeastern U.S. In: Proceedings of the 75th Annual Meeting of the Air
Pollution Control Association, New Orleans, Louisiana.
85
-------
U.S. Environmental Protection Agency. 1983. Northeast Corridor Regional
Modeling Project Annual Emission Inventory Compilation and Formatting,
Volume XVII: Development of Temporal, Spatial and Species Allocation
Factors. EPA-450/4-82-013q, Research Triangle Park, North Carolina.
118 pp.
U.S. Environmental Protection Agency. 1982. Northeast Corridor Regional
Modeling Projectf Annual Emission Inventory Compilation and Formatting,
Volume I: Project Approach. EPA-450/4-82-013a, Research Triangle Park,
North Carolina. 70 pp.
DISCUSSION
H. van Pop: ...those simulations were without further confusion?
R. Lamb; In these, there is a weak leakage, 1 ml/s. We thought 1 ml/s would be
small enough that there would be no appreciable leakage. However, after 50 h a
lot of stuff can leak out, even at that slow a rate. We account for that
leakage in a so-called true solution.
E. Runca; Is this a simulation then? Is there some accounting for the
subgrid-scale effects?
R. Lamb: The subgrid scale effects are handling this Layer 0, the bottom layer.
It is parameterized there. Taking into account the fact that some cells the
sources are mines or they are some network of absorption.
One of the recent pieces of data that goes into the model is a description of
how much the sources are segregated within any cell. If they are all within one
point, then we can take that into account, but it is a parameterization of a
subgrid effect.
E. Runca: So the parameterization takes place in the first layer?
R. Lamb: Layer 0.
E. Runca; Layer 0.
R. Lamb: Where the area emissions are in Layer 0, and the parameterization is
there. So, in effect, the top of Layer 0, the stuff that goes in Layer 1 now is
a function of the parameterization.
86
-------
APPENDIX. QUESTIONNAIRE ON THE CHARACTERISTICS OF
EXISTING REGIONAL-SCALE 03 MODELS
Depending on whether your model is a Lagrangian or Eulerian type, answer
the questions in either Part I or Part II of this questionnaire. Then, answer
the questions in Part III.
I. Lagrangian Models
1. How are the horizontal wind fields derived (e.g., from a model, from
r~n interpolation of data, etc.)?
2. How is the trajectory of the coordinate origin computed (e.g., from
wind at a fixed level)?
3. How are the effects of wind shear parameterized?
4. How is lateral diffusion parameterized?
5. How is vertical diffusion treated?
6. How ate species concentrations outside the plume determined?
7. Are plumes from neighboring sources allowed to mix and react?
8. What is the vertical resolution of the model (e.g., number of levels
and AZ)?
9. What numerical method is used to solve the equations that govern
vertical mixing and reaction processes?
10. Does the mixed layer depth vary with travel time? If so, how is it
determined?
11. Is mean (nonzero) vertical motion permitted? If so, how is the
vertical speed determined (e.g., from divergence of observed
horizontal winds)?
12. Is horizontal concentration variation resolved within the puff or
plume?
(a) With what resolution (give AX and Ay)?
(b) How many grid points in the horizontal plane?
87
-------
(c) What numerical scheme is used to solve the equations governing
horizontal mixing and reaction?
(d) Is horizontal variation of source emission rates resolved within
the puff or plume?
(e) If horizontal concentration variations within the puff or plume
are not resolved, how are the effects of these variations on
reactions rates parameterized?
13. Are convective cloud effects treated?
14. Is surface deposition treated? If yes:
(a) Are spatial variations in the rate permitted?
(b) Are temporal variations permitted?
15. Are terrain effects treated? Are land use effects treated?
16. Over how large an area are trajectories calculated?
17. What is the temporal resolution of the model?
18. How much machine time is required to compute 24-h averaged 03
measurements at a single receptor? (Which computer?)
19. Has the model been tested? If yes, cite report.
20. Is the model available for use? If not, when will it be available?
II. Eulerian Models
1. What is the temporal resolution (i.e., At)?
2. What is the horizontal resolution (AX and Ay)?
3. What are the present horizontal dimensions of the model domain?
4. What is the vertical resolution (i.e., number of levels and AZ)?
5. What is the elevation of the top level of the model?
6. How are the horizontal wind fields derived (e.g., from a model, from
r~" interpolation of data, etc.)?
-------
7. What numerical scheme is used to treat the horizontal transport terms
of the governing equations? If pseudo-spectral, how are inflow
boundary conditions handled?
8. Is mean (nonzero) vertical motion simulated? If so, how is the mean
vertical speed determined (e.g., from divergence of horizontal winds)?
9. What numerical method is used to treat the vertical transport and
diffusion terms in the governing equations?
10. How is horizontal diffusion parameterized?
11. How is vertical diffusion treated?
12. Is the mixed layer depth variable in space?
(a) Is it variable in time?
(b) How is the depth determined?
(c) How is it simulated in the model (e.g., in the form of the Kz(z)
profile)?
13. Are convective cloud effects treated?
14. Are terrain effects included?
15. Is surface deposition parameterized?
(a) Are the deposition velocities variable in space?
(b) Are they variable in time?
16. Are rainout and washout processes included?
(a) Are the rates spatially variable functions?
17. How much machine time is required to compute 24-h averaged 03
measurements? (Express estimate in terms of machine seconds divided
by the total number of surface grid points in the model domain, i.e.,
s/grid point.
18. How much computer memory would be required for a simulation of a 103 x
103 km region with a mesh size of 20 km? How many species does this
estimate include? Which computer?
19. Has the model been tested? If yes, cite report.
20. Is the model available for use? If not, when will it be available?
89
-------
III. Chemistry
1. Is the chemical kinetics scheme a fixed part of the model or can it be
interchanged with other schemes?
If fixed:
(a) List the species that are treated as dependent variables.
(b) List any species whose concentrations are prescribed.
If interchangeable:
(c) Which schemes have been tested?
(d) List the species that are treated as dependent variables in the
scheme currently used.
(e) List any species whose concentrations are prescribed.
(f) What is the largest number of species that the model can
accommodate?
2. Is the pseudo-steady-state approximation used in solving the chemical
rate equations?
3. Are the effects of subgrid-scale concentration variations
parameterized?
4. How are emissions of.major point sources treated?
5. Does the model provide any measure of how much the concentration at a
point may differ from the cell averaged value?
6. Do the photolytic rate constants vary in space as a function of cloud
cover?
Responses to Part II of Questionnaire on the ROM
1. Temporal resolution = 30 min.
2. Horizontal resolution = 1/4° longitude x 1/6° latitude.
3, Model dimensions are 60 x 42 cells = 1,100 km x 780 km.
90
-------
4. Vertical resolution is by four layers whose thicknesses vary in space
and time. Nominal values of the elevations of the tops of each layer
under clear sky conditions would be as follows: Layer 0—30 m,
Layer 1—300 m, Layer 2—1,500 m, Layer 3—2,000 m.
5. Top surface of the model is variable in space and time and is set at
each grid point to be just above the top of any convective clouds
present in that cell.
6. Horizontal wind fields can be derived by any method desired.
Currently the horizontal flows are generated in the form of function
sets (Lamb, I982b) derived jointly from observations and physical
laws.
7. Horizontal transport and diffusion terms in the governing equations
are approximated by the explicit, biquintic scheme described in
Chapter 9 of Lamb (1982a).
8. Mean vertical motion is included in the model. The method of
estimating its magnitude at each point in space and time is optional.
The calculation is presently based on the continuity equation and
computed divergence in the horizontal wind.
9. Vertical transport and diffusion are simulated by an analytic solution
of the linearized vertical equations (see Chapter 9 of Lamb, 1982a).
10. Horizontal diffusion due to small-scale wind fluctuations, viz.
convective and mechanical boundary-layer turbulence, is approximated
using K theory. The larger mesoscale fluctuation effects are
represented by the function sets of velocity fields, referred to above
and discussed in Chapter 7 of Lamb (I982a).
11. Vertical diffusion is treated by volume fluxes of material between
layers caused by turbulent fluctuations in the vertical wind.
12. The mixed-layer depth is variable in space and time. It is
represented in the model by the combined thicknesses of Layers 0, 1,
and 2. When convective clouds are present, the top of Layer 2 is
defined as the lifting condensation level and the effective mixed
layer extends into Layer 3. The method of estimating the mixed-layer
depth is optional. Currently, it is based on measured vertical
profiles of temperature and dew point and on the estimated vertical
profile of potential vorticity.
13. Convective clouds are treated explicitly, including vertical material
transfer, attenuation of sunlight, etc.
14. Terrain effects on surface deposition, horizontal winds, and mean
vertical motion are included.
91
-------
15. Surface deposition velocities are computed at each cell and each hour
by using the local friction velocity and surface deposition
resistences, estimated from land use data and empirical studies.
16. Rainout and washout are not included at present, but incorporation of
these processes would be straightforward.
17. Approximately 10 h of UNIVAC 1,182 time is needed to simulate 1/2-h 03
at 2,520 surface grid points over a 24-h period. Concentrations of 22
other species are also provided at all points and time steps. This is
approximately 15 s per grid point for a 24-h simulation.
18. Total computer memory requirement for 23 species; 7,560 grid points
(3 layers at 2,520 points per layer) is 110 K words. (Note: Layer 0
is handled diagnostically and does not require memory space.)
19. Analytical tests of the model are in progress, i.e., comparisons of
the predicted values with exact solutions.
20. The model is not available for use.
Responses to Part III of Questionnaire on the ROM
1. The chemical kinetics scheme is interchangeable.
(c) The scheme of Demerjian and Schere (1979) is currently being
used.
(d) This scheme treats 23 species: NO, N02, 03, paraffin, olefin,
aldehyde, aromatics, CO, HN02, HN03, PAN, RN03, H202, 0, N03, HO,
H02, H04N, RO, R02, R20, R102, R2O2.
(e) No species values are prescribed in the rate equation.
(f) The largest number of species that the model can handle is
limited only by the machine time one is willing to buy.
(g) Simulations of 2 days have been done. Longer simulations are
anticipated.
2. (a) No pseudo-steady-state assumptions are used.
(b) Nighttime chemistry kinetics is the same as the daytime, with the
photolytic rate constants all set to zero.
(c) Nighttime wind shear, stability stratification, and turbulence
are simulated.
92
-------
3. (a) Subgrid-scale chemistry phenomena are parameterized in the bottom
layer only, i.e., Layer 0.
(b) Subgrid-scale effects are those due to the segregation of fresh
emissions from line sources and aged pollutants brought down to
ground level by turbulence.
4. Emissions of major point sources are not now treated in a rigorous
way. A scheme will be developed.
5. Layer 0 contains the means of estimating how much the concentration at
any point in that cell might differ from the cell-averaged value.
6. (a) Rate constants vary in space and time due to sun angle and cloud
cover variations.
(b) Convective cloud effects on vertical material fluxes, mixed-layer
depth, and photochemical reaction rates are treated. Liquid-
phase chemistry is not treated at this time.
7. Natural emissions from vegetation are estimated on the basis of
biomass levels in each grid cell, temperature, and time of day.
Stratospheric 03 entrainment into the model domain is parameterized by
using estimates of 03 concentration aloft and mean vertical air speed.
93
-------
REGIONAL MODEL FOR OXIDANTS:
THE NORWEGIAN LAGRANGIAN LONG-RANGE TRANSPORT MODEL
WITH ATMOSPHERIC BOUNDARY LAYER CHEMISTRY*
Oystein Hov
Norwegian Institute for Air Research
Box 130, N-2001
Lillestrom, Norway
Anton Eliassen and Jorgen Saltbones
Norwegian Meteorological Institute
Box 320
Blindern, Oslo, Norway
Ivar S. A. Isaksen and Erode Stordal
Institute of Geophysics
University of Oslo
Box 1022
Blindern, Oslo, Norway
INTRODUCTION
The Norwegian Lagrangian long-range transport model for oxidants represents
the joint effort of three institutions—the Institute of Geophysics at the
University of Oslo, the Norwegian Meteorological Institute, and the Norwegian
Institute for Air Research. The model is actually a combination of a chemical
model describing the gas-phase chemistry of HCs, NOX, and SO2, together with a
simple parameterization of the gas-phase/aerosol-phase interaction, and a
meteorological model describing the long-range transport of air pollutants
(Eliassen et al., 1982). The chemical model was developed at the University of
Oslo (Hesstvedt et al., 1978; Hov et al., 1978a; Isaksen et al., 1978; Derwent
and Hov, 1979, 1980a). This model has been used to demonstrate that 03 can
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
94
-------
survive in the atmospheric boundary layer for a week or more and thus can be
transported over long distances (Hov et al., 1978b). The meteorological model
was originally developed as part of an OECD study on the long-range transport of
air pollutants (OECD, 1977; Eliassen, 1978). It is now being applied at the
Norwegian Meteorological Institute in connection with the Cooperative Programme
for Monitoring and Evaluation of the Long-Range Transmission of Air Pollutants
in Europe (EMEP), which is being sponsored by the United Nations Economic
Commission for Europe (ECE).
The EMEP project is divided into two areas, chemical activities and
meteorological activities. The Norwegian Institute for Air Research (NILU)
serves as the coordinator for the chemical activities; two other centers, one in
Moscow and the second at the Norwegian Meteorological Institute (NMl),
coordinate the meteorological activities. Some of the modeling efforts
conducted for EMEP have been described by Eliassen and Saltbones (1983), and
reports on monitoring and interpreting the long-range transport of oxidants in
Norway have also been published (Schjoldager et al,, 1978). Inventories on NOX
and S02 emissions in Europe have also been reported by Semb (1979) and by
Dovland and Saltbones (1979).
In 1978, a planning conference on the long-range transport of photochemical
oxidants was held by NILU under OECD patronage. The conference organizers
presented a list of questions on the characteristics of the model. The answers
to these questions are discussed throughout this text where appropriate.
95
-------
It is important to consider the construction of a model as a continuous
development, and the questions to be answered by the model should determine its
formulation. Thus, a model should be formulated with such flexibility that the
most important processes involved in answering a question can be identified and
quantified. Excessive complexity inevitably leads to several problems, such as
a heavy investment in computer capacity and model management. Such complexity
may be distracting when the impact of the various processes included in the
model must be assessed. The complexity may thus disguise the poor performance
of some components of the model.
MODEL DESCRIPTION
Meteorological Model
The Norwegian model presently employs 96-h, 850-mbar back trajectories,
which are assumed to be trajectories of polluted boundary-layer air parcels.
The pollutants are assumed to .be completely mixed vertically throughout a
boundary layer of variable depth; the concentrations of the various species are
therefore functions of the horizontal coordinates and time. The same assumption
is made for the horizontal wind. The top of this well-mixed layer is assumed to
be a material surface, through which no mass transport takes place. Lateral
diffusion is neglected, because the emission data are given in a 150-km grid;
hence, finer details in the concentration fields are already smoothed out.
In episode studies based on short sampling periods (i.e., periods much less
than 24 h), the rate at which instantaneous pollutant releases spread
96
-------
horizontally may be an important parameter to consider (Eliassen, 1982). For a
24-h sampling period, which is used for sulphur species in EMEP, the
instantaneous diffusion of pollutant releases is dominated by the diffusion due
to sampling time. The "synoptic swinging" (Smith, 1979) of trajectories is then
the dominating factor in plume spread (Eliassen, 1982).
During transport, pollutants are emitted into the air parcel according to
the NOX, HC, and S02 emissions maps, and the chemical reactions between the
various species proceed continuously. At the specified receptor points,
instantaneous concentrations are predicted upon arrival of a trajectory.
A receptor-oriented trajectory model, such as the one outlined above, has
an important advantage over a source-oriented trajectory model: Nonlinear air
chemistry can be included fairly easily. This is not true for a source-oriented
model in which the individual "puffs" emitted from the different sources are
followed separately. In practice, "puffs" emitted from different sources at
different times may overlap, and the chemical species present in the overlapping
should then be allowed to interact in a nonlinear chemical scheme. Because of
the nonlinearity, however, the distribution of pollutants back into the
individual puffs after the chemical interaction has occurred is indeterminate.
If one tries to handle this situation by letting all overlapping puffs merge,
the result is, in fact, a receptor-oriented model.
In a receptor-oriented model, the horizontal resolution of the
concentration fields is determined by the choice of emission grid and the
density of trajectory arrival points. The horizontal resolution that can be
97
-------
achieved is, to a large extent, a function of how well the horizontal
distribution of the emissions is known.
If. an Eulerian approach were taken, such problems would not arise. In
principle, one could construct multilayer models that would take into account
such factors as the combined effect of wind shear and vertical diffusion. This
situation would be difficult to handle in Lagrangian models. However,
Lagrangian or trajectory models have one important advantage over Eulerian
models: In Lagrangian models the integration of the equations is reduced to an
ordinary time integration along selected trajectories. In terms of the
numerical methods and the computer capacity required, this is a much simpler
problem to handle than a complete Eulerian integration over a large geographical
area.
Trajectory positions were calculated every 2 h by using the method of
Petterssen (1956). The calculations were based on wind observations made at the
850-mbar level at 0000, 0600, 1200, and 1800 GMT. The observed wind data were
analyzed objectively in the 150-km grid indicated in Figure 1. This is the EMEP
grid, which covers a net of 37 x 39 grid squares for an area of 150 km x
150 km2. The temporal resolution of the model is adjustable, but as mentioned
above, currently instantaneous concentration fields were calculated every 6 h.
In regions where wind observations were scarce, such as areas over seas, the
final wind analysis was heavily influenced by the quasi-geostrophic balanced
wind produced by NMI on a 300-km grid as part of its weather prediction routine.
98
-------
Figure 1. Air trajectories (850 mbar) followed for 4 days, arriving at one of
the five receptor points at 1200 GMT, April 6-13, 1979. The arrival
dates are indicated at the starting point of the trajectories. The
broken line represents the high NOX and HC emission rates for the
trajectory arriving April 12. Ozone was measured at Langesund (L).
The four receptor points used in the model calculations of 03 are
denoted by circles. Sulphur dioxide and particulate sulphate were
measured at Rorvik (R). The size of the 150-km grid squares is shown
in the lower right-hand corner. The meteorological variables are
given in the same grid.
Alternative trajectories may be calculated by backing the analyzed 850-mbar
winds (e.g., 15°) and by reducing the wind speed (e.g., 90%) when the radiosonde
observations in the vicinity of the trajectories indicate a significant turning
of the wind with height and change in wind speed. At present, the mixing of
pollutants associated with the diurnal variation of the atmospheric boundary
layer is not described. This would require a model with several layers, in
which "new" pollutants would be emitted into a shallow but growing convective
boundary layer and "old" pollutants, those emitted on previous days, would be
distributed within a deeper layer. The mixing height used in the Norwegian
99
-------
model is assumed to represent an "envelope" height, below which both old and new
pollutants are mixed. The 1200-GMT mixing height was chosen for this purpose.
Over most of Europe, 1200 GMT (which corresponds to 1300 or 1400 local time) is
the time by which a growing convective boundary layer has nearly reached its
maximum height.
This mixing height envelope is assumed to be a material surface through
which no mass transport takes place. Upward or downward movements of this
surface can be caused by horizontal convergence or divergence, and the density
of the air is assumed to be constant. These assumptions imply that the
concentrations of pollutants already present in the mixing layer should not be
diluted when the mixing height along a trajectory increases. The initial
concentration inputs from primary pollutant emissions are of course inversely
proportional to the mixing height.
The simplified description of vertical dispersion causes errors that are
difficult to quantify. The real situation is complicated and difficult to
handle in models. The pollutants are not only affected by the diurnal cycle of
the mixing layer, but they are also present in plumes during the initial phase
of dispersion. Thus, a nonlinear chemistry scheme using volume-averaged
concentrations may incorrectly describe the chemical development, especially
during the initial dispersion phase.
The basic data for the mixing-height analysis were taken from radiosonde
data. On the average, about 120 radiosonde reports are available within the
grid. The height up to the lowest stable layer (potential temperature
100
-------
increasing with height) was used as the mixing height for each radiosonde
report. Stable layers with a base lower than 200 ra were ignored. If no stable
layer was found, the mixing height was set equal to an upper limit, which was
arbitrarily selected as 2,500 m. The estimated mixing heights were thereafter
objectively analyzed to produce grid values at 1200 GMT. The individual
trajectories were assigned new mixing heights at 1200 GMT every day, according
to their position and the relevant mixing-height field. At intermediate times,
it is assumed that each trajectory conserves its mixing height.
A number of difficulties are associated with the objective analysis of
mixing height. For example, the mixing height estimated from a radiosonde
ascent will depend on the definition of a stable layer. Recent model tests
suggest that a lapse rate of about half the dry-adiabatic value, rather than the
dry adiabat itself, would give a better division between stable and unstable
layers (Eliassen and Saltbones, 1983).
In an experimental versioji of the EMEP model, .which may also be adopted in
the long-range transport of oxidant modeling work, the exchange of pollution
between the atmospheric boundary layer and the free troposphere was modeled by
Eliassen and Saltbones (1983).
Consider a parcel of boundary-layer air that follows a calculated
trajectory. The air parcel starts at t^ =12 GMT with a mixing height h, from
the objective analysis. Assuming that the mixing height behaves as a material
101
-------
surface, the air parcel will have a new mixing height h(t) after some time t,
given by:
t
h(t) = h, + / w(t)dt
t,
where w(t) is the vertical velocity at h(t), taken at the position of the air
parcel. The equation for h may be used to calculate the mixing height h(t2)
1 day later; i.e., t2 = ti + 24 h. In general, h(t2) will be different from the
mixing height h2, which is available from the analyzed mixing height field at
t2.
It is now assumed that h2 is the correct mixing height for the air parcel
at t2 and that the difference between h2 and h(t2) arises because the mixing
height is not in reality a material surface, but that there is a certain flux of
air through it. If h(t2) > b.2, then some boundary-layer air has been lost to
the free troposphere. If h(t2) < h2, then some air from the free troposphere
has penetrated the boundary layer. Using two methods, Eliassen and Saltbones
estimate the vertical velcocity at h(t). In the first, the velocity is
estimated from the divergence of the horizontal wind. In the second, w is
estimated as that due to frictional convergence in an Ekman boundary layer
(Eliassen and Saltbones, 1983).
An objective analysis of temperature, relative humidity, and absolute
humidity were carried out at 0000 and 1200 GMT in the 150-km grid. The
quantities analyzed were vertical averages between the surface and the 850-mbar
102
-------
level. A detailed description of the methods used can be found in Eliassen et
al. (1979). These analyses are also based on radiosonde data.
The temperature is used to evaluate certain temperature-dependent reaction
rate coefficients, such as the thermal decomposition of PAN, which influences
the 03 concentration through the production of N02.
The relative humidity is used as a rough indication of cloud cover, which
in turn influences the calculated photodissociation rate coefficients through an
"effective" albedo. Because no reflection is assumed at the ground level, any
cloud cover will reduce the photodissociation rate coefficients. The
parameterization shown in Table 1 was used.
The calculation of photodissociation rate coefficients, including the
influence of the cloud albedo, is described below. The relative humidity also
determines when certain components are removed by wet deposition.
The absolute humidity is a measure of the concentration of water vapor
molecules. This is used as input for the air chemistry model, in which HaO in
part determines the OH concentration.
The individual trajectories are assigned new temperature and absolute
humidity values when analyzed fields of these quantities are available, i.e., at
0000 and 1200 GMT. At intermediate positions, the temperature is estimated by
linear interpolation, whereas the absolute humidity is assumed to be conserved.
103
-------
TABLE 1. PARAMETERIZATION OF CLOUD
COVER ALBEDO
Relative
Humidity
>85%
75-85%
<75%
Cloud
Cover
1.0
0.5
0.0
"Effective"
Albedo
0.6
0.3
0.0
The removal rate k^ of any component due to dry deposition is:
where vd is the deposition velocity and h is the variable mixing height. The
calculated 03 concentration depends heavily on the deposition velocity assumed
for 03 (Eliassen et al., 1982). Garland and Derwent (1979) report a mean
deposition velocity over grassland of 0.58 cm/s by day and 0.29 cm/s by night.
In assigning values to this quantity, the following factors have been
considered:
• isolation of the 03 from the surface by nighttime inversions, and
• very little, if any, uptake by the sea.
Table 2 shows a set of values for the deposition velocities of 03 and other
compounds. These values were taken from a case study on the long-range
transport of oxidants to South Norway during April 1979. In this case, an
additional factor was considered when assigning values to the deposition
104
-------
TABLE 2. ASSUMED DRY DEPOSITION VELOCITIES FOR A CASE STUDY
OF LONG-RANGE TRANSPORT OF OXIDANTS TO SOUTH NORWAY DURING
APRIL 1979"
Component
03 (day, land)
03 (night, land)
03 (sea)
N02
PAN
S02
HN03
H2S04
0.1
0.2
0.3
0.4
0.5
10% of 03
(day,
land)
values
0.0
0.1
0.2
0.8
1.0
0.1
vd (cm/s)
«(/ > 65°N
60°N ~ $ < 65°N
55°N < * < 60°N
50°N •- } < 55°N
.// < 50°N
Boettger et al. (1978)
Garland and Penkett (1976)
Garland (1977)
Assumed
Value appropriate for sub-
micron particles
'From Eliassen et al., 1982.
More recent findings indicate that a value of 0.5 cm/s may be more appropriate
for the deposition velocity of N02 (Grennfelt, private communication; Galbally,
private communication).
Wet deposition is parameterized by using the relative humidity as it varies
along each trajectory. When the relative humidity exceeds 90%, precipitation is
assumed and a wet deposition rate of 1 x I0"4s~1 is applied to the H2S04, HN03,
H202, and CHa02H concentrations.
Ideally, the wet deposition should be calculated by storing the amount
removed from the atmosphere by each precipitation episode. In an experimental
105
-------
version of the EMEP model, applying scavenging ratios (W) for S02 and sulphate
resulted in wet deposition rate coefficients of the form:
kw = W h~
where h is the mixing height and P6 is the objectively analyzed 6-h
i
precipitation intensity (Eliassen and Saltbones, 1983). The objective analysis
produces a smoothing so that the time resolution is 6 h and the spatial
resolution is 150 km x 150 km. Such a grid square is either completely dry
during a 6-h period or completely wet, with an average precipitation intensity
P6. Eliassen and Saltbones (1983) state that, in reality, the space-time area
covered by a grid square will not be completely wet. The precipitation events
will generally cover parts of the area, whereas other parts will be dry. The
average precipitation intensity for the wet part of the space-time grid square
will be larger than P6. In the model, the trajectories will therefore have an
exaggerated probability of encountering rain, but the intensity of the rain will
be correspondingly lower. This reduces the probability that an air parcel will
be transported for many days without encountering precipitation at all, which
will therefore result in a reduced frequency of high concentrations in remote
areas. Eliassen and Saltbones (1983) report a method that obviates this
problem.
Chemical Model
Two types of approaches have been used in photochemical smog modeling.
One applies "lumped" kinetic mechanisms, in which the various HCs are not
106
-------
specifically included. Instead, the HC chemistry is described by using several
different classes of structure or reactivity (Hecht and Seinfeld, 1972; Reynolds
et al., 1973; Hecht et al., 1974; Falls and Seinfeld, 1978). The predictions of
such models depend strongly on the reaction rates and kinetic mechanisms used
for the different classes of HCs. A number of parameters cannot be measured and
must therefore be specified a priori. Such simple generalized mechanisms are
applied due to the lack of kinetic data and the heavy demand placed on computer
resources when combined schemes of transport and chemistry are to be integrated.
In recent years several models have been developed that apply specific
schemes describing the photooxidation of particular HCs (Demerjian et al., 1974;
Graedel et al., 1976; Hov et al., 1978a; Derwent and Hov, I980a). The accuracy
of these schemes can be determined directly from the uncertainty of the
measurable rate constants involved if the air-HC mixture has been incorporated
satisfactorily. The latter condition is not a trivial one, however, because
samples of polluted air are known to contain hundreds of different HCs.
A simplified approach must therefore be used in constructing a model system
that can be handled. The approach we have taken is to adopt a specific scheme
with a limited number of precursor HCs. This scheme reproduces reasonably well
the pollutant-generating capacity of a much more detailed specific scheme
developed earlier to represent average UK emissions (Derwent and Hov, 1979;
1980a).
After several model runs in which several different compositions of NMHC
emissions were tested, we decided to use a mixture of five different HCs to
107
-------
calculate pollutant transport. The selected mixture (30% C2H6, 20% n-
20% C2H4, 10% C3H6, and 20% m-xylene) represents fairly well the more detailed
Derwent and Hov mixture for species like 03 and OH. A compound like PAN,
however, may deviate by more than a factor of 2, due to its dependence on the
peroxyacetyl radical, which is derived only from certain HCs.
There are several reasons to use such a simplified mixture: the
uncertainty of the estimated NMHC emissions, the incomplete knowledge of the
kinetic mechanisms, and the heavy demand on computer capacity if a more complete
mixture is used.
As a result of this simplification, the chemical scheme consists of about
100 chemical reactions (including photochemical reactions) and 40 different
species. A list of reactions and rate coefficients is given in Table 3. The
concentration over time for all species involved in the reactions listed is
calculated, with the exception of a few organic radicals that react quickly with
molecular oxygen (e.g., formyl, HCO). (The 02 concentration is prescribed as
21% of M, by volume; the CH4 concentration is prescribed as 1.4 ppm).
The driving force behind photochemical air pollution is the stimulation of
free radical production by the photolysis of light-absorbing species. Thirteen
inorganic and organic species in the model are dissociated by sunlight. The
data for absorption cross sections, quantum yield, and solar fluxes are taken
from Calvert and Pitts (1967) and NASA (1979). The photochemical rate
coefficients at the earth's surface are calculated by the method of Isaksen
108
-------
TABLE 3. CHEMICAL REACTIONS AND REACTION RATE COEFFICIENTS
(cm3/molecule • s) FOR BIMOLECULAR REACTIONS
(Eliassen et al., 1982).
Reaction
Rate coefficient
Ethylene chemistry
Inorganic chemistry
O + Oj+ M -O, + M I.I x
O * NO + M - NO: + M 3.0 X
O'D + M — O + M 3.0 X
H-0 + O'D - 20H 2.3 X
O, + NO - NO; + O, 2.3 X
Oi * NO- — NO, + O: 1.2 x
O, + OH - HO; -HO, 1.8 x
O, f HOj — OH + O, + O, 1.4 x
NO + NO, - 2ND, 1.9 X
NO + HO, — NO, + OH 8.1 X
NO, + NO, - NO + NO, + O, 2.3 X
NO, + NO, — N:O, 1.48 X
NO, + OH - HNO, I.I X
NO, -f H,O, - HO, + HNO. 4.1 X
NO, + NO, - NO. + NO, •*• O, 8.5 X
N-O, - NO. + NO, 1.24 x
OH + HO, - H.O 5.1 X
OH + H,O, — HOj + H,O 2.7 x
OH + H,(+O,) - HO, + H,O 3.6 x
OH + HNO, - NO, + H.O 8 0 X
HO, + HO: - H-O, + O. 3 8 X
10°' exp(940/r)
IO-"exp<-l450/7")
10-" exp<-2450/r)
10'" exp(-930/r)
I0-"exp(-580/r)
10""
io-1
10-"ex(X-IOOO/D
I0-"exp(86l/r)
10-"
10-'
I0""exp(-2450/r)
10"exp(-10317/r)
io-"
I0-"exp(-l45/r)
lO'" exp(-2590/7")
10—
IO"Mexp( 1245/7")
Sulphur chemiury
OH - SO; - HSO, 1.1 X lO"'-'
CH,O. - SO; - SO, + CH,O <5 x 10""
HSO, + O. - HSO, I 0 X 10-"
HSO, + NO - HSO. + NO, 1.0 X 10""
HSO. + O, - SA' + HO, 1.0 x ID'"
SO, -r H.O - SA- 9.1 X 10-"
• SA sulphuric acid or sulphate aerosol.
Methane chemistry
H,O
OH +CH. -CH,O
CH, + 02 - CH,6,
CH,Oj + NO - CH,6 + NOj
CH,O, + CH,O, - CH,6
+ CH,O + O2
HO, + CH,O, - CH,O,H + O,
CH,6 + Oj - HCHO + HO,
OH + HCHO — H2O + HCO
NO, + HCHO - HNO, + HCO
OH t CO - COj t- H
HCO + O, — HO, + CO
* /\aim): pressure in atmospheres.
2.4 x I0-"exp(-l710/r)
5.1 x 10-"
6.5 x IO"2
'4 Ox 10""
7.7 x I0-"exp(l300/r)
1.8 x 10'"
1.25 x 10-" exp(-88/r>
8.0 x 10"'*
I 35 x 10'" 11 + />(atm)]'
5.1 x 10'"
Elhane chemistry
C-H. + OH - CH,
CjH, 4- O2 - C.HjO.
C-H.Oj + NO - C-H,6 + NO,
C.H.O - HCHO + CH,
C-H|O + O. - CH,CHO + HO.
CHjCHO + hv - CH, + CHO
CH,CHO + OH -CH,CO
+ H,O
CH,CO -r O; - CHjCOO;
CH,COO, + NO - CH, -t- COj
+ NO,
rH.coO. + NO, -
C'HjCOO-NO. (PAN)
PAN - CH ,COO. -r NO,
• Ver> f.isi rcjction step
1.86 x 10-" exp(-l236/r)
•
3 x 10-"
33.0
37 x 10-"
3 1 x 10-'
69 x I0-"e>.p(258/r>
2.6 x 10"" . „
1.4 X 10""
7.94 x I0"exp(-12530/r)
C,H. 4 OH -CHjCH-OH
CH,CH:OH + O, —
CH-O-CH-OH
CH-0-CH-OH 4 NO -
CH-OCH:OH 4 NO,
CH.OCH.OH 4 O, - HCHO
+ HCHO + HO,
C.H. 4 O, - HCHO 4 CH-O.
CH,O, 4 Os - HO, * HO,
4 CO,
* Very fast reaction step.
Propylene chemistry
C,H. + OH - CH,CHCH-OH
CH,CHCH.OH 4 O, —
CHiCHO;CH.OH
CH,CH6-CH.OH + NO -
CH,CHOCH:OH + NO;
CH.CHOCH.OH + O. -
CH,CHO + HCHO + HO:
C,H, + O, - HCHO + CH,6,
* HO: + CO.
— C'H,CHO * HO;
* HO: + CO.
• Ver> fasi fraction step
n-butane chemistry
_/~ tj i f\U **rS~ U * 1-1 f}
nv,4rjio T" \sn >«c\_«rt^ n*\j
secC.H, + O: — «rC.H.O;
NO + secC4H,0- — secC.H,6
+ NO.
secC«H,O + O, - HO,
- + CH,COC3H,
secC.H,6 - CH,CHO + C>H,
CH,COC,Hj -f hi — C,H,
+ CH,CO
CH,COCjH5 + OH —
CH,COCHCH, + H:O
CH,COCHCH, + O: —
CH.COCHb.CH,
CH,CbCHb.CH, -f NO —
CH.COCHOCH, 4 NO;
CH.COCHOCH, + O- —
CH,rOCOCH, 1 HO:
CH.COTOCH, 4 hi - CH.CO
4 CH.CO
• Ver> fast reaaion step
m-xvlcnc chewistrr
2.2 X I0-"cxp<385/r)
3.1 X 10'"
9.0 X
4 1 X I0-':exp(545/r)
•
3.1 x 10-'-*
•
305 X 10 '• e»p(-1900/7 :
305 x 10'" txp(- W(.lO/7
1 2 X lO'"
30 x 10'"
2.1 X 10-"
1.2 X 10s
6 5 x 10-*
3 4 x 10'i:
3 0 x lO'1-'
1.4 X 10-'
24 X ID'
Y*^
OH OH
NO- I ' - -o. 3.1 x 10"':
OM
fM .ft.o.
1 ' . Hil» i
n^N-IX) *
109
-------
TABLE 3. (continued)
HCOCCH.CHCHO + OH —
HCOCCH.CHOHCHO
HCOCCH,CHOHCHO + O, -
HCOC6,CH,CHOHCHO
HCOCO,CH,CHOHCHO
+ NO — HCOCOCHr
CHOHCHO + NOi
HCOCOCHjCHOHCHO
+ O, — CHOCHO
+ CH.COCHO + HO,
CHOCHO + h» — HCHO + CO
• Very fast reaction step.
Procen
10-"
3.1 X IQ-"
3.3 X I
Noon photolysis rate* (s~')
HCHO + A, - HO, + co
+ HO, 2 1 x |0'!
— H, + CO 4 5 X 10"'
CH.CHO * A. - CH, + HO,
* CO 3 | x |0"'
CH,COC,H, + A, ~ CH.COO,
+ C,H,6, 6.J X 10"*
CH,COCOCH, •» A, —
CH,COO, + CH,COO, 1.4 f ICTJ
CH,COCOH -f A. - CO
+ CH,CHO 6.5 X 10"
HCOCOH + h,• — CO + HCHO 3.3 X 10-
CH,O,H + A» — CH,0 + OH 4 4 X IO"*
• At SO'N latitude, pound level
Photochemical processes
O, + Av -Of'D)+ O,
d + he + O + O,
NO, 4 A> — NO + O
NO, + A* - NO, + O
— NO + Oj
N,O, + Ar - NO, + NO,
H,O, + Ar - 2OH
HNO, + A» - NO, + OH
2.1 X 10°
3.7 X 1C-1
5.7 X 10-'
IJ X 10-'
4.0 X ID'1
2 4 X IO-J
S.9 X IO-»
3.3 X 10°
et al. (1977), which assumes that scattering takes place only in a direction
parallel to the direct beam of solar radiation.
Reflection by a cloud layer above the mixing layer is modeled by assuming
that the cloud top and the base act as partially reflecting, homogeneous
surfaces with a specified albedo, which is assumed to be the same for radiation
from all directions and for all wavelengths (Derwent and Hov, 1979). In the
model so far, the albedo is specified as 0.0, 0.3, or 0.6, according to the
objectively analyzed relative humidity fields. Reflection at the surface is
ignored.
Dissociation rate coefficients are calculated for every 5° latitude and
every 15 min of the day. The total vertically integrated atmospheric 03 .column
is adjusted to correspond to the season and latitude in accordance with the data
110
-------
given by Duetsch (1978). Points along a given trajectory are allocated J values
through interpolation in time and space to the appropriate latitude and local
time.
The chemical scheme chosen (Table 3) and the composition of the HC
emissions adopted by no means represent a unique or formally optimized
selection. Rather, the choice is based on a subjective judgment of what kind of
chemical description is adequate to include in a model of long-range transport
of oxidants based on the present understanding of how meteorological, chemical,
and physical processes interact. Extending the chemical scheme to fill up any
computer capacity available, both with respect to memory and to CPU time, is
very simple. However, the important question is: To what extent would this
improve the model performance, or would it just increase the complexity and
intransparency of the model?
The modeling of diurnal behavior is very important in understanding the
mechanisms underlying the long-range transport of oxidants (Hov et al.,
1978a,b).
The concentrations assigned at the starting point of the 96-h trajectories
can be important for the development along the trajectory, particularly when the
photochemical activity is low and the chemical lifetime of both precursors and
secondary pollutants is long. Ground removal is the ultimate removal mechanism
for 03, and where low deposition occurs, the 03 lifetime is much longer than 4
days (Hov et al., 1978b). In such situations, even 4-day trajectories are
insufficient to trace the history of an air mass. If the weather is fair at the
111
-------
starting point, the air masses arriving there may have already accumulated
photochemically active pollution over a number of days. In such cases, air
chemistry calculations are initiated up to 4 days before the start of the
trajectory, depending on the length of the fair-weather period. The emission
rates, the radiation values, and the meteorological parameters are then taken as
averages over the 5 x 5 (25) grid squares surrounding the starting point of the
trajectory. In this way, the chemical development along a model trajectory is
made nearly independent of the initial conditions.
Air chemistry integration is started by using a set of concentrations
corresponding to a very slightly polluted atmosphere, with the removal processes
in equilibrium with NOX and NMHC emissions averages near the Northern Hemisphere
(2 x 10'a molecules/cm2/s for NOX and NMHC/NOX = 1.25). The initial
concentrations of the most important species obtained in this fashion are listed
in Table 4.
Natural sources of HCs are not accounted for in the model. Separate model
evaluations, indicate that natural HCs probably do not contribute significantly
TABLE 4. INITIAL CONCENTRATIONS
Species
NO
N02
S02
S04
Concentration
(ppbv)
0.14
0.34
1.27
0.58
Species
NMHC (C)
03
HN03
PAN
Concentration
3.38
29.50
0.19
0.05
112
-------
to oxidant formation on a regional scale in Europe (Derwent and Hov, I980b; Hov
et al., 1982). Natural sources of NOX are thought to be small compared to the
anthropogenic sources. Stratospheric 03 or the 03 concentrations in the free
troposphere do not affect atmospheric boundary-layer chemistry as long as the
upper boundary of the mixed layer is considered to be a material surface.
Emissions
Manmade NOX and SQ2 emissions are mostly due to the combustion of fossil
fuels. HC emissions arise mainly from incomplete combustion in motor vehicle
engines and from evaporative losses of gasoline and solvents from storage and
handling.
As a basis for the model calculations, emissions data for these gases were
needed for a grid covering Europe. In principle, such data can be estimated,
but the uncertainties are necessarily high. However, the degree of consistency
obtained between the calculations based on a case study and the actual
measurements (Eliassen et al., 1982) suggests that the estimates by and large
reflect the actual emissions.
An inventory of European sulphur emissions was prepared in connection with
EMEP (Dovland and Saltbones, 1979) giving the estimated annual (1978) emission
in 150-km grid squares (the grid indicated in the lower right-hand corner of
Figure 1).
113
-------
The estimated total national emission figures are listed in Table 5. The
uncertainty was 10% to 15% at best and considerably larger for many of the
countries.
The estimates of national NOX emissions in European OECD member countries
(i.e., Austria, Belgium, Denmark, Finland, France, Federal Republic of Germany,
Greece, Iceland, Ireland, Italy, Luxembourg, The Netherlands, Norway, Portugal,
Spain, Sweden, Switzerland, Turkey, and the United Kingdom) are based on
information obtained from ongoing OECD work to evaluate possible photochemical
oxidant control strategies. For the United Kingdom, however, the emissions data
given by Apling et al. (1979) are used. For the remaining European countries,
the emissions estimates are taken from Semb (1979), who has shown that, for OECD
countries, national NOX emissions are related to the total energy consumption,
if deductions are made for hydroelectricity and nuclear electricity production
and for noncombustion solid fuel used in the iron and steel industry. Data on
energy consumption for all countries of the world are published by the UN
Statistical Office. Thus, NOX emissions for the remaining European countries
have been estimated by assuming that the above-mentioned relationship is valid
for these countries as well (Semb, private communication). The resulting
estimated national emissions figures for NOX are listed in Table 5. Chemically,
NOX is assumed to be emitted as NO. Uncertainties are likely to be larger than
for the S02 emissions.
As a first approximation, NOX emissions data in the 150-km grid were
generated from the S02 emissions inventory by assuming that, for each country,
the distribution of NOX emissions on grid elements is identical to that for S02.
114
-------
TABLE 5. ASSUMED ANNUAL EMISSIONS OF S02, NOX, AND NMHC
FOR ALL COUNTRIES IN EUROPE*
Country
Albania
Austria
Belgium
Bulgaria
Czechoslovakia
Denmark
Finland
France
German Democratic Republic
German Federal Republic
Greece
Hungary
Iceland
Ireland
Italy
Luxembourg
The Netherlands
Norway
Poland
Portugal
Romania
Spain
Sweden
Switzerland
Turkey
USSR (within grid)
United Kingdom
Yugoslavia
Remaining area within grid
S02-Sb
50
215
380
500
1,500
228
270
1,800
2,000
1,815
352
750
6
87
2,200
24
240
75
1,500
84
1,000
1,000
275
58
483
8,100
2,490
1,475
256
Emissions
NOX-N02C
10
275
410
240
600
240
200
1,650
680
3,350
500
220
10
90
1,550
50
700
110
1,000
110
460
850
260
160
600
5,000
1,730
210
50
>
NMHC"
10
280
390
240
600
220
200
2,000
680
2,450
260
220
15
105
1,750
30
600
170
1,000
200
460
1,050
380
260
600
5,000
1,158
210
50
"Emissions expressed in 103 tonnes.
bS02 measured as S.
CNOX measured as N02.
dNMHCs measured by their total mass.
In grid elements where the sulphur emissions are thought to be anomalously high
relative to the energy consumption, lower NOX emissions were assumed. Due to
the relatively large uncertainties of the NOX emission allocated to the
115
-------
individual grid squares, a moving average over 3x3 (9) squares is employed in
the model calculations.
The estimates of NMHC emissions are again based on information obtained
from OECD (1982), except for the United Kingdom data, which are from Apling et
al. (1979). According to these data, the ratio between national NMHC and NOX
emissions in OECD Europe varies between 0.5 and 1.82 (NMHC measured by total
mass, NOX measured as N02). For non-OECD European countries, the NMHC emissions
were estimated to be roughly equal to the NOX emissions.
The resulting NMHC emissions estimates are listed in Table 5. The
uncertainty is thought to be considerably larger than that for S02 and may
approach a factor of 2, particularly for non-OECD European countries.
The emission grid data for NMHC were generated by distributing the national
emissions according to tire sulphur emissions inventory. In areas where the
concentrations from oil refineries and the petrochemical industry are high,
increased NMHC emissions are assumed. As with NOX emissions, a moving average
over 3x3 grid squares is employed in the model calculations.
Mathematical Formulation
Based on the assumptions made above, the mass conservation equation
determining the mass concentration GJ of species i can be written as:
DC i
dt
116
-------
where D/dt = the Lagrangian (total) time derivative along a trajectory;
Vfi(x,y,t) = the dry deposition velocity, assumed to be variable for 03;
h(x,y,t) = the variable mixing height;
kw(x,y,t) = the wet deposition rate, active only during the rain (i.e., at
a relative humidity >0.9%);
Ej(x,y) = the direct emission of pollutant, in mass per unit area and
time, from emissions inventories; and
Sj = chemical source or sink.
In general, S, consists of terms of the type:
N
ik II cmi and +njJj(x,y,t)Cj,
which describe production or destruction by gas-phase reactions and
photodissociation, respectively, where k is the reaction-rate coefficient, mj is
the order of the reaction with respect to the species number j, J, is the
dissociation rate, and n,- is a stoichiometric factor. The dissociation rate
depends on the cloud cover parameterized by the relative humidity, season,
latitude, local time, and the vertically integrated atmospheric 03 column.
The first step in the integration procedure is to calculate the appropriate
back trajectories from the analyzed wind fields. The second step is to convert
the quantities vd, h, kw, Ej, and Jj, originally given as Eulerian fields, into
Lagrangian information, i.e., as functions of transport time along the
trajectories. These operations transform the mass conservation equation into an
117
-------
ordinary differential equation in time. Thus, the third step is to integrate
the equation to obtain calculated instantaneous concentrations for the receptor
points at the arrival times of the trajectories.
The mass conservation equations for all species in the model were
integrated by using a quasi-steady-state approximation (QSSA) method, described
in detail by Hesstvedt et al. (1978). This method is explicit and applies a
fixed time step. When compared with Gear-type methods with automatic error
control (Hesstvedt et al., 1978; Derwent and Hov, 1979), the method gave
accurate predictions in a wide range of model calculations of atmospheric
chemistry.
A time step of 15 min is applied. Comparisons with results obtained with
shorter time steps showed that 5% is an upper limit for the computational error
and that the sign of the error is usually distributed quite evenly between plus
and minus.
The temporal resolution of the model is adjustable. Presently,
instantaneous concentrations are calculated along 96-h trajectories with an
arrival time every 6 h (0000, 0600, 1200, and 1800 GMT). The machine time
required to compute 24-h 03 levels on a CDC/CYBER 170/835 (i.e., four
trajectories, each 96 h, at a single receptor) is 9.5 s. This includes the time
it takes to compute the trajectories from the horizontal wind field. The
dissociation rate coefficients are calculated separately. At midsummer,
calculating 13 dissociation rate coefficients every 15 min of the day, on a grid
corresponding to every 5° latitude, for albedos 0.0, 0.3, and 0.6, requires
118
-------
approximately 1000 s of CPU time on the same computer. These calculations are
performed once and for all, however, and therefore contribute very little to the
total consumption of computer time during the experimental stage of the model.
Also, the dissociation rate coefficient calculations may be speeded up
significantly if the expense of computer time is an important constraint on
model work (Derwent and Hov, 1979).
A CASE STUDY
The model was tested against actual measured oxidant data in Southern
Scandinavia, April 6-13, 1979 (Eliassen et al., 1982). A brief summary of the
test is given here.
In south Norway a ground-based network of contimious Os recorders is
operated every summer by the Norwegian Pollution Control Authority at Langesund
(see Figure 1). Langesund is a coastal site that is not influenced by local
pollution when the air masses come in from the sea. Ozone measurements made at
Langesund and daily aerosol sulphate and S02 measurements made at Rorvik near
Gothenburg were used to test the model. The latter station is run by the
Swedish Water and Air Pollution Research Institute (IVL) and is part of the
Swedish EMEP station network.
The study period was April 6-13, 1979. During this period, there were no
frontal passages or rain over south Norway. The 96-h, 850-mbar, 1200-GMT
trajectories shown in Figure 1 indicate easterly transport over distances up to
4,000 km. On April 13, the synoptic situation changed, resulting in transport
119
-------
from the southwest. On April 6, 03 monitoring began. Maximum hourly
concentrations of over 75 ppbv of 03 were measured at Langesund from April 10-12
during the afternoon (see Figure 2), which strongly suggests that 03 generated
photochemically from pollution had accumulated in the atmospheric boundary
layer.
The reference model was used to calculate 03 concentrations at each of the
four receptor points at Langesund. Figure 2 shows the calculated mean
concentration and the standard deviation around the mean at 0000, 0600, 1200,
and 1800 GMT during the selected time period. The largest standard deviations
12 00 12 00 12 00 12 00 12 00 12 00 12 00 12 GMT
6 7 8 9 10 II 12 13 Dale
Figure 2. Mean and standard deviation of calculated 03 concentrations at the
four receptor points surrounding Langesund. The mean of the four
values is denoted by X. The measured 03 concentration is shown by a
solid line.
120
-------
are around 15 ppb, indicating that the calculated concentrations at the four
receptor points are clearly different in some cases. This is due to the
different trajectory paths and the pollutant emission fields, which exhibit
large spatial variations. In the following discussion, all the 03 model results
presented are averages over the four receptor points around Langesund. These
were averaged in order to smooth out variations in the calculations that are due
to random errors in the wind field.
Figure 2 also shows the measured 03 values at Langesund as a full line. In
general, 03 is underestimated during the selected period. The calculated 03
peaks for April 10, 11, and 12 have increasing maximum values instead of
recorded high peaks with a relatively constant maximum. The reduction of 03 on
April 13, associated with the shift in transport direction, is predicted by the
model. In some cases the measured 03 drops to a very low value at night. This
may be partly due to suppression of vertical mixing by nighttime stabilization
of the air near the ground, combined with dry deposition of 03 (Garland and
Derwent, 1979).
In a series of tests, the sensitivity of the calculated 03 concentrations
to variations in the initial pollutant concentrations, the 03 deposition
velocity, the advection wind, and the emission strength and composition was
investigated. "Reasonable" variations in each of these quantities affected the
calculations noticeably, suggesting that better knowledge might lead to improved
calculations. The emissions and the deposition velocity for 03 seemed to be the
most important quantities for determining the 03 concentration. The model
calculations further showed that the production of 03 from the primary
121
-------
pollutants emitted into the boundary layer can take several days, due to the
inhibition of sunlight by cloud cover and to the chemical stability of some HCs.
As an example of the sensitivity studies, Figure 3 shows the results of a
model run with zero emission outside Scandinavia. In this case, 03 is seriously
underestimated, and characteristic features in the measurements are not
reproduced. This strongly suggests that the 03 or its precursors to a large
extent originate outside Scandinavia.
In order to shed further light on the transport process and the origin of
the pollutants, we have studied the time development of the calculated pollutant
concentrations in the air parcel arriving at one of the receptor points near
Langesund at 1200 GMT on April 12. The trajectory of this air parcel is shown
in Figure 1. The parts of the trajectory that pass over very large NMHC and NOX
emisisons (more than 1011 molecules cm"2 s~1) are represented by a broken line.
About 75 ppb of 03 is measured at Langesund upon arrival of the trajectory. The
calculated 03 concentration in the air parcel at the time is also close to
75 ppb.
The time development of the 03 and NMHC concentrations and the accumulated
emission of NMHC into the air parcel are also shown in Figure 4. On the first
day, the initial 03 is slowly depleted by dry deposition until the air parcel
encounters the large emissions in the Donbass area of the USSR, after about 24 h
of transport. At night, the 03 is further depleted by reacting with the emitted
NO. There is no sunlight available so that 03 can be produced from the emitted
NMHC, which is therefore accumulated. The 03 production on the second day is
122
-------
100
80
S60
20
100
80
60
40
20
12 00 12 00 12 00 12 00 12 00 12 00 12 00 12 GMT
6 7 8 9 10 It 12 13 Dale
Figure 3. Sensitivity to emissions outside Scandinavia. The circles denote
calculated 03, with emissions of pollutants only from sources in
Denmark, Norway, and Sweden. The reference model run is denoted by
X, and measured 03 concentrations are denoted by a solid line.
o»
c
70
60
50
~ _' Ooy, clear thy
gjgj? Day, tOOVo cloud cover
MH Night
Cm,..M>n. NKHC
wfSPV)!"0'
Minnj hvight
C
OU 041 osl MUoJlui 4ll Ofi III ll
OK 0(< 07i ii.]?J7ttw: Jtl oil 23! :<
325m 1 7(
24 48
Timt (h)
• in i4n:s
-------
rather weak due to cloud cover, and only about 25% of the NMHC is consumed in
the reactions. The third day is clear, and a considerable amount of 03 is
produced. This production is due to the emissions encountered by the air parcel
about 1-1/2 days earlier. The high emissions encountered over Poland during the
evening of the third day and the following night result in further 03 production
on the fourth day.
ACKNOWLEDGMENTS
Parts of this work have been funded by the Norwegian Research Council for
Science and the Humanities (NAVF), the Royal Norwegian Research Council for
Science and Technology (NTNF), the Norwegian Pollution Control Authority (SFT),
and the Department of Environment (MD).
REFERENCES
Apling, A. J., E. J. Sullivan, M. L. Williams, D. J. Ball, R. E. Bernard, R. G.
Derwent, A. E. J. Eggleton, L. Hampton, and R. E. Waller. 1977. Ozone
concentrations in southeast England during the summer of 1976. Nature,
269:569-573.
Boettger, A., D. H. Ehhalt, and G. Gravenhorst. 1978. Atmosphaerische
Kreislaufe von Stickoxiden und Ammonia. Berichte der Kernforschungsanlage
Juelich, No. 1558, KFA Juelich, West Germany.
Calvert, J. G., and J. N. Pitts, Jr. 1967. Photochemistry. John Wiley and
Sons, New York. 899 pp.
Demerjian, K. L. , J. A. Kerr, and J. G. Calvert. 1974. The mechanism of
photochemical smog formation. Advances in Environmental Science and
Technology, 4:11-262.
Derwent, R. G., and 0. Hov. 1980a. Computer modelling studies of the impact of
vehicle exhaust emission controls on photochemical air pollution formation
in the United Kingdom. Environmental Science and Technology, 14:1360-1366.
124
-------
Derwent, R. G., and 0. Hov. 1980b. The contribution from natural hydrocarbons
to photochemical air pollution formation in the United Kingdom. In:
Proceedings of the First European Symposium on Physico-chemical Behaviour of
Atmospheric Pollutants, B. Versino and H. Ott, eds. Ispra, October 16-18,
1979. Commission of the European Communities, Directorate General for
Research, Science and Education.
Derwent, R. G., and 0. Hov. 1979. Computer Modelling Studies of Photochemical
Air Pollution Formation in North West Europe. AERE-R 9434, Her Majesty's
Stationery Office, London. 147 pp.
Dovland, H., and J. Saltbones. 1979. Emissions of sulphur dioxide in Europe in
1978. Report no. EMEP/CCC 2/79, Norwegian Institute for Air Research.
31 pp.
Duetsch, H. U. 1978. Vertical ozone distribution on a global scale. Pure and
Applied Geophysics, 116:511-529.
Eliassen, A., and J. Saltbones. In press. Modelling of long-range transport of
sulphur over Europe: A two-year model run and some model experiments.
Atmospheric Environment.
Eliassen, A. 1982. Aspects of Lagrangian Air Pollution Modelling. In:
Proceedings of the Thirteenth International Technical Meeting on Air
Pollution Modeling and Its Application, lie des Embiez, France, September
14-17, 1982. NATO CCMS, Brussels, Belgium.
Eliassen, A., 0. Hov, I. S. A. Isaksen, J. Saltbones, and F. Stordal. 1982. A
Lagrangian long-range transport model with atmospheric boundary layer
chemistry. Journal of Applied Meteorology, 21:1645-1661.
Eliassen, A., et al. 1979. An Operative Scheme for the Objective Analysis of
Relative and Absolute Humidity. Technical Report no. 42, Norwegian
Meteorological Institute. 20 pp.
Eliassen, A. 1978. The OECD study of long-range transport of pollutants: Long
range transport modelling. Atmospheric Environment, 12:479-487.
Falls, A. H., and J. H. Seinfeld. 1978. Continued development of a kinetic
mechanism for photochemical smog. Environmental Science and Technology,
12:1398-1406.
Garland, J. A., and R. G. Derwent. 1979. Destruction at the ground and the
diurnal cycle of concentration of ozone and other gases. Quarterly Journal
of the Royal Meteorological Society, 105:169-183.
Garland, J. A. 1977. The dry deposition of sulphur dioxide to land and water
surfaces. Proceedings of the Royal Society of London, A354:245-268.
Garland, J. A., and S. A. Penkett. 1976. Absorption of peroxyacetyl nitrate
and ozone by natural surfaces. Atmospheric Environment, 10:1127-1131.
125
-------
Graedel, T. E., L. A. Farrow, and T. A. Weber. 1976. Kinetic studies of the
photochemistry of the urban troposphere. Atmospheric Environment,
10:1095-1116.
Hecht, T. A., J. H. Seinfeld, and M. C. Dodge. 1974. Further development of
generalized kinetic mechanism for photochemical smog. Environmental Science
and Technology, 8:327-339.
Hecht, T. A., and J. H. Seinfeld. 1972. Development and validation of a
generalized mechanism for photochemical smog. Environmental Science and
Technology, 6:47-57.
Hesstvedt, E., 0. Hov, and I. S. A. Isaksen. 1978. Quasi-steady-state
approximations in air pollution modelling: Comparison of two numerical
schemes for oxidant prediction. International Journal of Chemical Kinetics,
10:971-994.
Hov, 0., J. Schjoldager, and B. M. Wathne. Submitted for publication.
Measurement and modelling of the concentrations of terpenes in coniferous
forest air. Journal of Geophysical Research.
Hov, 0., I. S. A. Isaksen, and E. Hesstvedt. 1978a. Diurnal variation of ozone
and other pollutants in an urban area. Atmospheric Environment,
12:2469-2479.
Hov, 0., E. Hesstvedt, and I. S. A. Isaksen. 1978b. Long range transport of
tropospheric ozone. Nature, 273:341-344.
Isaksen, I. S. A., E. Hesstvedt, and 0. Hov. 1978. A chemical model for urban
plumes: Test for ozone and particulate sulfur formation in St. Louis urban
plume. Atmospheric Environment, 12:599-604.
Isaksen, I. S. A., K. H. Midtbo, J. Sunde, and P. J. Crutzen. 1977. A
simplified method to include molecular scattering and reflection in
calculation of photon fluxes and photodissociation rates. Geophysica
Norvegica, 31:11-26.
National Aeronautics and Space Administration.' 1979. The Stratosphere:
Present and Future. Publication No. 1049, Scientific and Technical
Information Branch, Washington, DC. 432 pp.
NILU. 1978. The Long Range Transport of Oxidants: Report from a Planning
Conference on Future Research Co-operation, Oslo, September 12-14, 1978.
NILU TN 16/78, Lillestrom, Norway.
Organization of Economic Cooperation and Development. 1982. Photochemical
Smog. Contribution of Volatile Organic Compounds. Paris, France.
Organization of Economic Cooperation and Development. 1977. The OECD Programme
on Long Range Transport of Air Pollutants. Measurements and Findings.
Paris, France.
126
-------
Petterssen, S. 1956. Weather Analysis and Forecasting. McGraw-Hill, New York.
503 pp.
Reynolds, S. D., P. M. Roth, and J. H. Seinfeld. 1973. Mathematical modeling
of photochemical air pollution. I. Formulation of the model. Atmospheric
Environment, 7:1033-1061.
Schjoldager, J., B. Sivertsen, and J. E. Hanssen. 1978. On the occurrence of
photochemical oxidants at high latitudes. Atmospheric Environment,
12:2461-2468.
Semb, A. 1979. Emissions of Gaseous and Particulate Matter in Relation to
Long-Range Transport of Pollutants. In: Proceedings of the WHO Symposium
on Long Range Transport of Pollutants, Sofia, October 1-5, 1979, WMO No.
538, Geneva, Switzerland.
Smith, F. B. 1979. The character and importance of plume lateral spread
affecting the concentration downwind of isolated sources of hazardous
airborne material. In: Proceeding of the WMO Symposium on Long Range
Transport of Pollutants, Sofia, October 1-5, 1979, WMO No. 538, Geneva,
Switzerland, pp. 241-252.
DISCUSSION
R. Lamb: In your presentation you said that the receptor-oriented model has an
advantage over the source-oriented model because in the former case you can
treat nonlinear systems because you have no overlapping of plumes. In other
words, you are saying that the plumes overlap in the source-oriented models and
that this creates a problem treating nonlinear chemistry.
In the case of the receptor-oriented model, the counterpart of the overlapping
plumes is the diffusion of the material into the parcel? It seems that the
problem does not disappear by using the receptor-oriented approach. If you
allow any lateral mixing in the model, you have to compute the ambient
concentrations of the species outside the parcel.
0. Hov: That is probably right. At the moment, it does not take into account
lateral diffusion. However, neglecting diffusion is not a good approximation
for oxidant studies in which you are actually computing hourly average values.
127
-------
MODEL FOR THE REGIONAL TRANSPORT OF PHOTOCHEMICAL OXIDANTS
AND THEIR PRECURSORS IN THE UNITED KINGDOM*
Kenneth A. Brice
Environmental and Medical Sciences Division
AERE Harwell
Didcot, Oxon, 0X11 ORA, England
INTRODUCTION
Extensive monitoring programs in Europe and the United States have
established that 03, PAN, and visibility-reducing aerosol concentrations may
increase significantly during the summer over large areas in fair-weather
episodes. These episodes are associated with air masses that remain over, or
have passed through, industrialized or populated centers, and a significant,
anthropogenic influence has been demonstrated in simultaneous observations made
of elevated 03 concentrations and concentrations of manmade species, such as
trichlorofluoromethane (Cox et al., 1975) or acetylene (EPA, 1975).
Unlike pollutants in an urban area, primary and secondary pollutants
accumulate at different time and space scales in a well-developed summertime
high pressure situation. The time scale for oxidant production in an urban area
is typically a few hours, and high levels of both primary (precursor) and
secondary pollutants are already present on the first day of an episode. The
horizontal scale for such situations is typically 10 to 100 km. In a summertime
high pressure cell, however, accumulation occurs over a number of days, and
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
128
-------
quite uniform levels of pollutants, well-mixed throughout the atmospheric
boundary layer, may be found over an area of 100 to 1,000 km. At some distance
from the source regions, the precursors may be completely reacted, resulting in
high levels of 03, sulphate, and other secondary pollutants. In Europe, the
long-range transport of oxidants (03, sulphate, nitrate) in connection with high
pressure systems has been widely reported (Atkins et al., 1972; Cox et al.,
1975; Apling et al., 1977; Cuicherit and van Dop, 1977; Schjoldager et al.,
1978).
Theoretical investigations of the residence time of air parcels in a moving
high pressure system show that air masses in a circular high pressure cell with
a radius of approximately 1,000 km and moving at a constant speed of 5 m/s may
spend up to a week in the system (Vukovich, 1979; Vukovich et al., 1977). If
the system moves at slower speeds, the residence time may increase. Based on a
chemical formulation involving nine emitted HCs, photochemical models of such
episodes indicate that 03 generated from continental precursor emissions may
follow the mean air flow over the United Kingdom and remain at elevated levels
for several days (Hov et al., 1978). Over rural areas, 03 in excess of 100 ppbv
(1 part per billion by volume = 2.4 x 101* molecules/cm3) may build up over a
time scale of several days (isaksen et al., 1978). The physical formulation of
these models does not include vertical resolution; furthermore, the polluted air
masses are assumed to be well mixed and represented by boxes. Rather simple
assumptions are made about dilution and transport.
Derwent and Hov (1982) used this approach to investigate the potential for
oxidant formation from United Kingdom emissions of HCs, NOX, and S02, and to
129
-------
evaluate the possibility of long-range transport of secondary species generated
during anticyclonic weather conditions. The following sections provide a
description of the model used by Derwent and Hov, followed by a brief discussion
of some of the results they obtained.
MODEL DESCRIPTION
Physical Model
The most densely populated and industrialized part of the United Kingdom
was selected for study. A stationary model box with the horizontal dimensions
of 450 km x 360 km was situated over South Central England. No horizontal or
vertical grid was introduced; all emissions were assumed to be well mixed
instantaneously throughout the model volume. In anticyclonic weather
conditions, there is a suppression of vertical mixing of the atmosphere above a
certain height, where stable layers are formed. The base of the stable layer
may be as high as 2 or 3 km (Pasquill, 1974), but it is occasionally much lower.
In simulating the long-range transport of pollutants in anticyclonic
weather conditions, the bulk of pollutants in the boundary layer are of
interest, extending vertically to a potential mixing height. The actual mixing
height, however, exhibits a marked diurnal variation, building up from a very
small value at night to a potential value through the day (Pasquill, 1974).
Data from balloon ascents at Cardington in Bedfordshire, England, show
that, when a capped layer exists, the mean inversion height at midday, averaged
130
-------
over the whole year, is 800 ra. During the three summer months, the average
value is 1,300 m. During periods of anticyclonic weather conditions, the
nocturnal mixing height in rural areas is typically a few tens of meters.
The approach used to represent this situation in the model is as follows:
The mixing height is kept constant at 1,300 m day and night. During the day,
both primary and secondary species are deposited at rates corresponding to the
respective deposition velocities. During the night, when a shallow, stable
boundary layer is established, only primary species (i.e., those emitted) are
assumed to be deposited. Secondary pollutants such as 03 and PAN, which have
been generated during the previous day(s), are assumed to have a zero deposition
rate at night. Only a small fraction of the total boundary-layer column of
secondary species is trapped underneath the nocturnal inversion. Although these
compounds may be completely depleted in the shallow layer next to the ground
during the night (Garland and Derwent, 1978), the influence on the total budget
in the boundary layer is minimal (with a 40-m-deep nocturnal boundary layer,
only 40/1300, or 3%, is depleted through deposition), and it was disregarded in
the present model approach.
Horizontal homogeneity is an assumption that is justified only when the
sources of the precursor emissions are evenly distributed spatially. Traffic
and domestic emissions may satisfy this assumption rather well. In cases with
large single sources, such as power stations and oil refineries, a realistic
model picture should look more like an assembly of plumes, which eventually may
interact with each other, rather than a volume in which all emissions interact
all the time.
131
-------
Assuming that complete mixing may occur throughout may therefore seem
unrealistic. However, it is sufficient that the time scales of the various
physical and chemical processes are represented correctly relative to each
other. In the present model formulation, this means that vertical and
horizontal mixing must occur faster than chemical development. It is well
established that the time scale of oxidant generation is several hours or more.
The assumption of instantaneous mixing can therefore be reduced to the
assumption that complete mixing take less than a few hours to achieve. This
assumption may be partially satisfied in photochemical episodes.
Mathematical Formulation
Each model species satisifes the continuity equation, written as:
dC = Pe + Pch - (Lch + Ld)C (1)
dt
where C = the concentration of the compound in question,
Pe = the emission term,
Pch = chemical production and loss,
Lch'C = chemical production and loss, and
Ld-C = the loss rate due to ground removal.
The emission rate is defined as:
Pe = * (2)
H
132
-------
where * = the emission flux of the species in question, and
H = the mixing height.
LdC = Va C (3)
H
where Vd = the deposition velocity (see Table 1). S02 and N02 are assumed to be
removed by dry deposition at night, and the secondary pollutants are assumed to
be unaffected.
No transport term is included in Equation (1), indicating that there is no
interaction with surrounding air masses. Thus, depending on the direction of
the general mean air flow, the predicted oxidant generation may be overestimated
or underestimated. In 03 episodes, there is usually an easterly air flow over
the United Kingdom, carrying high concentrations of precursors and secondary
species from continental Europe (Cox et al., 1975; Apling et al., 1977).
TABLE 1. DEPOSITION
VELOCITES
Species
03
S02
HN03
N02
PAN and homologs
vd
(cm/i
0.6
0.8
1.0
0.1
0.2
133
-------
Emissions
Emissions data for NO, S02, CO, and various HCs (40 species in all) were
input into the model volume (see Table 2). The emissions of the various species
were split into eight source categories: (1) petrol engine motor vehicle
exhaust emissions, (2) diesel engine emissions, (3) petrol engine evaporative
emissions, (4) stationary fuel combustion, (5) solvent use, (6) industrial
processes, except petroleum industry, (7) petroleum industry, and (8) natural
gas leakage. The emissions, which were calculated from the annual rates for
1975, were generally based on a combination of emission factors, statistics of
total fuel consumption, or other relevant data. The detailed procedure and
results are outlined elsewhere (Derwent and Hov, 1979), and the average United
Kingdom emission fluxes for each species are given in Table 2. The emissions
were not given any diurnal variation in the model because the emphasis of the
study was on multiday features, not primarily on effects that occur on a time
scale of a few hours.
Chemistry
The chemical formulation of the model includes approximately 145
intermediate products and end products in addition to the 40 emitted species.
About 300 reactions are required to describe the degradation pathways. A
detailed discussion of the scheme is given by Derwent and Hov (1980). In the
present context, emphasis was placed on how the distribution of compounds within
the main groups of species—sulphur, nitrogen, HCs, 03—changes during the model
run.
134
-------
TABLE 2. AVERAGE UNITED KINGDOM EMISSION FLUXES
Species
NO
S02
CO
CH4
C2H6
C3Hs
n~ C^HI $
i— C4Hi in
n-C5H12
i-CsH12
C2H4
C3H6
C2H2
toluene
o-xylene
ra-xylene
p-xylene
ethyl benzene
HCHO
Emission
Flux
(molecules/
cm/s)
2.81 x 1011
4.23 x 1011
2.94 x 1012
1.22 x 1012
4.07 x 1018
8.64 x 109
2.19 x 109
9.88 x 109
1.80 x 1018
3.20 x I01fl
2.56 x 101
-------
the initial attack on a given HC molecule. Radical species are generated and
regenerated at various stages in the HC decomposition chains and on various time
scales. Secondary species such as aldehydes are formed; these are important
free radical sources when photodissociated. Increased radical concentrations
are an important feature of oxidant episodes. They favor rapid generation of
secondary pollutants and shorten the time scale for precursor degradation.
Hydrocarbons are oxidized to CO and C02 by way of aldehydes, ketones,
glyoxals, etc., as intermediates that are also eventually coverted into stable
products.
Oxides of nitrogen are converted into secondary products such as HN03 and
PAN and its homologs through reactions such as:
OH + N02 - HN03
CH3C002 + N02 - PAN
.C2H5C002 + N02 -> PPN
where the peroxyacetyl radicals, CH3C002, are formed from acetaldehyde by:
OH + CH3CHO - CH3C002.
136
-------
PAN is thermally unstable (Cox and Roffey, 1977); that is,
PAN - CH3C002 + N02 (7.9 x 10'4 exp (-12530/T))
and the same decomposition mechanism is assumed for all PAN homologs.
PAN and its homologs are also removed by ground deposition, which can be
estimated by using Equation (3) and the data in Table 1. The loss of gaseous
HN03 to the aerosol is considered to be slow, according to evidence reported by
Brosset (1978), and is set to a rate corresponding to a characteristic time of
2 days (Derwent and Hov, 1979).
The chemical conversion of S02 to sulphate aerosol is assumed to take place
through the following reactions:
Oil + S02 - HS03 1.1 x 10~12 (Calvert et al., 1978)
CH302 + S02 - S03 + CH30 5.3 x 1CT15 (Kan et al. , 1979)
H02 + S02 - <1 x 10~18 (Graham et al., 1979)
02 + HS03 - HS05
HS05 + NO -> N02 + HS04
HS04 + 02 ... - H02 + H2S04
S03 + H20 - H2S04
It should be noted that the recent data evaluation (NASA, 1981) recommends a
value of less than 5 x 10~17 cm3/molecule/s for S02 + CH302. The gas-phase
reactions of S02 compete with dry deposition, which is estimated by using
137
-------
Equation (3) and the deposition velocity given in Table 1. Droplet-phase
mechanisms for S02 oxidation by 02.03 and H202 were not considered in this
study. No removal processes for nitrate or sulphate aerosol were included in
the model.
Numerical Procedures
The system of differential equations describing the time evolution of the
chemical species in the model have been solved using a quasi-steady-state
approximation (QSSA) method (Hesstvedt et al., 1978). The accuracy of the
results have been assessed by comparing runs made with the Harwell computer
program FACSIMILE (Chance et al., 1977). This program employs a variable-order
Gears method. It is highly suited to the integration of large, stiff systems of
differential equations, and has an error limit of 0.1%. The agreement between
the QSSA method and FACSIMILE was better than 1% for most species, and the
required CPU time with QSSA was about one-third that for FACSIMILE. A typical
run using FACSIMILE on the IBM. 3081 required about 2 min of CPU time. The
computational error limit on the results presented here is thus about 1%.
A fully diurnal sun corresponding to summer, 50"N, was modeled. The
dissociation rates were calculated by using the scheme developed by Isaksen et
al. (1977). A value for the diurnal variation in temperature (daily maximum of
25°C, daily mean of 17°C) and a value for relative humidity were also included.
Relative humidity was given a maximum value close to 85% around dawn and a
minimum of 45% around noon. The initial conditions were established by midnight
138
-------
on the first day by running the model from noon till midnight on day "zero",
using the same set of model parameters specified for the main run.
RESULTS AND DISCUSSION
The potential for oxidant formation in a typical, fairly persistent high
pressure cell was evaluated by integrating the model equations over a period of
4 days with emissions. The emission fluxes were then set to zero, and the model
integration was continued for 3 days. The decay of the various species during
this period allowed an assessment of their respective lifetimes and demonstrated
the possibilities for long-range transport.
Ozone
Figure 1 shows how 03, which was of primary interest, accumulated to reach
130 ppbv on the fourth day. A maximum in the net gas-phase generation of 03
occurred on the second day, with gradually more and more being deposited due to
the increased 03 level. Even after emissions have been abolished, a substantial
03 production continues through days 5 and 6, demonstrated by the widening
divergence between the 03 curve and the dotted line for deposition only and
calculated bv assuming zero production and loss only by ground removal. On the
seventh day, the slopes of the two curves are similar, indicating that the
decline in 03 is then mainly due to deposition. The results shown in Figure 1
clearly demonstrate the long lifetime of 03 in old, weakly polluted air masses.
This supports the previous evidence on the long-range transport of 03 (Cox et
139
-------
\
\
O
2 4->
0) tO
M ~0 •-<
(0 >
C r-f
•H i-l
CO
>%
S (0
Q.T3
O
r-( M
ai ai
> t->
ai uj
a co
C
CO
•H ><
5 .-I
n o
ai
j= c
•u O
a; -H
GO 4->
o -H
*j <«
o
- Q.
p ai
3 -a
to oo
o o
10 tJ
^-1 J=
CO 4-1
a> c
>-i O
CO -H
4->
(0 01
C fH
o a
•r^ 0>
m -o
IA
•H t-(
e o
a* u-i
3
60
140
-------
al., 1975; Apling et al. , 1977; Hov et al., 1978). Such transport may be
particularly important over sea surfaces, where the 03 deposition is relatively
slow.
Nitrogen Budget
Figure 2 shows that total gas-phase N species decrease with time,
acompanied by an increase in nitrate aerosol and deposition of N species.
Figure 3 shows the relative distribution of the gas-phase N species with time.
Primary species (NO and N02) are replaced by secondary species (HN03, PANs), and
it is evident that PAN and its homologs are important N02 carriers in aged air
masses. Nitrate aerosol formation from HN03 is given a characteristic time of
2 days in the photochemical episode modeled here. Because the recycling of NOX
through dissociation of HN03 is slow, with a characteristic time of 10 days
(noon dissociation rate), HN03 is an efficient sink for NOX, while PAN and its
homologs are only temporary sinks.
Hydrogen Budget
The HC budget is shown in Figures 4 and 5. There is a decline in total
concentration from the fourth to the seventh day since TNMHC is defined here as:
TNMHC = ^olefins + iparaffins + ^aldehydes + laromatics + acetone + ketone
+ acetylene + lialcohols,
specifically excluding CO, PAN, and PAN homologs.
141
-------
to
CO
60
•• CO
Si C
•H O
4J
u u
0> CO
a. n
« y-i
oo
•H -H
C 4->
•H CO
CO ^
4-1
•H O .-I
T) Q. O
o) e
Ol T3 v_x
•H • 0>
4-1 QJ B
CO W 3
--I CO ^-1
CM
00
•H
142
-------
V)
>,
a
•a
4)
.a
E
•H
u
0)
ex
w
oo
c
c
o
o
I
c
ai
oo
o
c
o>
(0
a,
i
in
nj
M-4
O
O
•rl
4-1
•H
T3
>
•H
4-1
<0
CO
0)
d
oo
143
-------
Figure 4 shows there is a general increase in the relative fraction of
species with low reactivity (paraffins, C2H2, CH3COCH3), while olefins and
aromatic compounds virtually vanish. The relative distribution of the paraffins
(Figure 5) shows the same trend: The lower reactivity fractions dominate more
and more with time.
These calculations reveal that different mechanisms become important when
the multiday formation of oxidants on a regional scale is compared with oxidant
formation on an urban scale. During the first hours in a moderately polluted
air mass, the behaviour of the oxidant formation is quite similar to that for an
urban situation. After a few days, however, the less reactive HCs have
accumulated and contribute a relatively larger share to the oxidant generation
than the olefinic or aromatic HCs. Once emissions are abolished, the long-lived
species assume an even greater role. In an aging, moderately polluted air mass,
deposition is the most important constraint on the 03 accumulation.
Also,calculations have shown that the photochemical lifetime of 03 in old air is
very long, clearly demonstrating the potential of long-range transport from the
United Kingdom.
Model Validation
Typical concentrations for the fourth and seventh days of integration of
certain species are shown in Table 3. The fourth day concentrations represent
overall maxima for low-reactivity species such as n-CAH^, HNOa, and aerosols;
other species reach approximately the same concentrations on the second and
third day. The decline from the fourth to the seventh day reflects the chenical
144
-------
I/)
>s
o
.a
E
01
00
•o
0)
.e
o
w
•H
a>
>
•H
a>
M
3
60
•H
iu
145
-------
(A
>»
O
c
•1-1
o.
O)
B
O
3
•H
W
•H
TJ
o
W)
U)
i
IU
-------
ij
f^l
5
OS
»y
M
W
Q
s
CO
o
H
^
OS
W
CO
pa
o
EC
H
M
DC
00
2
O
M n
H Q
< 2
OS <
H J
2 O
W 2
U W
2
O 2
W
W H
H D
< O
i-J CO
g
J
£
M
O
s
En
o
r«2
o
to
M
OS
AJ
53 1
0 1
CJ 1
1
.
CO
W
i_3
PQ
H
"T3
...
>-<
in
ra
0)
25
CO
CO
rH
O
^ J
to
o
111
CJ
B
CO
r-1
CO
U-l
CO
OS
B
o
>r_j
to
l-i
j_j
p
0)
E
o
O
^
CO
T3
^
4J
Pv
X
CO
^
4-J
~*
10
co
•H
0
CU
a
to
*£>
ps.
rH
•.
.
rH
CO
4-1
CO
J^l
CO
B
CO
CM
i
o
rH
X
CO
CM
^
1
O
rH
X
ro
rH
O
rH
X
CM
CO
§
r^. PX
ty\ r\. r**
rH O^ ^ fV,
Px rH ^
•* C^ \£vO\£NO \£> \O \D v^rH
^H O^ O^ O^ O^ CT^ C5^ O^ • CT^ •*
(Q „ r-lr-jr-trH rHi-HrHrH rH*
CO rH
rHrHrHrH rHrHrHCO rH4J
rH *-> CO CO (0 CO CO (0 CO CO XXXX XXXB X rH
to co o o o o oooco o a.
O Q CJCJOO OOCJO^ CJ<
4J •» O »
« o w b 2
1 rH •» rH <
O O CM
rH X rH X ON
• IM
X \£> X P~> CO O
1 O
CO rH •& tO
OOOs*OrH«*coO«OO«»
OCMOOrHOOOOOrHsO
ON
OCOO^rsO^vOON^cN
OrHrHOCMrHCOOrHOvDO
CO
rH
*~ O
K _ ffi
•* * O O «
_ . _, F * fc_3 »*ri ^« *j +^ *y f\
OJ «M W [.TJ M^ ~ Wl F-+ fZ* \J
O O 1 *M CJ HM ^? O ^3 f*4 2 O
ON
ps.
ON
rH
O
03
•T"1
^3
C
CO
4-*
B
CU
M
CO
o
o
1
to
Ps.
^~
s*
rH
o
M
O
r-l
CO
CO
•o
•H
O
CO
O
•Jj
4-1
CO
ps.
ON
rH
•
rH
CO
OJ
10
B
•H
<
0
^
1-1
Ps,
vO
S
o
rH
0
10
o
M
CU
CO
CU
4-1
CO
rC
a
rH
3
00 1
1
1
1
1
1
1
1
1
1
i
1
1
1
I
1
1
.
n
e
~60
*
B
•H
•
^^
u">
(X.
rH
••
"
rH
to
4-»
CO
^
OJ
•g
QJ
PH
N^X
CO
• •
rH
O
4-1
VO
rH
"c
0
X
rH
CO
CO
3
to
•H
•rj
B
CO
in
jvJ
B
CU
e
CU
M
3
10
CO
cu
£
B
n
s
00
a.
B
M
u
147
-------
lifetimes of the various species. In general, primary species such as NOX,
and reactive HC almost vanish, while 03, PAN, and the PAN homologs remain at
elevated levels.
Table 3 also shows measured concentrations from these trace species. These
were determined at a rural site in southern England (Harwell), except for OH
measurements made in Juelich, West Germany (Perner et al., 1976), in Tennessee,
(Cambell et al., 1979), and in New Mexico (Davis et al., 1979). Ambient
measurements made in the United Kingdom exist for only a limited number of
species, and these are restricted to only a few sampling locations. In spite of
the numerous assumptions and simplifications involved, the modeled calculations
yielded concentrations that do at least appear to be realistic. Any refinement
of the model will require improved validation data for precursors,
intermediates, and secondary products.
SUMMARY
The results from this study indicate that photochemical oxidant formation
in a persistent high pressure cell can be approximated by using a simple
box-model formulation applied to United Kingdom precursor emissons data. The
photochemical lifetime of 03 in air that has left the source area is
approximately 10 days, being determined principally by deposition. This
demonstrates the potential for the long-range transport of 03, particularly over
surfaces such as water, where the deposition is inefficient. The calculations
also show that reactive HCs are more important to photochemical oxidant
148
-------
generation on a regional scale than in an urban situation. PAN and its homologs
are shown to be important carriers of photochemically active NOX in aged air
masses.
The simplified model described here is an initial attempt to simulate, on a
regional basis, the general features of photochemical oxidation in the United
Kingdom. It is not immediately applicable to the representation of oxidant
levels experienced in the United Kingdom during typical episodes, when transport
from continental Europe becomes important. The model runs can be regarded as an
extreme case of a trajectory model, in which the air parcel is advected over a
constant emissions field for a period of several days. Further work will
require the construction of spatially and temporally resolved emissions
inventories and the use of actual trajectories for past episodes to allow a more
complete model assessment.
ACKNOWLEDGMENTS
This work was supported by the United Kingdom, Department of the
Environment.
REFERENCES
Apling, A. J., E. J. Sullivan, M. L. Williams, D. J. Ball, R. E. Bernard, R. G.
Derwent, A. E. J. Eggleton, L. Hampton, and R. E. Waller. 1977. Ozone
concentrations in south-east England during the summer of 1976. Nature,
269:569-573.
Atkins, D. H. F., R. A. Cox, and A. E. J. Eggleton. 1972. Photochemical ozone
and sulphuric acid aerosol formation in the atmosphere over southern
England. Nature, 335:372-376.
149
-------
Brosset, C. 1978. Water soluble sulphur compounds in aerosols. Atmospheric
Environment, 12:25-38.
Calvert, J. G., F. Su, J. W. Bottenheim, and 0. P. Strausz. 1978. Mechanism of
the homgeneous oxidation of sulphur dioxide in the troposphere.
Atmospheric Environment, 12:197-226.
Campbell, M. J., J. C. Sheppard, and B. F. Au. 1979. Measurement of hydroxyl
concentration in boundary layer air by monitoring CO oxidation.
Geophysical Research Letters, 6:175-178.
Chance, E. M., A. R. Curtis, I. P. Jones, and C. R. Kirby. 1977. FACSIMILE: A
computer program for flow and chemistry simulation, and general initial
value problems. AERE-R 8775, Her Majesty's Stationery Office, London.
Cox, R. A., and M. J. Roffey. 1977. Thermal decomposition of peroxyacetyl-
nitrate in the presence of nitric oxide. Environmental Science and
Technology, 11:900-906.
Cox, R. A., R. G. Derwent, and F. J. Sandhalls. 1976. Some air pollution
measurements made at Harwell, Oxfordshire during 1973-1975. AERE-R 8324,
Her Majesty's Stationery Office, London.
Cox, R. A., A. E. J. Eggleton, R. G. Derwent, J. E. Lovelock, and D. H. Pack.
1975. Long range transport of photochemical ozone in north-western Europe.
Nature, 255:118-121.
Davis, D. D., W. Heaps, D. Philen, and T. McGee. 1979. Boundary layer
measurements of the OH radical in the vicinity of an isolated power plant
plume: S02 and N02 chemical conversion times. Atmospheric Environment,
13:1197-1203.
Derwent, R. G., and 0. Hov. 1982. The potential for secondary pollutant
formation in the atmospheric boundary layer in a high pressure situation
over England. Atmospheric Environment, 16:665-665.
Derwent, R. G., and 0. Hov. 1979. Computer modelling studies of photochemical
air pollution formation in north-west Europe. AERE-R 9434, Her Majesty's
Stationery Office, London.
Garland, J. A., and R. G. Derwent. 1978. Destruction at the ground and the
diurnal cycle of concentration of ozone and other gases. Quarterly Journal
of the Royal Meteorological Society, 105:169-183.
Graham, R. A., A. M. Winer, R. Atkinson, and J. N. Pitts, Jr. 1979. Rate
constants for the reaction of H02 with H02, S02, CO, N20, trans-2-butene
and 2,3-dimethyl-2-butene at 300K. Journal of Physical Chemistry,
83:1563-1567.
Guicherit, R., and H. van Dop. 1977. Photochemical production of ozone in
Western Europe, 1971-1975.
150
-------
Hcsstvedt, E., 0. Hov, and I. S. A. Isaksen. 1978. Quasi-steady-state
approximations in air pollution modelling: Comparison of two numerical
schemes for oxidation prediction. International Journal of Chemical
Kinetics, 10:971-994.
Hov, 0, E. Ilesstvedt, and I. S. A. Isaksen. 1978. Long range transport of
tropospheric ozone. Nature, 273:341-344.
Isaksen, I. S. A., 0. Hov, and E. Hesstvedt. 1978. Ozone generation over rural
areas. Environmental Science and Technology, 12:1279-1284.
Isaksen, I. S. A., K. H. Midtbo, J. Sunde, and P. J. Crutzen. 1977. A
simplified method to include molecular scattering and reflection in
calculations of photon fluxes and photodissocation rates. Geophysica
Norvegica, 31(5):ll-26.
Kan, C. S., R. D., McQuigg, M. R. Whitbeck, and J. G. Calvert. 1979. Kinetic
flash spectroscopic study of the CHaC^-CHaOa and CH302-SC>2 reactions.
International Journal of Chemical Kinetics, 11:921-933.
National Aeronautics and Space Administration. 1981. Chemical kinetic and
photochemical data for use in stratospheric modelling. Evaluation
Number 4: NASA Panel for Data Evaluation. NASA Jet Propulsion Laboratory,
Pasadena, California.
Pasquill. P. 1974. Atmospheric Diffusion. Ellis Horwood, Ltd., Chichester.
Penkett, S. A., F. J. Sandalls, J. E. Lovelock. 1975. Observations of
peroxy-acetyl nitrate (PAN) in air in southern England. Atmospheric
Environment, 9:139-140.
Perner, D., D. H. Ehhalt, H. Q. Patz, U. Platt, E. P. Roth, and A. Volz. 1976.
OH radicals in the lower troposphere. Geophysical Research Letters,
3:466-468.
Schjoldager, J., B. Sivertsen, and J. E. Hanssen. 1978. On the occurrence of
photochemical oxidants at high latitudes. Atmospheric Environment,
12:2461-2569.
U.S. Environmental Protection Agency. 1975. Investigation of rural oxidant
levels as related to urban hydrocarbon control strategies.
EPA-450/3-75-036, Research Triangle Park, North Carolina.
Vukovich, F. M. 1979. A note on air quality in high pressure systems.
Atmospheric Environment, 13:255-265.
Vukovich, F. M., W. D. Bach, Jr., B. W. Crissman, and W. J. King. 1977. On the
relationship between high ozone in the rural surface layer and high
pressure systems. Atmospheric Environment, 11:967-983.
151
-------
DISCUSSION
E. Runca: If I understand, your model is a one-dimensional model?
K. Brice: That is right.
E. Runca: There is almost no observation of meteorological factors affecting
the development of photochemistry.
K. Brice: That is partially true. We are assuming a stationary box across
southern England.
E. Runca: When you refer to a chemistry time scale of 10 h, exactly what do you
mean, the time scale of the oxidants?
K. Brice: This is really based upon an assessment of an urban-plume-type
approach where you look at the formation of, say 03 or PAN, to some extent as
well. However, I was interested in this major point for this discussion.
The time scale is really time taken as the maximum to be reached in that plume,
and it is about 10 h. It perhaps actually balances between the formation rate
and the k rate in that plume.
E. Runca: I think you would agree with me that the cycle of pollutants when
they meet in the mixing layer are strongly affecting the sequence of reactions
and the generation of secondary pollutants.
K. Brice: I agree with that point.
Unidentified Speaker: Your statment that, during the course of the multiday
stagnation, the light HCs were actually increasing was unclear. I understand
where the relative contribution would increase, but I was never aware that the
total concentration of the HCs increased. Is that what you were saying?
K. Brice: No, the type of HC concentrations will obviously slightly decrease.
Unidentified Speaker: Was that just the relative contribution?
K. Brice: That was the relative contribution, right.
Unidentified Speaker: Also, I think you quoted Fred Vukovich from RTI. He
indicated that the pollutants remain in the high-pressure system about 6 to
7 days, but that is the exception. I think ordinarily he is saying 3 to 4 days.
152
-------
APPLICATION OF A REGIONAL OXIDANT MODEL
TO THE NORTHEAST UNITED STATES*
James P. Killus
Ralph E. Morris
Mei-Kao Liu
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, California 94903 (USA)
INTRODUCTION
A major concern over regional air quality is the long-range transport of 03
and its precursors from upwind sources to rural areas. During several studies,
03 concentrations exceeding the air quality standards have been observed over
widespread areas beyond urban centers in the Eastern United States. Such
incidences have also been reported elsewhere in the United States and in Western
Europe. The movements of areas of high 03 concentrations, frequently associated
with increasing surface temperatures and decreasing visibility, generally
correspond to movements of synoptic-scale high pressure systems. Many
trajectory analyses seem to further implicate the long-distance transport of
oxidants and oxidant precursors. For example, urban plumes emanating from
St. Louis were tracked 160 km or more downwind. Air masses containing high 03
concentrations over the Atlantic Ocean 250 km east of New York City were traced
back to the metropolitan area. Similar observations were made in the Midwest
and in California.
*This report has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
153
-------
Current control strategies for photochemical oxidants are directed
primarily at emission sources in the general vicinity of an urban area where
excessive 03 levels are observed. Thus, the possibility of 03 or its precursors
being transported from distant upwind sources certainly compounds emission
control designs to reduce photochemial oxidants.
In order to arrive at an effective control strategy, contributions from
local and distant sources must be quantified. Mathematical models have emerged
as the most viable means for this assessment, based on the complex physical and
chemical processes involved on this scale.
The objective of this paper is to describe an application and evaluation of
a regional photochemical air quality model (RTM-IIl) developed by Systems
Applications, Inc. Evolved from a regional transport/dispersion code, this
model has undergone several iterations of improvements and modifications. The
present application focuses on a regional air pollution episode recorded in
July 1978 during the Sulfate Regional Experiment (SURE). A statistical
evaluation of model predictions and observations has been completed.
MODEL EQUATIONS
The regional transport model evaluated in this paper is a descendant of a
two-dimensional model developed by Liu and Reynolds (1983). This model, the
oxidant version of the Regional Transport Model oxidant version (RTM-IIl), is
154
-------
based on the time-dependent, raultispecies atmospheric diffusion equation for
multiple layers:
ac, ' + u ac,' + v ac,' = a_ /Kx ac,' \ + a_ (Ky ,
-------
through the surface layer to the ground, followed by absorption or adsorption at
the atmosphere/ground interface. The flux of pollutants of species i reaching
the ground can be expressed in terms of the mixed layer concentration c by the
expression:
(2)
I + (1/Ti)
where
hs ,t>( z .
I = J_ + / l L' dz , (3)
0u- z0 ku-z
and
u- = friction velocity,
k = the von Karman constant,
•i, = the velocity profile function,
L = the Monin-Obukhov length,
hs = the height of the surface layer,
z0 = the roughness length, and
T, = the surface reaction rate constant that, for the first-order reaction,
has dimensions of velocity.
Equation (2) can be derived by balancing the flux through the surface layer with
the flux through the surface layer/viscous sublayer interface and the removal
rate at the ground. The parameter 0, analogous to the Stanton number in heat
156
-------
transfer, is the inverse of a dimensionless resistance for the viscous sublayer.
Following Durran et al. (1979), 0 is given in the present study by:
& = u-1/3 . (4)
2.2
This parameterization is more realistic than those commonly used in
regional transport models, because it directly takes into account
stability-dependent concentration profiles near the surface. The sink term in
Equation (1) is related to the pollutant flux fi and the mixed layer depth h by:
if j = 1
S,j = <( (5)
if j = 2 .
An equation for the pollutant loss per unit of area per unit of time can be
written as (Benson, 1968):
f, = T, C|s (6)
where f, is the pollutant flux, C,s is the surface concentration of species i,
and T, represents surface reaction rate constants. Combining this equation with
Equation (2), we can thus determine the surface concentation C,s from the
average concentration in the mixed layer c by:
c . (7)
I-Ti
157
-------
Chemical Kinetic Mechanism
The chemical kinetic mechanism used in the regional-scale photochemical air
quality model is based on the Carbon-Bond-II mechanism (Whitten, Killus, and
Hogo, 1980). This mechanism is expressly carbon conservative, a feature of
particular importance in situations involving extended transport and residence
times. Hydrocarbon emissions used in the model are divided into five
categories, according to the individual bonding structure of carbon atoms within
each molecule. The carbon bond categories are: single-bonded carbon (PAR),
ethylene (ETH), aromatic rings (ARO), and carbon-fast (OLE) and double-bonded
carbonyl (GARB) groups.
The numerical solutions of the rate equations for chemical kinetic
mechanisms generally encounter the stiffness problem. Stiff systems are defined
as those with widely differing time constants. Classical methods for the
solution of differential equations require a time step sufficiently small to
avoid instability for the smallest time constant and may impose an enormous
computational burden. Although a numerical algorithm based on the method of
Gear (1971) can be used, the introduction of discontinuities into the forcing
function of a Gear solution—such as changes in boundary conditions-can cause
major inefficiencies in computation. These inefficiencies, together with the
additional storage requirements for a high-order predictor-corrector method,
make such a scheme unattractive in a grid model.
As an alternative, the invocation of the steady-state approximation can be
used as an important tool for the numerical solution of chemical rate equations.
158
-------
A transformation from differential equations to algebraic solutions allows the
reduction of the number of species requiring numerical solutions. The species
for which the steady-state approximation is valid are those with the smallest
time constants. Thus, an elimination of these species greatly reduces the
stiffness of the ordinary differential equations. The complexity of the
algebraic solutions increases as more steady-state species that react with each
other are included. For example, the implementation of the steady-state
approximation in the Carbon-Bond-II mechanism requires the solution of a quintic
polynomial for the selected radical species. One problem concerning the use of
the steady-state approximation is that its use in effect removes that species
from mass balance considerations; i.e., the steady-state expression assumes that
y = 0. If jydt is very small, then the removal of this species will not affect
the overall mass conservation. If, however, .fydt is sufficiently greater than
zero, so as to affect the overall mass balance, then the steady-state assumption
may produce invalid results.
This limitation has presented a potential problem in the use of the
steady-state relationship for simulating major species such as NO, N02, and 03.
Generally, these three species are in dynamic equilibrium, with a time constant
much faster than the rest of the nonradical chemical kinetic system. The
dynamic equilibrium established is very close to:
K3[03][NO] = K,[N02] (8)
Unfortunately, mass exchanges involving NO, N02, and 03 are not small, and the
conventional form of the steady-state approximation cannot be used. However,
159
-------
the simple steady-state values of NO, N02, and 03 my be modified in terras of a
correction factor:
ASS = - 1 [03] + [NO] + K + 1 t03] + [NO] +
- 4 [NO][031 - K [N02]
where
[N02]ss = [N02] - ASS
[NO]S5 = [NO] + ASS
[03]ss = [03] + Ass
If the state is variable, such as NOX (=NO + N02) and unpaired oxygen atoms
Ox (=03 + N02), the correction-factor equation becomes:
. 2
(10)
*> I
1/2
- A [NOJ [OJ
where
(N02]ss = - ASS
[N0]ss = NOX +
I03]s$ = Ox + a,
160
ASS = - 1 [Ox] + [NOX] + K, + 1 I (OJ + [NOJ + K,
-------
Mass conservation is thus maintained for the two redefined species, NOX and
Ox. The steady-state calculation does not affect the quantity of these two
species, it merely apportions them into the three molecular species—03, NO, and
N02. In this scheme, NO-to-N02 conversions become the source of unpaired oxygen
atoms.
This scheme has been tested against the Gear solution for a variety of
cases. As might be expected, the approximation tends to break down when peroxyl
radical concentrations are large relative to 03 (i.e., when there are very high
HC concentrations or very low NOX concentrations). The former condition is not
likely to be encountered, given the resolution of a regional-scale model. Under
the latter condition, photochemical 03 production is very slow relative to the
background. Thus, the approximation appears to be reasonable for regional-scale
applicat ions.
APPLICATION OF THE MODEL TO THE NORTHEAST UNITED STATES
A rigorous model evaluation is predicated on a comprehensive data base that
consists of meteorological data, an emissions inventory, and a spatially and
temporally dense air quality monitoring network. The Regional Transport Model
(RTM-III) described in the previous section has been used to simulate an episode
of relatively high regional 03 concentrations. The episode occurred over the
eastern third of the United States during July 16 to 23, 1978. In addition to
ample data coverage, the selection of this particular episode was also motivated
by similar modeling exercises also focusing on the same period (Lavry et al.,
1980; Niemann and Young, 1981; Bhumralkar et al., 1981; Stewart et al., 1983).
161
-------
The modeling region selected for this study is almost identical to the
EPRI/SURE grid, which is characterized by 80-km mesh squares defined over the
eastern third of the United States. Dimensions of the grid subset are 2,080 km
in the east-west direction by 1,840 km in the north-south direction. The grid
resolution selected for the model simulation is 40 km x 40 km.
Meteorological Data
Wind velocities and mixing depths were derived from the National Weather
Service (NWS) radiosonde network. The spatial resolution of the network is
approximately 200 km, whereas temporal resolution is 12 h. Data from the
monitoring network are spatially and temporally interpolated to generate the
necessary wind field on the 80-km grid and at 3-h intervals, as required by the
model.
For this modeling application, transport winds were defined as the
layer-averaged wind velocity between the ground and 1,500 m. This transport
wind was calculated for each radiosonde station and observation time. Linear
interpolation was then employed to compute the transport winds at intermediate
3-h intervals. The wind velocity at the center of each model grid was then
determined from an inverse-distance-squared interpolation of the irregularly
spaced radiosonde data.
The procedures for determining the depth of the mixed layer are similar to
those for determining the wind field. Following Benkley and Bass (1979), the
mixed-layer depth h is determined at 3-h intervals from linear interpolation
162
-------
between a nighttime minimum hmin and a daytime maximum hmax. The minimum depth
is determined from hmin = 53|v|, where |v| is the magnitude of the
layer-averaged wind. The maximum depth hmax is derived from the morning
sounding at height where the potential temperature is equal to the maximum
surface temperature of the following afternoon. Across the modeling region, a
minimum value of 200 m was imposed on the mixed-layer depth.
Although the model has finer resolution than the radiosonde network, only
those atmospheric features resolvable by the network can clearly influence the
concentration pattern simulated by the model. It should also be noted that
winds and mixed-layer depths are updated at 3-h intervals in a stepwise manner
in the model. No temporal interpolation on the scale of an integration time
step (i.e., approximately 20 min) has been attempted.
Air Quality Data
Air quality data used in this study consisted of ground-level 03, NO, N02,
S02, and sulfate concentration measurements, collected in the EPRI Sulfate
Regional Experiment (EPRI/SURE). A detailed description of this data base can
be found in Muller et al. (1982). The ground-level monitoring stations for S02
and sulfate consisted of 45 Class I and 9 Class II stations. The ground-level
monitoring stations for 03, NO, and N02 consisted of eight Class I stations
operating continuously during the modeling period.
163
-------
Emission Data
The EPRI/SURE program also established a detailed emissions inventory for
the study region. Anthropogenic emissions used in this study were compiled in
the following manner (Klemm and Brennan, 1981): Seasonal mean emission rates
were derived for various pollutants in each of several source categories (i.e.,
electric power plants, major industries, commercial industries, home heating,
and surface transportation). For each category, average temporal emission
variations were calculated for eight 3-h periods, considering factors such as
weekend/weekday and time of day. Emissions from all categories were then
combined into a major point-source file (including stack parameters) and an
area-source file aggregated into an 80 km x 80 km grid. Emission estimates
prepared in this manner, derived from the National Emissions Data System (NEDS)
of the U.S. Environmental Protection Agency, various state agencies, and the
provinces of Ontario and Quebec, are current through July 1977.
The SURE emissions inventory consists of more than 3,000 point soures
considered as major sources (>10,000 tons of SOX per year). To treat this
number of point sources in an economical manner, the emissions were aggregated
into 498 point sources. This was accomplished by combining all the emissions
from point sources with common plant codes and assigning the aggregated
emissions to a source with the highest NOX emissions rate.
Point and area source emission rates were available from the SURE emissions
inventory for S02, sulfate, NO, N02, high-reactive HCs, medium-reactive HCs, and
low-reactive HCs. The NO and N02 emissions were combined to form the NOX
164
-------
emission rate. The low-reactive HC emissions were low-reactive and nonreactive
compounds and were not included in this study. Both the medium- and
high-reactive HC emissions were composed of different reactive compounds, and
were combined and split into Carbon-Bond-II mechanism compatible compounds as
listed in Table 1. The total emission rates for the entire modeling region are
divided by point and area source categories for the seven emitted species in
Table 2. The spatial distributions of NOX and reactive HC emission rates for
the modeling region are displayed in Figures la and Ib.
Other Model Input Data
Exercise of the RTM-III requires additional information such as estimates
of the photolysis rate constants and the dry deposition rates for 03, NO, N02,
SO?, and SCu. The photolysis rate constant was diurnally and spatially varied
as a function of the solar zenith angle.
The parameterization of dry deposition rates requires a characterization of
the underlying surface. This is needed for estimating both the diffusion toward
the surface and the absorption rate. Table 3 lists the representative surface
roughness and the deposition velocities used for each type of surface
encountered in the model simulation. The geographical distribution of different
surface categories over the modeling region is illustrated in Figure 2.
165
-------
TABLE 1. HYDROCARBON SPLIT
USED FOR HYDROCARBON
EMISSIONS"
Normalized
Fraction
Compound (%)
ETH 5
OLE 3
ARO 22
PAR 65
CARB 5
Total 100
3From Killus and Whitten,
1981.
TABLE 2. TOTAL EMISSION RATE FOR THE MODELING REGION (tons/day)
Category NOX ETH PAR CARB ARO S02 Sulfate
Major point sources 18,180 193 2,510 309 850 61,880 1,665
Area sources 34,207 2,020 26,255 3,231 8,887 20,633 1,021
Total 52,387 2,213 28,765 3,540 9,737 82,513 2,686
166
-------
in
*
8
in
in
v
in
in
v
g
c
O
•H
to
VI
•l-l
E
0)
•O
OJ
-d
•a
-------
8
in
10
v
s
0)
C
CO
CO
0)
OJ
oi
tn
oo
C
O
in
•H
B
-------
1
1
Jl
II
il
W
PH
H
z
M
^
OS
>^
•H
U
O
I— (
> a
e 10
o -~
•H e
•M U
•H N_s
U)
o
a
a>
OS II Q
W II
H !i
>• H
PQ
Z li
O II
M I
H
M
OS
W
H
O
OS
•<
5
W
0
fa
g
CO
•
ro
U
r—3
PQ
^^
H
II
o
to
CM
O
to
o
OJ
o
o
o
o
„
to
^-
•
•H
rH
d
o
rH
•
rH
O
<3\
O
CM
•o
c
a
4,j
in
V
o a
*4-4 (Q
rH
TJ -0
V O
N O
<0 *
to
O
O
in
0
o
m
rH
in
«^f
o
0
o
in
•
rH
rH
O
CO
•
rH
O
o
rH
•*
^
I-l
C
(0
4-1
10
(U
to
o
e
•o
-------
E
O
•H
00
0)
t-l
00
S
0>
T3
O
e
4-1
«
a,
0)
-------
MODEL EXERCISES FOR THE JULY 1978 EPISODE
RTM-III was exercised for a period spanning eight consecutive days
(July 16-23, 1978), the same period analyzed by Stewart et al. (1983) using the
S02/sulfate version of the Regional Transport Model (RTM-II). Detailed
discussions on the prescription of the meteorological input to the model can be
found in Stewart et al. (1983). A brief synopsis of the 8-day sulfate episode
selected for model evaluation is presented in Figure 3. The surface weather
maps are representative patterns for 1200 GMT (0700 EST) of each day. Prior to
July 17, 1978, a ridge of a high pressure system extended southward behind an
advancing cold front. During July 18 and 19, 1978, the cold front slowly moved
offshore, and the surface high pressure intensified to 1,020 mbar. The high
pressure system drifted southward and expanded over the northern Atlantic Ocean.
The eastern and central states remained under the influence of the high pressure
system, which merged over the next few days with the Burmudian subtropical high.
During this period, air flow at 500 mbar exhibited a zonal-oriented weak ridge
that gradually evolved into a closed circulation over the eastern states.
Toward the northern and western portions of the modeling region, a cold
front advanced and became stationary while the high pressure system to the east
was strengthening. A low pressure system formed on this front over Iowa during
the morning of July 22, 1978. This low, with its accompanying cold front,
advanced across the modeling region late on July 23, 1978, displacing the high
pressure system.
171
-------
O
•H
OQ •
a> x-s
M 60
e
00 -H
C T3
•H CD
r-l J=
0> ^
B .0
-C 13
4J 0)
•H 4J
3 (0
u
01 M
CO
V
4_i o
o> B
<4-l — •
^H 00
3 a.
-< 01
3 C
•d x
0)
(A
a, to
co c
o
•H
U U
•r-l CO
•u U
a, u
o c
c ai
^ o
(0 C
o
Ol U
o
CO O>
y-i u
l-i (0
3 lu
e
01 T3
u a>
c c
a> -H
3 iH
cr u
0) 3
10 O
CO
0)
M
3
00
•H
172
-------
During Julv 17 and 18, 1978, winds in the lowest 1500 m were generally
light and northerly to the east and southeasterly to the southwest of the high
pressure system. The onset of the closed circulation at 500 mbar on
July 21, 1978, coincided with an increase in southwesterly winds northwest of
the surface high. Prior to July 20, 1978, high sulfate levels were associated
with stagnant or light wind conditions. A ducting situation developed after
July 20, 1978, with the elevated sulfate region becoming elongated in the
direction of the wind and gradually being transported eastward across southern
New England and out to sea.
To ensure that the pollutant mass within the modeling domain reaches a
quasi-steady state before evaluation, the model was exercised for 48 h prior to
July 16, 1978. Initial concentrations and boundary values for the nine modeled
species were held constant for the entire simulation and are listed in Table 4.
Tropospheric background concentrations of 03 have been studied by numerous
investigators (Jung, 1963; Aldaz, 1967; Ripperton and Worth, 1969; Rasmussen,
1975; Rutheir et al., 1980) and are established as 0.02 to 0.06 ppm. This range
is exceeded only under unusual conditions involving either stratospheric
intrusion episodes or anthropogenic contamination. For this modeling study, the
middle range of concentrations (0.04 ppm) was used for background 03. The
modeling region is downwind of numerous anthropogenic source areas, so the
boundary conditions may be affected by upwind sources.
Background concentrations of NOX are not precisely established (Singh
et al., 1980). However, they are known to be low, on the order of a few parts
173
-------
TABLE 4. INITIAL AND BOUNDARY CONDITIONS
USED BY RTM-III"
Species
NOX
Ox
ETH
PAR
CARB
ARO
PAN
S02
S04 =
Initial
Condition
1
40
1
35
15
0.8
0.1
2
2
Boundary
North.
South
and
Western
1
40
1
35
15
0.8
0.1
2
2
Conditions
Eastern
1
30
0.1
35
0.1
0.1
0.1
1
1.5
"All concentrations in ppb except S04~, which
are in jig/m3.
per billion at most. For the present model application, the value is set at
1 ppb. Use of this value may result in the generation of 10 to 20 ppb 03 when
combined with a background of reactive HCs.
On the basis of an extensive review of measurements, the total background
concentration for HCs in the model is set at approximately 0.05 ppmC. This
seemingly low concentration results in a HC loading of 4.7 x 104 metric tons
within the model. This is only slightly smaller than the daily anthropogenic HC
emission for the entire United States, about 6 x 104 metric tons in 1973 (EPA,
174
-------
1976). The source of this HC background is not well understood. Certainly, the
long-range transport from anthropogenic sources is a factor, along with natural
geogenic and biogenic sources. Since model results are sensitive to this
reactive background, a regional-scale modeling study will require a careful
assessment of the distribution and relative impact of the background sources.
EVALUATION OF MODEL PREDICTIONS
Hourly predictions of various pollutants species (03, NO, N02 , S02, and
S04=) from the RTM-III were compared with the corresponding observations.
Statistical comparison of predictions and observations for the 8-day episode are
summarized in Table 5. Reasonable agreement between the predicted and observed
03 concentrations can be seen from the small average residual and a high
correlation coefficient of 0.7. A scatter plot for the predicted and observed
TABLE 5. SUMMARY OF STATISTICS FOR PREDICTIONS/OBSERVATIONS'
Species
03
NO
N02
so?
S04=
Averaging
Time
Hourly
Hourly
Hourly
3-h
24-h
Sample
Size
1,508
679
662
2,838
362
Average Average
Observation Bias6
49 0 * 1.5
4 4 i 0.4
9 7-0.5
26 2 > 1.5
16 -7-1.5
Average
Absolute Correlation
Residual Coefficient
17 0.70
4 0.07
7 0.00
19 0.24
9 0.77
aConcentrations in ppb except S04~, which is in
bAt 95% confidence intervals.
175
-------
O3 concentrations is shown in Figure 4. A comparison of the predicted and
observed S04= concentrations is equally favorable with a correlation coefficient
of 0.77, comparable to an identical study using the S02/sulfate version (RTM-II)
reported by Stewart et al. (1983).
Comparisons of predictions and observations of N02 and primary pollutants
such as NO and S02 are, however, quite disappointing. Contributions from local
sources will certainly have significant effects on the model performance because
of the limited spatial resolution of the model. The poor NOj predictions, a
ie0.ee
i2e.ee
a
" ae.ee
i i i i i i i i i i i i i i
Correlation Coefficient » 0.699
Sample Size » 1508
ie.ee
o
o
I I I I I I I I I I I I I I I I I I I I I I I
40.00 se.ee ize.ee ise.ee
OBSERVED
Figure 4. Predicted and observed 03 concentrations (ppb),
176
-------
problem also present in the urban photochemical modeling, are apparently
attributable to a combination of its short half-life and local contributions.
Figure 5 illustrates the prediction of 03 distributions over the modeling
region. The predicted 03 patterns, generally corresponding to the prevailing
synoptic-scale flows, resemble a southwest-to-northeast 03 "river" (Figure 5).
In two areas, one along the Great Lakes and one near Long Island Sound, the
predicted 03 concentrations exceed 0.15 ppm.
A station-by-station evaluation of the model performance on 03 predictions
is even more impressive. The pertinent statistics are summarized in Table 6.
Of the eight stations reporting hourly 03 concentrations, the correlation
coefficients range from 0.67 to 0.88. Concentration histograms for each of the
eight stations over the entire 8-day episode are shown in Figures 6 through 13.
The station location is indicated in the map by the triangle symbol. The
abilitv of RTM-III to predict the diurnal 03 patterns at all locations seems to
be quite consistent.
177
-------
00
O
CM
C
O
s
0,
Q,
0)
to
r-l
T3
0)
X
•H
S
0)
(A
§
CO
M
4-1
e
0)
O
C
O
O
61
0)
u •
U ^
•H
T> O
o) o
ft* -H
"O
0)
60
•H
178
-------
TABLE 6. MODEL PERFORMANCE STATISTICS FOR HOURLY 03 BY
MONITORING STATIONS3
Station Sample
Number Size
1
2
3
4
5
6
7
8
181
171
195
197
181
198
187
198
Average
Observation
55
50
47
51
50
46
46
51
Average
Bias"
-1
-7
-13
6
7
1
7
8
i 5
• 4
t 4
* 4
* 4
i 4
i 4
. 3
Average
Absolute Correlation
Residual Coefficient
20
20
20
18
14
14
15
12
0.74
0.80
0.72
0.69
0.82
0.74
0.88
0.80
'Concentrations in ppb.
"At 95% confidence intervals.
179
-------
8
3
I
I
•
s*
:?
o
in
4J
CO
Cb
CL
tn
o
•H
U
C
O
u
•o
0)
(0
0)
s
-o
CO
01
4-1
CJ
•H
T3
(V
OJ
M
3
60
•H
Cb
5 2 s
NOIIMU/03M03
180
-------
§
•
•
•
•
a e
=5
e
o
(0
4J
.n
a.
ex
en
o
C
a;
o
o
u
3
W
CO
0)
S
•a
c
C8
u
•H
•a
a>
M
a,
rx
01
60
•H
181
-------
§
» H
O
•H
4-1
CO
4-1
0,
a
w
O
•H
0)
U
C
O
U
o"
to
0)
•a
C
n)
O)
00
01
lH
00
182
-------
d"
: §
iT
•
•
•
i
•
•
•
« i
IK-
2
li
S 8 S
MDIJMWB3N09
§ 5 5
o
to
4-1
00
Q,
CU
to
c
O
CO
l-i
4-1
G
60
•H
Ui
183
-------
in
in
§
5
' m •
N
•
f
•
. I
28
=?
HOUHUliOSHO
z s s
WUMUK33NQ3
G
O
4-1
C/3
o.
ex
M
O
•H
c
0)
O
O
a
•a
a>
M
a>
s
•n
a
a
•n
v
-o
a>
t>0
184
-------
a
8
y
i i i i i t i i i i i iTi i i
8 S
c
o
•H
4-1
0)
tfl
C
O
•H
4J
ffl
0)
u
o
a
01
to
ai
e
-d
C
(T)
0)
i-i
O
T3
a>
a>
u
e>o
NOI1HU
185
-------
I
•
•
• -r
t •
I I I I I I I I I I I I I 1
S S 8
NOI1MUNZM03
GO
O
•H
,0
0,
Q.
in
o
•H
4J
C8
1-1
4-J
C
O>
O
o
u
n
O
•o
Ol
M
3
V)
«
0)
S
•o
to
0)
4J
O
•H
-O
0)
t-l
o,
CM
iH
0)
i-t
3
60
•H
b
186
-------
• 1 - I
m o
• I
•
•
O
0.
a.
(0
o
C
0)
O
o
o
n
O
a;
u
3
(A
(0
at
E
C
m
•a
0)
4-1
u
•H
-a
a>
i-i
a,
a;
3
60
HOI
S S S S
NOIXMUM33N03
187
-------
REFERENCES
Aldaz L. 1969. Flux measurement of atmospheric ozone over land and water.
Journal of Geophysical Research, 74:6943-6946.
Benklev, C. W., and A. Bass. 1979. Development of Mesoscale Air Quality
Simulation Models. Vol. 6, User's Guide to MESOPAC, Mesoscale Meteorology
Package. EPA-600-7-79-061, U.S. Environmental Protection Agency,
Washington, DC.
Benson, S. W. 1968. Thermochemical Kinetics. John Wiley and Sons, New York.
Bhrumralker. C. M., R. L. Mancuso, D. E. Wolf, K. C. Nitz, W. B. Johnson, and
T. L. Clark. 1981. ENAMPA-1 Long-Term S02 and Sulfate Pollution Model:
Further Application of Eastern North America. EPA-600/57-81-102, U.S.
Environmental Protection Agency, Research Triangle Park, NC.
Briggs, G. A. 1971. Some Recent Analyses of Plume Rise Observations. In:
Proceedings of the Second Clean Air Congress, Academic Press, New York.
Durran, D., M. J. Meldgin, M. K. Liu, T. Thoem, and D. Henderson. 1979. A
study of long-range air pollution problems related to coal development in
the Northern Great Plains. Atmospheric Environment, 13:1021-1037.
Gear. C. W. 1971. The automatic integration of ordinary differential
equations. Communications of the ACM, 14:176-179.
Junge, C. E. 1963. Air Chemistry and Radioactivity. Academic Press, New York.
Klemm, H. A., and R. J. Brennan. 1981. Emissions Inventory for the SURE
Region. EPRI EA-1913, Electric Power Research Institute, Palo Alto,
California.
Lavery. T. F., R. L. Baskett, J. W. Thrasher, N. J. Lordi, A. C. Loyd, and G. M.
H.idy. 1980. Development and Validation of a Regional Model to Simulate
Atmospheric Concentration of Sulfur Dioxide and Sulfate. Proceedings of
the Second Joint Conference on Application of Air Pollution Meteorology.
American Meteorological Society, New Orleans, Louisiana, pp. 236-247.
Liu, M. K., and S. D. Reynolds. 1983. Development of a Regional-Scale Air
Quality Model. International Conference on Long-Range Transport Models for
Photochemical Oxidants and Their Precursors, Research Triangle Park, North
Carolina.
McMahon, T. A., and P. J. Denison. 1979. Empirical atmospheric deposition
parameters—a survey. Atmospheric Environment, 13:571-585.
Mueller, P. K., G. M. Hidy, K. K. Warren, H. M. Collins, and P. A. Hayden.
1982. The Sulfate Regional Experiment: Data Base Inventory and Summary of
Major Index File Programs. EPRI EA-1904, Electric Power Research
Institute, Palo Alto, California.
188
-------
Niemann, B. L., and J. W. S. Young. 1981. Modeling Subgroup Report of Work
Group 2—Atmospheric Science and Analysis (under the Memorandum of Intent
on Transboundary Air Pollution signed by the United Staes and Canada on
August 5, 1978). Report No. 2-13, National Oceanic and Atmospheric
Administration, Silver Spring, Maryland.
Rasmussen, K. H., M. Taheri, and R. L. Kabel. 1974. Sources and Natural Removal
Processes for Some Atmospheric Pollutants. EPA-650/4-74-032, U.S.
Environmental Protection Agency, Washington, DC.
Ripperton. L. A., and J. B. Worth. 1969. Chemical and Environmental Factors
Affecting Ozone Concentration in the Lower Atmosphere. NSF Grant GA-1022,
University of North Carolina, Chapel Hill, North Carolina.
Sheigh, C. M., M. L. Wesely, and B. B. Hicks. 1979. Estimated Dry Deposition
Velocities of Sulfur over the Eastern United States and Surrounding
Regions. Atmospheric Environment, 13:1361-1368.
Singh, H. B., F. L. Ludwig, and W. B. Johnson. 1979. Ozone in Clean Remote
Atmospheres. Project No. 5661, Stanford Research Institute, Palo Alto,
California.
Stewart, D. A., R. E. Morris, M. K. Liu, and D. Henderson. In press.
Evaluation of an episodic regional transport Model for a multi-day sulfate
episode. Atmospheric Environment.
Whitten, G. Z., H. Hogo, and J. P. Killus. 1980. The carbon-bond mechanism: a
condensed kinetic mechanism for photochemical smog. Environmental Science
and Technology, 4:690.
DISCUSSION
B. Luebkert: Do I understand that you do not use any nighttime chemistry other
than deposition or ground scavaging of N02 and 03?
S. Reynolds: That is not correct. We use a very simplified nighttime
chemistry. The 03-olefin reaction still operates. We have largely eliminated
most of the olefins from the model and keep those as part of the species anyway,
but ethylene is included and the 03-ethylene reaction does occur at night.
Furthermore, we essentially have a parameterized version of the N205-and-water
reaction operating at night as well.
E. Runca: Are you applying the box model to establish the boundary conditions
for the Philadelphia airshed?
S. Reynolds: Correct.
189
-------
E. Runca: Did you also run the model without these boundary conditions?
S. Reynolds: We did. I do not have those results here, and I am not even sure
that they have been released yet. If you arc interested in those, I suggest you
speak to Dr. Cole.
E. Runca: In what way are the results expected if you compare the simulations?
S. Reynolds: If you compare the simulations to the observations, the
simulations without the boundary conditions are obviously much smaller. As
Dr. Cole indicated this morning, without the transport, only one of the stations
showed an 03 excess. I can give you what generally happened, and it is as one
might expect. The 03 in this area, which is for the most part upwind of
Philadelphia, dropped rather precipitously. Slowly, as one goes farther and
farther down into Philadelphia, the 03 began to increase. The point at which
there is relatively little effect of switching off the boundaries is fairly far
downwind. I believe the farthest, Dowington Station, was the only one that
showed an exceedence about the boundary conditions.
190
-------
APPENDIX. SAI REGIONAL OXIDANT MODEL (ROM)
II. The SAI Regional Oxidant Model is an Eulerinn grid model based on the
numerical solution of a 2-1/2 layer multispecies diffusion equation:
1 . Variable — time steps vary with wind speed and horizontal resolution
(internally controlled by number stabilitv criteria). Output is
usually either hourly or 3-h averages. Inputs can be either hourly or
3-h averages.
2. Variable — typical resolution is on the order of 25 km x 25 km.
3. 1,000 km x 1,000 km.
4. Three physical layers — surface layer, generally 10 m to 50 m: mixed
layer, variable, 50 m to 1,500 m; inversion layer, variable 50 m to
1 .500 m.
5. Top of the inversion layer, generally on the order of 2 km.
6. Inversion layer — geostrophic winds derived from geopotential heights
(generally at 850 mbar level).
Mixed laver — geostrophic wind adjusted for surface effects according
;o the scheme of Hoxit (1973).
Surface layer — interpolation scheme of 1/r using surface observations.
7. SHASTA mothod (Boris and Book. 1973; Book, Boris, and Hain, 1975).
8. Yes. zi / ,ui^ +
-------
12. Yes.
a. Yes
b. Interpolation of T-sonde measurements
c. Defines the depth of a homogeneously mixed layer.
13. No.
14. Yes.
15. Yes (see Durran et al., 1979).
a. Yes.
b. Yes.
16. Yes.
a. Yes. Not exercised in the ROM.
17. 2.5 s (CDC 7600)/surface grid point.
18. 64 K small ore, 2.5 Mb disc space (nine species).
a. CDC 7600.
19. Yes.
Wojcik, M., T. Myers, J. Killus, 0. Serang, and M. K. Liu. 1978.
Development and Evaluation of a Mesoscale Photochemical Air Quality
Simulation Model. SAI Report No. E178-118, U.S. EPA, 195 pp.
Liu, M. K., and J. P. Killus. 1981. Development and Evaluation of a
Mesoscale Photochemical Air Quality Simulation Model. Proceedings of
the 62nd Annual Meeting of the American Association for the
Advancement of Science. Pacific Division, Eugene, Oregon.
Liu, M. K., and P. M. Roth. 1981. The Use of a Regional-Scale
Numerical Model in Addressing Certain Key Air Quality Issues
Anticipated in the 1980s, In: de Wispelaere, C., ed., Air Pollution
Modeling and Its Application, I.
20. Yes.
III. Chemistry
1. Fixed.
a. 03, NO or N02, CO, PAN, PAR (alkyl carbon atoms), ETH (ethylene),
OLE (olefinic bonds), CARB (carbonyl groups, ketones and
aldehydes), ARO (reactive aromatics).
192
-------
b. NO, N02, and 03 are lumped into two species: NOX (NO + N02) and
Ox (03 + N02). NO, N02, and N03 concentrations are then computed
via a steady-state relationship. Radical species 0, N03, N205,
OH, H02, Me02, AC03, and tour intermediate HC-specific peroxy
radicals are computed via a steady-state approximation.
2. a. Yes. See above.
b. Ozone-olefin product chemistry is used; 03-NOX chemistry is
deleted.
c. Wind shear and stability stratification are allowed to vary from
layer to layer.
3. a. No.
b. None.
4. Emissions from major point sources are followed by Gaussian puff
module (see Liu, M. K., D. A. Stewart, and D. Henderson, 1982, Journal
of Applied Meteorology, 21:859-813.
5. No.
6. a. Yes, if input as such.
b. Cumulus cloud effects can be treated via modifications of
photolytic rates.
7. Boundary conditions, initial conditions, source terms.
193
-------
DEVELOPMENT OF A REGIONAL-SCALE AIR QUALITY MODEL*
Mei-Kao Liu
Steven D. Reynolds
Systems Applications, Inc.
101 Lucas Valley Road
San Rafael, California 94903 (USA)
INTRODUCTION
Over the past few years, there has been a marked shift in the emphasis of
air pollution problems. Instead of the typically short or episodic problems of
a local nature, significant interest has been placed recently on the degradation
of air quality or related problems on regional scales caused by the transport of
a variety of air pollutants from a large agglomerate of sources over long
distances.
During the late 1970s, as part of the plan to promote the use of immense
coal reserves in the United States, various potential environmental problems
were assessed. Although impacts from primary emissions as a result of coal
burning, such as NOX, SO2, and particulates, are generally restricted to an area
immediately downwind of the sources, problems related to fine particulates and
secondary products such as sulfate and nitrate have been raised. These problems
have become particularly acute with the increasing use of tall stacks as the
control technique for ground-level concentrations, because increasing
atmospheric residence times tend to promote chemical reactions.
*This report has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
194
-------
Even the occurrence of photochemical smog, formerly considered an urban
problem, has recently taken on a regional character. For example, in the
Northeast Regional Oxidant Study (NEROS), a series of field measurements was
conducted to examine the role of pollutant precursors emitted from upwind
sources, such as NOX and reactive HCs, on the high 03 concentrations observed in
rural areas in the Northeast Corridor (Vaughan et al., 1982).
The various air-quality-related problems discussed above have motivated the
development of many regional-scale air quality models during the past few years.
Several reviews on this subject can be found in the literature (e.g., Eliassen,
1980; Stewart and Liu, 1982). The objective of this paper is to delineate the
development and application of a regional-scale air quality model developed by
Systems Applications, Inc.
DEVELOPMENT OF A REGIONAL-SCALE AIR QUALITY MODEL
A regional-scale air quality model capable of addressing issues related to
long-distance pollutant transport must have the following technical attributes:
• Ability to simulate pollutant concentrations at relatively low levels at
locations from 100 to 1,000 km away from the emission sources;
• Ability to simulate physical processes, such as dry deposition, which
are only important on large time scales;
• Ability to simulate wet deposition of S02, sulfate, and nitrogen-bearing
species due to precipitation; and
• Ability to simulate the diurnal formation of secondary pollutants, such
as sulfate and 03 as a result of chemical reactions.
195
-------
Through a series of research and development contracts, Systems
Applications, Inc., has developed a regional-scale air quality model. A brief
history of these research and development activities is described in Table 1.
This model development, initiated in early 1976, has drawn upon the extensive
experience accumulated during the successful development of a photochemical
kinetic mechanism (Whitten, Killus, and Hogo, 1980) and an urban airshed model
(Reynolds et al., 1973). Over the past 8 yr, the model has undergone
evaluation, improvement, and significant expansion. Two different versions of
the model now exist:
• A regional transport model for S02 and sulfate wet and dry deposition
(RTM-II), and
• A regional transport model for photochemical oxidants and their
precursors (RTM-III).
The formulations and numerical solutions for these two versions, as
described in the next section, are identical.
DESCRIPTION OF MODEL EQUATIONS
The regional transport model described in this paper adopts a puff-on-grid
approach to accommodate both point and area sources often encountered in a
regional air quality assessment study. The model equations, on an Eulerian grid
196
-------
TABLE 1.
Time Period
REGIONAL AIR QUALITY MODEL DEVELOPMENT ACTIVITY AT
SYSTEMS APPLICATIONS, INC.
Activity
Size of
Effort
(person-
years)
Sponsor
1976 - 1977
1977 - 1970
1977 - 1980
1981 - 1982
1982 - present
Development of a regional
transport model for S02 and
sulfate
Development of a regional
oxidant model
Development and application
of plume and regional
visibility model
Evaluation and improvement
of regional transport models
Application of the regional
transport model
U.S. EPA, Denver,
Colorado
4 U.S. EPA, Research
Triangle Park,
North Carolina
2-1/2 U.S. EPA, Washington,
D.C.
National Park Service,
1-1/2 Washington, D.C.
1-1/4 National Park Service,
Washington, Colorado
system, are based on the quasi-three-dimensional, time-dependent atmospheric
diffusion equations for multiple chemical species
oc,1 + u dc,' + dci1 = a /Kx ac,1 \ + ,. /Ky
(1)
t)t
'•,' + R,1 + Qi1 - S,
where u and v, Kx and Ky are defined as the horizontal wind velocities and
turbulent diffusivities in the x and y directions, respectively. The reaction
rate term and pollutant source and sink terms are denoted by R;', Qi1, and S,1,
respectively. £,' represents the interfacial transport resulting from
197
-------
entrainment/detrainment and F;' the effects of large-scale convergence or
divergence, respectively. In this equation, the subscript "i" refers to
pollutant species and superscipt "j" denotes the j-th vertical layer.
One of the unique features of this model is the adoption of physical layers
rather than predetermined equal- or variable-thickness layers in the vertical
direction. As shown in Figure 1, three different physical layers are invoked in
the present model. All three layers may vary temporally or spatially. These
layers include:
• A mixed layer beneath the inversion layer and above the surface layer
that accepts emissions from local sources and carries pollutants trapped
below the inversion layer.
• An inversion layer above the mixed layer that serves as a
"semipermanent" reservoir for pollutants released from the mixed layer.
(The model no longer keeps track of pollutants leaving the top of the
inversion layer.)
• A surface layer between the ground surface and the mixed layer that
accommodates surface modification of the vertical concentration profiles
due to processes such as surface deposition.
The advantage of the present approach is primarily the efficient use of the
computational resources. Because no significant concentration variations within
the mixed layer are observed at locations distant from the sources, additional
layers seem unwarranted. Considerable simplifications can also be realized as
coordinate transformations are not necessary when complex topography is
encountered.
Relative to horizontal transport by wind, vertical diffusion plays a
relatively more minor role than lateral diffusion in determining the fate of air
198
-------
9NI13QOW
\
\
\
\
\
\
\
\
\
\
\
\
\
3MI1AVQ
s^j:
XX^ ^
0)
•a
o
e
O
CX
-------
pollutants on regional scales. This can be seen from a simple dimensional
analysis. From the atmospheric diffusion equation, the following two ratios can
be formed:
Lateral Diffusion = KH
Horizontal Transport UAX
Vertical Diffusion = Kv /AXx2
IT. v V . „ '
Horizontal Transport UAX VAZ'
where U is the characteristic wind speed, and AX and AZ are the characteristic
lengths in the horizontal and vertical directions. According to Heffter (1965)
and Randerson (1972), a value of 105 m2/s appears to be the median horizontal
diffusivity for the spatial and temporal scales of interest. The vertical eddy
diffusivity is a strong function of height and atmospheric stability. For the
present analysis, a value of 102 m2/s can be viewed as respresentative
(Pasquill, 1974). Thus, using a 10 m/s average wind and AX = 100 m, the above
two ratios become:
Lateral Diffusion " 10"'
Horizontal Transport
Vertical Diffusion " 102
Horizontal Transport
Thus, on regional scales, vertical diffusion can be safely neglected. Vertical
exchange of pollutants across the mixed layer and the inversion layer can,
however, occur as a result of interfacial transport or entrainment/detrainment.
200
-------
The mixed layer generally expands or rises during the daytime and collapses
or falls in the evening. The resultant entrainment or detrainment of a
pollutant across the temporally varying mixed layer is simulated in Equation (1)
by the term E,1, which can be determined as:
-(c,' - c.-)
h
E;' = <
if j = 1, dh/dt > 0, or j = 2, dh/dt - 0
if j = 1, dh/dt < 0, or j = 2, dh/dt > 0
c,2 - c,A /dH\
(H - h) \dt)
if j = 2, dH/dt > 0,
(2)
where h denotes the depth of the mixed layer or the base of the inversion, while
H denotes the top of the inversion or the inversion or the height of the entire
modeling region and CIA is the concentration of pollutant i above the upper
boundary of the modeling region.
The fourth term on the right-hand side of Equation (2) represents
interfacial pollutant transport into or out of a layer that arises from the
large-scale convergence in the spatially varying wind fields. The
201
-------
two-dimensional divergence D is given as D = du/dx + dv/dy, and the function
Fj '(D) is defined by:
1)c,' if D < 0
FI' = 0, j = 1 (3)
J)c,A if D > 0, j = 2
SIMPLIFICATION OF THE TREATMENT FOR THE SURFACE LAYER
For pollutants originating from either elevated sources or distant
ground-level sources, most of the pollutant mass is contained in the mixing
layer. The removal processes, as discussed above, consist of diffusion of the
pollutants through the surface layer to the ground and absorption or adsorption
at the atmosphere-ground interface. As illustrated in Figure 2, through
atmospheric stabilities, the diurnal variation of temperature in the surface
layer affects the vertical pollutant distribution and, consequently, the rate of
surface uptake of pollutants (Hogstrom, 1975). Furthermore, Hill (1971)
observed that the adsorption of 03 by leaves does not vary linearly with
concentration at high concentration levels. As a result, a model that can
account for these variations must include diabatic atmospheric conditions and
nonlinear surface reactions.
202
-------
a>
u
03
UJ
M
3
in
t/i
c
o
aJ
n)
C
3
•H
U-l
O
C
o
to
•H
O
OJ
4=
U
3
«0
•H
203
-------
The model assumes that the transfer o£ pollutant gases from the atmosphere
to a surface is accomplished via three states (Sehmel, Sutter, and Dana, 1973;
Galbally, 1974):
• The gases are transported, primarily by turbulent diffusion, to a
laminar sublayer just above the surface.
• The gases are transported, primarily by molecular diffusion, through
this laminar sublayer.
• The gases interact by adsorption or chemical reaction with the surface.
Thus, as shown in Figure 3, the surface layer is divided into two parts;
the turbulent layer and the laminar sublayer. In the turbulent layer, after the
atmosphere reaches an equilibrium state, the atmospheric diffusion equation
becomes:
= 0, (4)
with the following boundary conditions:
c = c! a t z = i L|
Kv /^c\ * f at z = z0,
where c, is the cell-averaged concentration in the mixed layer, f is the
pollutant flux across the turbulent layer/laminar sublayer interface, and ze is
the height of the surface roughness element.
204
-------
o
4-1
2*
<
o
M
Of
LU
S
S
S»
S
s
s
s
s
s
01
X
CO
o
n)
3
in
0)
o
O
•i-l
4-1
to
01
JZ
y
to
0)
>-l
3
60
•H
i/l
205
-------
The vertical diffusivity Kv can be prescribed as:
ku-z, (5)
where k = the von Karman constant (0.35),
u- = the friction velocity,
z = the height, and
L = the Monin-Obukhov length.
This formula is the result of the similarity theory for the constant-flux
surface layer. For the neutral case, the ,A function equals unity. For the
stable and unstable cases, the ,/, function is greater and less than 1,
respectively. The following empirical expressions for the ,;, function were
proposed by Businger et al. (1971) based on the following observational data:
For the stable case (L > 0):
4.7 (z\ (6)
For the unstable case (L < 0)
'(!)
(7)
206
-------
For either the stable or unstable case, the solution of Equation (5) is simply:
c = c, - f. j ILI »(z) dz (8)
z ku-z
Across the laminar sublayer, it is assumed that the pollutant flux can be
written as:
f = /ju-(c8 - cs) (9)
where ce and cs denote the concentrations at the interface and the surface,
respectively; and a, analogous to the Stanton number in heat transfer, is the
inverse of a dimensionless resistance for the laminar sublayer.
If it is further assumed that mass and momentum are transferred in an
identical manner in the turbulent layer, but differently through the laminar
sublayer, then the relationships established by Owen and Thompson (1963) and
Thorn (1972) discussed above can be used:
(Owen-Thompson), (10)
at U
1/3
02 |
D
(Thompson) (11)
Equation (11) is used in the model. To complete the description of the
surface-layer model, a boundary condition is required at the surface. Uptake of
air pollutants occurs by chemical reaction with, or catalytic decomposition
within, either the soil or vegetation, or it can occur by these processes at
207
-------
soil or vegetation surfaces. These processes are generally dependent on the gas
concentration at the surface. A general equation for the gas loss per unit of
time can be written as (Benson, 1968):
f =
L
(12)
where f is the pollutant flux, 7 is a reaction rate constant, and cs is the
concentration of the gas at the soil or vegetation surface. The
denotes the reaction order. Eliminating ce and c, from Equations (8), (9), and
(10), the following transcendental equation is obtained for f:
I -f + 7~1" -f" - c, = 0,
(13)
where
s _L + /ILI ,/,(z) dz.
0u« z® ku-z
Although the reaction order is most likely to be 1, closed-form solutions can be
found for the cases of a = 1, 2, and 3:
f = <
I + 1
T
a = 1
+ 4Ic,\ 1/2 a = 2
\ 1
/
21
(A* + A-)3
(14)
a = 3
208
-------
where
AJ =-{3 Cj +
I
1 /?
It is interesting to note that these formulas reduce to that of Chamberlain
(1966) or Galbally (1974) for the special case of: (1) a first-order surface
reaction and (2) a neutrally stratified atmosphere.
CHEMICAL KINETIC CALCULATIONS
The chemical kinetic mechanism used in the regional-scale photochemical air
qualitv model is based on the Carbon-Bond-II mechanism (Whitten, Killus, and
Hogo, 1980). This mechanism is expressly carbon conservative, a feature of
particular importance in situations involving extended transport and residence
times. Hydrocarbon emissions used in the model are divided into five
categories, according to the individual bonding structure of carbon atoms within
each molecule. The carbon bond categories are: single-bonded carbon (PAR),
ethylene (ETH), aromatic rings (ARO), and carbon-fast (OLE) and double-bonded
carbonyl (CARB) groups.
The numerical solutions of the rate equations for chemical kinetic
mechanisms generally encounter the stiffness problem. Stiff systems are defined
as those with widely differing time constants. Classical methods for the
solution of differential equations require a time step sufficiently small to
avoid instability for the smallest time constant and may impose an enormous
computational burden. Although a numerical algorithm based on the method of
209
-------
Gear (1971) can be used, the introduction of discontinuities into the forcing
function of a Gear solution—such as changes in boundary conditions—can cause
major inefficiencies in computation. These inefficiencies, together with the
additional storage requirements for a high-order predictor-corrector method,
make such a scheme unattractive in a grid model.
As an alternative, the invocation of the steady-state aproximation can be
used as an important tool for the numerical solution of chemical rate equations.
A transformation from differential equations to algebraic solutions allows the
reduction of the number of species requiring numerical solutions. The species
for which the steady-state approximation is valid are those with the smallest
time constants. Thus, an elimination of these species greatly reduces the
stiffness of the ordinary differential equations. The complexity of the
algebraic solutions increases as more steady-state species that react with each
other are included. For example, the implementation of the steady-state
approximation in the Carbon-Bond-II mechanism requires the solution of a quintic
polynomial for the selected radical species. One problem concerning the use of
the steady-state approximation is that its use in effect removes that species
from mass balance considerations; i.e., the steady-state expression assumes that
y = 0. If /ydt is very small, then the removal of this species will not affect
the overall mass conservation. If, however, /ydt is sufficiently greater than
zero, so as to affect the overall mass balance, then the steady-state assumption
may produce invalid results.
This limitation has presented a potential problem in the use of the
steady-state relationship for simulating major species such as NO, NOz, and Oa.
210
-------
Generally, these three species are in dynamic equilibrium, with a time constant
much faster than the rest of the nonradical chemical kinetic system. The
dynamic equilibrium established is very close to:
K3[03][NO] = K,[N02]
Unfortunately, mass exchanges involving NO, N02, and 03 are not small, and the
conventional form of the steady-state approximation cannot be used. However,
the simple steady-state values of NO, N02, and 03 may be modified in terms of a
correction factor:
ASS = - 1 103] + [NO] + Ki
2
+ 1 \ [[03] + [NO] + K, \2
i\\ K-J
- 4 /[NO][031 - K! IN02]\ [ "2, (15)
K3
where
[N02]ss = [N02] - Ass
(N0]55 = [NO] + ASS
[03]$1 = [03] + ASS
211
-------
For state variables, such as NOX (=NO + NOj) and unpaired oxygen atoms Ox
(=03 -I- N02), the correction-factor equation becomes:
Ass = ~
2
1 /[Ox] + [NOX] + K±\
2\ K3/
+ 1 J ( [Ox] + [NOX] + K,_ V - A [NOX] [Ox] t1'2,
2 \ K3
where
IN02]SS = - ASS
[N0]ss = NOX + Ass
[03]ss = Ox + ASS
Mass conservation is thus maintained for the two redefined species, NOX and
Ox. The steady-state calculation does not affect the quantity of these two
species; it merely apportions them into the three molecular species 03, NO, and
N02. In this scheme, NO-to-N02 conversions become the source of unpaired oxygen
atoms.
This scheme has been tested against the Gear solution for a variety of
cases. As might be expected, the approximation tends to break down when peroxyl
radical concentrations are large relative to 03 (i.e., when there are very high
HC concentrations or very low NOX concentrations). The former condition is not
likely to be encountered, given the resolution of a regional-scale model. Under
the latter condition, photochemical 03 production is very slow relative to the
background. Thus, the approximation appears to be reasonable for regional-scale
212
-------
application. The incorporation of a similar scheme in the Regional Oxidant
Model (RTM-III) for simulating aerosol formation is currently being
contemplated. A description of the proposed aerosol module is described in the
appendix.
APPLICATIONS OF THE REGIONAL TRANSPORT MODEL
In this section, we will summarize two typical applications of the regional
transport model described in the previous sections. A preliminary version of
the regional photochemical air quality model (RTM-III) was applied to the
Northeast Regional Oxidant Study (NEROS) areas. The S02/sulfate version of the
regional transport model (RTM-Il) was applied to a sulfate episode in July 1978,
reported by the Sulfate Regional Experiment (SURE) in the Northeast United
States.
Application of the NEROS Areas
As a result of the long-range transport of 03 and its precursors, elevated
oxidant concentrations have been observed in many rural areas in the
northeastern part of the United States. Measurements collected at remote
locations, such as the White Mountains, indicate 03 concentrations significantly
higher than the background levels (Lonneman, 1977). Observations drawn from the
aircraft data reported by Simple et al. (1977) show that 03 concentrations at an
elevation of 2 km are consistently between 40 and 60 ppb, regardless of the
concentrations below this level. Exceptions occur when the base of the synoptic
subsidence inversion is above 2 km. On these occasions, 03 values at 2 km can
213
-------
reach 80 ppb. Both observations seem to imply that the high Os levels were
caused by the transport of pollutant precursors from distant upwind sources. A
preliminary version of the regional photochemical air quality model (RTM-III)
was applied to this area (Wojcik et al., 1978). The model region covered an
area approximately 800 km x 800 km, with a spatial resolution of 20 km and a
temporal resolution of 3 h. Data collected during the 1975 Northeast Oxidant
Transport Study were used for testing and validating the model.
For the modeling region, gridded area emission rates for particulates, S02,
NOX, HCs, and CO were derived from the National Emissions Data System (NEDS) on
a 40-km x 40-km grid with a 20-km resolution. To conform to the modeling grid,
these emissions were interpolated to the grid by proportionally allocating the
emissions from each cell. In addition to area sources, emission rates for point
sources exceeding 10,000 tons/yr were also included. Although most emissions
9
were yearly averages, some data were available for seasonal variations and the
number of hours of operation per day. The hours of operation were assumed to be
symmetrically distributed around noon and to have equal hourly emission rates.
For point sources having no data for hours of operation, the emission rates were
assumed to be constant, with the exception of power plants.
The model was exercised for 72 h between 0100 EOT on July 29, 1975, and
0100 EDT on July 31, 1975. The model predictions of primary pollutants such as
NO and S02 are easily discernible and generally show average concentrations in
the mixed layer that increase near major sources. These areas include the
industrial areas extending from New York City to Southern New Jersey, Northern
Delaware, and Southern Pennsylvania (Philadelphia). Also apparent are the
214
-------
Washington, DC/Baltimore urban complex and several major point sources in New
York, Pennsylvania, and West Virginia.
An analysis of the prediced 03 patterns is interesting. As shown in
Figure 4, 03 concentrations before 0600 EDT on July 30, 1975, are generally less
than 60 ppb. The level increases to a peak of 120 ppb during late afternoon,
with the high 03 concentrations areas extending from south Pennsylvania to the
Atlantic Ocean. After sunset, the predicted 03 concentrations begin to decrease
to a level generally lower than 80 ppb. With a similar diurnal pattern, the
predicted 03 concentrations reach a maximum of 140 ppb near New Jersey between
1800 and 1900 EDT on July 31, 1975.
For comparison, observed surface 03 concentrations from 1300 to 1500 EDT on
both days are displayed in Figure 5. On July 30, high 03 concentrations ranging
from 100 ppb to 150 ppb were reported along a corridor from Wilmington,
New Jersey, to New York City. On July 31, the observed 03 concentrations
generally increased, with high 03 areas extending further northeast. Thus, a
good qualitative comparison of the model predictions with the observation is
achieved.
Further application of RTM-III using the SURE data base will be discussed
by Killus et al. (1983).
215
-------
a.
a.
m
a
o
(0
c
O)
O
O
u
CO
0)
4->
to
4-1
10
•o
0>
4-1
•H
01
4-1
(-1
O
2
a>
x;
g
rs.
§
tn
c
o
•H
4-1
•3
•a
0)
•o
0)
M
Oi
x*
-------
o
IT)
O
O
o
01
3
C
O
u
-------
O
o
0)
3
C
C
O
o
r*.
O
UO
8
r*.
O
8
§
0)
t-l
D
60
218
-------
T)
0)
3
P
C
o
o
-------
0.
Q,
in
c
o
c
0)
o
o
o
in
0)
•o
o
o
u
o"
a
to
14-1
M
3
10
•a
a>
V)
J2
O
in
0)
M
60
•H
220
-------
o
ID
IT)
I
O
O
t -.
-o
0)
3
c
o
o
EC§
o
un
3
m
0)
3
•H
Cb
221
-------
Aj>jp_li_cat ion to a Sulf ate Episode
The S02/sulfate version of the Regional Transport Model (RTM-Il) was
applied to a su]fate episode in July 1978 using data from the EPRI Sulfate
Regional Experiment (Stewart et al., 1983). The purpose of this study was to
evaluate the performance of the model using S02 and sulfate data collected at
54 monitoring stations geographically distributed throughout the modeling
region.
The modeling region selected for this study was almost identical to the
EPRI/SURE grid, which is characterized by 80-km mesh squares defined over the
eastern third of the United States. Dimensions of the grid subset used for this
study were 2,080 km in the east-west direction by 1,840 km in the north-south
direction. The grid resolution selected for the model simulations was
40 km x 40 km. Wind velocities and mixing depths were derived from the National
Weather Service (NWS) radiosonde network. The emissions inventory used in this
study was prepared as part of the EPRI/SURE program.
RTM-il was exercised for a period spanning eight consecutive days
(July 16-23. 1978). To ensure that the sulfur mass budget within the modeling
domain reached a quasi-steady state before evaluation, the model was exercised
for 48 h prior to July 16, 1978. Temporally varying boundary conditions during
the simulation period were estimated by extrapolating average 3-h concentrations
from a few monitoring stations near the upwind boundaries.
222
-------
Figures 6 through 10 are a series of time-history plots for the model
predictions and corresponding observations at each of the SURE Class I and
Class II stations throughout the modeling region. The predicted 24-h average
concentrations are displayed as dotted lines, and the observed 24-h average
concentrations are indicated by square symbols. Successive figures include
stations grouped geographically from the western to eastern (generally downwind)
portion of the modeling region.
An examination of these plots suggests that the model's ability to predict
the sulfate trend is quite good. The ability of the model to simulate the S02
trend appears to be less favorable, a fact that may be attributable to the
influences of local emission sources. A common characteristic of both S02 and
sulfate predictions is that the temporal variations are smoother than their
observed counterparts. This is probably caused by several factors, including
the spatial resolution of the model, which tends to smooth out localized
concentration peaks.
The relatively smooth sulfate predictions produce a less pronounced jump in
concentrations, most noticeable at Station 29 (Figure 7) and Stations 30 and 14
(Figure 8). Similarly, the model tended to underpredict the sulfate peak in
Illinois on July 18, 1978 (Stations 26, 27, and 38 in Figure 6). Further
analysis indicates that the high sulfate region in Illinois (>30 (jg/m3) was
associated with a weak anticyclonic flow centered in western Kentucky and
Tennessee. An examination of the back-trajectories shows emissions from the
St. Louis area and the Ohio River Valley were advected over central Illinois at
223
-------
I I 111
*i
a
i
i
•
Ls
a
B
.e -
4J W)
M O
O -H
in H
C •*-• -O
O C a)
*pH ^ 1*4
t! ^ 3
Cu G t/1
M o <°
4-» U nj
(M CO
O
t/J '-V
C
TO ||
<1> (ft -a
•u C 01
TO O jj
3 TO -a
01 4J 0)
(0 M
•H C
TO -H .
T3 M (/)
° C
*" J-» O
O -H .H
C jj
" g TO
o e (j
>> o o
^ 6 c
O C o
W U o
»i a>
•|H U CM
J3
-------
r r t
I I I I t I i i I i_>
, S
S
S
I I
= 5 5S
N011IAUM33N03
. S
J j j I t I 'A L
8
S
'!
T3
OJ
3
c
c
o
u
0)
1-1
3
00
•H
fcu
NOUMUN33N03
225
-------
•
'I
S
S
s s •
,s
a
9
•
S
g 8 8 S 8 S S •_
i 11.
(U
J=
4J "
«
H C
o o
(ft CO ||
C >t
o *-• -a
n g s
CO O 3
M C 0)
4-> O CO
C U QJ
o> . e
o i>
c ^
O "J
T)
CO
I—1
..
>-> O 4-1
•—I -H O
3 4J .H
t/1 CO t)
•H 60
C8 C
•H -H
C 4-1
>^ O o
^ E c
O C o
.C W O
I 0) 00
cu S
e ^
•H O U3
H r-l S->
(U
M
3
00
•H
S S S S S
NOtiMUN33N03
S S S X S
NDI1MUNZIMB
t 8 S 5 » S S
N01iMUMSil03
226
-------
S
•
I I I I I I I I 1 'I I
i 1 1 I t r I r i
•
s
I
•
s
.8
i i i t i i i i i ,
S = 5 88 P SS S8
i 1 i i i i i i i
_ s
•
i i i i i i i i i
S SS 88 ? SS 98
I i i i i i i i i
_ s
| | • i i i i I I.
• •*9*p»**»
<4~C«E*#»»W«>P»
1 1 I i 1 1 1 1 4
*
V *
\
\
z \
m 1
f i
' • \*
f \
1 r I i i i i I I
1 >^
-
-
1 1
a = -
i It
• -
i
\ \
as •
-
i i
5 = -
I 1
H .
• *
l1 '
•r)
(D
o
o
-------
3 2 5 S S 2
NOllMUKDMn
^ • M» ^ * M ~ _
S
•
•
*i
8
I
_
V) •
(fl C II
c o
O- •<•! T3
•H J-> 0)
4J ") (-1
m n 3
M j-1 m
4-1 C (Q
c QJ
iu o g
u c
n o -o
o o c
O co
10 U-l
f-4
T3 3
c in
CO
cy'cr] "
4J v_x -Q
CO tti
U-l jj
-H •• U
3 tfl .H
T3 (fl
to
v4-i «>o n
O C o
4J O CO
O 4J ^
rH -H JJ
0. C C
O oj
>N E o
1-4 C
O t-l o
4J CO O
(-1
N
(A
•H
-c c o
I 0) C/)
CL> O
e ^
•H O XI
H ^H ~— •
CO
00
0)
t<
3
60
•H
JSSS5SS5-
S S g I S S
W11MUN39O3
B88SSS2-
228
-------
1 1 1 1 1 1 1 1 1
s
jj
i i i i i i i i i "
11)111)111
9
_ :
i i i i i i i i i
x
5
1 t 1 1 f 1 f 1 1
jtss:?ss ss
i i i i i i i i i
m
- m m
- •
™
"
i •
_•
•
• •
i
-
-
2 •
I
1
-o
01
3
c
c
o
a
CO
a;
u
D
00
•H
NOUWUM23N03
229
-------
»
•
s
,
! :
)
m
:
-•
i
-• 8
-
_ S
• «•
i
I
-
S
•
. •
• -
• -
• •*
• ^
1 1 1 1 ^
2 5 S S S •
1 1 1 1 1 ™
. ' • -
•
•
m. -
m
• -
1 1 1 1 1'
S 5 S S 2 '
i i i i i
. k
• • -
• • -
'M -
it
t i i i 'i
1* ^ *» IN •
1 1 1 1 i
• -
• -
• •
' • •
' • -
' • -
1 1 1 1 ' 1
•«
•*
M
•V
S
•
•
fl*
»
'«
M
««
M
M
•W
•
M
•»
•
•
«l
M
»t
«•
«•
%
•
f*
m
«w
••
•t
«M
S
^
•
^
*
I ^1 T^
I J I
s
s
g 8 g s a s s
i i i
m
i
0)
JS
4->
in
K g
o o
O to
(U
0)
C *-> T3
O *u C
o ^~* oi
c
tn S
x o o
M E c
o c o
u >-i u
tn a>
•H 4-1
JS
-------
3
•
I I I I I I I I Li
SSS8SSSS5SS5-
NOIJ.W4UO3N03
•
w
.
_ *
-".
1 1
M <- »
1 1
-
'3
. =
i l
3 = 5
l l
^
•
1 i
5 = 5
-
'•*
1 1
\H -
') m-
/ • ~
• i
\ •
t l i i i > t r t
II 1 II 1 1* 1 I
*<• -
• I -
f
V
1 t 1 1 1 1 lH't 1
$2 S 22 5S SS '
1 1 t 1 1 1 1 1 I
«
f •"
1 1 > 1 1 1 1 1
1 1 1 1 1 1 1 I
/ H.
fl
1 t 1 1 t 1 1
A
M
*
£
•t
t*
5
r-
1%
s
s
"
s
:
;
cu
3
c
o
o
3
(JO
•H
NOI1WUN33N03
NOtiMUN33H03
231
-------
5
•
k -
•
r_
S S
I I I T I
t : s s s
s
5
,S
I
g 8 S 3 8 S = ".
S
M
I I I -I
: s s s s 2
s : s
3 5 S S
NDI1MUK&N03
O O .
u-l -H _
tfl (0 II
c *"*
O •w T3
•r-l C O>
j_j Ol l_i
Cfl O 3
k-l C V)
u O to
C <-> OJ
o> E
O Q)
a >-> -a
O « c
US *
cy 3 *.
O e/i
to
-a to
C
O
•H oo
nj c
T3 -1-1
tn
c
Q
O
tn c
" O
O B
CL, u —
W Q;
>> O o
>-i e c
o c S
4-1 Ul (J
(A 01
•H OJ <\j
^C "l O
I (0 C/l
01 Ol
6 ^
•H O XI
H tH v^.
o>
1-1
60
232
-------
s
X
•
:ss
s
•
I I I I I I I II
8
S
1 1 t J
ai
S2SSS?SS5S5=
-I
a »••••••••••
•»«OT«^«MV
-------
that time. Although the model does produce a slight increase in sulfate levels,
the magnitude is underpredicted by more than 10 /ig/m3.
The model also appears to overpredict the sulfate minima in the northeast
region on July 21, 1978 (Stations 1, 10, 11, 31, and 50 in Figure 10). A
possible explanation for the overprediction of the sulfate minima at these
stations may derive from the fact that they are all located near the coast.
During July 20 and 21, 1978, the prevailing wind direction for air flow aloft
was approximately parallel to the coast. However, an analysis of the surface
weather maps of this period shows an onshore flow during daytime hours. A
slight deflection of the onshore surface wind, most likely caused by the diurnal
land/sea breezes, could conceivably account for the observed dip in sulfate
concentrations.
An examination of Figures 6b through lOb indicates that several of the S02
concentration trends are well simulated by the model (e.g., Stations 12, 14, and
43), whereas in other cases, the S02 concentration trends are completely missed
by the model. At one-fourth of the monitoring stations, there is no obvious
overprediction or underprediction in S02 concentrations. Of the remaining
stations, the majority of the predictions are higher than the observations. It
is, however, not clear whether this overprediction tendency is related to the
siting of the rural stations. An analysis recently carried out on hourly S02
concentrations observed in the St. Louis RAPS network (Hanna, 1982) suggests
that the natural variability of hourly average S02 concentrations is greater
than a factor of 2. One would expect the variability of 24-h average S02
concentrations to be somewhat less. Thus, the lack of agreement between the
234
-------
measured and predicted S02 concentrations cannot be considered to be solely a
result of natural variability. Because S02 is a primary emission, its
distribution should be closely associated with the distribution of major
emitting sources. Thus, inadequate subgrid-scale treatment of point sources in
the model could possibly lead to the discrepancies between the predicted and
measured SO2 concentrations.
Model performance statistics were computed to quantitatively assess its
predictive capability. In general, the model tends to overpredict low
concentrations and underpredict high concentrations. This trend is more
pronounced for S02 than for sulfate. Nevertheless, the statistics show that
sulfate predictions have an overall bias of -2.3 /ig/m3 and a correlation
coefficient of 0.80 when compared with the measurements. Similar statistics for
S02 indicate a bias of -8.6 jig/m3 and a correlation coefficient of 0.42.
235
-------
APPENDIX. INCORPORATION OF AN AEROSOL MODULE
General Considerations of Aerosol Dynamics and Chemistry
The environmental effects associated with aerosols at the regional scale
involve the dry and wet deposition of sulfate and nitrate species, visibility
impairment, and health effects. We will briefly address these three major
issues before we discuss the state of the art in modeling aerosol dynamics and
chemistry.
Aerosols that are formed in the atmosphere include primary aerosols that
have been directly emitted into the atmosphere from natural sources (e.g.,
volcanoes, plants, soil, sea salt) or anthropogenic sources, and secondary
aerosols that result from the gas-to-aerosol conversion of condensable chemical
species that often have been formed through gas phase chemical reactions. An
atmospheric, aerosol population consists of three main modes that differ in
concentration and relative importance according to the conditions considered:
a nuclei mode (aerosols ranging from 0.001 to 0.01 ^m in diameter) that
corresponds to new nucleated aerosols; an accumulation mode (aerosols ranging
form 0.1 to 2 /jm in diameter) that corresponds to secondary aerosols formed by
gas-to-aerosol conversion and growth of nuclei aerosols; and a coarse mode
(aerosols larger than 2 pin in diameter) that corresponds to mechanically
generated aerosols and interacts little with the other two modes. The size
distribution of the aerosol population depends on the relative importance of
these three modes and on the dynamic processes that affect its evolution. The
chemical composition of the chemical processes that have led to secondary
236
-------
aerosol formation. The chemical composition is a function of the aerosol size
and varies as gas-to-aerosol conversion, aerosol-phase chemical reactions,
coagulation, sedimentation, deposition, emission, washout and rainout of
aerosols occur. The environmental effects of aerosols depend on both aerosol
size distribution and aerosol chemical composition.
Dry and wet deposition of sulfate and nitrate aerosols contribute to the
acidification of soils and watersheds. A detailed treatment of these processes
in a regional model may require a description of the aerosol size distribution
and chemical composition. The dry deposition velocity of the aerosol will
depend on ils size. Wet deposition occurs when aerosols are rained out or
washed out. Their initial chemical composition will determine to a certain
extent the kinetics of the droplet chemistry, because many oxidation reactions
are pH dependent. It is likely that rainout efficiency does not depend notably
on particle size. However, washout by cloud droplets under the cloud base will
be more effective for aerosols larger than 1.0 /im in diameter (removal by
inertial impaction) and for aerosols smaller than 0.1 /im in diameter (removal by
Brownian motion) than for aerosols in the intermediate size range of 0.1 to
1.0 jjm in diameter.
The modeling of regional haze necessarily requires the modeling of the
aerosol size distribution and chemical composition, because light scattering and
absorption depend on these aerosol characteristics. Light scattering occurs
primarily for fine aerosols in the 0.1- to 3.0-pm-diameter range. Light
absorption by aerosols depends on the amount of carbonaceous aerosol present.
237
-------
The health effects of aerosols are size dependent. Large aerosols will
general!v be deposited in the extrathoracic region, and small aerosols will be
removed by Brownian motion. Aerosols in the 0.2- to 15-/im-diameter range are
inhaled bv humans and can be deposited in the tracheobronchial airways. Based
on these considerations, the State of California Air Resources Board has issued
an air quality standard for aerosols less than 10 fjm in diameter, and the U.S.
Environmental Protection Agency is also considering the promulgation of a
standard for fine aerosols.
State of the Art in Atmospheric Aerosol Modeling
The evolution of the aerosol size distribution in the atmosphere is
governed bv the so-called General Dynamic Equation, which may be expressed as
follows for the aerosol number distribution n (v,t):
;m = - v(u-n) + vKvn + 1 j 0(v - v, v) n(v - v) n(v) dv
3t 2 °
- .) 0(v, v) n(v) n(v) dv + a_ n/^tv\ (A-l)
O . A.
+ S(v, t) - R(v, t).
where v is the aerosol volume and t is the time. The term on the left-hand side
represents the change in the aerosol number distribution with time. The first
term on the right-hand side, where u is the wind field vector, represents
advection; the second term, where K is the eddy-diffusivity tensor, represents
atmospheric diffusion; the third term, where 0 is the coagulation of aerosols of
238
-------
volume v, represents production of aerosols of volume v; the fourth term
represents the coagulation of aerosols of volume v; the fifth term represents
the growth of aerosols by gas-to-aerosol conversion; the sixth term represents
sources of aerosols by emissions or nucleation of condensable species; and the
last term represents removal of aerosols by sedimentation, surface deposition,
washout, and rainout.
The numerical solution of the General Dynamic Equation has been presented
by Gelbard and Seinfeld (1979). It is, however, computationally expensive, and
one must introduce some simplifying assumptions for atmospheric applications.
Two major approaches have been considered. Eltgroth and Hobbs (1979)
represented three modes of the aerosol size distribution by lognormal sulfate
aerosol. The approach, introduced by Gelbard, Tambour, and Seinfield (1980),
consisted of approximating the size distribution by a discrete approximation,
the so-called sectional representation. This concept was extended at Systems
Applications, Inc., and applied to sulfate aerosol formation (Seigneur, 1982)
and nitrate aerosol formation (Seigneur, Saxena, and Hudischewskyj, 1982) in
power plant plumes. Comparisons of model predictions with smog chamber and
atmospheric data have been satisfactory.
These models considered the dynamics of chemically inert aerosols. The
growth law of an aerosol due to chemical reaction in the aerosol phase, in
addition to gas-to-aerosol conversion, has been modeled and, for example, may
include a detailed treatment of the thermodynamic equilibria between the gas
phase and liquid phase, diffusion-limited condensation, chemical reactions in
the aerosol, and the deliquescence of ammonium nitrate or ammonium sulfates
239
-------
(Saxena, Seigneur, and Peterson, 1983). This approach has been applied to the
chemistry of cloud or fog droplets (e.g., Seigneur, Saxena, and Roth, 1983;
Jacob and Hoffmann, 1983). The absorption of gases by falling raindrops below
the cloud base has also been considered in theoretical studies (Adewuji and
Carmichael, 1982: Reda and Carmichael, 1982; Durham, Overton, and Aneja, 1981).
The coupling of aerosol dynamics and chemistry can be achieved by solving
Equation (15) when the growth law term includes the growth due to chemical
reaction in the aerosol phase. The formulation of such a model has been
presented by Bassett, Gelbard, and Seinfeld (1981).
Regional Modeling of Aerosols
Until now, modeling of sulfate and nitrate formation in regional models has
been limited to parameterized representations (e.g., pseudo-first-order
oxidation of SOz to sulfate). However, the use of modeling to evaluate emission
control strategies will require a more accurate description of the nonlinear
chemical and physical processes involved.
The incorporation of aerosol dynamics and chemistry into a regional model
appears, therefore, necessary. This model development effort must take into
account many interacting modules. First, it is necessary to couple the
gas-phase and liquid-phase chemistry. In present models, liquid-phase chemistry
is assumed not to affect gas-phase chemistry. Then, it is necessary to
introduce the effect of aerosol chemistry in the General Dynamic Equation
according to the formulation advanced by Gelbard and Seinfeld (1980). In
240
-------
parallel to the model development effort, model evaluation studies must also be
continuously conducted to assure that the modeling approach is correct. These
efforts are presently underway at Systems Applications, Inc.
The computational requirment of a regional-scale model will necessitate the
simplification of the aerosol dynamics and chemistry module. This must be done
carefully in order to maintain an adequate balance between reasonable
computational costs and model prediction accuracy. The simplifying assumptions
that can be made will allow to reduce the complexity of the General Dynamic
Equation. For instance, the number of aerosol size sections that constitute the
size distribution can be chosen to be minimal while still preserving the
calculational accuracy. Coagulation may generally be neglected for
regional-scale aerosol concentrations, since other processes such as
gas-to-aerosol conversion are likely to prevail. The chemical kinetic mechanism
for the aerosol phase may be limited to only the most important chemical
reactions.
The result of the incorporation of aerosol dynamics and chemistry into a
regional model will be a model that can provide quantitative information on
sulfate, nitrate, and cloud pH, while taking into account the major chemical and
physical processes that govern the formation of acidic species in the
atmosphere.
241
-------
REFERENCES
Adewuyi, Y. G., and G. R. Carmichael. 1982. A theoretical investigation of
gaseous absorption by water droplets from SC^-NHOa-COz-HCl mixtures.
Atmospheric Environment, 16:719-729.
Basset, M. , F. Gelbard, and J. H. Seinfeld. 1981. Mathematical model for
multicomponent aerosol formation and growth in plumes. Atmospheric
Environment, 15:2395-2406.
Benson, S. W. 1968. Thermochemical Kinetics. John Wiley and Sons, New York.
Businger, J. A., J. C. Wyngaard, Y. Izumi, and E. F. Bradley. 1971.
Flux-profile relationship in the atmospheric surface layer. Journal of
Atmospheric Science, 28:181-189.
Chamberlain, A. C. 1966. Transport of gases to and from grass and grasslike
surfaces. Proceedings of the Royal Society, A290:236-260.
Durham, J. L., J. H. Overton, and V. P. Aneja. 1981. Influence of gaseous
nitric acid on sulfate production and acidity in rain. Atmospheric
Environment, 19:231-240.
Eliassen, A. 1980. A review of long-range transport modeling, Journal of
Applied Meteorology, 19:231-230.
Eltgroth, M. W., and P. V. Hobbs. 1979. Evolution of particles in the plumes
of coal-fired power plants, II. A numerical model and comparisons with
field measurements. Atmospheric Environment, 13:953-975.
Galbally, I. E. 1974. Gas transfer near the earth's surface. Advances in
Geophysics, 188:329-340.
Gear, C. W. The automatic integration of ordinary differential equations.
Communications of the ACM, 14:176-179.
Gelbard, F., and J. H. Seinfeld. 1980. Simulation of multicomponent aerosol
dynamics. Journal of Colloid and Interface Science, 78:485-501.
Gelbard, F, , Y. Tambour, and J. H. Seinfeld. 1980. Sectional representation
for simulating aerosol dynamics. Journal of Colloid and Interface
Science, 76:541-556.
Gelbard, F., and J. H. Seinfeld. 1979. The general dynamic equation for
aerosols—theory and application to aerosol formation and growth. Journal
of Colloid and Interface Science, 68:363-382.
Hanna, S. R. 1982. Natural variability of observed hourly SOZ and CO
concentrations in St. Louis. Atmospheric Environment, 16:1435-1440.
242
-------
Heffter, J. L. 1965. The variation of horizontal diffusion parameters with
time for travel periods of one hour or longer. Journal of Applied
Meteorology, 4:153-156.
Hill, A. C. 1971. A sink for atmospheric pollutants. Journal of Air Pollution
Control Association, 21:341-346.
Hoegstroem, U. 1975. Further comments on the long range transport of airborne
material and its removal by deposition and washout. Atmospheric
Environment, 9:946-947.
Jacob, D. J., and M. R. Hoffmann. In press. A dynamic model for the production
of M+, N03~ and S042~ in urban fog. Journal of Geophysical Research.
Killus, J. P., R. E. Morris, and M. K. Liu. 1983. Application of a Regional
Oxidant Model to the Northeast United States, International Conference on
Long Range Transport Models for Photochemical Oxidants and Their
Precursors. Research Triangle Park, North Carolina.
Lonneman, W. A. 1977. Ozone and Hydrocarbon Measurements in Recent Oxidant
Transport Studies. International Conference on Photochemical Oxidant
Pollution and Its Control—Proceedings, Volume I. EPA-600/3-77-001a, U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina.
Owen, P. R., and W. R. Thompson. 1963. Heat transfer across rough surfaces.
Journal of Fluid Mechanics, 15:321-324.
Pasquill, F. 1974. Limitations and prospects in the estimation of dispersion
of pollution on a regional scale. Advances in Geophysics, 188:1-14.
Reda, M., and G. R. Carmichael. 1982. Non-isothermal effects on S02 absorption
by water droplets, I. Model development. Atmospheric Environment,
16:145-150.
Reynolds, S. D., P. M. Roth, and J. H. Seinfield. 1973. Mathematical modeling
of photochemical air pollution. Atmospheric Environment, 7:1033-1061.
Saxena, P., C. Seigneur, and T. W. Peterson. In press. Modeling of multiphase
atmospheric aerosols. Atmospheric Environment.
Seigneur, C., P. Saxena, and P. M. Roth. 1983. Proceeding of the 76th Annual
Meeting of Air Pollution Association Annual Meeting, Atlanta, Georgia, June
19-24, 1983.
Seigneur, C. 1982. A model of sulfate aerosol dynamics in atmospheric plumes.
Atmospheric Environment, 16:2207-2228.
Seigneur, C., P. Saxena, and A. B. Hudischewskyj. 1982. Formation and
evolution of sulfate and nitrate aerosols in plumes. Science of the Total
Environment, 23:283-292.
243
-------
Sehmel, G. A., S. L. Sutter, and M. T. Dana. 1973. Dry Deposition Processes.
In: Pacific Northwest Laboratory Annual Report for 1971 to the U.S. Atomic
Energy Commission, Division of Biomedical and Environmental Research,
Volume II: Physical Sciences, Part 1, Atmospheric Science. BNWL-1751,
Battelle Northwest Laboratories, Richland, Washington, pp. 150-153.
Simple, G. W., et al. 1977. Long Range Airbone Measurements of Ozone off the
Coast of the Northeastern United States. International Conference on
Photochemical Oxidant Pollution and Its Control, EPA 600/3-77-OOla, U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina.
Stewart, D. A., et al. In press. Evaluation of an episodic regional transport
model for a multi-day sulfate episode. Atmospheric Environment.
Thorn, A. S. 1972. Momentum, mass and heat exchange of vegetation. Quarterly
Journal of the Royal Meteorological Society, 98:124-134.
Vaughan, W. M., et al. 1982. A study of persistent elevated pollution episodes
in the Northeastern United States. Bulletin of the American Meteorological
Society, 63:258-266.
Whitten, G. Z., J. P. Killus, and H. Hogo. 1980. Modeling of Simulated
Photochemical Smog with Kinetic Mechanisms. Report No. EF79-129, Systems
Applications, Inc., San Rafael, California.
Wojcik, M., et al. 1978. Development and Evaluation of a Mesoscale
Photochemical Air Quality Simulation Model. Report No. EI78-118, Systems
Applications, Inc., San Rafael, California.
244
-------
STEM MODEL FOR THE REGIONAL TRANSPORT OF PHOTOCHEMICAL OXIDANTS
AND THEIR PRECURSORS*
Gregory R. Carmichael
Chemical and Materials Engineering Program
University of Iowa
Iowa City, Iowa (USA) 52242
Toshihiro Kitada
School of Regional Planning
Toyohashi University of Technology
Toyohashi 440 (Japan)
Leonard K. Peters
Department of Chemical Engineering
University of Kentucky
Lexington, Kentucky (USA) 40506
INTRODUCTION
The relationships between emissions and the distribution of photochemical
oxidants and their precursors are complex and not fully understood. Many trace
species may be transported long distances (tens to thousands of kilometers) from
source areas, resulting in high ambient levels over broad regions. In addition,
many of these compounds have been related to a number of environmental problems,
including acid rain, tropospheric haze, and reduced visibility, and they have
been correlated with a variety of adverse health indicators.
The distribution of these species in the atmosphere is the result of
complex interactions among emissions (anthropogenic and natural), prevailing
meteorology, chemical transformation, and removal mechanisms. Regional-scale
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
245
-------
transport/chemistry models that describe the circulation of photochemical
oxidants and their precursors in the troposphere can be extremely beneficial in
understanding the physical and chemical processes occurring between the sources
and sinks of these pollutants.
In this paper, a combined transport/chemistry model for the regional-scale
transport of photochemical oxidants and their precursors is described. Called
STEM (sulfate transport emissions model), the model is Eulerian and three-
dimensional, and is an extension of an operational SOX transport model developed
by the authors (Carmichael and Peters, 1981). This second-generation model
handles 40 species, 19 of which are advected. The advected species are NO, N02,
S02, S042~, 03, HN03, NH3, PAN, H202, HCHO, HC, (alkanes), C2H4, HC2 (alkenes),
HC3 (aromatics), RCHO, ROOH, HN02, RON02, and R02N02. Both heterogeneous and
homogeneous chemical reactions and wet and dry removal processes are modeled. A
finite-element numerical method is used in the model. Model formulation and
initial test results are presented in the following section.
MODEL DESCRIPTION
The regional transport of photochemical oxidants and their precursors is
modeled within an Eulerian framework. A block description of the model is
presented in Figure 1. Forty chemical species are included in the analysis.
Nineteen of them are sufficiently long lived under certain circumstances that
they must be treated as advected species. The remaining species are short lived
and are modeled by steady-state methods. The mathematical model contains no
regional or area-specific information, and the
246
-------
Prognostic Eq.s. with B.C.s for 18 Species
Diagnostic Eqs. for 21 Species
Chemistry
( Homogeneous
and
Heterogeneous)
Advection/
Diffusion
Removal
Process at
Earth's
Surface
Objective Method for
Mass-Con. Wind Field
1-D Turbulent B.L.M.
Measured Met. Data
u,V, T, H20, Cloud (Cover index, height, average cloud drop size,
number density), Ram (intensity, average rain drop size ),
Solar Zenith Angle, Earth's Surface (type, roughness)
Figure 1. Schematic of model construction.
number of grids and the grid spacing can be chosen for a particular application.
The mathematical analysis is based on the coupled, three-dimensional advection-
diffusion equation:
6CJ! + 6(UlCe) = 5
fit fiX, 6Xj
(1)
where C is the concentration of speciesfi , U, is the velocity vector, KJJ is
the eddy diffusivity tensor (K,j=0 for i/j has been assumed), Rg is the rate of
formation or loss by chemical reaction, and SB is the emission rate.
247
-------
The model actually utilizes the surface-following coordinate system shown
in Figure 2. The vertical region, including topographical features, is mapped
into a dimensionless rectangular region according to:
Zk - h(x,y) h(y) < Zk < h(x,y) + H2(x,y,t),
H2 (x,y,t)
and
(2)
Pk
k - 1 xq 0 < pK < 1, k = 1, KGRID
'KGRID-1
(3)
where k (subscript) = the vertical grid number,
KGRID = the total number of grids, and
a = a parameter controlling the grid spacing.
When a = 1, the grid spacing in the dimensionless coordinate is uniform; when
« > 1, there is higher resolution near the surface. Our current application
uses « = 2 and H2(x,y,t) = 8 km. The height of the region is chosen so that
boundary-layer-free troposphere processes and interactions can be modeled.
"V°KGRID=1
-h(x,y)
Figure 2. Suface-following coordinate system used in the model.
248
-------
Chemistry
Both gas-phase and liquid-phase chemical reactions are treated in the
model. The homogeneous gas-phase mechanism used is summarized in Table 1. The
mechanism involves 84 reactions and 40 species. NO, N02, HN03, NH3, S02, S042~,
HC, (alkanes), C2H4, HC2 (alkenes), HC3 (aroraatics), 03, PAN, HCHO, RCHO, H202,
RON02, and R02NO2 are treated as transported species, and the remainder are
treated as pseudo-steady-state species. The extensiveness of the chemical
mechanism enables the modeling of urban chemistry as well as nonurban
tropospheric chemistry.
The model's treatment of the chemistry also includes the interactions
between the gas-phase chemistry and the heterogeneous removal processes. Thus,
additional reactions of the form noted below are added to the gas-phase chemical
mechanism. Two types of heterogeneous removal processes are treated in the
model: wet removal (both in-cloud and below-cloud) and deposition on aerosol
surfaces, i.e.,
i C, + Products, (4a)
(g) (8)
and
| C, + Products, (4b)
(g) (a)
249
-------
TABLE 1. HOMOGENEOUS GAS-PHASE KINETICS MECHANISM USED IN THE
STEM MODEL
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
13.
19.
20.
21.
22.
23.
24.
25.
:e.
27.
28.
29.
30.
31 .
32.
33.
34.
35.
36.
37.
38.
39.
40.
N02 4- hv - NO 4- 0 (3P)
0 (3P) + )2 + M - 0, 4- M
03 + NO - NOj 4- 02
NOj 4- 0 (3P) - NO 4- 0,
N02 4- 0 (3P) - N03
NO 4- 0 (3P) - N02
N02 4- 03 - N03 + 02
N03 4- NO - 2 N02
N03 4- N02 - Nj 05
N205 - N02 4- N03
N205 + H20 - 2HON02
NO + N02 4- H20 - 2HONO
HONO 4- HONO - NO 4- N02 + H20
03 4- hu - 0, + 0 CD)
03 4- hu - 02 4- 0 (3P)
0(1D) 4- M - 0 (3P) 4- M
0 CD) 4- H20 - 20H
HOZ 4- N02 - HONO + 02
HOj 4- N02 - H02N02
H02N02 - H02 + N02
HOZ 4- NO - N02 4- OH
OH + NO - HONO
OH 4- N02 - HON02
HONO 4- hu - OH -I- NO
CO 4- OH - C02 + H02
OH + HONO - H20 4- N02
H02 4- H02 - H202 -t- 02
H202 + hu - 20H
OH + H02 - H20 + 02
OH + 03 - H02 + 02
H02 -t- 03 - OH -(- 202
HCO + hu - H02 + HCO
HCO + hu - H2 + CO
HCHO + OH - HCO (+H20)
HCO + 02 - H02 (+CO)
RCHO + hu - R02 -f HCO
RCHO + OH - RC03
01 (olefin) + OH - RO2
01 + 0 - ROj + RC03
01 + 03 - 0.5 f HCHO + (1 - 0.5 () RCHO
4-0.5 (c) ({ +• 0.5 i,) R0t
+0.25 (c)(£ + r,) H02 f 0.25 (c{)OH
43.
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
56.
57.
58.
59.
60.
61.
62.
63.
64.
65.
66.
67.
68.
69.
70.
71.
72.
73.
74.
75.
76.
77.
78.
79.
80.
81.
82.
83.
84.
C2H. + OH - R02
C2H, + 0 - R02 + HCO
RO - « H02 + (1 - a) R02 + a HCHO + i RCHO
NO + RO - RONO
RONO + hu - RO +• NO
N02 + RO - RON02
N02 + RO - RCHO + HONO
N02 + R02 - R02N02
N02 4- R02 - RCHO + HON02
R02 N02 - N02 +• R02
NO -H R02 - N02 +• RO
NO +• RC03 - N02 + R02
N02 + RC03 - PAN
PAN - N02 + RC03
Arc + OH - R02 4- RCHO
S02 4- H02 - HO 4- S03
S02 4- HO - HS03
S02 4-0 (3P) - S03
S02 4- R02 - RO 4- S03
S02 4- RO - R S03
S02 4- RCHOO - RCHO 4- S03
S02 - Part
S03 4- H20 - H2 SO, (Part)
HS03 4- 02 - HS05
HS05 4- H20 - HS05 (H20) (Part)
RS03 4- 02 - RS05
RS05 4- H,0 - RS05 (H20)
RCHOO 4- H20 - RCOOH 4- H.,0
NH3 4- OH - NH2 4- H20
NK2 4- 02 'M NH202
NH2 02 4- NO - NH20 4- N02
NH2 02 4- OH - NH2 OH 4- O2
NH20 4- 02 - HNO 4- H02
HNO 4- OH - NO 4- H20
HNO 4- 02 - NO 4- H0?
NH3 4- HN03 M NH4 N03
HN03 4- OH - N03 4- H20
HN03 4- hu - N02 4- OH
N03 4- hu - N02 4- 0
R02 4- H02 - ROOH 4- 02
ROOH 4- n - RO 4- HO
ROOH 4- OH - R02 4- H20
4-0.5 (1 - c) RCHOO
4-0.25 (cij) RO 4- 0.25 («i,) HCO
41. Alk 4- OH - R02
42. Alk 4- 0 - R02 4- OH
250
-------
where k,wet = the wet removal coefficient, and
k.hei = the removal coefficient for deposition on aerosol surfaces.
The wet removal coefficient, kiwei, is calculated in the model from a
parameterization based on the liquid water content of the cloud, cloud
temperature, characteristic droplet size, number density of cloud droplets,
cloud pH and rainfall intensity, and the chemical and physical properties of the
absorbed species (i.e., Henry's Law constant, gas-phase diffusivity, and
dissociation and redox reactions in solution). Also included is the
liquid-phase generation of species like sulfate. The sulfate production rate in
clouds, which is calculated by using the above parameters, is based on the
reactions S(IV) + 03 and S(IV) + H202. (See Hong and Carmichael, 1982, for a
complete description of the calculation of kjwet)- The wet removal rates for
some soluble species, calculated by the above procedure for in-cloud conditions
with a cloud temperature of 0°C, a droplet size of 30 /jm, a pH of 4.7, and a
liquid water control of 1.5 g/m3, are summarized in Table 2. The rates reflect
the efficiency with which cloud droplets remove these highly soluble species.
The removal coefficient for deposition on aerosol surfaces is calculated by
using classical mass transfer theory and gas kinetic theory. It is described in
detail by Luther and Peters (1982). In this treatment, the aerosol particles
are assumed to be spherical and are described by the physical and chemical
properties of water; desorption phenomena are not included. Values of kjhet for
selected compounds, calculated by using the aerosol residence time distribution
of Jaenicke (1978) and the aerosol size distribution of Junge (1963) with a
251
-------
TABLE 2. TN-CLOUD REMOVAL RATES FOR SOLUBLE
SPECIES CALCULATED BY THE PARAMETERIZATION
PROCEDURE DISCUSSED IN THE TEXT
Species (s~1)
HN03 0.06
NH3 0.102
S02 0.28 x 10~4
S042~ -0.28 x 10"4 (production rates)
PAN 0.018
HCHO 0.086
RCHO 0.071
H202 0.088
HN02 0.07
surface area of 520 pm2/m3, and an accommodation coefficient of 1, are presented
in Table 3. These values represent upper limits.
The chemical reaction rate constants are calculated in the model and vary
with temperature and photon flux. The effects of clouds on the photon flux are
included in the calculation by using an empirical correlation relating photon
flux to cloud cover (Kaiser and Hill, 1976); i.e.,
G = Gc (1 - A C1-75), (5)
252
-------
TABLE 3. CALCULATED HETEROGENEOUS
LOSS CONSTANTS FOR AEROSOL/GAS
INTERACTIONS
khel
Species (s~1)
NO 1.4 x 10~6
03 6.7 x 10~6
OH 1.1 x 10-1
H202 6.2 x 10~2
HN03 6.5 x 10"2
N03 6.7 x 10~2
02 7.0 x 10~7
N02 7.6 x 10~2
where Or = the photon flux for clear sky,
C = the cloud cover fraction, and
A = a constant dependent on cloud type and has a value of 0.55 for
fair-weather cumulus clouds or patches of cirrus or cumulus clouds.
Furthermore, the enhancement of photon flux due to cloud albedo is included
in the analysis by multiplying the clear-sky flux at grid points directly above
the cloud level by a factor of 1.3.
253
-------
Numerics
The simulation of regional transport/chemistry described by Equation (1)
requires numerical integration. The method used in the model is a combination
of the concept of fractional time steps and that of one-dimensional finite
elements. This is referred to as the Locally-One-Dimensional, Finite-Element
Method (LOD-FEM). The LOD procedures (Mitchell, 1969; Yanenko, 1971) split the
multidimensional partial differential equation into time-dependent,
one-dimensional problems, which are solved sequentially.
The time-split equations are of the form:
6Ce + Lx CB = 0 (6)
Ly Cs = 0, (7)
fit
Lz Ca = Ss , (8)
St
254
-------
and
6Cg = RE , (9)
6t
where the L's represent the one-dimensional operators (e.g.,
LXCC = 6(U C8)/6X - S/&x(Kxx sCB/6x)).
The chemical reaction term is treated separately, because many of the
reactions have time scales much smaller than that for the transport. Splitting
out the reaction term allows different time steps to be used for the transport
and the chemistry processes. The time steps used in current applications are
0.1 min for the chemistry equations and 15 min for the transport.
The transport equations are solved by a modified Galerkin finite element
method (FEM) using asymmetric weighting functions and the Crank-Nicholson
approximation for the time derivative. The chemistry equations are solved by a
pseudo- linearization procedure, which gives analytical approximations to the
equations. The solution procedures for the transport and chemistry are
discussed in more detail below.
255
-------
Solution of the Chemistry Equations—
The technique used to integrate the chemistry of the transported species is
an adaption of the semi-implicit Euler method proposed by Preussner and Brand
(1981). With this technique, Equation (9) can be written in the following form:
dCe * -Ce £ dj '' n Ck + £ pg ' II Cm, (10)
dt j k#C i m
where de = the rate constant of the destruction of species £ by reaction j,
p, = the rate constant of the production of species fi by reaction i,
k and m = the reactant species involved in the destruction and production
reactions.
Under the assumption that all species concentrations except C are known
(either at the previous time) or have just been calculated, Equation (10) is of
form dCe/dt + DCV = P, which has the analytic solution:
' P
t, = ati D
-I- (C - P)
-DAt,,
e (11)
t,
where C^ It, at t = 0 is the initial concentration of species K. This procedure
has the important physical properties that Cc can never become negative provided
that the reaction rates are positive and the initial values are nonnegative, and
256
-------
that in the limit, as t - ->, the proper equilibrium concentrations are obtained,
i.e.,
C8eq " ? • i m . (12)
The above equation is used in the model to calculate only the advected
species. However, terms P and D also contain the short-lived species (e.g., OH,
H02, etc.). These species are calculated by using the pseudo-steady-state
approximations. The use of these equations results in a set of algebraic
equations that depend on the advected species concentrations. The calculation
procedure to advance from tj to t, + At, uses CE j t, to calculate the
short-lived species concentrations and then uses these concentrations to advance
Cs | t to Cs | t, + Ati using Equation (11).
The accuracy of this technique depends on the species of interest and
At,. For long-lived species (e.g., S02, CO, and CH4), relatively large Ati can
be used, but for relatively reactive species, much smaller time steps are
required. Figure 3 displays afternoon N02 concentrations, calculated by using
the kinetic mechanism summarized in Table 1. The results obtained by using
CSMP, the semi-implicit Eulerian method (Preussner and Brand, 1981), and the
"analytic" method just described are compared. After a 5-h simulation time, the
(the time step used to solve the transport equations), the results differed by
less than 1%. The analytic method with the free radical solver can execute
faster than CSMP by a factor of 2.
257
-------
o.ozo
0.015
\
- SEE
DETAIL A
E
Q.
O.
0.010
0005
13
CSMP RESULT
"ANALYTIC" Aim,, • 0.) min.
SEMI-IMPLICIT Atmo» > 0 lmm
"ANALYTIC" Almo,"05min
3O 14.30 1530 16.30
TIME
17.30
18.30
Figure 3. Afternoon N02 concentrations calculated by using the kinetic
mechanism shown in Table 1.
Solution of the Transport Equations--
Thp transport equations with the boundary conditions described below are
solved bv using a Crank-Nicolson/Galerkin finite element method with piecewise
linear trial functions, asymmetric weighting functions, and a filter to
eliminate high-frequency numerical noise. Selection of the finite element
method was based on the results of one- and two-dimensional numerical
experiments comparing available numerical methods (Carmichael et al., 1980;
Chock and Dunker, 1982; Pepper et al., 1980). The filter used is described by
McRae et al. (1982).
258
-------
The boundary conditions used in the model at the earth's surface and at the
upper boundary are, respectively:
-K vCe) • nh = Qe - Vd,eCc (13)
and
(VCP - K 7C ) • nH = F0 , (14)
where K = a diagonal matrix of eddy diffusivities:
Qe = the surface flux of species £ ;
V
-------
and
(if nb • V < 0) (VC - K
-------
Objective Analysis of Wind Field—
Typically, only the horizontal wind fields are measured, and these
measurements do not satisfy the continuity equation. A non-mass-conservative
wind field introduces an artificial pseudo-first-order loss or generation term
into the species mass transport equation (Kitada et al., 1982). In this model,
an objective analysis procedure based on variational calculus is being used to
obtain a three-dimensional mass-consistent wind field (Sasaki, 1970; Peters et
al., 1979). Kitada et al. (1983) has applied this analysis to the Mikawa Bay
area in Japan. The area, shown in Figure 4 is 60 km x 60 km. In this
particular application, a vertical region of 500 m was chosen. As shown, the
northwestern part of the region is blocked by mountains that extend beyond the
500-m vertical region. For the numerical calculations, the region was divided
into 20 x 20 horizontal grids and 10 vertical grids. Surface wind data were
available at the 27 observation locations, shown by the solid circles in
Figure 4. The initial horizontal surface wind was interpolated by using a
weighting factor of 1/r2 with a maximum radius of influence equal to twice the
average separation distance between the observations points. The initial upper
winds were obtained by extrapolating the surface data with a power law profile.
The derived winds at 1500 h local time on July 4, 1973, are shown in
Figure 4, as well as the horizontal wind field at 200 m above sea level and the
vertical wind fields along the A-A and B-B cross sections. These results
demonstrate that this analysis can reproduce reasonable three-dimensional flow
fields with substantial topographic features from limited horizontal input data.
This analysis is currently being used to generate regional-scale flow field.
261
-------
— 3
—3
60
_ S6
5
0 •
0 12
(80 M/S
8.0 M/S
24 36
X(KM)
48
60
(a)
(b)
_ 40O
2
200
O 12 24 36
^0 20 M/S X(KM)
BO M/S
60
_ 400
3
0 12
(O.20 M/S
8.0 M/S
24 36
V(KM)
60
(c)
(d)
Figure 4. Three-dimensional wind fields for Mikawa Bav, Japan, based on the
objective analysis procedure, (a) 20 x 20 horizontal grids, 27
observation points, (b) horizontal wind field at 200 m above sea
level. (c) vertical windfield along A-A cross section, (d) vertical
wind field along B-B cross section.
262
-------
Drv Deposition Velocities—
Dry deposition velocities are calculated from estimates of S02 aerodynamics
and surface resistances that are based on surface wind speed, surface roughness,
surface evaporation rate, and stability (Carmichael and Peters, in press). The
calculated values vary both temporally and spatially. The averaged calculated
value for the Eastern United States for the period July 4 to July 10, 1974
(average includes day and night values over 864 grid points) is VdS02 =0.44
cra/s (at 20 to 30 m). Deposition velocity values for the other trace gases are
obtained by scaling the S02 deposition velocity by a factor determined by
measured deposition velocities, estimated aerodynamic resistance, and Henry's
Law constants. Table 4 presents calculated values for selected species under
conditions when the dry deposition velocity of S02 is 0.5 cm/s.
Mixing Layer Analysis—
The distribution of trace gases in the atmosphere is greatly influenced by
the dynamic behavior of the boundary layer. In this model, the one-dimensional
boundary-layer model developed by Yamada and Mellor (1975) is used to describe
the diurnal boundary layer (specifically used to calculate Kv profiles).
Examples of vertical eddy diffusivity profiles generated by this analysis are
shown in Figure 5. This model can include large-scale meteorological features
263
-------
TABLE 4. CALCULATED DRY DEPOSITION VELOCITIES OF
SELECTED SPECIES FOR CONDITIONS WHERE VdS02 =0.5 cm/s
Species
NO
N02
HN03
NH3
S042"
Dry
Deposition
Velocity
(cm/s) Species
0.01 HC3 (aromatics)
0.01 03
1.0 PAN
0.67 HCHO
0.2 H202
Dry
Deposi t ion
Velocity
(cm/s)
0.01
0.01
0.80
0.52
1.0
for 10:00
0 '0 29 30 «0 JO «0
100 200 300 «OC
Vwtical Cdtfy
t, (m'/we)
I3:OO ondiTrOO
Figure 5. Example of vertical eddy diffusivity
profiles from the one-dimensional
turbulent boundary-layer model of
Yamada and Mellor (1975).
264
-------
(i.e., mesoscale or synoptic-scale) through a geostrophic wind term. The
one-dimensional calculation is performed at each grid point, but it is still
considerably less expensive than a three-dimensional boundary-layer model.
RESULTS AND DISCUSSION
The STEM model for the regional transport of photochemicals is currently
undergoing extensive testing. A one-dimensional (Z,t) version that includes all
processes except horizontal transport is being used to access the effects and
importance of the various tropospheric processes (i.e., heterogeneous removal,
homogeneous chemistry, stratospheric input, etc.) on the distribution of trace
gases in the troposphere.
Table 5 shows predicted Oa and OH values at selected times and elevations
for the following cases:
Without heterogeneous removal, with stratospheric sources, surface
sources excluding HC sources and no clouds;
With heterogeneous removal, with stratospheric sources, surface sources
excluding HC sources and no clouds;
With heterogeneous removal, with stratospheric sources, surface sources
including HC sources and no clouds; and
With heterogeneous removal, with stratospheric sources, surface sources
including HC sources and clouds.
As shown, the 03 and OH concentrations decrease when heterogeneous removal
is included in the model (see runs 1 and 2). Without heterogeneous removal,
there is a higher concentration of the soluble species (e.g., H202, HN03, HCHO,
265
-------
TABLE 5. CALCULATED 03 AND OH CONCENTRATIONS FOR VARIOUS CONDITIONS
USING A ONE-DIMENSIONAL VERSION OF STEM
Conditions
Oa at surface
(ppb)
03 at 1.25 km
(ppb)
OH at surface
(molecules/cm3)
I
OH at 3 km
(molecules /cm3)
Time
(h)
0700
1200
1800
2400
0700
1200
1800
0700
1200
1800
2400
0700
1200
1800
2400
1
17
58
176
~0
31
65
205
5.4(3)
3.8(5)
5.2(4)
1.3(3)
4.4(3)
9.0(4)
3.9(4)
5.5(3)
Case
2
17
23
13
-
31
31
26
3.2(3)
1.2(4)
1.0(3)
-
8.1(2)
5.8(3)
4.0(3)
-
;a
3
22
63
75
~0
31
49
86
1.3(5)
2.6(5)
5.4(4)
2.9(4)
3.3(2)
5.2(3)
5.1(3)
8.3(2)
4
16
42
43
~0
31
41
24
8.9(4)
1.4(5)
3.6(4)
3.0(4)
2.6(2)
7.3(3)
1.2(4)
5.5(2)
aCases are defined as follows: 1, without heterogeneous removal, no
clouds, with stratospheric sources and surface source excluding
HCs; 2, heterogeneous removal included; 3, HC sources included; and
4, cloud at level 6 included.
266
-------
Table 5 shows predicted 03 and OH values at selected times and elevations
for the following cases:
• Without heterogeneous removal, with stratospheric sources, surface
sources excluding HC sources and no clouds;
• With heterogeneous removal, with stratospheric sources, surface sources
excluding HC sources and no clouds;
• With heterogeneous removal, with stratospheric sources, surface sources
including HC sources and no clouds; and
• With heterogeneous removal, with stratospheric sources, surface sources
including HC sources and clouds.
As shown, the 03 and OH concentrations decrease when heterogeneous removal
is included in the model (see runs 1 and 2). Without heterogeneous removal,
there is a higher concentration of the soluble species (e.g., H202, HN03, HCHO,
PAN, etc.) and higher free radical concentrations (e.g., OH, H02, R02, etc.).
F.n addition, the NO concentration is lower and the N02 concentration is higher
without heterogeneous removal. Furthermore, the 03 and OH concentrations
increase when organic surface sources are included (see runs 2 and 3).
When clouds are included, the 03 and OH concentrations decrease below the
cloud due to a decrease in solar actinic flux in this region. The OH
concentration immediately above the cloud increases due to backscattering (see
runs 3 and 4).
Ozone concentrations predicted over a 48-h period £or the conditions
corresponding to case 4 in Table 5 constant surface sources) are shown in
Figure 6. In the lower troposphere (up to approximately 2 km), a diurnal
267
-------
"•**••*•""••;
Figure 6.
Predicted 03 concentrations over a 48-h period for conditions
given for Case 4 in Table 5.
profile is depicted, in which the 03 concentration reaches maximum level in the
late afternoon and minimum at night. Surface 03 concentrations gradually fall
to near zero in the early evening and increase sharply at sunrise.
A two-dimensional version of the STEM model for photochemicals is being
applied to a coastal region in Japan. Figure 7 shows sulfate profiles
calculated by using flow fields derived from a land/sea breeze model (Kitada et
al.. 1983). The land/sea interface is located at X= -5 km (X < - 5 km
represents the sea region), and initial profiles of the primary pollutants
representative of the land and sea regions were assigned. Shown are the wind
field (Figure 7) and the concentration profiles at 9:00 a.m. for conditions with
and without clouds (clouds located at z = 0.25). The without-cloud contours
268
-------
Tint 9.0
-70.0 -M.O -M.O -IO.C 10.0 M.O ' w'-O ' 70.0
»,5« X (KM)
(a)
-70.0 -5C 0 -30.0 -10.0 IC.O 30.0 50.0 70.0
X C KPN
-7C C -50.0 -30.0 -IC.C 10.C 3C.O SO.D 70.0
(b)
(c)
Figure 7. Predicted sulfate concentrations for a land/sea breeze situation.
Initial conditions were specified at 6:00 a.m. Concentrations are
x 102 ppb. (a) wind field, (b) concentration profile without
clouds, (c) concentration profile wi£h clouds.
269
-------
show the circulation patterns at 9:00 a.m. A strong surface inland flow is
predicted at the land/sea interface, with a strong circulation cell extending to
z =< 0.25. The resulting sulfate contours depict this general circulation. In
the absence of clouds, the sulfate peak is located about 10 km inland and z =*
0.1. In the presence of clouds, the peak is located at the land/sea interface
at cloud level. The sulfate levels are lower in the below-cloud region due to
the reduced solar actinic flux and the lower S02 concentrations resulting from
the removal of S02 from the air mass as it circulates through the clouds.
In-cloud sulfate production is shown along z = 0.25.
The computation time for this two-dimensional code with 330 grid cells,
that is, (v,z) = (30,11), for a 24-h simulation is approximately 940 s on a
FACOM M200 computer.
Three-dimensional simulations using the entire model have been successfully
performed for a short period over a limited region on an IBM-370. However, we
are currently using a simplified three-dimensional version for two transport
species and 27 chemical reactions) (actually the first-generation STEM) to study
S02/sulfate transport in the eastern United States (Carmichael and Peters,
1983). Figure 8 displays predicted 24-h averaged S02 and sulfate concentrations
calculated with July 4, 1974, meteorological data. Surface concentrations and
vertical profiles along the indicated slices are displayed. The CPU computation
time for 9,504 grid cells, that is, (x,y,z) = (27,32,11), is approximately 6 s
per time step, or = 576 s/24-h simulation on the NASA-Langley Cyber 203
computer.
270
-------
-1 -
^-r
"~"~i / ( ~' ^
T \ ^ Ns.X
Figure 8. Predicted 24-h averaged S02 and sulfate concentrations for conditions
on July 4, 1974. Top figures and Z-X slices correspond to Ly = 20
(top) and 25 (bottom).
271
-------
CONCLUSIONS
A combined transport/chemistry model for the regional-scale transport of
photochemicals and their precursors has been described. The model, which is
Eulerian and three-dimensional, is an extension of an operational SOX transport
model developed by the authors. This second-generation model treats 19 species
as transported species, including NO, N02, S02, S04=, 03, HN03, NH3, PAN, and
HC. Both heterogeneous and homogeneous chemical reactions are modeled, as are
both wet and dry removal processes. The model employs a finite element
numerical method and performs the nonlinear chemistry with a
pseudo-linearization scheme, resulting in virtually decoupled numerical
calculations.
The model is sufficiently detailed to simulate urban boundary layers as
well as non-urban-free troposphere chemistry, boundary-layer-free troposphere
exchange in cloud-free and cloudy environments, and in-cloud and below-cloud wet
removal and chemistry processes. The model is capable of addressing leading
regional-scale scientific as well as regulatory problems.
ACKNOWLEDGMENTS
This research was supported in part by the National Aeronautics and Space
Administration under Research Grant NAG 1-36, and the EPA/MAP3S Program through
the Battelle Northwest Laboratory. Travel to the United States for T. Kitada
was supported by the Japanese Ministry of Education through the Visiting
Research Fellowship at Foreign Countries Program. Special thanks go to
272
-------
Seong-Yeon Cho for helping with the computer analysis, Kay Chambers for making
the line drawings, and Bev Palmer for typing the manuscript.
REFERENCES
Carmichael, G. R., and L. K. Peters. In press. An Eulerian Transport/
Transformation/Removal Model for S02 and Sulfate, Part I and Part II.
Atmospheric Environment.
Carmichael, G. R., T. Kitada, and L. K. Peters. 1982. A Second Generation
Combined Transport/Chemistry Model for Regional Transport of SOX and NOX
Compounds. In: Proceedings of the Thirteenth International Technical
Meeting on Air Pollution Modelling and Its Application, Preprint volume,
pp. 180-190.
Carmichael, G. R., and L. K. Peters. 1981. Application of the Sulfur
Transported Eulerian Model (STEM) to a SURE Data Set. In: Proceedings of
the Twelfth International Technical Meeting on Air Pollution Modeling and
Its Application, pp. 324-347.
Carmichael, G. R., T. Kitada, and L. K. Peters, 1980. Application of a Galerkin
finite element method to atmospheric transport problems. Computers and
Fluids, 8:155.
Chock, D. P., and A. M. Dunker. In press. A comparison of numerical methods
for solving the advection equation. Atmospheric Environment.
Hong, M. S., and G. R. Carmichael. 1982. An investigation of sulfate
production using a flow-through chemical reactor approach. In:
Proceedings of the Second Symposium on the Composition of the Nonurban
Troposphere, Preprint volume, pp. 137-140.
Jaenicke, R. 1978. Uber die dynamik atmospharische-aitKenteilchen.
Bunsengesellschaft fuer Physikalische Chemie. Berichte, 82:1198-1202.
Junge, C. E. 1963. Air Chemistry and Radioactivity. Academic Press, New York.
382 pp.
Kaiser, J. A. C., and R. H. Hill. 1976. Irradiance at sea. Journal of
Geophysical Research, 81:395.
Kitada, T., A. Kaki, H. Ueda, and L. K. Peters. In press. Estimation of
vertical air motion from limited horizontal wind data—a numerical
experiment. Atmospheric Environment.
273
-------
Kitada, T., G. R. Carmichael, and L. K. Peters. 1983. The Locally-One-
Dimensional Finite Element Method for Atmospheric Transport/Chemistry
Calculations. Presented at the Third International Symposium on Numerical
Methods for Engineering, March 1983.
Luther, C. J., and L. K. Peters. 1982. The Possible Role of Heterogeneous
Aerosol Processes in the Chemistry of CH4 and CO in the Troposphere.
Conference on Heterogeneous Catalysis-Its Importance to Atmospheric
Chemistry, Albany, New York, 1981.
McRae, G. , W. Goodwin, and J. Seinfeld. 1982. Numerical solution of the
atmospheric diffusion equation for chemically reacting flows. Journal of
Computational Physics, 45:1.
Mitchell, A. R.. 1969. Computational Methods in Partial Differential
Equations. John Wiley and Sons, New York.
Pepper, D., E. Cooper, and A. Baker. 1980. In: Developments in Theoretical
and Applied Mechanics, 10:397.
Peters, L. K., J. Yamanis, and W. Akhtar. 1979. An Algorithm to Generate Input
Data from Meteorological and Space Shuttle Observations to Validate a
CH4-CO Model. Status report. Grant no. NSG 1501, National Aeronautics and
Space Administration.
Preussner, P. R., and K. P. Brand. 1981. Application of a semi-implicit Euler
method to mass action kinetics. Chemical Engineering Science, 10:1633.
Sasaki, Y. 1970. Some basic formulisms in numerical variational analysis.
Monthly Weather Review, 98:875.
Yamada, T., and G. Mellor. 1975. A simulation of the Wangara atmospheric
boundary layer data. Journal of Atmospheric Science, 32:2309.
Yanenko, N. N. 1971. The Method of Fractional Steps. Springer-Verlag,
New York.
DISCUSSION
J. Novak; You mentioned a lot about cloud parameterization and the temperature
in cloud water. How did you get that data?
G. Carmichael; We are currently testing this model and, if it is available,
that is the parameter. There are two directions to go with our plans. The
first is to incorporate a cloud-type model, say a pluvial-type model, in which
you can generate cloud from surface parameters and the rainfall grade. And from
the lookup table if Uranius model (not the transport chemistry model but the
cloud variation model for a number of cases), you could then, instead of running
274
-------
a model, have a lookup model approach. From those parameters, you would get the
statistics for that storm or that cloud.
Maybe the third approach is to couple them with a dynamic model where you would
be predicting and to tag this along with the dynamic model. There is no easy
way of doing it.
B. Luebkert: Is your model based on the principle that you have different
submodels like the Lamb model such that you can basically interchange modules,
like the chemistry or the cloud module?
G. Carmichael: Yes, this model is modular, so you can interchange these
processes.
J. Killus; Was I correct in reading the slide you had for surface deposition
rates for 03, that there is something like 50 times less than that residue?
G. Carmichael; Ozone is—I cannot remember the numbers.
J. Killus: It looked like 0.01 and 0.1.
G. Carmichael: Yes, it was low for 03.
J. Killus: That seems to conflict with practically every bit of data that I am
aware of for 03 and N02 surface deposition rates. Those have actually been
measured in closed chambers and so forth, and they were always comparable to
S02. Was this based on any existing data?
G. Carmichael; This was the procedure that we used, based on calculating the
surface resistance for a species that we know a lot about like S02 and then
charaterizing and modifying that surface resistance—according to the
chemical-physical nature of the module. However, that is not hot-wired into the
program.
G. Whitten; Earlier, you illustrated the dramatic effect of N03 with cloud
cover. Later, you illustrated the dramatic effect of cloud cover on nitric
acid, but you used the expression "N03." Do you mean the nitrate radical or the
nitrate ion?
G. Carmichael: Nitrate ion.
G. Whitten: In all cases?
G. Carmichael: No, the earlier slide was of the gas phase of N03.
G. Whitten: The N03 gas-phase radical?
G. Carmichael: Yes.
275
-------
ACID DEPOSITION AND OXIDANT MODEL*
P.K. Misra
Ontario Ministry of the Environment
880 Bay Street, Fourth Floor
Toronto, Ontario M5S IZ8 Canada
MODEL DESCRIPTION
The Acid Deposition and Oxidant Model (ADOM) is used primarily to simulate
acid deposition. However, oxidant transport is also simulated, since oxidants
are products of the chemistry model.
The issue of nonlinearity of the transformation of S02 to S042~ in the
atmosphere is important when one wishes to quantify long-distance,
source-receptor relationships. Modeling efforts on the long-range transport of
acidic pollutants have today been confined to the linear parameterization of the
transformation processes, due in part to the lack of data and the inadequate
modeling techniques of these processes.
Over the past years, the quantity and quality of atmospheric chemistry data
have improved significantly. Also, atmospheric chemical transformation modeling
techniques have shown improvements and promise. It would therefore be
desirable to develop a long-range transport model that includes nonlinear
chemical reactions in the atmosphere, with the aim of studying the effects of
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
276
-------
the nonlinear processes on the source-receptor relationship at distances greater
than 100 km. The model would be designed for regulatory applications.
Three organizations, the Ontario Ministry of the Environment, Environment
Canada, and the Federal Republic of Germany are jointly funding the development
of such a model. The spatial domains of interest are different for the three
sponsors. The attached map, a polar stereographic projection true at 60° N
latitude and used by CMC for numerical weather forecasting, delineates the
boundaries of the domains of interest to the Ontario Ministry of the Environment
(inner domain) and Environment Canada (outer domain). The grid size (X,Y) is
127 km x 127 km. The Federal Republic of Germany is interested in a
geographical domain in Europe and has a somewhat lesser grid size.
The pollutants of interest are SOX, NOX, and 03. The model has a time
resolution of 1 h and is designed to simulate pollutant concentrations and
deposition over the domains of interest for episodes lasting from 24 to 96 h.
The problems to be addressed by the model are: the source-receptor
relationships for S042~, N03_ depositions for a travel distance of 1,000 km, and
the formation and transport of oxidants in the same space and time scales.
Because the model is Eulerian by design, these problems can easily be addressed
by executing the model with appropriate emissions inventory and input data in
the given domains. Although the model is aimed at episode simulations,
long-term averages can be simulated either by averaging over a finite number of
typical episodes or by executing a simpler version of the model for a period of
1 yr.
277
-------
The model is being developed by Environmental Research and Technology, Inc.
(Concord, Massachusetts, U.S.A.) and Meteorological and Environmental Planning
Ltd. (Downsview, Ontario, Canada). A scientific review panel consisting of 12
internationally reputed scientists is overseeing the model development.
Development of the model should be complete by March 31, 1986. A 1-yr time
period will then be required to evaluate model performance.
The joint development of the model by the Ontario Ministry of the
Environment, Environment Canada, and the Federal Republic of Germany is based on
a Memorandum of Understanding mutually signed by these agencies. Cooperation
from other interested agencies is welcomed and can be arranged through similar
institutional agreements.
Difficulties in achieving the established goals are expected to be
primarily of a technical nature, particularly in obtaining adequate data to
specify the flux fields at the required spatial and temporal resolution and to
give useful evaluation data sets. Specifically, we can identify potential
problems in the specification of:
Mass consistent wind, humidity, cloud, precipitation, and radiation
fields with a 6- to 12-layer vertical resolution and horizontal grid
size of 150 km or less appropriate for oxidant modeling.
Emissions inventories of NOX and speciated HCs together with
concentration distributions (in three dimensions) of appropriate gas and
aerosol-reacting consitituents required for the complex nonlinear
chemistry models.
This will surely introduce uncertainties to model results; e.g., one may
have to specify typical values to execute the model. The availability and
278
-------
quality of such data will likely improve in the future, making it possible for
the model, in view of its modular nature, to perform better. In fact, model
runs may provide both incentive and information on the nature of the improved
data sets required. This should perhaps be a major item for discussion at this
conference.
ACKNOWLEDGMENTS
This model is jointly funded by the Ontario Ministry of the Environment,
Environment Canada, and the Federal Republic of Germany.
REFERENCES
Atkinson, R., A. C. Lloyd, and L. Winges. 1982. An updated chemical mechanism
for hydrocarbon/NOx/S02 photooxidations suitable for inclusion in
atmospheric simulation models. Atmospheric Environment, 16(6):1341-1355.
Environmental Research & Technology, Inc., and MEP, Inc. 1982. Models for Long
Range and Mesoscale Transport and Deposition of Atmospheric Pollutants.
Phase I: Modeling System Design. Report SYMAP-101, Ontario Ministry of the
Environment, Toronto, Ontario.
Lloyd, A. C., R. Atkinson, F. W. Lormann, and B. Nitta. In press. Model
potential ozone impacts from natural hydrocarbons, Part I: Development and
testing of a chemical mechanism for the N0x-air photooxidation of isoprene
and a-pinene under ambient conditions. Atmospheric Environment.
Stelson, A. W. and J. H. Seinfeld. 1982. Thermodynamic prediction of the water
activity, NH4N03 dissociation constant, density and refractive index for
the NH4N03~(NH4)2SO4~H20 system at 25°C. Atmospheric Environment,
16:2507-2514.
279
-------
DISCUSSION
B. Luebkert: Do you have any particular area in mind for validating this model
and what plans are there to gather the emissions inventory?
P. Misra: We plan to develop the model for three special studies that were done
in the past 2 or 3 yr. The studies were done in order to create an emissions
inventory for that period and to collect data for that period.
B. Luebkert; Do you have all the emissions inventories you need, all of the HCs
for example?
P. Misra; The development has already started and that includes comparison of
emissions data for the model. That is going on at this time. So, we will see
what is available.
E. Runca: Is the grid size of the model, the total grid size, 110 km?
P. Misra: That is right.
E. Runca: Is this a size that is enforced by the size of the region in which
the model is to be applied?
P. Misra: That went into the selection of the model grid size.
E. Runca: If you want to do exposures and factors like cloud formation with the
model. could the grid size be less than it is now?
P. Misra; It could, but we have not found any smaller grid size at this time.
F. Smith: Would you say a little about the horizontal diffusion? Does this
really represent the mixing or is it really just for deformation?
P. Misra: That is right.
R. Yamartino: The deformation basically represents the differences in the
horizontal wind field across the grid and to some extent the unresolved wind
field.
F. Smith: Strictly speaking, a deformation is not a diffusion process. You are
just squashing a lump of material and expanding it; it is a smashing of area.
R. Yamartino; Right, but we are essentially representing the advection by an
mean flow. The rest of it at the smaller scales you choose to represent is
incorporated into the diffusion process to allow for diffusion.
A. Venkatram: The deformation is nothing but the velocity gradient. The
unresolved velocity is the velocity gradient multiplied by some land scale. It
was used by Morenski in his numerical weather-prediction model. He has just
taken it and put it into looking at the diffusion of the concentrations. In
280
-------
fact, the alternative way to do it is Bob Lamb's way, to run these latest
realizations with ensembles and then look at the diffusion.
F. Smith: Also empirical diffusion.
A. Venkatram: This is empirical.
F Smith: Actually, it is a diffusion process. You say you actually work out a
x,y and plug it into a diffusion equation somehow.
R. Lamb: I think the intent of it, as in this Morenski model, is that the
horizontal shear is supposed to somehow generate small-scale fluctuations and a
sort of variance velocity based on what the model-scale horizontal shear is
doing. We come up with an estimate of a velocity variance, in other words, as
that horizontal shear is beginning to break down into some sort of mesoscale
fluctuation.
281
-------
II. Eulerian model
(1) Adjustable but planned as 1 h.
(2) Adjustable but planned as 127 km on polar stereographc grid (true at
60°N).
(3) Two applications planned: 33 km x 33 km and 65 km x 65 km (i.e.,
approximately 3,500 and 7,000 km, respectively).
(4) Presently operational as 1-layer model but 12-layer and 20-layer
versions are planned. Variable vertical resolution, approximately
logarithmic.
(5) 5 km and 10 km, respectively.
(6) From a wind field model that uses PEL theory.
(7) Eulerian splines are interpolated to compute fluxes, which are then
altered to avoid negative concentrations. Finite element scheme with
filtering also available. Both schemes conserve mass, have low
distortion, low phase error, and good peak preservation properties.
(8) Yes, predicted from w-equation.
(9) Transport as in (7). Diffusion from fully-implicit finite element
algorithm.
(10) Proportional to velocity deformation tensor according to Smagorinsky.
(11) Kz obtained from similarity and convective scaling for convective
conditions and from' Brost and Wyngaard (1978) for stable conditions.
(12)
(a) Mixed depth variable in space and time.
(b) Sensible heat flux used to compute convective depth according to
Maul (1980). Empirical relation z, equals approximately
2,400 u- 3/2 of Venkatram (1980) used for stable depth.
(c) Via the K2 profile.
(13) All cloud types to be treated but not in version to date.
(14) Model uses terrain following coordinates.
282
-------
(15) Resistance modeling approach used so that deposition velocity depends
on local surface roughness, canopy resistance, wind speed, and
stability.
(16) Both rainout and washout effects to be computed from cloud module and
heterogeneous chemistry module. Removal rates will be functions of
space, time, and other variables.
(17) 7 s/grid point in the present 1-layer model.
(18) Estimated 1.7 x 106 bytes assuming 15 species and 1-layer but
including storage of wet and dry surface fluxes and flux through top.
(19) Preliminary tests but on unfinished model.
(20) Full-scale tests expected in 1-2 yr.
III. Chemistry
(1) The chemical mechanism is interchangeable. The current chemical
mechanism consists of four parts.
(a) Anthropogenic RHC/NOX/SOX based on Atkinson et al., 1982. Biogenic
RHC/SOX based on Lloyd et al., 1983. SO,=/N03~/NH/ aerosol
composition based on Stelson and Seinfeld, 1982. Aqueous mechanism
based on ERT, 1982. See references for species list.
(b) Concentrations of H20 vapor and CO? are prescribed.
(c) Atkinson et al., 1982, Carbon II and III and Dodge, 1977, have
been tested in urban scale simulations.
(d) See references for species list.
(e) H?0 vapor and C02.
(f) The number of species is user selected.
(g) Four- to seven-day runs are anticipated.
(2)
(a) The pseudo-steady-state approximation is used for very fast
reacting species such as OH.
(b) Nighttime chemistry is included in the mechanisms. See
references.
283
-------
(3)
(c) Nighttime windshear, stability stratification and turbulence are
accounted for through the 12-layer windfield and K-theory
diffusion. The winds in layers between 850 and 500 mbar are
derived from the numerical weather model and the lower wind from
a numerical boundary layer model.
(a) Horizontal grid resolution is variable, with finer grid
resolution in source areas which helps resolve subgrid
concentrations. Explicit parameterization beyond this have not
yet been developed.
(b) The key subrid scale chemical process of concern is fast
reactions such as NO + 03 - NOa + 02.
(4) Major point source emissions are injected into the layer of their
final plume rise, based on the hourly meteorological condition. They
are currently pseudo-dif fused horizontally into the injection layer.
(5) The model does not provide a measure of how predicted grid average
concentrations may vary from point observations. Sensitivity
analysis with fine horizontal resolution is being performed to
address this question.
(6)
(a) Photolytic rate constants vary with latitude, longitude, time of
day, elevation, and cloud cover.
(b) Cumulus and frontal clouds, vertical velocities, and entrainment
rates are obtained from a one-dimensional cloud model and are
applied to the fraction of the grid column covered by clouds.
(7) Natural sources of HCs and NOX are simulated explicitly by input of
emissions and chemistry based on isoprene and a-pinene oxidation.
Stratospheric 03 and NOX can be entrained from aloft.
284
-------
The NATO/CCMS AIR POLLUTION MODEL COMPARISON*
Han van Dop
Royal Netherlands Meteorological Institute
3730 AE De Bilt, The Netherlands
INTRODUCTION
In 1980, the North Atlantic Treaty Organization/Committee on the Challenges
of a Modern Society (NATO/CCMS) initiated a pilot study on air pollution control
strategies and impact modeling. The study was undertaken by three panels
covering three separate areas—emissions, air quality prediction, and
environmental impact.
The air quality prediction panel, which was chaired by representatives from
The Netherlands, was charged with producing documents describing and comparing
models for interregional transport of air pollutants, with an emphasis on the
transport, removal, and transformation processes of S02 and NOX. The results
will be reported in four separate documents covering the following topics:
• state of the art of interregional modeling,
• comparison of interregional models,
• removal and transformation processes, and
• interpretation and summary of documents 1 through 3.
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
285
-------
Documents 1 and 3 have now been completed and are available as NATO/CCMS Reports
126 and 127, respectively. Documents 2 and 4 are in preparation.
This paper briefly reports the main results and progress of the panel to
date.
STATE OF THE ART OF INTERREGIONAL MODELING
This document (NATO/CCMS Report 126) reviews operational models that
describe dispersion, deposition, and chemical transformation processes on a
spatial scale ranging from 50 to 5,000 km. This range is subdivided into the
mesoscale (50 to 500 km) and the synoptic scale (500 to 5,000 km). Depending on
the spatial scale, different approaches to observations and physical/
mathematical descriptions for the models are required, as indicated in Table 1.
All transport models are based on the mass conservation equation and use
either a Lagrangian or an Eulerian framework. The Lagrangian approach, in which
any desired polluted air parcel is followed (or traced) along its path through
the atmosphere, is much simpler to consider from a numerical point of view. It
has, however, certain disadvantages when applied to multiple source areas or
(complex) chemistry. Eulerian models require a medium- or large-sized computer
and considerable computer time, and thus incur high development and
operationalcosts. Moreover, their mathematical complexity hinders fast
implementation in operational air pollution abatement programs.
286
-------
C/J
W
^2
u
t/5
<£
h-l
H
Ai
V]
8
O
Z
M
2
o
u
w
^j
u
o
o
X
U.
O
z
o
H-l
H
U
M
tt,
M
V)
VI
rf
I_J
u
.-1
s
OQ
^
H
^».
U)
>> 4J
U B
(B £ B
4J 3 01
nj tr n
O 0> 3
u «
U. to
11
B
B
O
•rt
n
3
14-1
•H
a
B
O
JJ
fl
U
3 -d
aCiH
•H 41
B b4
0
U "d
B
r* -H
01 3
•e
0
X
I-l
flj
AJ
0
H -a
0)
U!
^
B
O
•i-l
4-1
a
14
01
n
jj
O
01
6* "9
ai*^
"T3
0
X
^-*
-
a.
41
>N U
M TJ e
o •« o
01 U
an
0) B 4-1 « !K
3 O B 3 w
O -rt «l O -rt
W I) -H 01 >
B l-i 0 6 -H
41 01 fH 01 «
40 CL Mrj ftO 3
O M >w O «-l
o -o o o -H
.B^ 0 X •«
0)
U
nj
a.
V4 B W
O iH
13
S 40 C
>-i B m
O -H
MJ >, 41
•H M S
B ID -H
3 > 4-.
U
•H i-l
U (9
Q.-H
O O 01
B 01 4-1
x a B
n n 01
B
0) M 0)
U O k4
a 3
«-l « 01
>* 4-1 n
3 « 01
«o -o e
O
0
4-t ~
0
01
a
u
a
0
u
•H
X
m
d
i
i-H
>^
^,j
W *H
3 >
O -H
01 «
B 3
rtj (^j
00 «J --N
O i-l u
a -o a
o 01
JS X 0
c -a c
o -a o
B 01 U
s
•H
4j
B
•i-l 41
U
oc B 00
3 -H iH B
on -H
0) 3 H X
6 «J O -rt
41 «-< Q
40 *H ^^
o -o u w
o x oi o
j: -o o oi
B -0 B B
O 41 O <0
B ^•^ O W
41
s
•H
4J
B
•H 41
U
00 nj
B a.
•H n
*^ T3
tn s
> m
-------
While suspended in the atmosphere, emitted matter is exposed to
photochemistry and the removal processes, dry and wet deposition. The direct,
adverse effects of high air pollution levels (e.g., on human health) have been
reduced considerably in urbanized, industrial areas via site planning and
high-stack policies. It has become increasingly clear that the deposition
process is the major environmental impact of air pollution. No operational
models exist, partly because a quantitative description of dry and wet
deposition requires a detailed knowledge of both atmospheric dynamics and
chemistry.
REMOVAL AND TRANSFORMATION PROCESSES
The document on removal and transformation processes (NATO/CCMS Report 127)
focuses on SOX and NOX in the atmosphere.
Dry deposition can be satisfactorily parameterized by using the deposition
velocity concept, and a wealth, of definitive data already exists on dry
deposition for various pollutants and surfaces. Although similar concepts exist
for wet deposition, the underlying physical and chemical processes that lead to
the wet deposition of matter are neither fully understood nor fully validated.
For reasons of simplicity, chemical conversion is often described by linear
decay. Current experimental evidence allows an estimation of the constant of
proportionality.
288
-------
The underlying series of chemical processes is extremely complicated; the
most relevant chemical processes involving NOX consist of more than 20
reactions. The study of these reactions is a separate discipline still far from
completion. Because of limitations imposed by current computer capacity, the
transport and chemistry of no more than one or two dozen air pollution
components can be described satisfactorily on an operational basis.
COMPARISON OF INTERREGIONAL TRANSPORT MODELS
This document, which is now in preparation, is restricted to a comparison
of models that operate on a mesoscale not exceeding 500 km. The models
evaluated meet the following criteria, which were determined by the panel of
investigators:
• The model output consists of ambient concentrations and deposition (dry
and wet) estimates;
• The time resolution is less than 3 h;
• The horizontal spatial resolution is 10 to 25 km;
• The vertical atmospheric structure is resolved (i.e., no single-layer
models);
• Advection, turbulent diffusion and deposition (dry and wet), and linear
chemistry is incorporated;
• A complex source area is described; and
• Only routinely available meteorological input data are used.
Four models were selected for comparison: (1) the TDMB model (Klug et al.,
FRG); (2) the SAI model, Dutch version (Reynolds and Builtjes, The Netherlands);
289
-------
(3) the RIV model (Van Egmond et al., Institute for Public Health, The
Netherlands); and (4) the KNMI model (Van Dop et al., The Netherlands).
The comparison was conducted in Darmstadt, FRG, under the coordination of
Klug, and supported by NATO/CCMS. The Netherlands and surrounding areas were
selected as the test area because:
• The region contains a dense network of meteorological surfaces and
radiosonde stations;
• SO2, NOX, and 63 are routinely monitored bv networks in the involved
countries, and data can thus be easily made available; and
• A detailed emissions inventory for the area exists.
The comparison consisted of three phases. In Phase I, the elementary
numerical procedures in all models were compared. For this purpose, the
dispersion of a single point source under uniform meteorological conditions was
simulated. In Phase 2, the dispersion of a single source under real
meteorological conditions was simulated. For this purpose, three 48-h episodes
with varying meteorological conditions were used. In Phase 3, the standard
emissions inventory was input to each model and runs were made for the three
selected episodes. A sensitivity analysis was performed, and the differences in
model outcome were compared.
290
-------
PRELIMINARY RESULTS AND CONCLUSIONS
The study, which is now in the final stage, will be reported towards the
end of 1983. Some preliminary results and conclusions are as follows:
Schemes should be chosen with care. Saving computer time by increasing
grid size or simplifying numerical procedures leads to unacceptable
numerical errors.
Ideally, inflow conditions at model boundaries should be specified.
However, the lack of data prevents this in most cases. The best
alternative is to choose boundaries in regions where no high
concentrations of involved compounds are expected and to assume zero
inflow. In all other cases, artificial boundary conditions will lead to
errors, which are often difficult to estimate.
The rate of vertical turbulent diffusion (and thus the height to which
released matter may extend) is of paramount importance in determining the
concentration of a single pollutant. For operational applications, this
diffusion rate should be estimated from routinely available weather data.
Although some progress has been made in this area, it remains a matter of
concern.
291
-------
GENERAL DISCUSSION FOLLOWING
SESSION I
Jack Shreffler, Chairman
J. Shreffler: Does anyone have questions for the people who presented papers
today or questions for the OECD people?
E. Runca: I would like to discuss a concept that was brought to the attention
of the audience several times today. That is the concept of a simplified,
sophisticated model.
This meeting is addressing the long-range transport of air pollutants and the
formation of oxidants. I think that priority should at least be given to the
processes that have to be considered in order to describe at least what is
happening and what we can assimilate. Then, we come to the issue of application
and the parameterization of these processes in order to reach a temporary
solution and to reach a greater solution.
I think it is more appropriate to look at what we want to describe and to
identify the elemental processes. After we have identified these processes, we
can consider how we can eventually simplify them in such a way to "steer" the
model with at least a description of the processes you want to assimilate.
D. Jost: I agree with Mr. Runca's more philosophical point. However, in
preparing for this workshop, our U.S. colleagues sent questionnaires to the
people who were presenting models requesting that they follow some scheme to
describe which processes are handled in their model. Additionally, questions
were prepared concerning the objectives of this workshop.
Thus, I am asking the participants, especially the chairmen for the small group
discussions to occur tomorrow, to keep in mind Mr. Runca's comments, the
questionnaire that has been prepared, and the objectives that have been
mentioned.
S. Zwerver: One point is not clear to me. We will be discussing these models
by starting with the objectives and looking at how capable the models are of
meeting these objectives.
However, there is another way. The objectives can be discussed as possibilities
for models. Can the modelers say something about how changing the objectives
would influence the simplicity of their models? Perhaps that is impossible, but
I think the politicians or policymakers in the different countries are not
completely aware of the detailed objectives they have. It would be helpful to
determine the exact definitions of the objectives and to look how these
objectives are obtained and what it means so that we can find the most optimum
point.
A. Venkatram: A lot models have been described today, ranging from the
basically simple to the overwhelmingly complex model. It occurs to me that the
reason for complexity is really understanding, more specifically the quality of
understanding, because we can never really hope to quantitatively simulate the
292
-------
atmosphere, because we can never hope to have the input. Thus, it seems to me
that the simple models are obviously missing something. That is the implication
of complexity. I would like for the complex modelers to identify those missing
pieces in the simple models.
A. Venkatram: Why should we make models complex, unless the simple models are
missing something—missing something qualitatively?
E. Runca: That is exactly my point. The things we have seen in complex models
must identify with the processes that are relevant in relation to modeling
regional air pollution problems.
My interpretation of these comments is that, by discussing the simple models
with the authors, we can have an understanding again of the the processes to be
taken into account when modeling long-range transport.
J. Shreffler: That philosophy drives you to a complex model with high resource
requirements and high data requirements. As I understand it, part of the
purpose of this meeting is to make recommendations to OECD on the way they
should go. Certainly the resource requirements are part of it.
A. Galli; We are getting bogged down with the modelers issue and that is just
part of what might be a bigger problem with emissions inventory data bases.
Before we can answer the questions just presented, we need to get the European
community, Canada, and the U.S. to upgrade their emissions inventories and their
data bases, in order to build a better model, a simpler model or a more complex
model, and then have another data base in order to test them.
So, it might be premature to even discuss here the simple model versus the
complex model. The first thing we have to discuss here are data bases, the
availability of data bases and how good they are. Then we can consider how
simple or how complex the models should be. A big argument is taking place in
this country as to whether building more complex models buys you anything, as
you are inferring here, versus spending the money on emissions inventories and
data bases. That is not going to make the modelers happy; nonetheless, it has
to be recognized, because the models are only as good as the input data.
A. Christie: There is no question about that, but the emissions inventories are
no good unless you know what is going to happen to them once you have admitted
them. Surely a part of our consideration should be the state of emissions
inventories and concentration measurements at the present time, and what is
likely to be required by some of the more complex models that are on line and
are likely to require data bases. We really should not be approaching this from
the point of view of designing our emissions inventories to what we think we can
produce and then bending the models to fit this, because we may lose something
in the process.
A. Galli: You already have a number of models. Those models tell you that you
need some kind of input data, whether they are emissions inventories or
aerometric data. You really do not have enough good, scientifically valid data,
with which to build models and later test them throughout Europe and North
America, regardless of whether they are simple models or not.
293
-------
I am not suggesting that the emissions inventories drive the models. I am
saying that vou have to have a number of models now that require certain types
of information.
A. Christie: We are talking about how one validates a model in its fullest
sense. In discussing his model, Gregory Carmichael showed the aqueous-phase
process. However, these models are based on a knowledge of droplet size
distribution and a variety of parameters that may or may not be known.
How can we parameterize the laboratory-scale model that puts this into it until
we at least have some measurements that can validate even a one-dimensional or
two-dimensional model on a smaller scale? We have to agree on what you have to
have parameterized to be able to put anything into the models on the scale most
of us are looking at, that is, 100 km.
J. Killus: Are we certain that we know what we are talking about when we say
complex models versus simple models? For example, we heard today that the Hov
model or the U.K. model were certainly meteorologically simpler than any of the
grid models being discussed. On the other hand, the chemistry they are using is
extraordinarily complex, vastly more so than you could possibly include in any
sort of grid formulation.
In the urban modeling problem and in the regional modeling problem, we have
found that transport, in developing a wind field and mixing layers, is the
critical factor where one can get away with fairly simple biochemistry.
However, that may not always be true. In certain circumstances, like the
stratosphere, it is not true because the chemistry must become quite complex to
deal with the situation.
I would really like for us to solve for model application in conjunction with
development. Only when you apply the model, only when you are looking for real
data and understanding exactly what sort of situation you are dealing with do
you decide the appropriate scale and complexity required for the model.
J. Shreffler: That is a very good point. Only four, five, or six models were
presented, so that probably sums up the efforts on the regional scale. None of
them is a simple model.
D. Jost: Within OECD, all this business started about 15 yr ago with long-range
transport and acid deposition. Within this field, we knew which pollutant we
were looking at, which was mainly S02. All this business has been done during
the last year, mainly for S02 and as it occurs.
In the field of oxidants, I get the feeling that we are not even very sure which
pollutants are the important ones or which pollutant needs to be abated.
During this meeting several chemistries have been presented, more chemistries
even than models. Do the people who are using these different chemistries know
if they all give the same results or if they are different? Do the chemistries
all indicate that we need to abate primarily HCs instead of NOX, or are there
also applied chemistries that give comparable results?
294
-------
R. van Aalst: We have been trying to compare different chemical models and,
indeed, they tend to give different answers. To my knowledge, several examples
of such differences have been reported in the literature.
Most of the problem is identifying the kinds of concentration regimes these
differences show. I agree that putting in a complex chemistry will lead you in
a specific consideration to get rid of some difficult intermediate reaction,
which could be vital in other specific situations.
So, the policy probably is that you should get agreement upon the fundamental
chemistry and decide in the actual situation you are modeling what you should
leave out. Even with respect to the fundamental chemistry, there seems to be
differences at the present between the chemical models in operation.
G. Whitten: The testing of chemical modules is something you can do outside of
the atmosphere with the large data base that exists for smog chambers. We are
now reaching the point where we can characterize the background effects within
smog chambers. When we see different control strategy effects among different
chemical mechanisms, we can analyze the specific reactions and parts of the
chemistry that have lead to these effects, and we can design smog chamber
effects to emphasize that part of the chemistry and then test them at that
level.
EPA has been sponsoring this ongoing research for many years. Scientists at
Systems Applications, Inc., as well as Dr. van Aalst have also been doing some
of this research. Mainly, you have to get into a position where you can see the
specific reactions to the different parts of the chemistry because you quite
often have a very complex chemistry and many, many reactions. Yet, the
particular reaction that is involved might be a very simple one and might not
even be in the complex chemistry. So the complexity in the chemistry is not
always realistic to the problem involved in the differences in control strategy
and how they are used.
R. Lamb: Assuming for the moment that we have a rough definition of complex and
simple, the answer to the question lies in what you are trying to predict. If
you are interested in predicting hourly average 03 at any place in a region,
then you have to look at all the processes that are instrumental in affecting
that level and then you have to determine what it is you needed in the way of
data and parameterizations and the things needed to predict that. If you are
interested in the annual average 03 over all of Europe, then you can obviously
get away with several models.
In my view, the answer to the question has to be posed by starting with what it
is you are trying to get at and then by going back. You can also ask the
question in this way: Given the data that we have, given the knowledge we have
at this moment, what can we say?
What we can say may be quite limited. If it is limited, if we cannot make
meaningful statements about the things we are interested in, then we have to
determine what more we need to know and what it is going to cost to get that
information?
295
-------
So, the answer is quite simple. It starts with what you are trying to predict.
If you want to predict an hourly average concentration and you insist on a
simple model, the prediction of that model is going to be unreliable. Thus, we
are trying to minimize that error by going to more complexity and so on. We
really need a sophisticated model to determine the level of sophistication we
need, because we cannot guess which of. the processes is dominating and which of
them is important. We almost have to have a complex model to answer that
question itself. It seems to me that all roads lead back to a complex model.
B. Dimitriades; Obviously, you can get to the annual average by using the
complex model and averaging the hourly data, or you can get to the same answer
by usng a simple model, which gives you the annual average. The key question
here is: How do you accomplish significantly more accuracy by choosing a simple
model or a complex model? This is really relative by going the simpler route.
That is the key question.
R. Lamb: If you write out the equation to predict annual average O3 in the case
you just raised, you find turns in the equation that you do not know how to
resolve. So, you begin guessing at them. When you make a guess, you may be
wrong and you may end up with a prediction that is unreliable.
If you grind through hour by hour, you get around to the assumption you made to
go directly to this number. So, you are again back to a complex model. You do
not know whether the guess you made about those terms is accurate.
This is the problem of closure. Some closure schemes work well for some
problems, but fail for others. So a priori in these complex problems with
nonlinear chemistry, diffusion, transport, all these interaction mechanisms, a
guess at a closure scheme in advance is a very speculative thing.
S. Zwerver: The question is not to predict hourly averages of 03 or annual
averages of 03. The question is: Will a 50% reduction of NOX, an overall
reduction, also result in a 10%, 15%, or 25% reduction of 03. That is a
different question? It could also lead to different ojectives for our models.
B. DimitrJades: A 50% reduction of what, 03? The annual average?
S. Zwerver: I should say for the overall pattern of 03, and that includes the
probability of having high values in the overall areas.
A. Eliassen: After listening to this, I think that it would be useful in
tomorrow's sessions to discuss the different models that we are trying to
identify, what each can do, and where it can be applied. All of the models have
their assumptions, and there are situations where they can say something useful
and others where they cannot.
G. Whitten; Quite often in physics, the utilization of what is known as
perturbation theory leads to a different philosophical approach. You have a
mainline effect, and then you are aware of certain subtleties that would perturb
that mainline effect.
These can be handled very accurately; it is not a matter of guessing.
296
-------
Furthermore, there are many powerful mathematical treatments to handle such
things that are small effects, once you have the main thing.
One of the risks that you run by going lo a very large complex scheme and
grinding away is that there is a certain sort of numerical diffusion. Over a
long period of time, that can also add to a large error. Taking a very
simplistic view and adding small perturbations tend to cut down on the
accumulated numerical problem.
A. Christie; When you are addressing a prediction of the mean, you can
certainly parameterize the area for which you want to predict the mean. If in
fact you want to produce hourly values, you are not going to get that from a
perturbation approach.
297
-------
SESSION II
AVAILABLE EMISSIONS INVENTORY DATA BASES
April 13, 1983
299
-------
NORTHEAST CORRIDOR REGIONAL MODELING PROJECT EMISSIONS INVENTORY*
Joan H. Novak*
Environmental Sciences Research Laboratory, MD-80
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
James H. Southerland
Office of Air Quality Planning and Standards, MD-14
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
INTRODUCTION
The U.S. Environmental Protection Agency (EPA), recognizing the potential
impact of high ambient 03 concentrations in the populous and industrialized
Northeast United States, has initiated the Northeast Corridor Regional Modeling
Project (NECRMP) in order to develop optimally effective and equitable
strategies for 03 control in the Northeast. The project includes several
critical elements: (1) the development of comprehensive emissions, air quality,
and meteorological data bases required for adequate technical analysis; (2) the
development of a regional-scale photochemical oxidant model to consider
interurban pollutant transport and photochemical transformation processes; and
(3) the combined application of regional- and urban-scale models to evaluate
alternative control stategies.
*This paper has been reviewed by the Environmental Sciences Research Laboratory,
U.S. Environmental Protection Agency, and approved for publication. Mention of
trade names or commercial products does not constitute endorsement or
recommendation for use.
'''On assignment from the National Oceanic and Atmospheric Administration, U.S.
Department of Commerce.
300
-------
The emissions inventory requirements for the EPA Regional Oxidant Model
(ROM) are more detailed and extensive than those for urban models or control
strategy analyses. Existing emissions inventories available to the modeling
community are inadequate for regional model testing, refinement, and validation.
Thus, EPA, in conjunction with the Northeast Corridor States, local agencies,
and Metropolitan Planning Organizations (MPOs), has compiled an improved
emissions inventory, which is believed to be a reasonably comprehensive and
accurate 1979/1980 data base.
GCA Corporation's Technology Division was contracted to compile, provide
quality assurance for, and correct the U.S. annual point and area source
emissions inventory, including mobile sources. This has been accomplished
through direct liason with the assistance from the state and local agencies.
The Canadian emissions data were supplied by Environment Canada and the Ministry
of the Environment.
BACKGROUND
The NECRMP emissions inventory study area includes all or parts of
14 states, the District of Columbia, and portions of Quebec and Ontario
Provinces in Canada. The area lies between the boundaries of 69° to 82° west
longitude and 38° to 45° north latitude, as shown in Figure 1. It includes in
total Connecticut, Delaware, Maryland, Massachuetts, New Jersey, New York,
Vermont, Pennsylvania, Rhode Island, and Washington, D.C., as well as portions
of Maine, New Hampshire, Ohio, Virginia, West Virginia, and Quebec and Ontario,
Canada. A substantial portion of the southeastern corner of this area
301
-------
-------
(about 175,000 km2) is covered by the Atlantic Ocean. No emission estimates
were made for this portion.
The data were allocated to a longitude/latitude system within the
coordinates indicated above. A grid system of 1/6° latitude by 1/4° longitude
was laid over the area and used as a basis for allocating area source emissions.
The regional model domain extends 2° further west than the NECRMP emissions
study area, to 84° west longitude. Thus, the standard regional model grid
system consists of 2,520 grid cells (60 columns x 42 rows) of slightly varying
area, approximately 18.5 km x 18.5 km. Emissions from the Ohio counties of
Franklin, Licking, Perry, and Fairfield are the only values available within
this extended 2° sector. Landsat data, population (census) data, and other
similar information were also apportioned on the same basis for subsequent use
in allocating emissions into these subcounty grids.
The major pollutants of interest were VOCs and NOX, with CO included as a
potentially important tracer. Particulate and SOX emissions were also included,
but no quality assurance or separate correction of source/emission information
has been carried out for these pollutants, so their quality is undetermined. A
distinction was made between VOC data supplied by states as "total" data and
that supplied by states as "reactive" data, as the final modeler's file is
required to be in a speciated format. A higher emphasis for completeness and
accuracy of data was placed on point sources with VOC emissions above
500 tons/yr and NOX emissions above 750 tons/yr, as they are generally found to
be responsible for a large portion of the point source emissions and as they are
likely to have distinct and isolatable impact on the model results.
303
-------
The collection of raw emissions inventory data in the United States is
generally the responsibility of individual states. The collection process
includes sending questionnaires, filing permits, etc., and data are generally
compiled via the state's own computerized data system or one supplied to the
state and supported by EPA. EPA's state system is the Emissions Inventory
System (EIS), a subsystem of the Comprehensive Data Handling System (CDHS). EIS
is broken down in two major components, EIS/AS and EIS/PS, which handle area and
point source data, respectiveley. After generating the raw data and the
associated data files, states are required to annually supply the updated data
to EPA's National Emissions Data System (NEDS), which is the national repository
and data base upon which many national and state analyses are done.
Due to conflicts in resources, priorities, and differences in emphasis, not
all states have kept their NEDS data files in a sufficiently detailed and
up-to-date form for use in highly complex modeling exercises. Updates tend to
reflect current program priority areas such as the requirement to update the
data base for developing a revision to the State Implementation Plan (SIP), as
required by the Clean Air Act. Hence, when it was determined that a data base
of a more current and complete nature than available was required, it was
necessary to identify the NECRMP effort as a state emphasis program.
Consequently, EPA negotiated with each of the states in the study area to
provide increased assistance and priority for updating the basic data base and
for responding to the EPA contractor's questions and comments relating to
compilation and quality assurance efforts. The basic NEDS/EIS formats and data
were maintained as the starting point for continued improvements and error
detection/correction. An additional emphasis to the effort for many
304
-------
jurisdictions included in NECRMP was the states' concurrent preparation of
revised SIPs for 03.
As indicated previously, states are normally required to update the NEDS
data base annually. They may choose to complete updates more frequently,
however. In the case of the NECRMP data, the data base compilation and quality
assurance effort was planned as a one-time emphasis, and no further organized
compilation thrusts are envisioned. However, emissions inventory is by nature
dynamic; thus, further updates are to be expected form time to time to reflect
detected errors or new/better information on sources and their emissions. Such
updates would normally be improvements in estimates for the original base years
of 1979/1980 as opposed to later "new" year data. As the program progresses,
some separate activity may develop to project the 1979/1980 data base to a
future year for strategy scenario for determining requirements to attain and
maintain the National Ambient Air Quality Standards (NAAQS).
Legal constraints on emissions inventory data vary from state to state.
EPA's authority under the Clean Air Act states that "emission data" shall not be
deemed confidential, but determination of the legal meaning of emission data and
its relevance to statutes on the protection of proprietary information are
subject to complex, time-consuming procedures. During compilation of the NECRMP
inventory, questions of confidentiality were not beyond reason. In cases where
state statutes or state/industry agreements raised questions of confidentiality
(and thus the release of data), EPA, the affected states, and the contractor
were able to work out screens, selective deletions, etc., such that the data
given to EPA and now in the NECRMP data base are not deemed confidential.
305
-------
Specific data items that were claimed as confidential generally could be avoided
or worked around in a manner that was not detrimental to the modeling aspects of
the data collection program.
In Canada, the provincial agencies and regional offices are responsible for
collecting emissions information that is primarily based on company-supplied
data. Through surveys and questionnaires, other government agencies such as
Statistics Canada; the Ministry of Industry, Trade and Commerce; and the
Department of Energy, Mines, and Resources collect basic data that can be used
in estimating emissions from major sources categories. There are no legal
requirements for the submission of emissions data; however, many of the
regulations controlling the release of toxic pollutants require quarterly or at
least annual submissions of data that are used to develop emission estimates.
The Canadian National Emissions Inventory is primarily used to assess the
effectiveness of control programs. There are not formal restrictions on the
release of company-supplied data, but informal agreements have been made
regarding confidential processes. The Canadian inventory is updated every 2 yr.
Background information on the Canadian emissions inventory was provided through
a personal communication with Frank Vena, Air Pollution Control Directorate,
Environmental Protection Service, Environment Canada, Quebec. Environment
Canada has been extremely cooperative in supplying emission information for the
NECRMP modeling efforts.
306
-------
POINT SOURCE DATA
U.S. Inventory
A point source is typically defined as a stationary source large enough to
be identified and tracked individually, usually emitting greater than
100 tons/yr of any pollutant of interest. However, all states in the Northeast
Corridor did not strictly adhere to this specific definition. Each state
submitted the most current point source inventory, typically for 1979/1980, in a
computer-readable form compatible with either NEDS or the Emissions Inventory
Subsystem/Permits and Registration (EIS/P&R), predecessor of the current EIS/PS
and EIS/AS systems.
The following key parameters were among those reported:
• UTM coordinates,
• Stack parameters,
• Emission control equipment,
• Efficiency of primary and secondary control equipment,
• Operating schedule,
• Annual production rate,
• Sulfur and ash content of fuels, and
• Annual emissions by source classification code (SCC).
Stack parameters included height, diameter, temperature, and exhaust flow rate.
Source location information, which was reported as UTM zone and coordinates to
307
-------
the nearest tenth of a kilometer, was converted to latitude and longitude in
degrees, minutes, and seconds. Point source emissions data were reported for
over 1,400 different SCCs. However, a large percentage of 03 precursor
emissions occurred within 10 to 20 categories. Table 1 lists the percentage of
emissions from the major precursor-emitting point source categories. The
categories listed account for 84% of the VOCs and 98% of the NOX emitted from
point soures in the NECRMP region.
Annual emissions from electric utilities were distributed seasonally by
applying fuel and state-specific seasonal factors derived from U.S. Department
of Energy (DOE) power generation statistics. Hourly allocation of electric
utility emissions was based on hourly fuel use data collected from approximately
300 power plants during the Electric Power Research Institutes (EPRl) Sulfate
Regional Experiment (SURE). Because of the lack of detailed data for developing
specific temporal allocation factors for other point source categories,
individual source operating pattern data were used to calculate temporal
factors. Default values providing a uniform distribution were assumed when no
operating data were available.
308
-------
TABLE 1. MAJOR CATEGORIES OF OXIDANT PRECURSOR
EMISSIONS IN THE NECRMP REGION
Emissions
Category
Gasoline handling
Petroleum refineries
Other chemical manufacture
Iron and steel manufacture
Stone, lay, glass, and concrete
In-process fuel use
Others industrial processes
Industrial surface coating
Degreasing
Graphic arts
% NOX
<0.1
0.8
0.2
0.8
1.1
0.2
0.5
<0.1
0.6
<0.1
% voc
4.3
7.4
3.5
4.4
1.0
12.9
18.0
17.2
2.0
4.8
External combustion boilers,
electric generation 79.9 3.8
External combustion boilers,
industrial 12.2 4.4
External combustion, space
heaters - industrial 1.7 0.3
309
-------
The following algorithms were used to perform the temporal allocation of
annual point source emissions:
Standard Algorithm —
Hourly Emission = Annual Emission x Daily Fraction x Hourly Fraction
Daily Fraction = 1
No. of Operating Weeks/Season x Days
Hourly Fraction = 1
Hours
Seasonally Weighted Operating Pattern Algorithm —
Hourly Emission = Annual Emission x Seasonal Factor x
13 x Days x Hours
where seasonal factor = fraction of annual operation occurring in the season of
study,
13 = assumed number of weeks of operation in a season,
days = number of operating days per week, and
hours = number of operating hours per flay.
Uniform Default Algorithm —
Hourly Emission = Annual Emission x 1
52 wk/yr x 5 days/wk x 8 h/day
where the hours of operation are assumed to begin in the eighth hour (0700-0800)
local time.
310
-------
The annual values reported by the states are estimates of actual pollutant
emissions as emitted. The effects of control equipment have been incorporated
by using type and efficiency information for primary and secondary control
equipment. Approximately 40% of the emissions are estimated by using the NEDS
standard emission factors with plant-specific information. About 25% of the
emissions are calculated by using nonstandard emission factors and
plant-specific data. Material balance (20%), source test (5%), and other
methods (10%) account for the remaining reported emissions.
VOC Methodologies —
The chemistry mechanism in most photochemical grid models requires the
lumping of individual VOC species into the reactive classes specifically treated
by the mechanism. The current chemical mechanism in ROM handles four reactive
classes: olefins, paraffins, aldehydes, and aromatics. Future mechanisms under
development may require more classes or different distributions of individual
chemical species in the designated classes. Thus, a general methodology was
developed to calculate factors that, when applied to annual total HC or NMHC
values, produce emissions for any reactive class defined to meet the
requirements of the chosen chemical mechanism.
The basis of this flexibility is a set of 119 species profiles, each of
which lists the typical organic compounds associated with a particular source
category or process. An index file assigns each source category, represented by
a unique SCC, to the most appropriate profile. Thus, given the definition of
which organic compounds make up each reactive class, the mole fraction of each
311
-------
class in any profile is calculated. Mole factors for each profile and class
combination are then determined by dividing the mole fraction for each class by
the profile average molecular weight. These factors are then adjusted for
aldehydes and for compatibility with HC that is reported as nonmethane or
reactive VOC.
This generalized process, applicable for both point and area sources,
enables the calculation of emissions for each reactive class of HCs required for
any chemical mechanism a modeler may choose. These profiles also include
specific information for splitting the NO and N02 components of NOX for each
applicable source category.
CANADIAN INVENTORY
The Canadian emissions data were submitted to EPA by Environment Canada in
computer-compatible form. Point sources are considered to be major individual
pollutant emitters. The 1976 .point source data base contains the following key
parameters: latitude, longitude, stack parameters, annual emissions by standard
industrial classification (SIC), and seasonal variation of SOX and NOX. Stack
parameters are height, diameter, exit temperature, and flow rate. The 1978 data
base includes additional parameters such as base quantity, sulfur and ash
content, and emission factor. Source location is reported in latitude/longitude
by degrees, minutes, and seconds. Point source emissions data were reported for
62 different SICs. Table 2 list the major contributing point source categories.
These nine categories account for over 99% of the 1976 Canadian HC and NOX
emissions.
312
-------
TABLE 2. MAJOR CANADIAN POINT SOURCES
CATEGORIES
Category
Petroleum refining
Natural gas production
Electric power generation
Nitric acid production
Tar sands operations
Sulphate pulping
Emissions
1 NOX %
12
26
52
2
2
A
HC
95
2
2
,0
<1
-0
Manufacture of carbon and
graphite electrodes 1 <1
Seasonal data were obtained for NOX and SOX only. Therefore, any further
temporal resolution is currently provided by applying the temporal factors
developed for the U.S. inventory to appropriate source categories.
Speciation of Canadian HC data can be accomplished by using the U.S.
methodologies where appropriate correlations can be made between U.S., SCC, and
Canadian SIC. Additional profiles may be introduced to define more
realistically the chemical species associated with the Canadian processes.
Most point source emissions data supplied by the provincial agencies and
regional offices are obtained from individual companies through stack tests,
engineering analyses, or mass balance methods. Most other emission estimates
313
-------
are developed by using standard U.S. emission factors (EPA, 1977) and either
plant-specific or general fuel use production and consumption data.
AREA SOURCE DATA
U.S. Inventory
Area sources comprise stationary and mobile sources that typically emit
less than 100 tons/yr and are too small and/or too numerous to record
individually. Many states submitted their current area source inventory either
in hard copy or in computer format compatible with EIS/AS. Standard
methodologies (EPA, 1982) were recommended to ensure consistency from state to
state. Annual county emissions of VOCs and NOX were reported for the 54 area
source categories shown in Table 3. Raw data from 1970 to 1980, which are used
in the annual county emission calculations, are extremely disparate: gasoline
and fuel oil sales; industry employment levels; county population; asphalt
paving; diluent content of asphalt; pesticide use or acres harvested; highway
vehicle classes and vehicle miles traveled; fuel consumption; registration data;
vessel movement; and agricultural, highway, and business statistics. Because of
their significant contribution (62% of NOX emissions and 48.3% of VOC
emissions), highway mobile sources divided into the following categories:
light-duty vehicles (LDV); light-duty gasoline-powered trucks, 0 to 6,000 Ibs
GVW (LDT1); light-duty gasoline-powered trucks, 6,000 to 8,500 Ibs GVW (LDT2);
heavy-duty gasoline-powered trucks (HDG); heavy-duty diesel-powered vehicles
(HDD); and motorcycles. For some of the states, area source inventories were
314
-------
either unavailable or sufficiently out of date or incomplete, so that they had
to be replaced by contractor-generated data by the procedures recommended above.
Annual county emissions for the 54 source categories were apportioned to
the previously defined grid system to satisify the spatial requirements of
regional-scale modeling. Spatial apportionment factors were developed to
allocate a portion of a particular county's area emission to a specific grid
cell according to the known distribution of some surrogate indicator. Surrogate
indicators used in NECRMP are: housing, population, urban land, agricultural
land, composite forest, land area, airport location, and park location. The
particular surrogate indicator assigned for allocation of emissions in each area
source category is listed in Table 3. Distribution of the surrogate indicators
over the NECRMP grid was obtained from various U.S. Bureau of the Census
statistics, land use classification data derived from Landsat imagery, the U.S.
Geological Survey maps, and Waterborne Commerce of the United States - 1978. A
homogeneous distribution of land use within a grid was assumed.
Temporal distribution of annual area source emissions is achieved by
applying seasonal, daily, and hourly factors for each source category. The
standard algorithm described in the U.S. point source section is applied.
Table 3 summarizes the temporal patterns used for the NECRMP emissions. Details
and comprehensive source references for developing spatial and temporal
allocation factors are listed at the end of this paper (EPA, 1983, 1979, 1978).
315
-------
•z.
o
5
3
.—1
*£
t—>
•tf
Cn CO
;/: W
M
Z O
w
CO H
Z <
e: o
w
E- W
H O
< Oi
z to
o
M <
H W
«C ftS
J OS
J 0
j to
-£ W
II
pu W
s:
&
o
w
z
.
3
ca
H
j
~ I
i 8
—
o
3 ~ a'
= Q
SI - * 2
§
o
i £ 1 «•- I r
xl *° '^ "^ ,5 " JZ ^^
2) i-j^ii^^1
..
vt
<0
4)
t/1
4* U
W O
S»-0
k. ~
12
•*- UJ *-» f— 19 f»
** 4> « "X. 41 *»
_0. 4*^ 41 Ot U «
T — Z«— 21 5
u ^- o -w o
M O »« ^ * O* 4(
41 V*'QCOBCU
0 3<3ff-JSe£
X _— 00 rn k
O) 4) QJ IQ ^ c U
4i o» e\ i. e « je
o ^S^3<^Qi
o o
ss
0 0
i§
*M **«
83
«• t«
LD trt
§§
r^.
i t
§ §*
O O w» o
i i **- "^ J: i
§ S »< fr* — 0
•S * -^
M
c
o
s
VI
«
o
u
4*
s
E
0
u-
c
3
1- tj
4i •—
4J **-
EEEE^'
fS!l««
if *
-3 e
•^ •> > »—
£ *S "* iS *
U W wt «A
fa < vi ^ 41
<^ c -n
' S" « "
2 ~" °- i 5 ^
3cJ5^ooo!
1-5
_ « 3
"" ^£
< "*-
«W (A
SL. -^
41 O
wi > 4l
3 »- C
0 £ O
£ g 4> r»~
U E >«
O "* 4)
4} -o u
k wt o oi
<« 3 4) */»
> ca L. *-
«1S^^«^^^^-Si
^-S-^-S-^^-^^^^-S'-S'^
"^
^
J £
5 , "
1 0 4)
41 U 41
S1!^
"" " £{
« 4> *a
Ok > M
S5H
> 4->
-------
o
zl £
3 §
= 1 tsi
. •
o
2 ^
ll
O O — M<
) U) 4> ff>
|W* -S
1 l§ SS ."JgS
k k fc. k k O Of* '
«^«^i*. **" ^ ^ ^ £* 2
«« H »4 »• e* « *
rf» O i £ i J
V Q 9k ^> O f- f- O O O O * O '
-o
c
3
C
•r-l
C
0
u
ro
W
CQ
H
N
1
*«
£
O
VI
«
*
0)
i1
o
s.
3
k
u
W
«
a
«
•0
^
4
t*
W>
1
t*
o>
e
a.
•«
un
k
= sciee
»» *»•«] O O O O O
ui e e e e c
k
3
«
^
O
E
.- 5 ^ v»
2 V» 4» VIO «J »
**3 ** O 4t^»O^*
S*^ 2 *« ^! ^- £ « •* 3 ^I *<«
«•• 3 *<~ *• O k • "O ^ k
k«^-^k -^J= 5-^ *> 3
53— ** 3 ^ 4_» *J Vt trt **
^Vt vt *> o O C ^» «*••-«
c *• 4> •• ««euO*.r.^i
"
^
H
CNJ
k
V
••
e
k
£ ** V
o ** » —
£»Sril
«M^3
ii S* SS
e e — -2 o *«
— ~ I CJOQ:
3
^
*«O
H»
••
f
Wl
«
§
2
C
••
h
vi O
§s
CO CO
^_
s
e ^-
s^
35
« OJ
**
^^
4_
O£
S
*«•
E"
£S
§**
VI
00 U)
M
•*• 4t
U
« JS
l»
£ 1
™J
If
41 "*•
VI
•&
^^
^
i
41
vi
<*-
i
VI
41
-
m
1
M
«l
o
41
u
I
?
X
<*»
o
Vt Wt VI VI VI
•^s.
tn
^_
*"* *~
•*- <•
w
O •*
k *
vi 9
VI
s.s
t -
If
^
S-
LO
k •
Is
o o o"5 -S
C C CO »•
a 3 a ~ i/>
C ft** -^ O
O O (/I Ik U.
VI "O
If
i^
N 41
T3
41 "*- •
£ 2§
^Jg
VI **
|| si
|| 1§
•«- £ c
VI k ** 41
s*s*
e T) 43 4i
a 4) m ^s
^B??
• 0.-S i
J k« 4>
Ol 41 -^ 2
41 > *• *>
k O •*•
nil
VI •*• k •
* 3 0
c e o
O O '"CO
II -J
ii-J
Vt VI tj^*"
z;^ o
i i i
317
-------
Volatile organic compound emissions for area sources are estimated by using
the same methodology as that discussed Cor point sources, by assigning a pseudo
SCC to each area source category.
The spatial, temporal, and speciation factors are appropriately applied to
the annual source emissions through the use of the Regional Model Data Handling
System (RMDHS) (EPA, 1981), a COBOL software package specifically designed to
interface with the ROM.
Canadian Inventory
Canadian area sources consist of minor point sources, mobile sources, and
other sources too small or too numerous to track individually. Emission totals
have been calculated for 54 SICs. However, these do not correspond directly to
the 54 U.S. area source categories. The following major categories account for
over 85% of both NOX and HC emissions in Canada: diesel engines, diesel and gas
marketing, forest fires, fuel -combustion, and general-purpose motor vehicles.
The Canadian National Emissions Inventory available to the U.S. currently does
not record area source emissions by source category. Only total emissions for
each pollutant are reported on the polar stereographic grid system used by the
Canadian Meteorological Center (CMC). The side length of a grid cell is 127 km,
and each grid is identified by the x-y coordinates of the south-west corner.
Latitude and longitude information is also available. Emissions are reported on
the CMC grid system for all of Canada; however, only those portions falling
within the NECRMP region are included in the modeling inventory. Written
communication from A. Sheffield, Air Pollution Control Directorate, Environment
318
-------
Canada, Ottawa, Ontario, provided the percentage of contribution for each area
source category. Currently, this source distribution and a surrogate indicator
of population by census district are the only data available for source category
and spatial apportionment. More detailed source and spatial distribution data
such as population, housing counts, and fuel type by enumeration district,
employment statistics, and industry location data will be available from
Environment Canada in the future. When appropriate, the U.S. temporal
apportionment and VOC speciation methodologies will be applied to obtain the
data resolution required for Eulerian regional-scale models.
DATA QUALITY
An emissions inventory is a compilation of estimates. As such, it does not
possess the capability of comparison to true and firm standards. Nevertheless,
il standard, proven procedures are followed and the resulting data are subjected
to logical quality assurance and validation checks, one can develop a reasonable
assurance that an inventory data base is reliable for the purposes intended.
The NECRMP inventory is being composed of input from several states and
other jurisdictions that has existed in several, often incompatible formats.
Thus, an early stage of the data compilation and quality assurance effort was to
reduce the data to a common format. The NEDS/EIS format was chosen. (NEDS and
EIS are compatible, and output from one may generally be used to update the
other.) For most states in the NECRMP area, the reduction to a common format
for point sources was done either largely or entirely by computer. Therefore,
319
-------
as the data were processed into a standard format, they were edited and became
amenable to further computerized logic checks and validation procedures.
The following list is fairlv typical of the checks that were done during
the edit step.
• Determine whether valid numeric characters are in numeric fields.
• Check UTM coordinates against a range of possible valid values for the
state.
• Flag excessive values when compared with expected norms.
• Cross-check for specific missing data.
• Cross-check information on stack height, diameter, plume rise,
temperature, velocity, boiler firing rates, etc.
Several programs were developed that supplement the EIS edit and summary
reporting capabilities, and provide "fixes" to the various inventory files.
These programs produce files and format printouts that organize the data into a
form whereby easily managed manual checking and "common or generic" error
repairs can be performed.
The NEDS-to-EIS-conversion editor program and supplemental programs that
were developed provided lists of all severe and conditional warning errors.
Errors considered important for NECRMP modeling were documented for submittal to
the states. Similar listings were generated from the master file creation
program checks, which identified such problems as duplicate transactions and
incomplete records. Perhaps one of the most valuable steps was the manual
review of the major point sources. All sources with VOC or NOX emissions
320
-------
estimated at _-500 tons/yr were so reviewed. Inconsistencies, nonsensical
entries, engineering unreasonableness and inaccuracies, etc., were indentified
and flagged for further investigation and resolution. External data bases
(e.g., lists of all sources in an industry category as available from a trade
association listing) were used to cross-check and identify missing source
possibilities.
Upon completion of this review and consolidation of all questions and
identified errors, the lists and supporting information were forwarded to the
states for resolution. The state agencies reviewed the problem lists and
searched their records for answers. In most instances, the contractor and the
state personnel met to assure that the problems and responses were properly
communciated. Many telephone calls were made to follow up individual problems.
In some instances, the contractor also spot-checked the agency's file data to
gain an added measure of understanding and confidence in the completeness and
accuracy of the data.
After revising the data base to add, delete, or correct information found
in error, the data were then reprocessed through the system to ensure that the
updates were done properly and that the same or new errors did not exist.
Resulting data were made available to the states for their use in the normal
updating of NEDS, SIPs, etc.
Similar procedures were followed for the area source data base. However,
there was one major distinction in that the states in many cases did not
321
-------
maintain a current area source data base. In these cases, the contractor
compiled the data according to standardized procedures (EPA. 1980, 1982).
Thus, the NECRMP emissions inventory is believed to be as complete as is
practical under the resources available and within the realm of current
responsibilities. Undoubtedly, errors remain and some sources may be omitted.
Since the data base can be continually updated, it is likelv that some
additional corrections may be warranted in the future. For example, through the
modeling process itself, peculiar results will often surface that will culminate
in the identification of a missing source or erroneous information on an
included source.
As alluded to earlier, the estimated data are for 1979/1980 (or at least
adjusted to that time period). Inventories are of necessity retrospective
(unless projected in some manner). Thus, the data compiled in NECRMP are as
appropriate for the time period of estimate as can now be determined.
In summary, an emissions inventory can never be proven precise and
accurate. The NECRMP inventory, however, has undergone rigorous checks and
validation efforts and is believed to be capable of driving the ROM. Any errors
or inconsistencies identified in the future can be subsequently corrected.
The Canadian inventory is as complete as possible in terms of major source
contributions. However, the lack of detailed data for determining temporal
distribution of point and area annual emissions affects the accuracy of the
hourly resolved emissions estimates. The use of U.S. factors, when appropriate,
322
-------
assuages that effect. The aggregate nature of the reported area source
emissions definitely weakens any speciation or spatial allocation methodologies.
Future updates must concentrate on providing more resolution in these areas,
particularly for major contributing source categories.
The currentness of Canadian emissions data varies by pollutant. Data for
particulates, CO, and HC from petroleum refineries are reported for 1976. Data
for NOX, SOX, and the remaining HC emissions have been updated to 1978. A few
additional updates for 1980 have been incorporated into the National Canadian
Emissions Inventory. Further updates toward a 1980 base year will be
incorporated into the NECRMP inventory when received from Environment Canada.
The Environmental Protection Service (EPS) of Canada maintains the quality
of the emissions inventories. However, since there are no direct reporting
requirements, the province-supplied data are typically accepted unless a large
discrepancy is noted. EPS compiles, reviews, and processes via computer the
emissions data that ultimately become part of the Canadian National Emissions
Inventory.
BIBLIOGRAPHY
Paddock, R. E., S. K. Burt, and R. C. Haws. 1981. The Regional Model Data
Handling System (RMDHS) User's Guide. Research Triangle Institute for the
U.S. Environmental Protection Agency, EPA Contract 68-02-3052, Research
Triangle Park, North Carolina. 250 pp.
U.S. Environmental Protection Agency. 1983. Northeast Corridor Regional
Modeling Project Annual Emission Inventory Compilation and Formatting.
Volume XVII, Development of Temporal, Spatial and Species Allocation
Factors. EPA-450/4-82-013q, Research Triangle Park, North Carolina.
118 pp.
323
-------
U.S. Environmental Protection Agency. 1982a. Emissions Inventories for Urban
Airshed Model Application in the Philadelphia AQCR. EPA-450/4-82-005,
Research Triangle Park, North Carolina. 384 pp.
U.S. Environmental Protection Agency. 1982b. Northeast Corridor Regional
Modeling Project Annual Emission Inventory Compilation and Formatting.
Volume I, Project Approach. EPA-450/4-82-013a, Research Triangle Park,
North Carolina. 70 pp.
U. S. Environmental Protection Agency. 1980. Procedures for the Preparation of
Emission Inventories for Volatile Organic Compounds. Volume I, Second
Edition. EPA-450/2-77-028, Research Triangle Park, North Carolina. 232 pp.
U.S. Environmental Protection Agency. 1979. Procedures for the Preparation of
Emission Inventories for Volatile Organic Compounds. Volume II. Emission
Inventory Requirements for Photochemical Air Quality Simulation Models.
EPA-450/4-79-018, Research Triangle Park, North Carolina. 232 pp.
U.S. Environmental Protection Agency. 1978. Seasonal Variations in Organic
Emissions for Significant Sources of Volatile Compounds. EPA-450/3-78-023,
Research Triangle Park, North Carolina. 58 pp.
U.S. Environmental Protection Agency. 1977. Compilation of Air Pollutant
Emission Factors. Third Edition and Supplements, AP-42, Research Triangle
Park, North Carolina.
U.S. Environmental Protection Agency. 1975. Residential and Commercial Area
Source Emission Inventory Methodology for the Regional Air Pollutant Study.
EPA-450/3-75-078, Research Triangle Park, North Carolina. 50 pp.
DISCUSSION
S. Reynolds: What, if anything, has EPA done to quantify the uncertainties in
the emissions inventory?
J. Novak: There is a report out. I am not totally familiar with everything
that was done in terms of quantification. All I can say is that the procedures
that were undertaken were done to actually verify the comprehensiveness. Joe
Southerland was responsible for that inventory development. Maybe he could
better answer that question.
J. Southerland: The job of attaching some statistical uncertainty to an
inventory is a very difficult, if not impossible, task. A number of studies and
papers have looked at this kind of thing. Essentially, it comes down to the
fact that an inventory is kind of an iterative process. It is dynamic in that
you are always finding additional uncertainties or errors. For some of the
things, like an omitted source, how do you quantify the degree of uncertainty
when you do not realize that the source is there?
324
-------
So, for this particular inventory, we have a report in draft form. It is
essentially an evaluation document that will look at the improvements we have
made in the inventory and the uncertainties of the unknowns, some of the things
that we except or suspect to be additional things that could be improved in the
inventory. It is a kind of evaluation document that will enable additional
steps to be taken if so desired to continue this iterative process.
L. Lindau: I think we will come back to these problems in the general
discussion. I think this is a very important question.
E. Runca; Can you define what you mean when you say that these data can be used
to develop emission scenarios? Are these projections in the future?
J. Novak: In my understanding of what types of emission scenarios would want to
be developed, possibly certain source categories would be earmarked for
reduction. For instance, you have 14-area 54-source categories. Highway
sources would be a major contributor we might like to reduce. This system has
the flexibility such that factors could be put in if you wanted a 50% reduction
or 20% reduction: The entire system could be executed and run to produce an
emission scenario that then included reduced highway sources or whatever other
combinations of reductions for individual sources that you wanted. You could
come up with various strategies for how you are going to control certain source
types, develop the strategies, put them into the system, and come out with
emission scenarios that would reflect those control strategies.
E. Runca: Are economic and technological factors taken into account?
J. Novak: No. They would have to be taken into account in terms of inputting
the appropriate reduction. This is not an economic analysis; it is strictly a
software package to produce the required emissions to the right levels for the
different sources.
325
-------
EMISSION INVENTORY DATA BASES FOR THE UNITED STATES*
Charles 0. Mann
Monitoring and Data Analysis Division, MD-14
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
INTRODUCTION
In the United States, most emissions inventory activity has been performed
by state and local air pollution control and planning agencies. These agencies
develop emissions inventories to support the State Implementation Plans (SIPs)
required by the U.S. Clean Air Act. Point source emissions inventory data are
required by U.S. Environmental Protection Agency (EPA) regulations, to be
submitted by the states to EPA in the format of the National Emissions Data
System (NEDS). The NEDS data base is thus the most readilv available emissions
inventory information in the United States.
NEDS has been in existence since 1972. Currently, it contains data for
about 55,000 establishments defined as point sources. NEDS also reports
estimated emissions for all other sources not represented in the point source
file as area source emissions. The data in the point and area source files are
described briefly below.
*This paper has been reviewed by the Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, and approved for publication.
Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
326
-------
NEDS POINT SOURCE DATA
NEDS point source data are all submitted to EPA by state and local air
pollution agencies. EPA regulations require that data be submitted on an annual
basis for all establishments with annual actual emissions of more than 100 tons
(90.7 metric tons) of particulate matter, S02, N02, or VOCs or more than 250
tons (227 metric tons) of CO. Most states report more data than are actually
required. Only about 13,000 of the 55,000 establishments currently in NEDS
actually emit over 100 tons per year. All 50 states plus the District of
Columbia, Puerto Rico, the Virgin Islands, and Guam submit point source data.
Data are collected by each state using its own procedures. Most rely upon a
combination of source permit systems, emissions inventory questionnaires, and
on-site facility inspections to obtain data.
Point source data collected by a state are normally maintained in an
automated data system by the state agency. About 20 agencies use the Emissions
Inventory System (EIS) designed by EPA specifically for state use.
Approximately 15 other agencies use other automated systems. The remaining
agencies, usually those with a relatively small number of sources, maintain
their records manually.
Data are submitted to EPA in the standard NEDS format, which is illustrated
in Figure 1. The basic data elements are the establishment name and address,
the Standard Industrial Classification (SIC) code, the Universal Transverse
Mercator (UTM) grid coordinates, stack parameters, control equipment data,
source operating schedules, source operating rates and fuel characteristics, and
327
-------
6
(-1
O
Ol
u
>-<
3
O
VI
O
a,
a
w
2
cu
3
00
328
-------
estimated emissions. In NEDS, all emissions are reported as annual actual
emissions estimates. These estimates, which may be based on stack results,
material balance calculations, or special state-defined emission factors, may be
hand calculated and entered by the state agency. Alternatively, the state may
specify that emissions be calculated by EPA by using a standard file of emission
factors along with reported source operating rates and control efficiency data.
These standard emission factors are available in Compilation of Air Pollution
Emission Factors (EPA, 1978).
NEDS AREA SOURCE DATA
In NEDS, the county or county equivalent in each state is represented as an
area source. Thus, about 3,200 area source records are maintained in NEDS.
Area source data are updated annually by EPA. States are not required to submit
area source data, and very little data are voluntarily submitted. EPA develops
area source emissions estimates based on the standard NEDS area source
categories shown on the NEDS area source form (Figure 2). For each of these
categories, county-level estimates of source activity are derived. These are
normally based on data published by other Federal agencies, such as the
Department of Energy, Department of Transportation, etc. Often, needed data are
available only at the state level. EPA has developed a set of procedures to
allocate state-level data to individual counties, based on the use of
population, employment, or other data that are available at the county level
(EPA, 1978). All county-level estimates of area source activity are converted
to emissions by using standard emission factors. For most source categories,
these data are taken from the EPA reference document cited above (EPA, 1978).
329
-------
orm
NEDS area source
0)
(J
3
00
330
-------
Highway vehicle emission factors for each county are calculated by using a
simplified version of the MOBIL2 model (EPA, 1981), which takes into account
vehicle age distribution, applicable emission standards, average vehicle speeds,
average ambient temperature, and other factors. As with point source data, area
source emissions in NEDS are reported only as annual actual emissions.
AVAILABILITY OF NEDS DATA
The NEDS point and area source files reside in EPA's Univac computer at
Research Triangle Park, North Carolina. Data are available in a variety of
standard computer printout formats or on magnetic tape files (EPA, 1980). In
addition, qualified EPA Univac users may access the files directly. However,
because of confidentiality claims made by states, free access to the NEDS point
source file is not granted to all potential users. In addition, data items
identified as confidential are excluded from computer printouts and data files
made available to non-EPA users. Data items sometimes identified as
confidential are source operating rates, capacities, and emission estimation
method codes that, if revealed, would possibly allow calculation of confidential
source operating rates. No emissions estimates can be claimed as confidential,
however, under the provisions of the Clean Air Act.
OTHER DATA BASES
Recent activities in photochemical oxidant and acid deposition research
have promoted the development of a number of other emissions data bases. Many
of these have been developed by using the existing NEDS data as a starting point
331
-------
to improve the data by additional data collection or additional software to
process the data. These data bases have been reviewed in another paper (Bosch,
1982). A few of these other data bases are briefly identified below.
Northeast Corridor Regional Modeling Project (NECRMP) Inventory
These data were developed for 15 Eastern United States to provide input to
photochemical oxidant models and will eventually be included in NEDS. This
project will be discussed further in a separate presentation at this conference.
Sulfate Regional Experiment (SURE)
These data were developed by the Electric Power Research Institute (EPRl)
for use in sulfate episode modeling, long-term transport of sulfates, and the
Utility Simulation Model. The data base covers the Eastern United States,
emphasizing emissions of SOX and related species for the period 1977 to 1978.
EPRI is currently sponsoring a major new effort to develop emissions data for a
1982 base year (Heisler, 1982). These data are to include emissions estimates
for SOa, sulfates, NO, N02, total particulates, and VOCs by reactivity class for
the United States, excluding Alaska and Hawaii. Seasonal and daily average
emissions estimates are also to be included.
Hazardous and Trace Emissions System (HATREMS)
This is a companion system to NEDS that is maintained by EPA for
calculating emissions estimates of other pollutants not included in NEDS.
332
-------
Basically, HATREMS provides a separate emission factor file that may be used to
calculate emissions by using NEDS source parameters. Presently, HATREMS is
routinely used only for reporting lead emissions.
Multistate Power Production Pollution Study (MAP-3S)
These data were developed by Brookhaven National Laboratory to provide data
for air quality forecasting and econometric models. The data base started with
1976 base-year NEDS data and has been upated to include data for major Canadian
sources, as provided by Environment Canada and improvements for the SURE data
base. The data now represent a 1978 base year for annual emissions of
particulates, S02, N02, VOCs, and CO.
Historical Trends
National trends in the emission of particulates, S02, N02, VOCs, and CO
have been reported to EPA (1982a,b). These data are intended primarily to
identify, on a national basis, the long-term trends in emissions as well as
recent changes caused as the result of air pollution control efforts. These
reports show trends in emissions by source category for general management and
public information purposes. More detailed historical emissions estimates have
been developed for S02 and N02 (Gschwandtner et al., 1981). State-level
estimates of these emissions for major source categories have been developed for
the period 1950 to 1978 for states in the Eastern United States. Work is now in
progress to expand the data base to include all states and all years, in 5-yr
intervals, for the period 1900 to 1980.
333
-------
CURRENT DEVELOPMENTS
To meet the needs of the Federal National Acid Precipitation Assessment
Program (NAPAP), EPA is developing a new emissions inventory data base. The
data are being installed in an Emissions Inventory System file on EPA's Univac
computer at Research Triangle Park, North Carolina. The data base will
represent the base year 1980 and will cover all states except Alaska and Hawaii.
The initial data file has been created from current NEDS data. These data are
being improved by using input from the users and quality assurance activities
conducted by the data base manager. Data for major Canadian sources are also
being included in the file.
The objective of creating the data base is to provide a single, common set
of 1980 base-year data to be used as the starting point for acid deposition
related analyses requiring emissions data. Atmospheric modeling activities
involving the development of both Lagrangian and Eulerian transport models will
be supported. The data base may be expanded to include additional pollutants
such as sulfates, ammonia, and chlorides as required. Thorough documentation of
the NAPAP inventory is planned. Updates to the data base will be subject to
peer review and must meet the needs of the models, procedures for temporal and
spatial allocation of emissions will be developed. A procedure for allocating
VOC species into reactivity classes, drawing upon the information available from
the NECRMP project, is also planned. These efforts are expected to take place
over the next few years. Plans also call for the eventual development of a 1984
base-year inventory by FY1988. Further information on the NAPAP data may be
334
-------
obtained by contacting the data base administrator, David Mobley, at
919-541-2578 (FTS 629-2578) or the author of this paper.
NEDS is expected to continue normal operation for at least 2 yr. EPA is
developing a new data base management system, the Aerometric Information
Retrieval System (AIRS), that is intended to replace NEDS, HATREMS, EIS, and
other systems for storing air quality data. AIRS is scheduled to be completed
in 1985 at projected budget levels. When completed, AIRS will make available to
state and local agencies, as well as to EPA users, a more advanced
state-of-the-art software system for processing emissions and air quality data.
REFERENCES
Bosch, J. C. 1982. Emission Inventories for Acid Rain. Presented at Specialty
Conference on Emission Inventories and Air Quality Management, Air
Pollution Control Association, Midwest Section, Kansas City, Missouri.
Gschwandtner, G. C., C. Mann et al. 1981. Historical Emissions of Sulfur and
Nitrogen Oxides in the Eastern United States. Presented at the Air
Pollution Control Association Annual Meeting, Philadelphia, Pennsylvania.
Heisler, S. L. 1982. United States Emissions: NEDS, MAP3S, and the 1982 EPRI
Inventories. Presented at the Air Pollution Control Association Specialty
Conference on Atmospheric Deposition, Detroit, Michigan.
U.S. Environmental Protection Agency. 1982a, National Air Pollution Emission
Estimates, 1970-1981. EPA-450/4-82-012, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1982b. National Air Pollution Emission
Estimates, 1940-1980. EPA 450/4-82-001, U.S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1981. Compilation of Air Pollution
Emission Factors: Highway Mobile Sources. EPA-460/3-81-005, U.S.
Environmental Protection Agency, Ann Arbor, Michigan.
335
-------
U.S. Environmental Protection Agency. 1980. NEDS Information.
EPA-450/4-80-013, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina.
U.S. Environmental Protection Agency. 1978. Compilation of Air Pollutant
Emission Factors, Third Edition including Supplements 1-13. EPA-AP-42/8,
U.S. Environmental Protection Agency, Research Triangle Park, North
Carolina.
U.S. Environmental Protection Agency. 1976. AEROS Users Manual, including
Updates 1-5. EPA-450/2-76-029, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina.
DISCUSSION
P. Misra: Could you explain the differences between the different emissions
inventories, such as SURE, NECRMP, and RTS?
C. Mann: I am not sure I can explain the differences in a brief time period.
Each of the data files was developed to serve a slightly different purpose.
They started at different points in time, using different NEDS files, and
changes were made in each inventory based on different data that were collected,
different assumptions that were made, and different emission factors that were
used. Technically, there are a number of reasons why these things are
different.
One of the things that we are trying to avoid in creating the NAPAP file is the
proliferation of multiple sets of data bases such as those people have been
using in the past. Ideally, we want one single set of data for everybody to
start with as the basic emissions inventory data that goes into their analysis.
336
-------
EMISSIONS INVENTORIES AND
THE NATIONAL EMISSIONS INVENTORY SYSTEM*
Arthur Sheffield
Inventory Management Division
Program Management Branch
Environment Canada
Ottawa, Ontario, KlA 1C8 (Canada)
INTRODUCTION
The National Emissions Inventory System (NEIS) is a data storage and
retrieval system maintained by the Inventory Management Division, Program
Management Branch of Environment Canada. Such an inventory of sources and
emissions is a prerequisite to any national program on air pollution and
abatement. It provides the information that enables problem areas to be
identified and priorities to be set. The information identifies and locates all
sources, types, and quantities of emissions. Although an emissions inventory
may be regarded as simply an information storage, processing, and retrieval
system, it is nonetheless the most important planning tool in a comprehensive
program of air pollution control. The data that a complete, up-to-date
inventory provides may be used, for example, in the design of an air sampling
network to predict ambient air quality in a region or to evaluate or modify a
control program.
Assessing national air pollution requires accurate data on the quantity and
characteristics of emissions from all sources that contribute to the problem.
The number and the diverse types of sources make field measurements of emissions
on a source-by-source basis impractical. Other means for collecting emissions
data for each source include: questionnaires, permits, telephone or written
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
337
-------
communications, and individual company records. When specific information is
not available through any of these means, data from published reports dealing
with production and/or consumption statistics and emission factors are used.
Data of this nature are obtained from various sources, such as
company-specific information from questionnaires. However, most information is
collected from various Federal departments (e.g., Statistics Canada) or from
provincial regulatory agencies, as in the case with permit data. In certain
situations, data are collected bv private consultants who mav obtain their
information from any of the above-mentioned sources.
Data in NEIS are updated every 2 yr and are summarized in the publication,
A Nationwide Inventory of Emissions of Air Contaminants. These updates are not
required under law, but they are performed for the purpose of comparing
emissions data on a year-to-year basis. In order to preserve confidentiality,
data are summarized in the publication; emissions for individual plants are not
given. The same restriction does not necessarily apply to requests for
information from the NEIS. The release of data depends upon the source of the
request. That is, in the case of public or private consultants, data are
"rolled up" in order to maintain the rules of confidentiality, but plant names
may be given to, for example, Environment Canada regional personnel.
The inventory covers the entire country (i.e., 10 provinces and 2
territories) and five major source categories: industrial processes, stationary
source fuel combustion, transportation, solid waste incineration, and
miscellaneous. Emissions from approximately 80 sectors are calculated and can
be resolved spatially in various ways, including a 127 km x 127 km grid (derived
from latitude/longitude coordinates).
338
-------
POINT SOURCE DATA
A point source can be defined as a plant consisting of a number of stacks
and any number of processes. The exact location of the plant and site-specific
data are available for these sources.
Raw Data
Within the hierarchy of NEIS, various levels contain detailed information
on the plant: specific location, stack information, process description, and
emissions.
Available stack information consists of the following: height, diameter,
temperature of the flue gas at the point of exit, flow rate of flue gas, exit
velocity of the flue gas, and dust concentration.
The industries contributing the most to the pollution burden vary according
to the contaminant under consideration. Based on the 1978 inventory, the major
contributing industrial 'sources for each contaminant were as follows:
• Particulate matter—iron ore mining and beneficiation, mining, rock
quarrying, and power plants;
• SOz—primary copper and nickel production power plants and natural gas
processing;
• NOX—power plants, sulphate pulping, and tar sands operations;
• HCs—petrochemical industry, petroleum refining, and crude oil
production; and
• CO—petroleum refining, iron and steel production, and carbon black.
The types of raw data that are collected vary. For example, plant
production or capacity are usually available from various sources. Other types
of data that are collected include: fuel consumption, electrical generation in
339
-------
the case of power plants, sulphur/ash content of fuels, operating schedules,
pollution 'ontrol equipment and their efficiencies, stack information, emission
factors, and a> tual emissions. The temporal resolution of these data is usually
annual but at times is seasonal (quarterly). Point source data, if received
through questionnaires, may contain latitude/longitude coordinates, thus
enabling in-house staff to calculate grid coordinates by using a software
package Other forms of its location (e.g., city or census division) are easily
determined.
Data are available on the NE1S for 1974, 1976, and 1978; major point soures
have been updated to 1980. Additional information for 1970 and 1972 is
available in hard copy.
Emissions Data
The contaminants for which data are stored in the NEIS were discussed in
the previous section, with the exception of primary sulphates. In addition,
emissions data for particulate matter are distributed by particle size, and HCs
are available on a compound class basis. Emissions data for particulate matter
are divided into three distributions: <2.5 (jm, 2.5 to 15 pm, and >15 ^m. The
10 classes of HC emissions data are: methane, paraffins/alkenes,
olefins/alkenes, aromatics, carbonyls, oxygenated HCs, mercaptans
(sulphur-containing compounds), halogenated aliphatics, other, and unidentified.
Other than the activities mentioned earlier, very little work has been done
with respect to the temporal resolution of emissions data (i.e., quarterly).
Data are limited to only a few sectors, and the present estimates are extremely
crude. The major source of information for apportioning these data is quarterly
340
-------
fuel consumption statistics. Data for such activities as monthly sulphur or
material balances and power plant electrial generation are scarce.
Emissions are estimated by any of the following methods (ranked in order of
decreasing use):
• Standard emission factors on a sector basis, applied with plant-specific
information;
• Consultants' reports, questionnaires, and other governmental data;
• Plant-specific emission factors used with other plant data;
• Material balances; and
• Source testing.
On a year-to-year basis, this ranking changes, but one can assume that the above
list is generally correct.
Emission factors, the major tool used in developing emission estimates, are
obtained from a number of sources: the U.S. Environmental Protection Agency's
Compilation of Air Pollutant Emission Factors (Report AP-42), internally
developed factors, company-specific factors, and other available literature.
Data for both controlled and uncontrolled emissions are reported. Data on
pollution control equipment and equipment efficiency are available for some
plants. Several bytes of computer capacity are made available in the NEIS per
plant for storing these data along with either controlled or uncontrolled
emission factors.
Hydrocarbon and VOC emissions are generally estimated by using emission
factors (from AP-42 and other literature) and plant production statistics. In
the case of petroleum refineries, data are available on process-specific
operations from questionnaires. The major problem with using these data occurs
341
-------
when updating; emission factors must then be back calculated in order to be used
with updated process flows.
A copy of the detailed record formats for point sources is attached.
AREA SOURCE DATA
The area source file is composed of hierarchical records containing data
from nonfixed sources such as automobiles, fuel combustion, forest fires, etc.
Raw Data
As with point sources, the area sources (sectors) contributing the most
pollution to the total burden vary according to the contaminant. The following
area sources were the major polluters based on the 1978 inventory:
• Particluate matter—forest fires, slash burning, and industrial fuel
combustion;
• S02—industrial fuel combustion, commercial fuel combustion, and
residential fuel combustion;
• NOX—gasoline-powered motor vehicles, diesel-powered engines, and
industrial fuel combustion;
• HCs—gasoline-powered motor vehicles, gasoline and diesel marketing, and
application of surface coatings; and
• CO—gasoline-powered motor vehicles, diesel-powered engines, and forest
fires.
Transportation sources include the following: gasoline-powered motor
vehicles (cars, motorcycles, snowmobiles, and light-, medium-, and heavy-duty
trucks), railroads, aircraft, marine vehicles, offroad use of gasoline
(agricultural equipment, heavy-duty construction, and industrial engines),
diesel-powered engines (heavy-duty, agricultural, construction, and other diesel
vehicles), and tire wear.
342
-------
The types of raw data collected include: fuel consumption statistics,
vehicle registrations, ships' calls in ports, landing/takeoff cycles for
aircraft, refuse burned in commercial and industrial establishments, acres of
forest burned, quantity of slash burned, fertilizer and pesticide application,
solvent consumption, number of structural fires, and tobacco consumption. These
data are usually available on a national or provincial basis. Depending on the
base quantity, raw data are available either annually, seasonally (quarterly),
or monthly. For example, fuel consumption statistics are quarterly, whereas
data on forest acreage burned is annual. As with point source data, area source
data are available on the NEIS for 1974, 1976, and 1978; additional information
is available in hard copy for 1970 and 1972.
Emissions Data
The same contaminants discussed in the section on point sources apply in
the discussion of area sources. Other than annual data, very little emissions
information is available. Some statistics other than annual data are available,
but the data have not been correlated. A first approximation of seasonal
emissions would be the allocation of annual emissions evenly (25%) over all four
seasons.
Area source emissions data can be resolved spatially in any of the
following manners: national, provincial, census division, 127 km x 127 km grid,
and census metropolitan area (CMA) for the 15 largest metropolitan areas in
Canada. The starting point for apportioning area source data is the province.
In order to obtain emissions on any other spatial resolution, a method of
343
-------
proration is used. The following parameters are used to prorate the provincial
emissions: population, land area, number of dwelling units, labor force, acres
of fertilized land, plant capacity, number of employees, landing/takeoff cycles
for aircraft, and ship departures. These data are stored in proration tables in
the NEIS for each year for which data are available. Emissions can then be
determined by using the simple general equation:
Resolved
Parameter Value for Resolved Area x Provincial Emissions = Area
Parameter Value for Province Emissions
This standard technique is not in published form, although the argument values
of the proration parameters are accessible through an NEIS printout.
Provincial emissions generated for area source sectors are calculated by
using emission factors and base quantity statistics. In a few cases, control
efficiencies are applied to uncontrolled emission factors (e.g., particulate
emissions from coal combustion. Emission factors are mostly taken from EPA
Report AP-42, but they can also be taken from other literature or derived
internally. The cast approach is used widely for motor vehicle emissions, where
emission factors are calculated by using actual data from domestically tested
vehicles and emission standards applicable solely to Canada.
Emissions for VOCs are estimated by using two major references, EPA Report
AP-42 and an internal report on the sources and emissions of VOCs in Canada.
The latter document was not cited in the discussion of point sources, because it
was not generally used as a reference for estimating point source emissions.
344
-------
Whether they are total VOC emissions or emissions by compound class, all VOC
area source emissions are calculated by using emission factors.
A copy of the detailed record formats for area sources is attached.
GENERAL INFORMATION
As discussed in the introduction, accurate data on the quantity and
characteristics of emissions from all sources contributing to air pollution are
required to assess the problem. Thus, an attempt has been made to incorporate
as many sources of air pollution into the inventory as possible. Based on
experience in developing the inventory over the past few years, data are
complete to the extent possible. There are individual gaps in the plant file
for many point sources, (e.g., stack data), but data have generally been
sufficient for retrieval requirements.
The most recent data on file is the 1978 inventory; data on major point
sources have been updated to 1980. Present plans call for completion of the
1980 update within a year. Ideally, inventories should be as up-to-date as
possible. However, due to the actual time required to assemble and complete
data for an inventory, there is going to be a considerable lag in the completed
product. For example, Statistics Canada, a major source of production and fuel
consumption statistics, does not publish its results for 12 to 18 mo after the
calendar year. Thus, 1980 statistics are usually not available until mid-1982.
345
-------
There is no formalized quality assurance program for the emissions
reporting developed in the inventory. Information on a particular plant may be
obtained from multiple sources (e.g., provincial agency, company questionnaire),
and there is always the possibility that emissions calculated from these sources
may be significantly different. It is then the project engineer's
responsibility to use his/her expertise and judgment to determine which source
is correct. This can be achieved in different ways (e.g., direct contact with
company, personal experience in that industry sector).
Estimates of the overall precision of the inventory have been made for two
contaminants, S02 and NOX. A critical review of the methodologies used to
determine information sources was made prior to establishing which factors
determine the confidence level of the inventory. Spot checks and systematic
reviews of emissions versus production and/or previous emission rates were also
made. The results of this exercise showed that the precisions of the two
inventories were <6.3% and ±10.3% for SO2 and NOX, respectively.
Maintaining the quality of the data is the responsibility of the Inventory
Management Division. Random checks of the data prior to and following input to
the NEIS are necessary in order to ensure that emissions data are reasonable
relative to historical information. Both an internal and external review of
inventory data is necessary to maintain the quality of the data. When a draft
report is completed, it is forwarded to Federal regional and provincial offices
for technical comment, as well as to other divisions within Environment Canada,
to ensure that data have been used correctly. Both summarized reports and
346
-------
individual plant files (in computer format) are forwarded. Reviewer comments
are analyzed and, if necessary, the report is revised.
Presently, data (point and area sources) are manually input to the NEIS
through the use of updated files. Manual entry necessitates strict control of
the inputs from the original keypunching. Once it is ensured that all inputted
data are correct, retrievals on any level are possible, for any parameters
required. For example, plant emissions can be determined by stack or summed for
the entire plant, or total emissions can be retrieved by sector for each
province. There is great flexibility in requesting retrievals at any level
needed.
347
-------
Ji,
5-
4-1 aj
C -O
•H to
o c
O, CO
0) 4-1
4~t C
, h
o ">
•u C
C r'1
> ^
C 4->
M C
0)
3
CO O
25 W)
(U
3
(30
348
-------
Sac w'
ui z.
O « m
W) Q. —
< « a
z a.
o o
U iw
O »'
£ 5
u. -
s!
e
a>
in
>>
to
C
O)
C
10
C
o
•r-l
to
m
•1-1
e e
C
-------
"I!"
t- U
2g
e E
o> u
J-> O
O.
l-i C
O -H
4-1
C -
0) ^-1
> m
c
M r-l
0>
« >
C 0)
O .-I
•H
in -
w «
•H (!)
e u
U !-i
3
r-4 O
(U tn
O 4->
•H C
iJ -H
CO O
i-i
3
oo
•H
b-,
350
-------
0
Ol
O 4J
AJ C
c a>
a) e
> 3
c u
M O
to
O
•H
(A
U)
a
c
c
O
•H
>*
-------
(0
^
in
C
Ol
6
V) )-l
C O
o y-j
•H
to 4-1
>
05 0)
2 .H
3
60
b-,
352
-------
OECD PRESENTATION*
Anton Eliassen
Norwegian Meteorological Institute
Blindern, Oslo 3, Norway
Although I am not an emissions expert, I have been using emissions data for some
time. It is therefore perhaps easier for me than the emissions experts to say
that not everything is fine in the field of emissions inventories.
I work at the Norwegian Meteorological Institute, which is one of two institutes
responsible for meteorological dispersion modeling in the European program on
the long-range transport of air pollutants. This is a United Nations program,
which operates under the United Nations Economic Commission for Europe (ECE).
It is also part of the convention we have in Europe on long-range transboundary
air pollution that has just entered into force.
This program is now an activity under the convention, where it is referred to as
EMEP. Basically, we have been studying sulfur transport up to now, but we
expect that since the convention has entered into force, we will include other
species.
To start with the sulfur emissions, these are given in a report here, which is
available upon request from the Institute. This is 1978 emissions data for S02
in Europe. These are data from both the Eastern and Western European countries.
I think the emissions from Eastern Europe are bound to be relevant in the
oxidant problem because, whenever there is a high pressure area with good
chances of oxidant formation over Europe, you will have mostly easterly winds
and you will have transport from the Eastern countries towards the west, with
time for oxidant production along the way.
The problem is that Europe consists of several independent countries, each of
which decides what sort of data to provide for this work. Up to now, we have
received at least one number for sulfur emission from each country. However,
some of the Western European countries have given us fairly detailed emissions
inventories for S02. So, this is the information we have to work with.
This is an example of an emissions grid for Czechoslovakia. We received one
number from Czechoslovakia, which was submitted to ECE. However, we know the
location and type of industries found there, and it is amazing how much
information it is possible to find. We know the locations of the cities and the
populations of the cities. So, we try to distribute as best as we can this
total number over Czechoslovakia. This is how we work.
*This text is the transcript of a presentation made at the EPA-OECD workshop.
It has not been reviewed by the U.S. Environmental Protection Agency and
therefore does not necessarily reflect the views of the Agency, and no official
endorsement should be inferred.
353
-------
Other countries, for example the Federal Republic of Germany, have provided very
detailed data. In such cases, we obviously get much better information.
However, there is no point in using such detailed information from one country
when we may have only one number for several other countries. Therefore, we
grid the data from the Federal Republic of Germany onto the 150-km emissions
grid that we are using to run the model. So, the trouble is the vast difference
in the quality of data submitted by the different countries.
On the map shown here, which illustrates total S02 emissions, you do not see
numbers but colors to indicate the most important emissions squares. The
highest emission in the whole grid is in the Soviet Union. In face, we have
some additional data for that particular area. The only other thing we have is
a total emission number for S02 , which is for the whole country, including
Siberia. It is quite difficult to grid that information in.
This is the situation that we have to work with. We hope that it will improve,
but it appears to be a quite sensitive area.
As for NOX, this is basically the data that we used to perform the model
calculations Dr. Hov discussed. The NOX emissions data are available in this
report from the Norwegian Institute for Air Research.
try to outline briefly how these numbers were estimated. For
OECD-Europe , we have fairly good data, based on emission factors and fuel
consumption. From there, we obtained national NOX emission estimates, which are
available from OECD reports.
As for Eastern Europe, we obtained one number, an annual NOX emissions number
for each country in the OECD region. Then, we relate that number to energy
consumption, using energy consumption data that were available from the United
Nations Statistical Office. That included data for both Eastern and Western
European countries.
If you look at the data for OECD-Europe, you can relate energy consumption on
the horizontal axis to the NOX emissions on the vertical axis. First, you have
to remove the energy consumption that obviously does not produce NOX. After
that, you get quite a nice relation. So, we think that the best thing to do for
Eastern Europe is to take the energy consumption data from the United Nations
Statistical Office and to use this same relation to estimate their NOX
emissions .
We needed these data in the grid to perform calculations with the model. We
distributed the NOX emissions according to the S02 emissions that we had
estimated earlier, with a few exceptions.
Then comes the most difficult part, the HCs. From OECD, we again have some
estimates, NMHCs , and we use those that the different countries have submitted
to OECD. Of course, we do not always know the basis for those numbers.
The difference in the ratio of hydrocarbon-to-nitrogen emissions among OECD
European countries varies between 0.5 and 1.8. The only thing that we can
354
-------
possibly do at present for Eastern Europe is to assume a ratio of 1 for these
emissions. Perhaps it is a little too low.
To summarize, I can present a table showing the national estmated emissions of
the species I have talked about. If we get any complaints for countries who
disagree with these estimates, we assume that we were wrong and ask for
suggestions. For example, in the case of Romania, we were informed that we
should divide our sulfur emission estimate by 10.
These data form the basis of any model calculation to include all of Europe.
The experience is that, even if you have countries with very good emissions data
and you try to run sophisticated models within these countries, many of the
calculated values are determined by the flux across the air model boundary. In
such cases, it is very difficult to avoid the problem of data availability. So,
I thought that I would point out this problem, so we do not forget it under our
next discussions. I think this also shows that there is no point in using an
extremely sophisticated model with very bad emissions data.
355
-------
OECD PRESENTATION*
Peter Builtjes
MT-TNO, Department of Fluid Mechanics
The Netherlands
The emissions inventory in The Netherlands was made by TNO by order of the
Ministry of the Environment. It contains air pollution data as well as water
pollution data. The first phase started about 1974 and it is about complete
now. We are now in the second phase of updating.
The last year for which emissions data are available is 1980. The information
for large industries was obtained by inspection and the information for small
industries was obtained by questionnaire. Inspection can, of course, be easily
done for a small country like The Netherlands.
The issue of confidentiality presents some limitations for real numbers for the
industry, but there are old tapes on file somewhere. These are updated yearly
by questionnaire for large industries. For other industries, these are updated
every 3 yr.
Something must be said about point sources, stack information, location, and
stack height, temperature, etc. Otherwise, we cannot use the information for
dispersion calculations. Area sources are divided into 1 x 1 km grids, which
include things like heating. The species are NOX, equivalent N02, S02, CO, and
HCs. There is also some information about things like heavy metals,
particulates, and ammonia.
As to accuracy, particularly with regard to total NOX emissions in The
Netherlands, the guess is that it is accurate within 10%. For HCs, it is
accurate within 30%.
There are other things like annual average values and information for traffic,
days, etc, to be considered. All of this information is available on file or on
tape, so we can use it, as we did for the chemical studies. We can process it
and we can directly use the information connected with our dispersion model.
Let's say something about scenarios. When we do a real control strategy, we use
scenarios and abatement reference cases. We use abatement reference cases to
say we could put down this industry or have this requirement for traffic. There
is also a scenario system, which means you can assume a certain gross national
product and then allocate in a balanced way which industry could increase and
which industry could decrease.
*This text is the transcript of a presentation made at the EPA-OECD workshop.
It has not been reviewed by the U.S. Environmental Protection Agency and
therefore does not necessarily reflect the views of Agency, and no official
endorsement should be inferred.
356
-------
I will conclude my remarks by showing you two slides. The first slide shows NOX
emissions for 1980 in The Netherlands. Because, The Netherlands is so small,
you cannot do any real calculations for the surrounding countries. So, we also
used information from Germany.
You can clearly see the area including the Rhine, Uie Ruhr, and Antwerp. This
is an area of approximately 305 km2. These are grids of 10 km x 10 km. This is
just for the NO values of the total NOX values for 1977, and specifically just
for mobile sources, traffic, for The Netherlands. Road traffic and auto
traffic, 270 stationary sources—power stations, industry, we interpreted around
500 x 106 kg.
For the Federal Republic of Germany, it is only about 1,500 kg and for Belgium
and Northern France it is about 370 kg.
357
-------
DISCUSSION FOR OECD PAPERS
D. Jost; Towards the end of the year, we will have available a nationwide
emissions inventory for NOX, which is based on energy consumption and energy
production. It will be divided into grid sizes in the range of 20 km, and it
will be based on data from villages.
B. Luebkert: Will that only be available on a grid basis or will it also be
available with respect to source categories?
D. Jost; The original data will be available on a grid base of 1 km x 1 km.
Within that framework, one is free to locate the known point source. We know
there are point sources within this base, but we do not give the exact
locations.
B. Luebkert; That is not quite what I mean. Will it give information all
together on a nationwide basis? Will it classify how much is emitted by each
industry source category?
D. Jost; Yes.
B. Luebkert; So, you have one figure for traffic and you have one figure for the
chemical industry. Will that be available too?
D. Jost: This will be available. It will not be printed out at once, but it
could be obtained from the system.
H. van Pop: Will there also be an indication of source height in that emissions
inventory?
D. Jost: There will be source height categories, although not exactly for each
source height. I do not know the numbers, but there will be several categories
within this system, which we thought would be good enough for calculating
medium-range transport. I cannot give you the area at this moment.
L. Kropp: To add to that, from the emissions declarations that have to be given
by large-power-plant operators, we will be able to get all of these source
heights for the individual sources of the heavily polluted areas.
358
-------
GENERAL DISCUSSION FOLLOWING
SESSION II
Lars Lindau, Chairman
L. Lindau: I would like to review some questions that I posed this morning.
The new point is, from the discussion in OECD, whether we are going to have a
project on total chemical oxidant modeling and emissions inventories. What are
the main difficulties and where are the data errors when you are establishing
inventories? You could list a lot of different questions from that.
In addition, no one has talked about emissions from vegetation and forests and
how they have to be dealt with.
There have been some discussions about currentness, time variation,
confidentiality, things like that.
Some of these questions have already been brought up during the discussion, but
how big are the errors in making emission estimates? You have to know whether
these are 10%, 30%, as was talked about from the Dutch side, but perhaps they
are much more. Are they of a factor of 2 or a factor of 5? How big are the
errors?
What does it mean to the output of a model if there are large error variations
in emissions data from different areas? That important question was raised here
before by Anton Eliassen, especially for the European case where you have
trouble getting data from the Eastern countries.
Is it worthwhile to raise emissions data from the Western countries?
What is the cost of establishing an emissions inventory for NO* and different
process HCs? For example, would a grid size of 100 km, some sort of area could
be Germany or some other area, and some sort of error, some sort of reliability,
and then some sort of breakdown of the emissions.
Another question that is important is: Which error in emissions data can you
accept when you are discussing the end result, the control strategy? What is
the need for reliability in the emissions data we are using when we are using
the models for control strategy purposes?
H. van Pop; Anton Eliassen mentioned an interesting thing, that it might be
possible to construct emissions inventories in an indirect way. This obviously
requires industries and so on, but it can be done by just making smart guesses.
It would be useful to compare these guesses with emissions inventories, with
well-known emissions inventories, to prove if this method can be used anywhere
you want. I would like to ask Anton if that has been done. Have you compared
your guesses with existing emissions data?
359
-------
A. Eliassen: Unfortunately we did not think of that. For countries supplying
data that we thought were accurate, we used those. We did not try to make
so-called intelligent guesses.
L. Kropp: In this context, I compared the data you just mentioned in the first
and second slide (between) the NOX emissions of European countries presented in
the paper. I realized that these data were about 30% less than the data I found
for different emissions inventories. I first thought that NOX was given as NO.
J. Bosch: Our responsibility is the National Emission Data System and national
estimates. We have two means of estimating emissions. One is a "bottom-up"
approach whereby we obtain individual plant-by-plant data nationwide. The other
is a "top-down" approach whereby we use Department of Commerce and national fuel
consumption figures.
Actually, in comparing the two of them, the difference in variance is less than
5%, generally about 2%. So, this is a rather independent means of estimation
that could actually be applied to any nation that is obtaining national data on
fuel consumption, which is quite common.
E. Runca: I would like to make a general comment on model application.
Considering the difficulties in obtaining emissions data in Europe, models must
be adequate to the data available. This is true if you want to apply the model
to the whole of Europe, including Eastern and Western counties.
On the other hand, for other regions of Europe, it might also be interesting to
have a monitored data base, monitored data, and to apply more sophisticated
models to verify that some assumptions are valid and some strategies are
correct. So, I do not think that we have only to look at the general
description of the problem in Europe, including Eastern and Western countries.
G. Wh i 11 e n: As a modeler and a user of some of these emissions data, a powerful
cross-check amongst emissions.inventories is a per-capita emissions of HCs, a
per-capita emissions of NOX, and perhaps an HC-to-NOx ratio determined on a
large scale. One might subtract the power plant emissions, since they are a
large elevated source of NOX. Sometimes, when you go to a per-capita emission
of HCs based on total fuel consumption, you lose some of the range in the HC-NOX
ratio that you see in emissions inventories and you also provide a cross-check
within the inventory itself.
Another procedure that we have used is to look at the overall speciation. An
important aspect at any emissions inventory is the emission of carbonyl
compounds like aldehydes. Occasionally, they get left out, and the models are
sometimes very responsive to these things. Also, vou can look at a cross-check
between other emissions inventories, a very useful procedure.
360
-------
SESSION III
AVAILABLE AEROMETRIC DATA BASES
April 13, 1983
361
-------
AVAILABILITY OF OZONE AND OZONE PRECURSOR DATA FROM THE SAROAD SYSTEM*
Jacob G. Summers
Monitoring and Data Analysis Division, MD-14
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
INTRODUCTION
The Storage and Retrieval of Aerometric Data (SAROAD) system was developed
in 1966 by the Federal Government when it became apparent that the large volume
of data generated by ambient air quality monitoring required management by a
computerized system. The volume of data used to establish SAROAD initially
involved a network of approximately 300 sites. This network, the National Air
Sampling Network, began operation in 1957, generating approximately 100,000
pollutant volumes nationwide a year.
SAROAD has evolved from this initial system to a complex system that
currently stores over 200 million pollutant values collected from over
16,000 sites operated by Federal, state, and local agencies. The 4,000 to
5,000 active sites in SAROAD report approximately 20 million data values
annually. The remaining 11,000 to 12,000 sites are no longer active, but they
are retained for historic information.
*This paper has been reviewed by the Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, and approved for publication.
Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
362
-------
Growth in the volume of data collected has mandated standardized procedures
for site characterization, sampling method identification, data editing and
validation, and data analysis and summarization. In order to use these data for
analysis, over 30 programs have been standardized and are available for
retrieval and display of both raw data and summary statistics.
DATA COLLECTION AND REPORTING REQUIREMENTS
The change from a small Federal sampling network to a large nationwide
network operated by Federal as well as state and local agencies was a result of
the 1970 Clean Air Act Amendments. This legislation, which established the U.S.
Environmental Protection Agency (EPA), required EPA to develop and implement
National Ambient Air Quality Standards (NAAQS) for six pollutants: TSP, S02,
CO, 03, HC, and N02. (The NAAQS for HC was recently rescinded because it is
used only as a guide in devising control programs for 03.)
The Clean Air Act Amendments required that primary ambient air quality
standards, designed to protect the public health, be met nationally by 1975
unless a 2-yr extension was granted by the EPA Administrator. Secondary
standards, designed to protect the public welfare, had to be achieved within a
reasonable time. Each state was required to develop and submit to EPA a State
Implementation Plan (SIP) to identify corrective actions to be taken to reduce
air pollution to meet the NAAQS.
Each SIP was to include the design and operation of an ambient air
monitoring network to determine compliance with the NAAQS. In addition to
363
-------
operating the monitoring network and collecting ambient air quality data, state
and local agencies were required to report the concentrations to SAROAD each
quarter, with 45 days after the end of the quarter. Historic data collected and
utilized to develop the SIP were also to be reported. These data were to be
used to:
• Judge compliance with and/or progress made toward meeting ambient air
quality standards;
• Activate emergency control procedures to prevent or alleviate air
pollution episodes;
• Observe pollution trends; and
• Provide a data base for evaluating the effects of urbanization, land
use, and/or transportation planning; for developing and evaluating
abatement strategies; and for developing and validating diffusion
models.
When the NAAQS were promulgated by EPA in 1971, standard monitoring methods
(reference methods) were also promulgated for each pollutant. These reference
methods were the methods most often used by EPA to collect ambient
concentrations and were thought to be the most reliable. As agencies began
routinely utilizing the reference methods, methodological problems were
encountered. Also, differences were noted in the sample collection time,
sampling procedures, and sampling instrumentation required for the NAAQS.
Automated instrumentation was developed to sample for criteria pollutants, which
created in turn the problem of comparing sample results from different chemical
analysis procedures.
These problems were resolved in 1975 when EPA regulations established an
"equivalency" method between reference methods and "candidate" methods. The
364
-------
regulations prohibited state and local agencies from purchasing any new
instrumentation that was not equivalent and permitted the utilization of
existing nonequivalent instrumentation until February 1980. The equivalency
regulations identified test procedures and reporting requirements that each
instrument manufacturer must complete to .allow EPA to determine whether the
candidate method was equivalent. The manufacturer was also required to supply
an instrument manual to the purchaser to identify mandatory operating
parameters.
Table 1 lists criteria pollutants and the reference methods and number of
equivalent instruments available for each. Table 1 does not include HC because
the HC standard is only used as a guide in developing the SIP to meet the 03
standard. HC monitoring is not currently required but is performed by some
state and local agencies.
Although the regulations for SIPs specified an approximate number of sites
for a given geographical area,, based on population, they did not specify
location. This resulted in inconsistencies in site locations and specific probe
locations, such as height above ground and distance from the nearest street or
road. These variables have been studied, and specific guidelines have since
been developed and implemented to standardize these parameters.
Specific quality assurance procedures were not specified as part of the
initial SIP regulations. Quality assurance procedures were left up to the
individual state and local agencies to develop and implement.
365
-------
TABLE 1. CRITERIA POLLUTANTS AND MEASUREMENT METHODS
Pollutant
Suspended particulate matter
Pb
S02
CO
03
N02
Reference Method
High-volume
High-volume/atomic absorption
Pararosaniline
Nondispersive infrared
Chemiluminescence
Chemiluminescence
Number of
Reference
Equivalent
Instruments
None
4
15"
9b
14C
13d
"Two equivalent procedures for analyzing samples collected by the reference
method and 13 equivalent instruments sampling continuously.
bAll instruments use the reference method measurement principle.
°Nine instruments use the reference method measurement principle.
dTen instruments use the reference method measurement principle; three use a
manual sampling technique and laboratory analysis.
In 1975, the Quality Assurance and Environmental Monitoring Laboratory of
the EPA Office of Research and Development (ORD) began compiling quality
assurance manuals. These manuals consolidated previous guidelines into a
centralized source of quality assurance information. Volume I of the manual
defined the quality assurance function in the air pollution control agency and
identified the procedures required for training, preventive maintenance, sample
collection, sample analysis, data reporting, calibration, audit procedures, data
validation, etc. Volume II defined the quality assurance procedures for the
specific reference and equivalent methods for each criteria pollutant:
suspended particulate, N02, S02, CO, 03, and Pb.
366
-------
NAMS/SLAMS REPORTING REQUIREMENTS
In 1979, as a result of deficiencies in state and local agency air quality
monitoring and data reporting programs and of revisions to the 1977 Clean Air
Act Amendments, EPA promulgated new regulations that changed the air quality
data collection and reporting requirements. These changes were made to improve
the timeliness and quality of air monitoring data. The revised regulations
identified the National Air Monitoring Stations (NAMS) and the State and Local
Air Monitoring Stations (SLAMS).
NAMS is a limited network of approximately 1,200 stations required by EPA
to perform national data analysis. In this network, sites are located in large
urban areas, with the objective of measuring ambient concentrations in areas of
high pollutant concentrations and/or high population exposure. These sites
monitor for gaseous pollutants, using only continuous instruments. The raw data
values are submitted quarterly, within 90 days after the end of the quarter.
SLAMS is an expanded network of approximately 5,000 sites required by
individual states to determine violations of NAAQS. This network was designed
by the states primarily to meet their needs. The size is based on such factors
as meteorology, geography, population, and emission density. The NAMS network
is a subset of the SLAMS network. For SLAMS, the states are required to submit
an annual summary, within 6 mo after the end of the year.
The regulations promulgated in 1979 required that the NAMS network be
established and operational by January 1, 1981, and that the SLAMS network be
367
-------
established and operational by January 1, 1983. The regulations involved the
following additional specifications for both NAMS and SLAMS:
• A quality assurance plan by state and local agencies, approved by
EPA—Precision and accuracy information must be collected and reported
to assess the sampling and analysis procedures. The agency must review
and evaluate this information to identify possible problem areas and to
initiate corrective action such as more frequent calibration or
maintenance.
• Ambient monitoring methodology—For the NAMS network, only reference or
equivalent methods that monitor continuously were permitted for gases
and only the reference method was permitted for suspended particulates.
For the SLAMS network, only reference or equivalent methods were
preferred, but procedures were defined to permit states to obtain
limited approval of other methods.
• Network designs for SLAMS and NAMS—Design criteria included parameters
such as emission sources, meteorology, and geography to evaluate the
existing sites. Other parameters, such as the different measurement
scales (micro, middle, neighborhood, urban, or regional) that are
appropriate for SLAMS and NAMS and the number of NAMS sites that are
required, based on population and approximate concentration ranges, were
also defined.
• Probe siting criteria—The material used to make the probe should be
nonreactive. In addition, a maximum residence time for the sample in
the probe was established, and restrictions on the probe location at the
sampling site were made, including vertical and horizontal probe
placement, distance from obstructions, and distance from roads as a
function of traffic.
NAMS/SLAMS reporting requirements have been implemented as scheduled to
provide air monitoring data of acceptable quality, comparable data from all
monitoring stations, and timely data submission for national assessment
purposes.
368
-------
THE SAROAD SYSTEM
As previously discussed, SAROAD was intially developed to store small
volumes of data. As requirements were implemented for collecting and reporting
air quality data by state and local agencies, the volume of data acquired
mandated the development of uniform coding procedures and formats, codes to
identify the data, and procedures to edit, validate, summarize, and report the
data.
SAROAD is operated by the National Air Data Branch (NADB). The
organizations supplying air quality data to SAROAD include state and local
agencies and the 10 EPA Regional Offices. These organizations are responsible
for the following functions:
• State or local agencies establish the monitoring sites, operate the
equipment to sample the ambient air, convert the ambient air data to a
format compatible for storage in SAROAD, and submit these data to their
respective EPA Regional Office.
• The EPA Regional Office approves the sampling sites established by the
state or local agency; edits, validates, and corrects data with
assistance from the submitting agency; submits the data to NADB for
update; and returns data for standards violation and trends in air
quality.
• NADB manages SAROAD by updating data obtained from the EPA Regional
Offices; by developing and maintaining SAROAD software necessary to
edit, store, and analyze the data; by developing procedures and codes
for processing data; and by providing data to users when requested.
Before air quality data from a site can be stored in SAROAD, the site must
be registered. This involves assigning an identifier to the site and using the
identifier when any data are reported. The identifier includes the state and
369
-------
city or county where the site is located and the site number within the city or
county. Other data items that are stored for the site include: the
geographical coordinates, the time zone, the Standard Metropolitan Statistical
Area (SMSA) code, the agency name, the site address, the station type (center
city-commercial), the elevation above ground and above sea level, the Air
Quality Control Region (AQCR) code, the site type (NAMS or SLAMS), and details
describing the site. The population of the city and AQCR are also included.
SAROAD permits the storage of ambient concentrations for any pollutant or
parameter that can be defined and measured with an acceptable analytical
procedure. Each pollutant is assigned a code, and each different sampling and
analysis procedure is assigned a method code. Other codes that are assigned
include the units code, which identifies the units of measure, and the interval
code, which identifies the sampling frequency. The sampling frequency can vary
from long-term monthly exposures to short-term continuous sampling reported as
hourly averages. The complete identifier for a data value includes the site
identifier; the year, month, day, and hour; and the pollutant-method-interval-
unit codes.
In addition to storing site information and raw data, SAROAD computes and
stores summary statistics on a quarterly and annual basis. These summary
statistics are site- and pollutant-specific and include statistics such as
arithmetic and geometric means, standard deviations, maximum and minimum values,
violation counts for criteria pollutants, and the percentile distribution.
370
-------
AMBIENT DATA AVAILABLE FOR TRANSPORT MODELS
As discussed in previous sections, state and local agencies are required to
monitor and report ambient air quality concentrations to SAROAD for the NAAQS
pollutants. In addition to these criteria pollutants, many state and local
agencies collect and report data for several meteorological parameters. These
data are not required but are collected and reported at sites monitoring for
other pollutants.
Table 2 lists pollutant and meteorological data reported to SAROAD that
might be of value for input to 03 transport models. Although most state and
local agencies began reporting data in 1971 or 1972, only data for the last 5 yr
are summarized. The count represents the number of sites reporting data for
each pollutant each year. The number of sites reporting data for N02 and 03 is
most consistent from year to year. The number of sites reporting data for the
other pollutants and meteorological parameters peaked in 1980 and has now begun
to decrease. The reduction in. required monitoring data probably reflects the
implementation of NAMS/SLAMS regulations that are decreasing the total resources
available for monitoring.
In order to reduce the volume of data presented in Table 2, the third
quarters for 1980 and 1981 were selected for additional analysis. These time
periods were selected because the third quarter is the 03 season for all states.
The third-quarter data for 1982 were not utilized because the data were not
complete for all states. Also, the data for 1980 and 1981 were selected because
371
-------
TABLE 2. AVAILABILITY OF DATA BY POLLUTANT AND YEAR
Number of
Pollutant
NO
N02
NOX
THCs
NMHCs
CH4
03
Wind speed
Wind direction
Lapse rate
Mixing height
Temperature
Temperature difference
Solar radiation
1978
200
298
114
126
83
67
581
232
219
0
0
138
4
28
1979
183
314
153
103
76
60
651
242
244
0
2
103
12
18
Sites for Each Year
1980
219
341
185
88
86
72
796
317
310
0
1
135
13
25
1981
175
344
164
60
64
42
773
252
251
0
0
129
12
8
1982
138
278
144
36
28
14
656
211
210
0
0
106
11
5
the quality of data is expected to be better than that for previous years as a
result of implementing NAMS/SLAMS monitoring requirements.
Tables 3 and 4 present the number of sites reporting data to SAROAD by
state for the pollutants listed in Table 2. The counts reflect the number of
sites collecting and reporting not only 03 data but also other general
372
-------
TABLE 3. NUMBER OF SITES REPORTING DATA FOR THIRD QUARTER, 1980
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
03
7
8
1
68
5
1
5
11
5
1
36
12
3
3
11
5
6
1
. 7
14
8
2
5
2
3
2
1
5
5
17
11
24
1
03,
NOX
1
1
35
2
1
2
1
3
2
1
1
1
3
1
1
1
7
1
5
3
5
Type of Data Rcpor
03,
NOX , 03 ,
HC MET
1
1 5
34
1 1
7
7
2
1
1
8
3
1
1
1
1
1
3
1 2
tcda
03,
MET,
NOX,
HC
1
1
1
1
2
3
1
7
11
7
4
2
5
4
2
2
03,
NOX,
HC,
MET
1
1
1
7
3
(continued)
373
-------
TABLE 3. (continued)
Type of Data Reportf-d3
State
03,
03 NOX
03,
NOX,
HC
03,
MET,
03, NOX,
MET HC
03,
NOX,
HC,
MET
Oregon 7
Pennsylvania 9312 2 16
Rhode Island 1 1
South Carolina 6 2
South Dakota
Tennessee 6 1
Texas 9 2 7 10 6
Utah 4
Vermont 3
Virginia 10 7 1
Washington 8
West Virginia 2 1
Wisconsin 14 5 1
Wyoming
TOTAL 371 99 43 56 68 35
aSites were counted as reporting NOX if NO, N02, NOX, or any
combination was reported; HC if THCs, NMHCs, CH4, or any combination
was reported; and MET if any meteorological parameter from Table 2 was
reported.
374
-------
TABLE 4. NUMBER OF SITES REPORTING DATA FOR THIRD QUARTER, 1981
State
Alabama
Alaska
Arizona
Arkansas
California
Colorado
Connecticut
Delaware
District of Columbia
Florida
Georgia
Hawaii
Idaho
Illinois
Indiana
Iowa
Kansas
Kentucky
Louisiana
Maine
Maryland
Massachusetts
Michigan
Minnesota
Mississippi
Missouri
Montana
Nebraska
Nevada
New Hampshire
New Jersey
New Mexico
New York
North Carolina
North Dakota
Ohio
Oklahoma
03
8
13
59
7
1
4
8
4
1
33
13
4
2
22
12
5
9
. 7
13
9
3
2
1
4
5
5
7
7
23
11
24
4
Type of Data Reported"
03,
03, MET,
03, NOX, 03, NOX,
NOX HC MET HC
2
1
2 2
2
36 32
21 2
4 6
1 1
3 91
2 3
2 1
1
1
3 5
1
1
9
1 4
3
1
2
1
1 54
1
7
1 1
2
2
3122
3
03,
NOX,
HC,
MET
1
1
1
2
2
2
(continued)
375
-------
TABLE 4. (continued)
Type of Data Reported3
State
Oregon
Pennsylvania
Rhode Island
South Carolina
03,
03 , NOX ,
03 NOX HC
7
14 5
1
7
03, 03,
MET , NOX ,
03, NOX, HC,
MET HC MET
3 14
1
South Dakota
Tennessee 8
Texas 5 7 16
Utah - 4 4
Vermont 3
Virginia 11 7 1
Washington 8
West Virginia 2
Wisconsin 20 4 3 2
Wyoming
TOTAL 419 109 40 38 47 25
'Sites were counted as reporting NOX if NO, N02, NOX, or any
combination was reported; HC if THCs, NMHC, CH,v, or any combination
was reported; and MET if any meteorological parameter from Table 2 was
reported.
376
-------
combinations of pollutants. The general combinations of pollutants were used to
reduce the number of possible combinations.
A review of Tables 3 and 4 indicates that approximately the same number of
sites reported 03 in 1980 (672) and 1981 (678). In 1980, the number of sites
reporting data for meteorological parameters, HC, and NOX was higher than the
number for 1981. This is especially evident for Maryland, Massachusetts,
New Jersey, New York, and Pennsylvania. During 1980, these and other states in
EPA Regions I, II, and III participated in the Northeast Corridor Regional
Modeling Project (NECRMP), a project involving the collection of data to assist
states in developing control strategies for the 1982 Ozone State Implementations
Plans.
Figure 1 present the locations of sites that reported 03 data for 1981. As
indicated by Table 3, these sites are concentrated around the most populous
areas of the country, with only Idaho, South Dakota, and Wyoming reporting no
data.
DATA AVAILABILITY
SAROAD is designed to store, analyze, and retrieve air quality and
meteorological data. At least 30 different reports provide the user with
quarterly or yealy summary statistics or individual hourly data values. The
hourly data values are available in a hard copy format that displays data values
for 1 mo per page or in a computer-readable format on magnetic tape. These
377
-------
Figure 1. Sites reporting 03 data for 1981.
378
-------
Figure 1. (continued),
379
-------
Figure 1. (continued),
380
-------
reports permit data selection based on geographical area, pollutant, and time
period.
Publications that identify the most useful report programs are available.
Requests for publications and data may be addressed to the author.
BIBLIOGRAPHY
Duggan, G. M. 1982. SASD Computer Graphics. U. S. Environmental Protection
Agency, Research Triangle Park, North Carolina.
Nehls, G. J., and G. A. Gerald. 1973. Procedures for handling aerometric data.
Journal of the Air Pollution Control Association, 23(3):180-184.
Office of the Federal Register. 1982. Requirements for Preparation, Adoption,
and Submittal of Implementation Plans. Code of Federal Regulations, 40,
Part 51.
Office of the Federal Register. 1982. Ambient Air Monitoring Reference and
Equivalent Methods. Code of Federal Regulations, 40, Part 53.
Office of the Federal Register. 1982. Ambient Air Quality Surveillance. Code
of Federal Regulations, 40, Part 58.
U.S. Environmental Protection Agency. 1983. SAROAD Retrievals. National Air
Data Branch.
U.S. Environmental Protection Agency. 1982. List of Designated Reference and
Equivalent Methods. Department E, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1979. SAROAD Information.
EPA-450/4-79-005, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1976. AEROS Users Manual.
EPA-450/2-76-029, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1975. Quality Assurance Handbook for Air
Pollution Measurement Systems, Volume I - Principles. Office of Research
and Development, Research Triangle Park, North Carolina.
U.S. Environmental Protection Agency. 1975. Quality Assurance Handbook for Air
Pollution Measurement Systems, Volume II - Ambient Air Specific Methods.
Office of Research and Development, Research Triangle Park, North Carolina.
381
-------
DISCUSSION
D. Jost: Does SAROAD allow for data extraction also? For example, if we are
asking for the high and low concentrations during situations with high HC
situations, do we need then to extract HC and 03 data?
J. Summers: That could not be done directly. An overview would have to be done
to identify high sites of 03 or HCs and then retrieve the data separately.
There are other studies that may have been done by other EPA groups that would
already identify some of those, so that may already be available.
P. Misra; Do these reports give an estimate of the errors in these data?
J. Summers: As I mentioned, the precision-accuracy data were required to be
reported beginning in 1981. When these reports were designed, of course,
precision-accuracy reporting was required. So, we have a separate report after
the report that gives the precision-accuracy data. Right now a working group is
trying to evaluate the precision-accuracy data and come up with an exact way of
how it can best be used.
So, although we have got about 2 yr of it already, we are not completely sure
how it is going to be used.
382
-------
APPENDIX. GUIDELINES FOR AEROMETRIC DATA PRESENTATIONS
Surface Air Quality
Data base name/source
Area of coverage
Total number of monitoring sites
Spatial distribution (attach site map)
Year of record
Check available site information:
physical location (lat-long, UTM)
geographic location (state/province/
department, other sublevels)
elevation (MSL, AC)
classification (i.e., urban, rural
suburban, remote)
environment of site
descriptive information
dominating influence (i.e., industrial
residential, mobile)
other, specify:
Storage and Retrieval of Aerometric Data
United States
Approximately 700 ozone sites for 1981
1981
Both sets of coordinates
State, county, city
Both
Station type combine this and dominating influence
Only site address unless information in comments
See classification
Parameters (attach table of measured
parameters, associated equipment
type/analysis method and temporal
resolution)
Upper Air Quality
Data base name/source
Spatial distribution (attach standard
of flight paths)
Year/date of record
Parameters (attach table of measured
parameters, associated equipment
type/analysis method and temporal
resolution)
Spatial resolution
Surface Meteorology
Data base name/source
Area of coverage
Total number of stations
Spatial distribution (attach site map)
Site information available
Parameters measured
Time interval of measurements
Year/date of record
Storage and Retrieval of Aerometric Data
United States
Varies by parameters and year
Same as air quality
Win'd speed, direction, temperature, radiation
Hourly averages
1981
383
-------
Data Quality
Is data suitable for model evaluation?
Are standard quality assurance procedures
implemented?
Comments on data reliability.
List criteria for acceptance of data.
Present summary of quantity of ozone
measurements of available.
The air quality data is collected utilizing
consistent sampling and analysis procedures and is
suitable for model evaluation. Meteorological
data is not reported bv all States and is o£
unknown quality.
Quality assurance procedures exist and were
implemented in 1981 for all States and probably
before 1981 for many States.
All procedures for sampling, data processing and
analysis have been standardized and utilized for
several years to ensure reliable data.Recent
data are the most reliable.
Sampling site must be registered and sampling
performed utilizing EPA approved procedures.
Study must be longer than three months and sites
usually operate for several years.
Quality of ozone data is good especially for
1981 - present as quality assurance continues to
improve.
384
-------
NORTHEAST CORRIDOR REGIONAL MODELING PROJECT: DATA BASE OF
REGIONAL AMBIENT CHEMICAL AND METEOROLOGICAL MEASUREMENTS*
Norman C. Possiel+
Office of Air Quality Planning and Standards
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
Francis S. Binkowskif
Environmental Sciences Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711 (USA)
INTRODUCTION
An extensive data base of regional ambient chemical and meteorological
measurements is available for the Northeast United States from the Northeast
Corridor Regional Modeling Project (NECRMP). The NECRMP is a multiphased
program conducted by the U.S. Environmental Protection Agency (EPA) in
conjunction with state/local air pollution control agencies to support the
development and use of models to evaluate control strategies for reducing 03
concentration levels in the Northeast. The geographical domain for NECRMP is
shown in Figure 1. The focal point of this program is the combined application
of the Regional Oxidant Model (ROM) (Lamb, 1983) and urban models for this
region.
*This paper has been reviewed by the Office of Air Quality Planning and
Standards, U.S. Environmental Protection Agency, and approved for publication.
Mention of trade names or commercial products does not constitute endorsement
or recommendation for use.
ton assignment from the National Oceanic and Atmospheric Administration.
385
-------
m
o
-o
oo
c
O
e
cfl
CO
c
o
•H
00
0)
u
O)
00
u,
386
-------
The NECRMP ambient data base is the result of field measurement programs
conducted in 1979 and 1980 to provide the chemical and meteorological
measurements needed to develop and apply the ROM. These programs include the
1979 Northeast Regional Oxidant Study (NEROS I), the 1980 NEROS II, and the 1980
Urban Field Studies. Collectively, these studies have provided a variety of
chemical and meteorological measurements on regional, urban, and site-specific
scales. This paper discusses several of the regional-scale measurements and
those urban-scale measurements that, when combined, also provide data on a
regional scale. These include measurements aloft, obtained via instrumented
aircraft; HC species measurements; meteorological measurements, obtained from
the National Weather Service (NWS); and supplemental upper air meteorological
measurements, obtained specifically for NECRMP. A companion paper at this
conference by Summers (1983) discusses chemical and meteorological measurements
available from surface networks operated by state/local agencies or contractors.
Additional information on the scope of NECRMP and the ancillary field
experiments is provided by Possiel et al. (1982).
REGIONAL DATA BASE COMPONENTS
Aircraft Measurements
Regional-scale chemical and meteorological measurements aloft were obtained
during August 1979 as part of NEROS I. Instrumented aircraft operated by the
Research Triangle Institute (RTI), Washington State University (WSU), and
Brookhaven National Laboratory (BNL) provided continuous measurements of 03, NO,
NOX, S02, light-scattering coefficient (b-scat), cloud condensation nuclei
387
-------
(CCN), temperature, and relative humidity/dew point temperature. The specific
instrumentation used for measuring these elements are listed in Table 1. During
NEROS I, Regional Air Mass Characterization (RAMC) aerial sampling flights
provided measurements at several altitudes within the daytime boundary layer and
aloft above the nocturnal inversion at night. The typical sampling scenario, as
TABLE 1. NEROS I AIRCRAFT INSTRUMENTATION
Operator
Research Triangle Institute
Brookhaven National Laboratory
Washington State University
Element
03
NO/NOX
S02
CCN
b-scat
AT"
DPTb
03
NO/NOX
S02
b-scat
AT
DPT
RH°
TSRd
UVRe
03
NO/NOX
S02
CCN
b-scat
AT
RH
Equipment
Bendix 8002
Monitor Labs 8440
Meloy 285
Environment One Rich 100
MRI 1550 B
Rosemount 102
EG & G 880
AID 560
TECO 14d
Meloy 165
MRI 1550
Yellow Springs 705
EG & G 1.37-3C
Weather Measure HM 111
Eppley pyranometer 8-48 A
Eppley UV radiometer
Bendix 8002
Monitor Labs 8440 HP
TECO 43
Environment One Rich 100
MRI 1550
Metrodata M-8
Metrodata M-8
"AT = air temperature.
bDPT = dew point temperature.
°RH = relative humidity.
dTSR = total solar radiation.
eUVR = ultraviolet radiation.
388
-------
shown in Figure 2a, began with a traverse in the western portion of the domain
at about 1200 EST, followed by four traverses at approximately 6-h intervals.
These were performed at downwind distances speciiied to approximate Lagrangian
sampling of the initial traverse. The traverses were approximately 500 km wide,
flown in the wave pattern shown in Figure 2b. Occasionally, spirals were made
from approximately 500 m to approximately 3,000 m to provide additional vertical
resolution. In all, regional flights were conducted during six 24- to 48-h
experimental periods.
Regional-scale chemical and meteorological measurements were also obtained
via instrumented aircraft during 1980 as part of NEROS II and the Urban Field
Studies. Instrumentation for the elements measured during these experiments are
listed in Table 2. Aircraft were operated in Washington, DC, Baltimore, New
York City, and Boston. Regional-scale sampling was conducted in a Lagrangian
fashion similar to that for NEROS I. Urban-scale sampling was conducted in both
Lagrangian and Eulerian modes in Columbus and Baltimore and in an Eulerian mode
in the other Corridor cities. As shown in Figure 3, a Lagrangian sampling
scenario typically began with measurements upwind of the city during the early
morning, followed by a series of traverses perpendicular to the horizontal
movement of a tetroon that had been released during the morning near the city.
Traverses were performed at, below, and above the altitude of the tetroon at
various distances downwind, to approximately 200 km from the city. In many
cases, a forecast trajectory was used instead of a tetroon to estimate air
parcel positions.
389
-------
Figure 2a. Example of NEROS I regional
sampling flight tracks.
Mixing Height
Ground
r
Residual
Mixed Layer I
Ground
Day
Night
Figure 2b. NEROS I regional flight patterns.
390
-------
TABLE 2. AIRCRAFT INSTRUMENTATION USED DURING 1980 FIELD PROGRAMS
Operator
Element
Instrumentation
NEROS II Aircraft
Environmental Monitoring, Inc.
AeroVironment
Stanford Research Institute
Washington, DC, Aircraft
EPA-Las Vegas
03
NO/NOX
S02
S04
b-scat
ATb
DPTC
03
NO/NOX
S02
S04
b-scat
AT
DPT
03
NO/NOX
S02
S04
b-scat
AT
DPT
03
NO/NOX
S02
S04
b-scat
AT
DPT
Dasibi
Monitor Labs 8440
Meloy 285
Meloy SA285
NAa
NA
NA
Bendix 8002
Monitor Labs 8440
TECO 43
Not measured
NA
NA
NA
Dasibi 1003 AAS
Not measured
Meloy SA285E
Not measured
NA
NA
NA
Bendix 8002
Monitor Labs 8440
TECO 43
Not measured
MRI 1550 B
Rosemount 102U2U
General Eastern 1011
(continued)
391
-------
TABLE 2. (continued)
Operator
Element
Instrumentation
Baltimore Aircraft
Brookhaven National Laboratory
Washington State University
New York City Aircraft
Battelle Northwest Laboratory
03
NO/NOX
S02
b-scat
AT
DPT
RHd
TSRe
UVRf
03
NO/NOX
S02
CCN
b-scat
AT
RIT
03
NO/NOX
S02
S04
b-scat
AT
DPT
AID 560
TECO 14d
Meloy 165
MRI 1550
Yellow Springs 705
EG & G 137-3C
Weather Measure HM 111
Eppley pyranometer 8-48 A
Eppley UV radiometer
Bendix 8002
Monitor Labs 8440 HP
TECO 43
Environment One Rich 100
MRI 1550
Metrodata M-8
Metrodata M-8
Bendix 8000
Monitor Labs 8440 HP
Not measured
Not measured
MRI 1550
Rosemount 102U2U
EG & G 137-C
(continued)
392
-------
TABLE 2. (continued)
Operator Element Instrumentation
Boston Aircraft
Battelle Columbus Laboratory 03 Bendix 8000
NO/NOX Monitor Labs 8440 HP
S02 Not measured
S04 Not measured
b-scat MRI 1550
AT Rosemount 102U2U
DPt EG & G 137-C
SAT = air temperature.
bNA = not available.
°DPT = dew point temperature.
dRH = relative humidity.
eTSR= total solar radiation.
'UVR = ultraviolet radiation.
393
-------
Figure 3. Typical Lagrangian sampling flight track. Times indicate position
of tetroon; letters identify aircraft traverse locations.
The Eulerian flight scenarios included measurements 20 km to 40 km upwind
between 0500 and 0600 EST, followed by a traverse and spiral pattern over the
city and downwind from mid-morning through early evening (approximately 1700
EST). A typical afternoon Eulerian flight track is shown for New York City in
Figure 4. Spatial resolution of the urban plume was obtained from traverses at
several altitudes perpendicular to the plume, and/or traverses at a fixed,
mid-boundary layer altitude perpendicular to the plume, with spirals to provide
vertical resolution. On more flights, traverses were 50 km to 100 km wide.
Also, Eulerian flights in Columbus were occasionally extended to include
394
-------
nighttime sampling of the urban plume. The total number of flight days for each
city was 17 in Columbus, 13 in Washington, DC, 22 in Baltimore, 19 in New York
City, and 15 in Boston. Of these, there were 2 days with flights in all areas,
5 days with flights in all Corridor cities, 12 days with flights in both
Washington, DC, and Baltimore, and 8 days with flights in both New York City and
Boston.
The quality assurance program for aircraft operations included audits of
the chemical instrumentation onboard each aircraft. Audits were conducted in
the field during the first days of the program. The audit results are given by
Murdoch et al. (1979) for NEROS I and Arey et al. (1980) for the 1980 aircraft
programs. Routine quality assurance procedures included zero and/or span checks
of chemical instrumentation prior to, during, and/or following each flight. All
gaseous chemical measurements are in parts per million by volume; other elements
are in metric units.
Hydrocarbon Species Sampling
A major effort to sample for HC species was incorporated in NECRMP. Grab
samples (1 to 3 min) were collected from aircraft during sampling flights, and
1-h integrated samples were collected at fixed surface sites. All samples were
analyzed in the laboratory by gas chromatography (GC) for species
concentrations. The species analyzed are listed in Table 3. Concentrations are
reported as parts per billion of carbon.
395
-------
o
£
O
8
I
to
N
§
m
33
pa
H
Z M
UJ Z
10 UJ
I CO
ui Z ;:
— uj <
UJ-JZ
ct >- o
>-XZ
a a
OO
u u
a.C4.
i i
y
UJ
UJ CO
Z i
I LJ -J UJ
'0^3
I— UJ -J I
I = O
_J ~! h-
>- a i
X I- -•
I- I >•
UJ IA X
UJ
UJ UJ I
•z. z :
in
G
.
o
u
VU UJ
a z z
uj uj
UJ UJ Z Z
Ul UJ LJ UJ
ZOO*
r- UJ I I
u. Z
*-* uj
Z^
uj rj
O i
Z I UJ UJ LJ Z
oooooooo
<
M CJ
- UJ
cj a i
- I I
** z :
3 x
»-
Lw LJ
O Z
uj^l— uj LJ O Q. O <
a-zzu-arircj-i;
I UJ uj I I O —I LJ
«j a. a. «j CM -J u >-
I I I I I X >- I i
-J IH CM -J _J CJ CJ
>- I I >- >- IH I
x -i -J x x a: -J
X
>- x :
x u i
i a 1-1
i a.
ui ct
5 H u,
. cr 2 o
o -i x
UJ I UJ UJ
Z Kl X 0
LJ i a. —i
z «i i <
O I- < I- >-
:: -J i— uj u
UJ Z Z
IT uj UJ
o a. a.
l i-l i-l 3 U> I-
«O 83
o o
O- O
03 O-
O O
•"TCEKiKliHlliHinuJLb.*-
I I I I I -ICC-l-Juj
CJcMcvih-cjiHiHF-cjKioiHCMKt*ru>isoF''-a~ooi
O-O-OO-OOOOOOOOOOOfH,
»-H UJ UJ
-J O O <
UJ IM -
X
uj :
uj
i uj Z i
Z <
< X
a >H z
r- a <
i i i-
•r u a.
» i u
• • I UJ h-
I —
a -J H-
O >• uj
Oh-IHiHCMKI^in^lNeO
>• iu a i i i i i i i i
z
X
UJ Z *
X < <
-j a £
>- LJ UJ
XXX
I I
Ul
< Ul
X Z
Ul <
X X
I I UJ
-J X
>- o
I— U
UJ 't-
z
X
UJ
5
UJ
UJ
ii UJ-J_JQ:-IUJ>-IIIIIIIIIIUJUJUJ
-Ji5i22 !: 2 ":"**£ r4lt2 = SE-
>-_io^c)OpooobooouJoi-p
sststiSiStstesiittn
555555555=
; ui CM CM
Ojooi-iojKi^in4i>cooo>-icMK>'>r
I*IKI.T ^^-
• UJ
Z U
<
CJ >-
O X ^
o o
i I/I O
o o
ooooooooooooooo
ui Z
Z ul
Ul IU -J
Z -t >•
«»>-»-
X X Ul
>- t- U
UJ iu <
e -< M
o o a
o o o
uj i
uj ui z :
z z -X
"J. UJ t-
c. n. 3
o o co
u. ct i
UJ
Ul
CO Z
Z Z cu ui i o ul
D i
i b 3
I O O
>- >- uj Z v-
i§5'
I I U
; KI a
i
>MZ- a.
>- uj
X a.
^- l
i >- — >•
X »H X
(- O »-
) I UJ
Kl I"
» I
CM CM
I CM Kl
I CM C J
> o O
UJ _J Ul >^
a. u; u< u — x
•^UJIUUJUICJQIUJ
>- z z x x i t- i: ui
— Ul < UJ UJ -J I MZ
I-XXXX>-KIQUJ
UJ UJ UJ I I X * I M
HXirj(j|-rj^Z
i i l I l bi • » uj
KliHZHOZcvlcJO
•^in«jj>f*>c3o>OfHCM
CM CJ CJ CM CJ CJ Kl Kl 1*1
OOOOOO^C* H.
fc
ui ul
z a
< i
x -i
Ul Ul >-
z x x
< I >-
X -I UJ
iu >- ::
5?S
-J IU I
u x: M
>- i -
CJ CJ CM
m * in ~o i
— C s. C,
o.
o
X X
u t-
>• a
H >-
UJ I
c m
Kl ^1
396
-------
AMIMr MM •
Figure 4. Typical afternoon Eulerian flight track. Letters
identify aircraft traverse locations.
were obtained within the mixed layer, the residual mixed layer at night, and
above the mixed layer. Similar spatial sampling for HCs in the Northeast (but
on a much smaller scale) was conducted during the 1980 NEROS II.
Hydrocarbon samples were also obtained over urban areas and within the
urban plumes of Columbus, Washington, DC, Baltimore, New York City, and Boston
during 1980. Aircraft grab samples were collected both within and above the
surface stable layer on early morning flights and, in Columbus and Baltimore,
within and above the mixed layer on later flights. Samples were collected
upwind, over urban centers, and out to 100 km downwind. In all, 223 samples
were analyzed for flights in Columbus, 45 in Washington, DC, 136 in Baltimore,
65 in New York City, and 87 in Boston. Also, as part of the 1980 program, over
162 quality assurance samples were collected for use in evaluating the
397
-------
comparability of ambient samples analyzed by the three GC laboratories operating
during the program.
Ground-level 1-h samples were normally collected from 0500 to 0600 EST and
from 0700 and 0800 EST at two sites in the urbanized portion of each city. In
addition, upwind surface samples were collected in Columbus. A total of 794
surface samples were analyzed during the 1980 study.
Meteorological Measurements
NWS Surface Data--
Surface observations of meteorological conditions are made hourly at fixed
locations by NWS. Figure 5 shows a plot of locations from which hourly data
were received during the 1979 and 1980 field experiments. No distinction has
been made between stations reporting on a 24-h basis and those reporting only
during daylight hours. As indicated by the figure, NWS data were obtained not
only for the NECRMP domain but also for the remainder of the contiguous United
States and the adjacent areas of Canada and Mexico. The observations from these
locations consists of a standard set of meteorological elements, some of which
are determined with calibrated instruments and others that are subjectively
determined by trained observers. The measured elements include temperature, dew
point, station pressure, wind speed and direction, and the amount of
precipitation. Elements that are subjectively determined are the sky cover,
cloud type, and prevailing visibility. Some elements, such as the height of
clouds, are either measured with instruments or are subjectively determined,
398
-------
o
o
,o oe A
,0-06 .ti
00'00l
o
•H
4J
ro
U
a)
M
JD
O
x^
(0
>»
f8
U
•H
(0
0)
U
fl
3
(A
X
a
o
60
c
e
OT
C
(fl
O
•H
4-1
(0
4-1
(A
OJ
X
4-1
TO
0)
(rt
O
(0
U
3
in
a>
M
3
00
399
-------
depending upon the value. Specific information on individual elements important
to regional modeling is presented in Table 4 in the order in which these
elements appear in the North American hourly airways sequence for data
transmittal. All of the instruments used for meteorological measurements are
maintained to operational calibration tolerances by a routine preventive
maintenance program. The procedures for making, recording, and transmitting
surface observations are described in detail in the Federal Meteorological
Handbook, Number 1 Surface Observations (DOC), which is the standard reference
for the United States. Other nations have similar documents.
NWS Upper Air Data —
In situ air data were taken by balloon-borne instruments packages twice
daily at 0000 GMT (1700 EST) and 1200 GMT (0700 EST) on a routine basis by the
rawinsonde stations shown in Figure 6. Additional soundings were conducted at
0600 GMT (0100 EST) and 1800 GMT (1300 EST) from July 15 through August 15,
1979, and from July 1 through August 31, 1980, at the stations within the NECRMP
domain, except New York City. The following elements were reported at standard
pressure surfaces: geopotential height (meters), temperature, dew point
depression (degrees Celsius), wind direction (nearest five degrees of arc) and
wind speed (to the nearest knot). Significant levels of temperature, dew point
depression, and wind speed and direction are also reported, when required, to
define the shape of the temperature/dew point profile. Details of operational
procedures and quality assurance checks for the radiosonde and upper wind data
are available in the Federal Meteorological Handbook, Number 3 (DOC).
400
-------
TABLE 4. SURFACE METEOROLOGICAL DATA REPORTED BY NWS
Element
Comments
Cloud height
Sky cover
Prevailing visibility
Observed atmospheric phenomena
Sea level pressure
Temperature/dew point temperature
Wind direction/wind speed
Precipitation
Measured by a ceilometer for altitudes up
to approximately 5,000 ft; estimated for
higher altitudes; units in feet above
ground level.
Estimated using a rule of summation, which
states that clouds at a higher altitude
may not be reported as covering a smaller
fraction of the sky than those at a lower
altitude; reporting basis is tenths or
fractions thereof.
The greatest visibility that is equaled or
exceeded throughout at least half of the
horizon circle surrounding the observer;
units in miles or fractions thereof.
Hydrometeors (rain, snow, fog, drizzle,
etc.) and lithometeors (haze, smoke, dust,
etc.).
Air pressure measured by a barometer at
the station analytically modified to be
the pressure that would be observed if the
station were at sea level; units in
millibars.
Instantaneous measurements at time of
observation from instrument in shelter or
shielded housing; units in degrees
Fahrenheit.
Reported to the nearest 10 degrees of arc;
units in knots.
Amount of hundredths of an inch reported
at 3-h intervals.
401
-------
o
o
in
o
o/--"
,0 08 *
a°
0 06 .H
in
jo
o
§
w
•o
c
•H
3
CO
M
Q,
Q.
3
60
C
E
(A
nj
en
o
m
4-1
(A
4-1
a
s
in
c
o
o
o
vO
3
oo
402
-------
NECRMP Supplemental Upper Air Data —
An intensive program of upper air meteorological measurements was conducted
during 1980 to supplement existing NWS measurements. Basically, three types of
measurement procedures were used: (1) rawinsonde soundings for temperature/dew
point temperature and winds aloft; (2) pilot balloon observations for winds
aloft; and (3) sodars for low-level mixing heights. The rawinsonde soundings
were conducted at seven locations, sodars were operated at six locations, and
pibal observations were taken at seven locations, as shown in Figure 7.
Measurements were made at these sites from Juy 15 through September 12, 1980,
except for all the pibals; the rawinsonde soundings were made at State College,
Pennsylvania; and the sodars, which were terminated on August 15, were operated
near Columbus.
The rawinsonde soundings at State College were conducted daily at 0100,
0700, 1300, and 1900 EST. Soundings at the other six locations were obtained
5 days per week, typically Monday through Friday, although the schedule was
adjusted to obtain soundings on weekend days when aircraft flights were planned.
The daily schedules of soundings at these sites are listed in Table 5.
Slow-rise balloons were used for the soundings, and an average ascent rate of
400 ft/min was achieved on most soundings. Data were recorded between ground
level and 700 mbar (approximately 3,500 m) only.
Temperature/dew point temperature and height values for mandatory and
significant levels were determined by using standard NWS data reduction
procedures, except that ±1/2°C rather than ±1°C was used for selecting
403
-------
0)
3
60
•H
404
-------
TABLE 5. SCHEDULE OF NECRMP SUPPLEMENTAL RAWINSONDE
SOUNDINGS
Sounding Site
Start/End Date
Launch
Time
(EST)
Washington, DC (Site 16) July 24-September 12
0500
0900
1300
Baltimore (Site 11)
July 16-September 12
0700
0900
1000a
1200a
1300
1400"
1500a
1600"
Marlboro (Site 15)
July 17-August 29
0500
0700
0900
1300
Newark (Site 14)
July 19-September 12
0500
0900
1100b
1300°
Derby (Site 13)
July 16-September 12
0700
0900
1100
1300
1500
Boston (Site 12)
July 18-September 12
0500
0700d
0900
1100d
1300
1500d
"Soundings on July 16 and 31 and August 1, 5, 6, 7, and 14
only.
"Soundings effective September 3.
cSounding discontinued September 2.
dWinds only—no temperature measurements.
405
-------
significant temperature points. The balloons were tracked by using LORAN-C
navigational aids, and wind speed and wind direction were determined at 30-s
intervals for the entire sounding. Standard NWS radiosondes were used for the
temperature and humidity measurements.
Pilot balloon observations of wind speed and wind direction aloft were made
almost daily during the period July 14 through August 15, 1980. Thirty-gram
balloons were released hourly from 0400 EST through 1700 EST and tracked with a
single theodolite. Readings of azimuth and elevation were recorded every 30 s
for a total of 20 min. A constant-rise rate was assumed for reducing the raw
data into wind speed and wind direction values. The reduced wind data were
recorded at 110 m above ground level and continued at approximately 90-m
intervals up to 3,690 m above ground level.
The AeroVironment model 300 monostatic acoustic radar (sodar) was used to
provide a near continuous record of the thermal structure of the lower
atmosphere from 30 m to 1,000 .m AGL (the respective lower upper detectable
limits of the instrument). Data from the sodar were reduced into 30-min average
mixing height values determined as the height of the base of the lowest stable
layer detected by the instrument.
Remotely Observed Data —
Satellite data were obtained during 1980 for the NECRMP domain from the
National Oceanic and Atmospheric Administration. Cloud cover and cloud top
406
-------
heights have been produced by using the method of Reynolds and Vonder Haar
(1977) for selected cases.
DATA BASE AVAILABILITY
The repository for the NECRMP data base will be the Environmental Sciences
Research Laboratory, Meteorology Division, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina. At present, the various data base
components, including those described in the paper, remain as separate entities.
The data are currently undergoing various quality control checks and will be
merged into a single chronologically ordered data base on magnetic tape.
Additional information on the NECRMP data base can be obtained from Ms. Joan H.
Novak, Mail Drop 80, U.S. Environmental Protection Agency, Research Triangle
Park, North Carolina 27711.
REFERENCES
Arey, F. K., R. C. Shores, and R. W. Murdoch. 1980. Performance Audits of the
NEROS/PEPE Sites (Revised Report). Research Triangle Institute for EPA
Contract No. 68-02-3222, Technical Directive No. 98. pp 71.
Lamb, R. G. 1983. A Regional Scale (1000 km) Model of Photochemical Air
Pollution. Part 1, Theoretical Formulation. To be published by the U.S.
Environmental Protection Agency, Research Triangle Park, North Carolina.
Murdoch, R. W., and C. N. Dimmick. 1979a. First Audit of the Northeast
Regional Oxidant Study. Research Triangle Institute for EPA Contract No.
68-02-3222, Technical Directive No. 15.
Murdoch, R. W., and C. N. Dimmick. 1979b. Second Audit of the Northeast
Regional Oxidant Study. Research Triangle Institute for EPA Contract No.
68-02-3222, Technical Directive No. 15.
407
-------
Possiel, N. C., T. K. Clarke, J. L. Clark, J. K. Ching, and E. L. Martinez.
1982. Recent EPA Urban and Regional Scale Field Programs in the
Northeastern U.S. In: Proceedings of the 75th Annual Meeting of the Air
Pollution Control Association, New Orleans, Louisiana.
Reynolds, D. W., and T. H. Vender Haar. 1977. A bispcctral method for cloud
parameter determination. Monthly Weather Review, 105:446-457.
Summers, J. G. 1983. Availability of Data in SAROAD for Ozone and Its
Precursors. In: Proceedings of the International Conference on Long-Range
Transport Models for Photochemical Oxidants and Their Precursors,
Organization of Economic Cooperation and Development, April 12-14, 1983,
Research Triangle Park, North Carolina.
U.S. Department of Commerce. Federal Meteorological Handbook No. 1, Surface
Observations. Washington, D.C.
U.S. Department of Commerce. Federal Meteorological Handbook No. 3, Radiosonde
Observations. Washington, D.C.
DISCUSSION
P. Misra: In your aircraft measurements, do you have some kind of averaging
procedure or are these instantaneous measurements?
N. Possiel: The aircraft measurements were made instantaneously and then
averaged in 10-s to 25-s intervals, depending upon the individual aircraft.
P. Misra: It works out to some kind of an aerial average?
N. Possiel: It turns out to be some kind of an aerial average. The type of
aircraft we used generally flew at about 130 knots, 120 to 130 knots, which is
pretty standard.
S. Reynolds: You were out for about 1 mo in each of 2 yr. Were any interesting
regional 03 episodes observed while you were in the field?
N. Possiel; Yes. As a matter of fact, I showed you the regional sampling
scenario from NEROS 1979, which had some very high 03 concentrations transported
across the region. During 1980, we were very fortunate in that 03 was
relatively high in that particular year, and we did obtain some very successful
emissions in terms of seeing urban plume and regional transport.
A. Christie; You mentioned a meteorological data base. You have quite an
extensive data base. When you are inputting that to the model, how is it going
to be put in?
408
-------
N. Possiel: I am not sure of the exact scheme. Perhaps Joan Novak can answer
that.
J. Novak: As part of the modeling system, there is a series ot preprocessors,
which will take the data as it is received from the station. Then it will be
transformed through the different meteorological preprocessors and actually put
into the model in a series of matrix coefficients for submission and
differential solving.
We do not just input the raw data as taken off the aircraft. We take each of
the data sets independently and reformat it into the standard format for that
type of data. Then we put it through extensive quality assurance/quality
control procedures to validate the data, do extensive graphics to look at
consistencies, or look for inconsistencies in a generated field. From that point
we sort them, put them in chronological order, so they will be available to the
model.
H. van Pop: In regard to this issue, I think that it is fruitful to keep in
mind that it will be a difficult point to determine where the aerometric data
ends and where the modeling starts. There is an interface between input data
and a model, and that point of interface can be established more or less
arbitrarily. Your preprocessing, does it belong to the model or does it belong
to'the input data? That is a difficult thing to distinguish.
409
-------
CANADIAN SURFACE AIR QUALITY
MONITORING NETWORKS*
T. Dann
Environmental Protection Service
River Road Environmental Centre
River Road, Ottawa, Ontario (Canada)
D. Balsillie
Ontario Ministry of the Environment
880 Bay Street
Toronto, Ontario (Canada)
INTRODUCTION
The majority of continuous surface air quality measurements in Canada are
carried out in urban-oriented air quality monitoring networks, of which the
National Air Pollution Surveillance (NAPS) network is the largest. Some
regional and background sites have been established to obtain daily measurements
of S02, HNOs, particulate nitrates, and trace elements by using filter pack
techniques. Networks incorporating these measurements include CAPMON
(Atmospheric Environment Service, Environment Canada) and APIOS (Ontario
Ministry of the Environment). These will be discussed elsewhere. Regionally,
representative monitoring sites for 03 have been established only in
southwestern Ontario and south-central Nova Scotia.
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
410
-------
NAPS Network
The NAPS network includes most of the air quality monitoring stations
operating in Canada. Environment Canada coordinates the operation of the
network and provides most of the monitoring equipment used. Actual day-to-day
operation, calibration, and equipment maintenance are handled by municipal and
provincial air pollution agencies.
The network was established in 1970 and was expanded until 1978, when it
reached its present size. NAPS now consists of 500 instruments located in 52
cities.
Parameters Monitored and Site Locations—
Table 1 lists the stations, station addresses, and parameters measured in
the NAPS network. (Geographical coverage is shown in Figure 1). Entries in the
table indicate equipment ownership (F-Federal, P-Provincial) and sampling height
(in meters) above the ground. The letter following the five-digit station code
indicates land use at the sampling site (R-rural or residential, C-city core or
commercial, I-industrial). A description of the monitoring equipment used,
units of reported data, and sampling frequency are given in Table 2. Most of
the sites listed have been in operation since 1976.
411
-------
TABLE 1. LIST OF STATIONS IN NAPS NETWORK, 1981
Station8
10101C
20101C
30101C
30102R
30114R
301151
30116C
30301R
303091
30310C
30311R
30401R
30405C
30408R
40102C
40201C
40202C
40301C
Location
St. John's, Newfoundland
Duckworth & Ordinance
Charlottetown, Prince
Edward Island
56 Fitzroy Street
Halifax, Nova Scotia
Nova Scotia Technical
College
Dalhousie University
Mt. St. Vincent University
CFB Shearwater
Barrington & Duke
Sydney, Nova Scotia
Murphy Road
Pt. Edward, Richmond Plst
County Jail
Whitney Pier Fire Station
Glace Bay, Nova Scotia
Lake Road
General Hospital
South Street
Fredericton, New Brunswick
York Street
Saint John, New Brunswick
110 Charlotte Street
Post Office
Moncton, New Brunswick
774 Main Street
Parameters Measured*1
Partic-
ulates
S02 CO N02 03 COH and Pb
F,09 F,09 F,09
F,10 F,09
F,12
F.18
F,10 F,10 F,10
F,08 F,08 F.08 F,08 F,ll
F,02 F,02 F,02 F,02 F,02
F,04 F,04 F,02
F,05 F,05 F,02
F,03 F,03 F,02
F,07
F,18
F,09 F,09 F,09 F,09
F,22
Dust-
fall
and
S03
04
04
04
04
04
04
(continued)
412
-------
TABLE 1. (continued)
Parameters Measured1"
Station*
50101R
50102R
50103R
50104C
50105C
50106R
50107R
50108R
50109C
50110C
50111C
50112C
50113R
50114C
50115C
50116R
50203R
50301C
50302R
503031
50304C
50306R
50402R
50403C
50502R
50503C
50601C
Location
Montreal, Quebec
Pare Jarry
Jardin Botanique
Pointe-Aux-Trembles
1125 Ontario Est
1212 Drummond
Ville St-Laurent
Ville Lasalle
1700 Bourassa, Longueuil
Duncan & Decarie
Pare Pilon, Mtl-Nord
2900 Boul. Concorde
Boul. Laurentides
Pie X & Cardinal
677 Ste-Catherine 0
Metcalfe & Maisonneuve
3161 Joseph, Verdun
Hull, Quebec
Gamelin & Joffre
Quebec, Quebec
Parc-Autos Paq.Laliberte
Pare Bardy
Centre Loisirs Limoilou
325 Dorchester Sud
2026 Blvd. St-Cyrille
Sherbrooke, Quebec
Casserne De Pompiers #5
Wellington & Albert
Chicoutimi, Quebec
Usine De Filtration
222 Racine
Rouyn, Quebec
Hotel de Ville
Partic-
ulates
S02 CO N02 03 COH and Pb
F,03 F,03
F,03 F,03 F.03
F,05 F,05 F,05 F,05
F,14 F,14 F,14 F,14 F,U
M,13 F,13
F,03 F,03
P,05 F,05 F,05 F,05 F,05
P,05 P,05 F,05 F,05 F,05
F,05 F,05 F,05 F,05
F,03 F,03 F,03 F,03
F.05 F,05 F,05 F,05
M,07
F,05 F,05 F,05 F.05 F,05
F,16 F,16 F,16 F.16 F.16
F,05 F,05 F,05 F,05
F,15 P,15
F,05 F,05 F,05 F,05
F,05 F,05
F,08
F,03
F,05
M,U
F.ll
F,03
F,04
F,14
P,03
P,05
P,05
P,03
P,05
F.05
F,25
F,05
F,15
F,05
F,12
F.12
F,05
F.09
F,17
F,06
F,06
P,08
Dust-
fall
and
S03
05
03
04
03
04
(continued)
413
-------
TABLE 1. (continued)
Station8
50701C
50801R
50901R
51001R
51002R
51101C
512011
51301R
60101C
60103C
60104C
60105R
602011
60202C
60203R
60204C
60211R
60301R
60302R
Location
Sept-lies, Quebec
Hotel de Ville
Trois-Rivieres, Quebec
Hart & Ste-Cecile
Arvida, Quebec
Powell & Hoopes
Tracy, Quebec
225 Ave Courshesne
Garneau & Rte 132
Thetford Mines, Quebec
Boul, Ste-Marthe
Shawinigan, Quebec
Frigon & Laval
Baie Comeau, Quebec
39 Ave. Marguette
Ottawa, Ontario
88 Slater Street
Gilmour Street
Rideau & Wurtemburg
NRG, Montreal Road
Windsor, Ontario
Morton Terminal Dock
City Hall
Tecumseh Water Works
471 University Avenue
College & Prince
Kingston, Ontario
Queen's University
Napier Street
Parameters Measured11
Partic-
ulates
S02 CO N02 03 COH and Pb
F,08 P,08
F,05 P,04 P,06
F,05 P,06
F,08
F,04
F,05
F,05 F,05 F,05
F,12 F,12
F,17 F,05 F,05 F,13 F,13 F,16
F.12
P,04 F,04 P,04 P,04 P,04 P,04
F.05 F.05 F,05 F,05
P,04 F,04
P,15
P,03
P,12 P,12 F,12 F,12 P,12 P,12
P,04 P.04
F,15
F,06 F,06
(continued)
414
Dust-
fall
and
S03
-------
TABLE 1. (continued)
Parameters Measured*"
Station8
60401C
60402R
604031
60404R
604051
60406R
60408C
60409R
60410R
60412R
60413R
604141
60415R
60416C
60417C
60501C
605021
605031
60505R
60507C
605101
60602R
60605C
60606C
60702R
607031
60704C
60705R
60801C
60806R
Location
Toronto, Ontario
67 College Street
Don Mills, Science Cntr.
Evans & Arnold
5126 Yonge Street
John Street Pump Station
Rosehill Reservoir
Danforth & Robinson
Redlands Crescent
Lawrence & Kennedy
Bathurst & Wilson
Elmcrest Road
Sherbourne & Wilton St.
Queensway W & Hurontario
381 Yonge Street
26 Breadalbane
Hamilton, Ontario
Barton & Sanford
Burlington & Gage
Chatham & Frid
North Park
Hughson & Hunter
Strathearn
Sudbury, Ontario
Ash Street
19 Lisgar Street
Kennedy Street
Sault Ste. Marie, Ontario
Anna Mcrea Public School
Bayview & Young
Queen & Elgin
550 Queen Street West
Thunder Bay, Ontario
14 Algoma Street
435 James S.
S02
P,16
P.09
P,04
P.03
P,05
P.06
P,04
F,05
F.04
P.15
P,04
P,04
P,04
P,05
F,06
F,17
CO N02 03 COH
F,04 P,19 F,19 P.19
P,09 F,09 P,09 P.09
F,04 P,04 P,04 P,04
P.03
F,05 P,05 P,05 P,05
F,06 F,06 P,06
F,04 P,04 F,04 P,04
F,05 F.05 F,05 F,05
F,04 F,04 F,04 F,04
P,04
F,15 P.15 F.15 P.15
F,04 F,04 F,04 P,04
F,04 P,04
F,04 F,04 F,04 P,04
F,17 F,17
Partic-
ulates
and Pb
P,17
P,09
P,01
P,01
P,04
P,01
P,04
P,05
F,04
P.13
P,04
P,06
P,04
P,04
P.ll
F,05
P,12
F.17
Dust-
fall
and
S03
19
01
05
05
04
05
04
05
06
10
05
05
06
12
(continued)
415
-------
TABLE 1. (continued)
Parameters Measured1"
Station8
Location
S02
CO
N02
03
COH
Partic-
ulates
and Pb
Dust-
fall
and
S03
London, Ontario
60901C King & Rectory
60902C 372 Dundas
Sarnia, Ontario
61004R Front St. at C.N.Tracks
Petersborough, Ontario
61103C 500 George Street
Cornwall, Ontario
61201R Memorial Park
St. Catharines, Ontario
61301C North St. & Geneva St.
Kitchener, Ontario
61501C Edna and Frederick
Oakville, Ontario
61602R Bronte and Woburn Ores.
Oshawa, Ontario
61701R Ritson Rd. & Olive Ave.
Guelph, Ontario
P,04 F,04 P,04 P,04 P,04 P.04 04
P,17 P,17
P,03 F,03 P,03 P,03 P,03 P,03
P.18
P,18 P,18
P,04 F,04 F,05 P,04 P,05 P,04
F,06 F,05 F,06 F,06 F,06 F,06
F,05 F,05 F,05 F,05 F,05 F,05
P,05 P,05 P,05 P,05 P,05 P,05
F,05 F,05 F,05 F,05 F,05 F,05
61801C
70102R
701041
70105R
70110C
701131
70115C
701161
70118R
70119C
70120R
Farquhar & Wyndham
Winnipeg, Manitoba
Portage & Woodlawn
Union Stock Yards
Martin & Henderson Hwy
Kennedy & St. Mary's
Windermere & Rockman
Portage & Minto
Smith & King
Jefferson & Scotia
65 Ellen Street
604 St. Mary's Rd.
F,04 F,04 F,04
F,07
F,17
F,04 F,04 F,04 F.04 F,04 F,05
F,03 F,03 F,03 F,03 F,03 F,03
F,04 F,04 F,04 F,04
05
04
03
06
03
(continued)
416
-------
TABLE 1. (continued)
Station3
70201C
80102R
80108C
80109C
80203R
80209C
80301C
80401C
901211
90122R
90125C
90126R
901271
90128R
90130C
90204C
902181
90219C
90221R
90222R
90223C
902241
90225R
902261
90227C
Location S02 CO
Brandon, Manitoba
llth St. & Princess Ave.
Regina, Saskatchewan
3211 Albert Street
12th Ave. & Smith St.
1620 Albert Street F,14 F,14
Saskatoon, Saskatchewan
30th St. & 833 P Ave.
Idylwyld Dr. & 33rd St. F,ll F,ll
Moose Jaw, Saskatchewan
Fairford St. & 1st Ave. F,15
Prince Albert,
Saskatchewan
1257-lst Ave. East F,07
Edmonton, Alberta
17 Street & 105 Avenue F,05 P,04
127 St. & 133 Avenue F,04 F,04
Prins Elizabeth & 108 St.
77 Avenue & 85 Street
115 Avenue & 159 Street
99 Avenue & 160 Street
10255 - 104th Street F,09 F,09
Calgary, Alberta
316-7th Avenue
Bonny Brk & 18A St. S.E. F,04
620-7th Avenue, S.W.
Dalhousie Dr & Dalham Dr
39 St. & 29 Ave. N.W. F,04 F,04
11 St. & 38 Ave. S.E.
Ogdendale & 71 Ave. S.E.
Palliser Dr. & Oakwood S.W.
Sheppard & 84 Ave. S.E.
1611-4th Street, S.W. F,06 F,06
Parameters Measured1"
Partic-
ulates
N02 03 COH and Pb
F.12
F,06
F,12
F,14 F,14 F.14 F,12
F,14
F,ll F,ll
F,12
F,09
F.04
F,04 F,04 F,04 F,04
F,09 F,09 P,09 F.06
F,09
P,04 F,04
F,04 F,04 F,04 F,04
F,06 F,06 P,06
Dust-
fall
and
S03
02
04
02
03
09
02
02
02
04
04
(continued)
417
-------
TABLE 1. (continued)
Station8
90301C
90501C
99001C
00102R
00104R
00106R
001081
00109C
00110R
001111
00112C
001131
00114C
00115R
00116R
00117R
00202C
00302C
00401C
Location S02
Red Deer, Alberta
4747 50th Street
Lethbridge, Alberta
13 St. & 9 Avenue S.
Yellowknife, Northwest
Territories
50th Ave. & 49th Street
Vancouver, British Columbia
100 Richmond Street
27th & Ontario
2294 West 10th Avenue F,03
250 West 70th Avenue
970 Burrard
E. Hastings & Kensington F,05
Rocky Pt. Park F,04
Robson/Hornby F,05
Annacis Island, Delta
Municipal Hall, Richmond
Newton Elem. Sch., Surrey
Fire Hall, N. Vancouver
Beit Burnaby
Prince George, British
Columbia
1011 4th Avenue
Victoria, British
Columbia
1106 Cook St. F,ll
Kamloops, British
Columbia
301 Seyumour St. F,15
Parameters Measured6
Partic-
ulates
CO N02 03 COH and Pb
F,08
F,15
F,07
F,05 F,05
F,18
F,03 M,03 F,03 F,03 F,17
F,05 F,05 F,05 F,05 F,05
F,06
F,05 F,05 F,05 F,05 F,04
F,04 F,04 F,04 F,04 F,04
F,05 F,05 F,05 F,05
F,04
F,14
F.08
F,06
F,12
F,18 F,18
F.ll F,ll F.ll F.ll F,ll
F,ll
Dust-
fall
and
S03
(continued)
418
-------
TABLE 1. (continued)
Station3
Location
Parameters Measured6
Dust-
Partic- fall
ulates and
S02 CO N02 03 COH and Pb S03
Whitehorse, Yukon
09001C Federal Building
F,04 F,08
"Letter following station code indicates land use, i.e., R = residential or rural,
C = city core or commercial, I = industrial.
bEntries indicate equipment ownership (F = Federal, P = provincial) and sampling
height above the ground (in meters).
419
-------
Ul
in
o
SV
i .• i
,./r
*
/
/
z*
O o,0
0 i ??-§../
CO
,0
E
0)
u
o
4-t
0>
z
a;
u
c
03
O)
c
o
3
i—I
.—I
O
a)
c
o
a)
2
(1)
U
3
£.=,
420
-------
TABLE 2. DESCRIPTION OF NAPS NETWORK INSTRUMENTATION
Pollutant
SO,
CO
NOj
03
Suspended
Particulates
(Soiling index)
TSP
Pb (Particulate)
Dustfall
Sulphation Rate
Detection
Principle
Coulometry
Ultraviolet
fluorescence
Nondispersive
infrared
spectrometry
Chemiluminescence
Chemiluminescence
Ultraviolet
photometry
Photometry
Gravinetry
Atomic absorption
spectroscopy
X-ray fluorescence
Deposition by
gravity
Reaction with lead
dioxide
Concentration Tvpe of
Unit of Measurement* Reported Monitoring
pphm 1 pphm Continuous
ppro 0.5 ppra Continuous
ppha 1 pphra Continuous
pphm 0.1 pphm Continuous
COH' 0.1 COH Intermittent
(12 2-h or
or 24 1-h)
samples
daily)
Mg/m3 1 jjg/m3 Intermittent
(24-h sample
everv 6th
day)
>ig/m3 0.1 jig/m3 Intermittent
(24-h sample
every 6th
day)
g/m2/30 days 0.1/g/m2/ Intermittent
30 dav (12 30-d
samples per
year)
0g SOj/100 cm'Vday 0.1 mg S03/ Intermittent
100 cm2/dav (12 30-d
samples per
vear )
'Coefficient of haze.
421
-------
Site Descriptions—
Comprehensive site documentation is available for most sites in the
network. Documentation includes geographic and street location, UTM
coordinates, latitude and longitude, time zone, length of record, instrument
details, scale of representativeness, land use by sector (within a 2-km radius
of station), site elevation, airflow restrictions, manifold type, source
influences on station (local, roadway, and major point sources), topographic
map, and site photographs.
Data Quality—
A comprehensive quality assurance program has been established for the NAPS
network. Monitoring and calibration equipment has been standardized, as have
operational, zero/span, and multipoint calibration techniques. The Federal
Government conducts an audit program for all continuous monitors reporting data
to the network; data are rejeqted if the difference between monitor response and
audit concentration is greater than ±15%. Compressed gas cylinders for
zero/span calibrations are analyzed by a central Federal laboratory and
distributed to all operating agencies. Validated hourly data submitted by the
provincial and municipal cooperating agencies are subjected to a number of
data-screening routines before incorporation in the data bank.
All archived data can be retrieved for any time period and distributed in
magnetic tape format. Data are normally archived within 6 mo of collection.
422
-------
ONTARIO MINISTRY OF THE ENVIRONMENT MONITORING NETWORK
The majority of monitoring instruments in the Ontario network have been
incorporated into the NAPS network; however, supplementary monitoring data for
NO, THC, and NMHC are available from the province. Additionally, a number of
rural sites for Oa have been established in the southwest portion of the
province, as shown in Table 3 and Figure 2.
The province compiles site documentation information similar to that of the
NAPS network. All monitoring is carried out under a very comprehensive quality
assurance program incorporating all the elements of the NAPS program. Data are
available in magnetic tape format for all sites, beginning with 1976 data.
ADDITIONAL OZONE MONITORING SITES
Supplementary 03 monitoring data are available from a site in south-central
Nova Scotia (Kejimkujic National Park), beginning with 1982 data, and for two
sites within 30 km of Montreal (Tracy and Beauharnois).
423
-------
TABLE 3. ONTARIO MINISTRY OF THE ENVIRONMENT—SUPPLEMENTARY MONITORING STATION
Region
Station
Number
Station Address
Air
Intake Pollutant
UTM Grid Above Monitored
Ground
East North (m) 03 CO HC N0y
Southwest 10001 College of Agriculture
Tech., Huron Park W.
West
Central
12008
13021
14064s
a 467 University Ave. W.
Windsor
MOE Pump Station
Middle Road
Merlin
Centennial Park
Front St/CN Tracks
Sarnia
14118 PUC Water Pump Station
Highway 21
Petrolia
14903 Virgil LaSalle
Froomfield
Corrunna
14904 East Sombra P.S.
Wilkesport
15001" King-Rectory
London
18007 Concession Road 2
Lot A
Tiverton
22071 Experimental Farm
Simcoe
22086 Cheapside Road
3 km S of Highway 3
Nanticoke
04600 47931 5 x
03316 46867 11 x x x x
03991 46776 3 x
03854 47592 3 x x x x
04027 47564 4 x
03814 47520 3
03892 47285 3 x
x x
04818 47595 4 x x x x
04541 49053 4 x
05597 47449 4 x
05821 47472 5
(continued)
424
-------
Region
TABLE 3. (continued)
Station
Number
Station Address
Air
Intake Pollutant
UTM Grid Above Monitored
Ground
East North (m) 03 CO HC NOX
26029"
270378
29008"
29025s
Central 31001"
31086
31104"
31105s
31120
33003"
34002s
34007"
Edna/Frederick Street
Kitchener
North/Geneva Street
St. Catharines
North Park
Hamilton
Barton-Wentworth
Hamilton
67 College Street
5th Floor
Toronto
15 Breadalbane
Toronto
26 Breadalbone
Toronto API
Sherbourne/Wilton
Toronto
Junction Triangle
Perth Avenue
Toronto
Lawrence-Kennedy
Scarborough
Science Centre
Don Mills Road
North York
Bathurst-Wilson
North York
05427 48116 5 x x
06431 47805 5 x x x x
05984 47927 3 x x
05939 47900 4 x x
06300 48352 20 x x x x
06302 48355 4 x x x x
06302 48356 15 x x x x
06317 48338 5 x x
06248 48344 9 x x
06389 48452 3 x x x x
06338 48419 9 x x x x
06261 48437
XXX
(continued)
425
-------
TABLE 3. (continued)
Air
Intake Pollutant
UTM Grid Above Monitored
Station Ground
Region Number Station Address East North (m) 03 CO HC NO,,
35003" Elmcrest Road
Etobicoke 06142 48338 4 x x x x
35033" Evans-Arnold
Etobicoke 06192 48302 3 x x x x
44008 Highway 2/North Shore
Boulevard East
Burlington 05972 47964 17 x x
44015" Bronte Road/Woburn Cres.
Oakville 06031 48059 5 x x x x
45025" Ritson Road/Olive Avenue
Oshawa 06724 48624 4 x x x
46110" Queensway West/
Hurontario Street
Mississauga 06122 48249 5 x x x
47035 M. of A.
509 Victoria Street,
East
48002
49010
Southeast 51001"
56051"
Alliston
MTC Yard, Highway 47
Stouffville
Hwy 11 7 /Paint Lake Road
Dorset
McDonald Gardens
Ottawa
Memorial Park
Cornwall
05918
06391
06624
04471
05208
48898
48694
50096
50312
49846
4 x
4 x
3 x
4 x x x
4 x x x
(continued)
426
-------
TABLE 3. (continued)
Region
Northwest
Northeast
North
NEMP
Station
Number
63022"
71049s
71057
77016°
22901
22902
22903
22904
22905
Station Address
Hospital
35 Algoma Street North
Thunder Bay
Land Regulation Office
Queen Street
Sault-Saint Marie
Michipicoten Avenue
Lot 38
Town of Mission
Ash Street
Sudbury
100 M.S. Highway 59
Long Point Prov. Park
Tyneside/Chippewa Road
Binbrook West
Cheapside Road/Walpole
Cone . 5
Nanticoke
Walpole South PS
Sandusk Road
Nanticoke
Nanticoke Road/
Walpole Cone. 5
Nanticoke
Air
Intake Pollutant
UTM Grid Above Monitored
Ground
East North (m) 03 CO HC NOX
03356 53672 22 x
07047 51543 6 x
06625 53110 4 x
04994 51486 3 x x x
05502 47138 4 x x x
05914 47746 5 x x x
05821 47472 4 x
05794 47434 4 x
05744 47456 4 x
"Part of NAPS network.
427
-------
oo
o
cu
u
o
c/1
CD
4-1
00
c
•H
l-l
O
o
e
o
ro
o
o
0)
V-i
OO
•H
428
-------
AEROMETRIC DATA BASES IN THE NETHERLANDS*
R.M. van Aalst
Division of Technology
TNO Society
P.O. Box 217
2600 AE Delft, The Netherlands
INTRODUCTION
Major air quality data bases in The Netherlands are available from the
National Air Quality Monitoring Network and from the FLAT network. The former
network, which is operated by The Netherlands Institute of Pulic Health, RIV (RIV,
1982; Van Egmond en Onderdelinden, 1981), comprises a number of fully automated,
fixed monitoring stations that provide hourly averaged concentrations, 24 h a day,
for NO and N02 (92 stations), 03 (30 stations), CO (41 stations), and S02
(220 stations). The S02 and CO concentrations are measured by coulometry; the NO,
N02, and 03 concentrations are measured by chemiluminescence.
These measurements are made at a height of 3.8 m; at one station,
measurements are made at 100 m and 200 m. Figures 1 and 2 show the stations for
NOX and O3. The network for O3, which has been in operation since 1978, comprises
30 stations that can be characterized as rural or semirural. For the other
components, urban, suburban, industrial, and rural stations are available.
*This report has not been reviewed by the U.S. Environmental Protection Agency and
therefore does not necessarily reflect the views of the Agency, and no official
endorsement should be inferred.
429
-------
Figure 1. National Air Quality Monitoring Network: Measuring stations for NO*
430
-------
Figure 2. National Air Quality Monitoring Network: Measuring stations for 03.
431
-------
These fixed stations are supplemented with mobile vans that measure not only
concentrations of the components mentioned above but also horizontal fluxes of N02
and S02 inflow by Barringer correlation spectrometry. Incidental airplane
measurements are also carried out.
Future development of this network is aimed on the one hand at reducing its
density and on the other hand at improving its flexibility with regard to the
choice of component and sampling strategy. There is a need for data on HCs,
sulphates, nitrates, and other components during air pollution episodes or under
selected meteorological conditions. Remote operation of the sampling stations is
under development.
A second major network, operated by TNO is called FLAT. This is the Dutch
acronym for the project on Photochemical Air Pollution, Aerosols, and Toxirity,
carried out by TNO from 1979 to 1982. The network was etablished for the specific
purpose of evaluating mathematical air pollution models (Van Aalst and Guiherit,
1980). Photochemical precursory and products, aerosol components, and physical
aerosol parameters were measured at six stations from 1979 to 1981. These are
shown in Figure 3 and listed in Table 1 (Diederen et al., 1981). Table 2
summarizes the parameters measured, the temporal resolution, the sampling
frequency, and the monitoring period.
In addition to these two major data bases, 03 (half-hour) and PAN
(quarter-hour) concentrations have been measured by TNO at Delft, the former since
1971 and the latter since 1973 (TNO, 1978; Heidema et al., 1981). All of these
data are suitable for model evaluation. Standard quality assurance
432
-------
TVschiUi'ng
cz*
Figure 3. FLAT monitoring stations (underlined) in The Netherlands
(see Table 1).
433
-------
TABLE 1. FLAT MONITORING STATIONS8
Station
Type
Delft
Terschelling
Eindhoven
Vlaardingen
Hellevoetsluis
Ypenburg
Suburan
Rural
Urban
Suburban/industrial
Rural
Suburban
'See also Figure 3.
434
-------
TABLE 2. PARAMETERS MEASURED AT FLAT STATIONS
Parameter
In aerosols
S04=
N03~
ci-
NH4+
Na+
H2S04
TSP
Carbon
Organic matter in aerosols
HN03
F~
NH3
Aldehydes
Formic acid
Acetic acid
Propionic acid
Organic compounds
C6-C16 as gas
C6-C16 in aerosol
Pesticides
Scattering coefficient
Particle size distribution (EAA)
Temporal
resolution
(h)
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
24
1
24
24
1
1
Frequency
daily
daily
daily
daily
daily
daily
once/6 d
once/6 d
once/6 d
daily
daily
daily
daily
once/6 d
once/6 d
once/6 d
once/6 d
twice/day
incidental
once/6 d
once/18 d
hourly
hourly
Location3
1,2,3,4
1,2
2
1,2
3,4
1,2
1,2
3,4
1
1
1
1
1,2,4,5
1
1
1
6
6
Year
of
record
1979/1981
1981
1979/1981
1979/1980
1979/1980
1981
1979/1981
1979/1981
1981
1979/1981
1979/1980
1979/1981
1979/1981
1979/1981
1980/1981
1979
1980/1981
1979/1981
1979/1981
'For location, see Table 1.
435
-------
procedures are implemented for both sampling and analysis, and most of the data
are very reliable. Measurements of H2S04, HN03, and organic matter in aerosols
are less reliable due to possible interference.
Data on surface meteorology in The Netherlands can be obtained from a
synoptic meteorological network of 22 stations (Figure 4) covering The
Netherlands (Personal communication, Van Dop, 1983). This network is operated
by the Royal Netherlands Meteorological Institute, KNMI, and provides hourly
standard synoptic observations. One station is part of an upper air radiosonde
network. Rawinsondes to measure pressure, temperature, humidity, and wind are
released twice a day (at 0000 and 1200 GMT) to a maximum height of 20 km.
Pibals for wind measurements are released daily at 0600 and 1800 GMT.
In The Netherlands (as elsewhere), concern is growing over acid deposition.
Since 1978, KNMI and RIV (1981) have operated a network of 12 stations,
measuring the concentrations in rainwater of several components as well as those
of some metals and organic compounds. These components are H+, NH,,"1', K+, Ca2"1",
Mg2+, Zn2"1", F~, Cl~, NO3", SO4=, HCO3~, and PO4=. The National Institute for
Water Supply, RID, has operated a similar network of 26 stations since 1978
(RID, 1980). Recently, these networks were integrated. However, estimates have
shown that at least half of the acid deposition in The Netherlands is dry
deposition, and there is an urgent need to develop and apply methods for
measuring dry deposition fluxes.
436
-------
St>rtoorotiK>tt praKCtM School t ' HO.OOO
Figure 4. Surface meterological stations.
437
-------
ACKNOWLEDGMENTS
Dr. D. Onderdelinden (RIV), Ir H.S.M.A. Diederen (TNO), Dr. H.F.R.
Reijnders (RIV), and Dr. H. van OOP (KNMl) provided information for this paper.
Presentation of the paper was made possible by financial support from The
Netherlands Ministry of Housing, Physical Planning and the Environment.
REFERENCES
Diederen H.S.M.A., et al. 1981. Niveaus van Luchtverontreiniging Gemeten in de
Periode Januari 1979-Maart 1981, FLAT-project. (Levels of Air Pollution as
Measured in the Period January 1979-March 1981, FLAT project). Report CMP
81/02, TNO, Delft, The Netherlands.
Guicherit, R., editor. 1978. Photochemical Smog Formation in The Netherlands.
TNO, Delft, The Netherlands.
Heidema L. F., et al. 1981. Rapport ter Afsluiting van het Oxidantia Project
(Final Report of the Oxidants Project). IMG-TNO, Delft, The Netherlands.
KNMI/RIV. 1981. Chemical Composition of Precipitation over The Netherlands,
Survey, 1978-1980. KNMI Report 156-3a, KNMl, DC Bilt, The Netherlands.
RID. 1980. RID Regenwater Meetnet, Verslag over de periode Juli 1978-Januari
1980; Verdere Verslagen (RID Rain Water Measurement Network, Report for the
Period July 1978-January 1980; Other Reports). Report cbh 80-13, RID,
Leidschendara, The Netherlands.
RIV. 1982. Nationaal Meetnet voor Luchtverontreiniging, Verslagen, 1978-1982
(National Air Quality Monitoring Network, Progress Reports for 1978-1982.
RIV, Bilthoven, The Netherlands.
Van Aalst, R. M., and R. Guicherit. 1980. Het Project FLAT in het Kader van
het Onderzoek van Luchtverontreiniging bij TNO en de Samenhang Tussen de
Deelprojecten (The FLAT Project, Its Place in TNO's Air Pollution Research
Program and the Interrelations Between Its Subprojects). Report CMP
80/18, TNO, Delft, The Netherlands.
Van Egmond, N. D., and D. Onderdelinden. 1981. Objective analysis of air
pollution monitoring network data; spatial interpolation and network
density. Atmospheric Environment, 15:1035.
438
-------
PHOTOCHEMICAL OXIDANTS IN
NORTHWESTERN EUROPE, 1976-1979
A PILOT STUDY*
Jorgen Schjoldager and Harald Dovland
Norwegian Institute for Air Research
P. 0. Box 130
N-2001 Lillestrom, Norway
Peringe Grennfelt
Swedish Water and Air Pollution Research Institute
S-402 24 Gothenburg, Sweden
Jorgen Saltbones
Norwegian Meteorological Institute
P. 0. Box 320, Blindern
N-Oslo 3, Norway
INTRODUCTION
This pilot study resulted from the growing concern about photochemical air
pollution in Europe during the last decade. From other studies conducted in
several countries, it had become evident that oxidants and their precursors
could be transported over many hundreds of kilometers and that they could affect
countries other than those of the precursor sources. In a report from the
Organization of Economic Cooperation and Development (OECD, 1978), the Ad Hoc
Group of Experts on Photochemical Oxidants and Their Precursors in the
Atmosphere concluded that, in view of long-range transport, emission control on
a local scale may be grossly insufficient in Europe and Eastern North America
(OECD, 1978).
*This paper has not been reviewed by the-U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
439
-------
In 1978, the Norwegian Institute for Air Research (NILU) hosted a planning
conference on future cooperative research efforts on the long-range transport of
photochemical oxidants (NILU, 1978). Participants from 12 countries in Europe
and North America attended the conference, which concluded with several
proposals for future research, both on a national basis and in the form of
international cooperation. The proposals covered the following major research
areas: emissions, transformation, ambient measurements, effects, and integrated
modeling studies. The pilot study discussed here, which was conducted as a
result of these recommendations, focused on large-scale oxidant episodes in
Northwestern Europe during the years 1976 to 1979. The term "pilot study" is
used in order to emphasize that the subject has by no means been covered in full
detail.
In December 1979, a questionnaire was sent to individuals and institutions
in 10 countries in Northwestern Europe, requesting information on measurements
of 03 and other secondary air pollutants for 1976 to 1979. Data were available
from eight countries: Austria, Belgium, the Federal Republic of Germany (FRG),
Finland, The Netherlands, Norway, Sweden, and the United Kingdom (UK). Requests
for more specific data were sent in the spring and summer of 1980.
The formation and transport of photochemical oxidants in northwestern
Europe have been frequently discussed in the literature over the last 10 yr.
This research has been conducted mainly in the FRG, The Netherlands, the UK, and
Scandinavia. Some of the relevant references are Atkins et al. (1972), Cox
et al. (1976), Grennfelt (1975, 1976), Fricke and Rudolf (1977), Guicherit and
van Dop (1977), Apling et al. (1977), Guicherit (1978), Becker et al. (1979),
440
-------
and Schjoldager (1979, 1980). A relatively detailed literature survey is
provided in the project report for the pilot study (Schjoldager et al., 1981).
OZONE MONITORING STATIONS
Eight countries provided information on their 03 monitoring stations for
the pilot study. These stations are described below. Included in each
description is information on the type of station, location of the station
(altitude, latitude, longitude), period for which the data were collected, the
measurement method, the calibration method, and additional comments regarding
the availability of data on secondary pollutants.
Station; IMP, Vienna, Austria
Type: Urban, 1 m above street level.
Altitude: 180 m.
Coordinates: Latitude 48°13' N, longitude 16°22' E.
Period: July-September 1976; March-September 1977; 1978 and 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 197]).
Comments: HC and NOX measurements also available.
Station: AfL, Vienna, Austria
Type: Urban, 14 m above street level.
Altitude: 193 m.
Coordinates: Latitude 48°13* N, longitude 16°22" E.
Period: March-July 1976, May-September 1977, March-July 1978,
March-September 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 1971).
Comments: HC and NOX measurements also available.
441
-------
Station: Illmitz, Austria
Type: Rural, 65 km southeast of Vienna.
Altitude: 119 m.
Coordinates: Latitude 47°46' N, longitude 16°46' E.
Period: May-September 1978, April-September 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 1971).
Comments: NOX measurements also available.
Station: Roeschitz, Austria
Type: Rural, 65 km northwest of Vienna.
Altitude: 282 m.
Coordinates: Latitude 48°40* N, longitude 15°53' E.
Period: April-September 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 1971).
Comments: NOX measurements also available.
Station: R 822, Antwerp area, Belgium
Type: Suburban/industrial, 3 m above surface.
Altitude: 8 m.
Coordinates: Latitude 51°16' N, longitude 4°22' E.
Period: March-September 1979.
Measurement: Chemiluminescence.
Station: R 801. Antwerp area, Belgium
Type: Urban.
Coordinates: Latitude 51°13' N, longitude 4°26' E.
Period: March-September 1979.
Measurement: Chemiluminescence.
Comments: HC and NOX data available from 13 and 18 Belgian stations,
respectively.
Station; Zentralstation, Frankfurt, FRG
Type: Urban.
Period: June-August 1976, June-July 1977, June-August 1978, May-July
1979.
Measurement: Chemiluminescence (ethylene).
Calibration: UV/absorption.
442
-------
Station: Feldberg, Frankfurt, FRG
Type: Rural, mountain station.
Altitude: 805 m.
Period: March-September 1976, 1977, and 1978.
Measurement: Chemiluminescence.
Calibration: UV/absorption.
Station: Venusberg, Bonn, FRG
Type:
Altitude:
Period:
Measurement:
Calibration:
Comments:
Suburban.
220 m.
March-September 1976, 1977, and 1978.
Chemiluminescence.
UV/absorption.
03 data from five other stations in the Cologne-Bonn area
(Eifelwall, Godorf, Bonn Universitat, Olberg, and Michelsberg);
NOX and HC data available from several stations.
Station: Helsinki, Finland
Type:
Coordinates:
Period:
Measurement:
E.
Urban.
Latitude 60° N, longitude 25C
April-August 1979.
Chemiluminescence (rhodamine B).
Station; Delft, Netherlands
Type: Suburban.
Altitude: 1.5 m.
Coordinates: Latitude 52°00'.N, longitude 4°23' E.
Period: March-September 1976, 1977, 1978, and 1979.
Measurement: Galvanometric (1976, 1977), colorimetric (1978),
Chemiluminescence, ethylene (1979).
Calibration: Electrochemistry (1976), gas-phase titration (1977 to 1979).
Station: Terschelling, Netherlands
Type:
Altitude:
Coordinates:
Period:
Measurement:
Calibration:
Comments:
Rural.
4 m.
Latitude 53°24' N, longitude 5°21' E.
June-August 1978.
Chemiluminescence (ethylene).
Gas-phase titration.
03, HC, and NOX data available from many other stations in the
Netherlands. Measurements of peroxyacetyl nitrate (PAN) and
peroxybenzoyl nitrate (PBzN) have also been performed (Guicherit,
1978).
443
-------
Station; Maridalen, Oslo, Norway
Type: Rural, 15 km north of Oslo.
Altitude: 165 m.
Coordinates: Latitude 60°00' N, longitude 10°48' E.
Period: May-September 1977, June-September 1978, May-September 1979,
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 1971).
Station; Bjornstad, Telemark, Norway
Type: Suburban/industrial.
Altitude: 30 m.
Coordinates: Latitude 59°09' N, longitude 9°38' E.
Period: May-September 1976, May-August 1977, May-September 1978.
Mesurement: Chemiluminescence (rhodamine B).
Calibration: Same as Maridalen.
Station; Langesund, Telemark, Norway
Type: Suburban/coastal.
Altitude: 10 m.
Coordinates: Latitude 59°01' N, longitude 9°45* E.
Period: April-September 1979.
Measurement: Chemiluminescence (rhodamine B).
Calibration: Same as Maridalen.
Station: Haukenes, Telemark, Norway
Type: Rural.
Altitude: 30 m.
Coordinates: Latitude 59°12' N, longitude 9°29' E.
Period: April-September 1979.
Measurement: Chemiluminescence (rhodamine B).
Calibration: Same as Maridalen.
Comments: 03 data from three other stations available for 1978 to 1979.
HC and NOX data available from some of the Telemark stations.
Station; Roervik, Sweden
Type: Rural/coastal.
Altitude: 20 m.
Coordinates: Latitude 57°25' N, longitude 11°56' E.
Period: May-September 1976, 1977, 1978, and 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: KI, EPA (Federal Register, 1971).
Comments: NOX data also available.
444
-------
Station: Goeteborg, Sweden
Type: Urban, 20 m above street level.
Altitude: 25 m.
Coordinates: Latitude 57°43" N, longitude 12°00' E.
Period: May-September 1976, 1977, 1978, and 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: Same as Roervik.
Comments: HC and NOX data also available. 03 data from two stations on the
Swedish east coast (Stockholm and Mollergren, 1978).
Station: WSL, UK
Type: Suburban/rural.
Altitude: 100 m.
Coordinates: Latitude 51°53' N, longitude 00°12' W.
Period: April-September 1977, March-September for 1978 and 1979.
Measurement: Chemiluminescence (ethylene).
Calibration: Neutral buffered KI, cross-referenced with UV/absorption.
Station: London, UK
Type: Urban.
Altitude: 6 m.
Coordinates: Latitude 51°29' N, longitude 00°08' W.
Period: March-September 1976, March-September 1977, March-September 1978,
March-September 1979.
Measurement: Same as WSL.
Calibration: Same as WSL.
Station; Islington, UK
Type: Urban.
Altitude: 20 m.
Coordinates: Latitude 51°32' N, longitude 00°06' W.
Period: March-September for 1976, 1977, 1978, and 1979.
Measurement: Same as WSL.
Calibration: Same as WSL.
Station: Sibton, UK
Type: Rural.
Altitude: 46 m.
Coordinates: Latitude 52°18' N, longitude 01°28* E.
Period: July-September 1976, March-September 1977, March-September 1978,
April-September 1979.
Measurement: Same as WSL.
Calibration: Same as WSL.
445
-------
Station: Canvey, UK
Type: Suburban/rural.
Altitude: 3 m
Coordinates: Latitude 51°32' N, longitude 00°34' E.
Period: May-September 1977, March-September 1978, and 1979.
Measurement: Same as WSL.
Calibration: Same as WSL.
Station: Harrow, UK
Type: Suburban.
Altitude: 60 m.
Coordinates: Latitude 51°34' N, longitude 00°21' W.
Period: August-September 1979.
Measurement: Same as WSL.
Calibration: Same as WSL.
In addition to the six UK stations listed above, four other stations
provided data for the oxidant episode occurring June to July 1976: MRC City,
GLC County Hall, GLC Hainault, and GLC Teddington (Ball and Bernard, 1978).
Also, data from Harwell were made available for certain episodes in 1977 and
1978, and some data from Lancaster for 1977 and 1978 are available in the
literature (Harrison and Holman, 1979; Harrison and McCartney, 1980). Harwell
is a rural site with an altitude of 130 m, a latitude of 51°34' N, and a
longitude of 1°19' W.
SUMMARY OF OZONE MEASUREMENTS
High Concentrations of Ozone
Many factors make a comparison from year to year or from country to country
difficult. Both the number of stations and the type (urban, suburban, rural)
vary from year to year. Furthermore, the meteorological conditions favorable
446
-------
for oxidant formation vary considerably from one year to the next. Finally, the
calibration methods used to collect data are not consistent.
Data on concentrations ?200 ppb are given in Table 1. The
"Grosswetterlagen" (GWL) categories, which are daily categories published by the
FRG's Meteorological Service, are also given.
Data on maximum hourly concentrations exceeding the reference values of
100 ppb and 150 ppb are given on a country basis for the period May to August
each year in Table 2. High concentrations during the warm, dry summer of 1976
are evident, as well as the large number of high concentrations in Austria in
1979. As mentioned previously, a comparison from year to year or from country
to country should not be made because of inconsistencies in the data base. The
number of days with high concentrations may have been underestimated for The
Netherlands and the FRG, because these countries collected more 03 data than
those discussed in this paper.
Covariation with the Large-Scale Weather Pattern, GWL
In Table 3, the number of days with maximum hourly 03 concentrations
>100 ppb is listed, as well as the total number of days for each GWL category.
The English data are grouped under "Great Britain"; the Belgian, Dutch, and
German data are grouped under "European Continent"; and the Norwegian and
Swedish data are grouped under "Scandinavia". The Austrian data are not
included in Table 3.
447
-------
TABLE 1. HOURLY OZONE CONCENTRATIONS _-200 ppb.
Station
Illmitz, Austria
Venusberg, Bonn, FRG
Delft, Neth.
Vlaardingen, Neth.
Vlissingen, Neth.
Haamstede, Neth.
WSL, UK
Harwell, UK
MRC City, UK
GLC Teddington, UK
Maximum
Concentration
Date (ppb)
04-15-79
06-07-79
06-11-79
08-15-79
08-22-79
09-14-79
07-12-77
1976
05-08-76
06-25-76
07-03-76
07-03-76
07-02-76
07-03-76
07-04-76
07-05-76
07-06-76
07-07-76
06-25-76
06-26-76
06-27-76
07-03-76
06-28-76
205
220
213
249
203
208
202
200
270
208
261
207
>220
>220
230
258
204
212
201
203
200
>200
211
Hours with
Concentration
200 ppb GWLa
3
2
2
3
2
2
1
1
1
2
6
7
4
6
1
2
2
2
1
1
1
U
BM
BM
HFA
TRW
NWA
HNA
SEA
HM
HNA
HNA
BM
HNA
HNA
HNA
HNA
HNA
HM
HM
HM
HNA
HM
"Grosswetterlagen.
448
-------
TABLE 2. OZONE CONCENTRATIONS .-100 ppb AND 150 ppb,
MAY TO AUGUST 1976-1979
Country
Austria
Belgium
FRG
Finland
Netherlands
Norway
Sweden
UK
Reference
Value
(ppb)
100
150
100
150
100
150
100
150
100
150
100
150
100
150
100
150
1976
5
0
31
6
29
5
6
0
12
0
28
14
1977
15
2
6
2
2
0
1
0
2
0
13
1
1978
30
1
9
1
3
0
3
0
14
0
7
0
1979
115
65
6
1
1
0
1
0
7
0
4
2
7
2
3
1
449
-------
TABLE 3. MAXIMUM 1-h 03 CONCENTRATIONS >100 ppb FOR VARIOUS LARGE-SCALE
WEATHER PATTERNS, MAY-AUGUST, 1976-79
Description
Grosswetterlagen der zonalen
Zirkulationsform
Westlage, antizyklonal
Westlage, zyklonal
Suedliche Westlage
Winkelf oermige Westlage
Grosswetterlagen der gemischten
Zirkulationsform
Suedwestlage, antizyklonal
Suedwestlage, zyklonl
Nordwestlage , antizlyklonal
Nordwestlage, zyklonal
Hoch ueber Mitteleuropa
Hochdruckbrucke (Ruecken)
ueber Mitteleuropa
Tief Mitteleuropa
Grosswetterlagen der meridionalen
Zirkulationsform
Nordlage, antizyklonal
Nordlage, zyklonal
Hoch Nordmeer-Island,
antizyklonal
Hoch Nordmeer-Island, zyklonal
Hoch Britische Inseln
Trog Mitteleuropa
Nordostlage, antizyklonal
Nordostlage, zyklonal
Weather ECa
WA 1
WZ 3
WS
WW 1
SWA
SWZ
NWA
NWZ
HM 16
BM 16
TM
NA 1
NZ
HNA 10
HNZ
HB
TRM
NEA 2
NEZ 1
Total
GBb SC° Days
1 27
1 61
6
1 7
4
4
11
19
16 8 27
81 54
1 15
8
16
10 23
1 17
12 16
17
21 22
21 26
(continued)
450
-------
TABLE 3. (continued)
Description
Hoch Fennoskandien,
antizyklonal
Hoch Fennoskandien, zyklonal
Hoch Nordmeer-Fennoskandien,
antizyklonal
Hoch Nordmeer-Fennoskandien,
zyklonal
Suedostlage, antizyklonal
Suedostlage, zyklonal
Suedlage, antizyklonal
Suedlage, zyklonal
Tief Britische Inseln
Trog Westeruopa
Uebergang
Weather ECa GBb
HFA 8 3
HFZ
HNFA 4 2
HNFZ 1
SEA 3 3
SEZ
SA
SZ
TB
TRW 1 1
U 21
SC°
5
4
2
1
2
2
4
1
Total
Days
24
13
12
11
4
2
13
34
3
Total
69 51
38
492
'EC = European Continent (Belgium, Netherlands, Federal Republic of Germany).
bGB = Great Britain (United Kingdom).
°SC = Scandinavia (Norway, Sweden).
Although high 03 levels were strongly associated with some weather events
(e.g., HM, BM, HNA, HFA and HNFA), most of these events did not necessarily
imply high 03 concentrations. Thus, it seems that large-scale weather patterns
alone do not determine conditions sufficient for 03 formation. An exception to
this is the HM category, showing high 03 concentrations in UK and in Europe
during 16 of 27 days.
451
-------
Certain weather patterns are grouped together in Table 4 in order to
examine differences between Scandinavia and the rest of Europe. The relative
occurrence of high 03 levels for the categories HM, HFA, and HNFA was similar
for the three regions. For the categories BM, HNA, and SEA, there were many
high values for Europe and the UK, but there were considerably fewer high values
for Scandinavia. The opposite was the case for the categories HNFZ, SZ, TB, and
TRW; there were few high values for Europe and the UK and high values for
Scandinavia. Some of these categories may be associated with pollutant
transport to Scandinavia from other parts of Europe.
TABLE 4. MAXIMUM 1-h 03 CONCENTRATION >100 ppb FOR VARIOUS LARGE-SCALE
WEATHER PATTERN CATEGORIES, MAY-AUGUST, 1976-1979
Category
HM, HFA, HNFA
BM, HNA, SEA
HNFZ, SZ, TB, TRW
Other
Total
EC8
(No.) (%)
28 41
29 42
1 1
11 16
69 100
GBb
(No.) (%)
21 41
21 41
2 4
7 14
51 100
SCC
(No.)
17
2
10
9
38
CD
45
5
26
24
100
aEC = European Continent (Belgium, Netherlands, Federal Republic of
Germany).
bGB = Great Britain (United Kingdom).
°SC = Scadinavia (Norway, Sweden).
452
-------
SELECTED EPISODES
This section contains a discussion of selected time periods during which
the 03 concentrations exceeded 100 ppb at several stations. The discussion is
based on daily weather maps, including synoptic weather situations and local
meteorological conditions, and on 850-mbar air trajectories. The trajectories
were calculated as part of OECD's study, Long-Range Transport of Air Pollutants
(LRTAP) in 1976 and 1977 (OECD, 1977) and as part of ECE's study, European
Monitoring and Evaluation Programme (EMEP) in 1978 and 1979 (ECE, 1977). The
850-mbar trajectories should not be used to identify definite precursor source
regions, but as rough indicators of the air flow aloft, especially during high
pressure situations when the trajectories are more uncertain.
This paper presents information on the following two episodes, June 19 to
July 17, 1976, and May 30 to June 8, 1979. In the project report, six other
episodes are also discussed (Schjoldager et al., 1981):
• August 16 to August 30, 1976,
• June 12 to June 15, 1977,
• July 2 to July 12, 1977,
• July 28 to August 1, 1978,
• August 20 to August 23, 1978, and
• May 12 to May 20, 1979.
For each of the selected episodes, the daily (1200 GMT) weather maps are
given for every second day. The weather maps are from Weather Log, published by
453
-------
the British Meteorological Office. Air trajectories at the 850-mbar level are
presented. For 1976 and 1977, 48-h trajectories are available; for 1978 and
1979, 96-h trajectories are available.
June 19 to July 17, 1976, Episode
This episode, which was discussed in the literature by Apling et al. (1977)
and Ball and Bernard (1978), was characterized by high pressure centers over
various parts of Europe with abnormally warm and dry weather. According to
Weather Log (1976), the hot spell in the UK was "probably unprecedented in
length and intensity since the eighteenth century."
The weather maps for every second day of the period are given in Figure 1.
At the beginning of the period, the high pressure center moved eastward from the
Atlantic Ocean, covering large parts of central Europe, while a low pressure
area was located south of Iceland. The high pressure center later moved slowly
towards the Norwegian Sea. During most of the period, wind speed was low and
maximum temperatures exceeded 25°C. Wind direction was often variable. The
skies were mostly clear, except for the first and last days in the period.
In Figure 2, the 48-h air trajectories at the 850-mbar level, arriving at
1200 GMT on every second day, are presented. The air trajectories indicate
transport aloft from the west during the first days of the period. Towards the
end of June, the transport to Scandinavia and the UK was from the southwest,
while there was variable transport on the continent. In the beginning of July,
the air aloft moved clockwise around the high pressure center in the North Sea.
454
-------
Figure 1. Daily weather maps at 1200 GMT for every second day, June 19-
July 17, 1976 (British Meteorological Office, 1976).
455
-------
JUNE 1976
Figure 2. The 48-h air trajectories at the 850-mbar level arriving at 1200 GMT
on every second day, June 19-July 17, 1976.
456
-------
f^'"-
1 JULY 1976
F--V-- c
ka^ ? /* •-. --\
i\ K^
JULY 1976
,-•• •-. 7 JULY t976
Figure 2. (continued).
457
-------
Figure 2. (continued),
458
-------
Between July 5 and July 15, there was generally no large-scale transport aloft;
this lasted until the end of the period when a cold front approached from the
Atlantic Ocean.
The maximum hourly concentrations are given on a daily basis in Table 5.
The 03 concentrations were high over all of Europe, reaching 129 ppb in Austria,
186 ppb in Germany, 191 ppb in The Netherlands, 258 ppb in Great Britain, and
125 ppb in Sweden.
As explained earlier, the highest concentrations in the UK resulted from a
combination of 03 transport from the European continent and local and mesoscale
formation (Apling et al., 1977; Ball and Bernard, 1978). This explanation may
in general be valid for other countries as well. During the long-lasting high
pressure situation, transboundary air pollution probably affected large parts of
Europe, interacting with the locally emitted oxidant precursors.
May 30 to June 8, 1979, Episode
An anticyclone stagnated over Scandinavia during most of the period, and
the pressure gradients over Europe were generally small (Figure 3). The local
weather was fair in southern Scandinavia and on the continent. -In England, it
was much more cloudy.
The air trajectories indicate transport from the south to Western Europe in
the beginning. Later, the transport had a significant easterly component,
changing to west at the end of the period (Figure 4).
459
-------
o
f>
u
H
H
CU
co
CM
fOOOOOvt
X
«
w
CM
oo
CM
U
CO
w
o
vO
o>
C
3
vO
CM
in
CM
in
O
00
CM
o-i «* ,H
O fH O
CM rH iH
CM
CM
§
CM
CM
CM
N)
3
0)
3
B
B
O
O
w
u
o
0
o
CM
CJ^
OO-CM
CM
in
N
§
X
m
w
j
CO
H
B
O
•H
4-1
CD
4-1
-------
1
•d
V
r
c
•H
C
o
cu
ij
H
•— t
3
fx.
•H
r-t
2
p-4
f)
rH
CM
t— 1
O
CO
r^
«0
in
-
«
CM
Station
CM m r^ CM f** ft
2 2 S 2 S S
c? 3 « .2 S £
*— ' «^ «^ <^ O 00
oo oo CM oo no
5 SoS2S2S"S
82 2 S2SS3S
fOM^OO cnCMvOOOOO
o^ooco oocM^-a-^r.
~ 5 ! § * * 5 S s -
^H f^ CM ^ ^4 »^
f-H CM f-H r-4
OMcMm O oo oo so co
^r^c^im xj(*immoo
•H *H fH fH CVJ «^ i-^
c^ O 00 ^f *^ O f°** O c*^
•^ ••* *-^ CM CM CM CM
A A
— 1 !•* -4 CM ^ ^
A
CMO ^SgS^ggS
^ ^ ^ -*
U
3 in
J* B
C (0
as u — i
U kl « hi ><
u, 3 -on -a a
•" 6 J= T) CCS
"^CIQ-^C (9£>ki
CCB-HUU ---8O
O <5 0 li Z ^ 00 > ^ w
•Hx^fUai c>-4Bwu«
- ij -a 3 u-i 01 a - h
-------
Figure 3. Daily weather maps at 1200 GMT for every second day, May 30-
June 1979 (British Meteorological Office, 1979).
462
-------
•,'-•-, 3 JUNE 1979
5 JUNE 1979
Figure 4. The 96-h air trajectories at the 850-mbar level, arriving at 1200
GMT on every second day, May 30-June 9, 1979.
463
-------
The O3 concentrations were high in Austria and low in England throughout
the period (Table 6). Maximum concentrations exceeding 100 ppb also occurred in
Belgium and the FRG. In Sweden, the maximum hourly concentrations were 80 to
50 ppb. In Norway, the maximum concentration of 197 ppb occurred downwind of an
industrial area (Schjoldager, 1980).
It may be the high concentrations in Central Europe were due to local and
mesoscale formation with some enhancement from long-range transport. For
southern Scandinavia, transport from distant sources was probably important in
the first part of the period, but local and mesoscale production appear to have
contributed significantly during last part of the period.
CONCLUSIONS
With one exception, all the hourly 03 concentrations exceeding 200 ppb were
measured in England or The Netherlands during 1976, or at Illmitz, Austria,
during 1979. The highest 1-h 03 concentration discussed in this report is
258 ppb, measured at Harwell, England, on July 5, 1976. However, concentrations
up to 0.27 ppm were reported earlier at Vlaardingen, Netherlands, on
May 8, 1976.
Most high 03 concentrations occurred with stagnating anticyclones. When a
high pressure area was located over Central Europe, Scandinavia, or Finland, the
concentrations were often high over all the examined regions in Northwestern
Europe. With a high pressure ridge over Central Europe or with the high
pressure center over the North Sea, the concentrations could be high in Europe
464
-------
TABLE 6. MAXIMUM HOURLY 03 CONCENTRATIONS (ppb) AND LARGE-SCALE WEATHER PATTERNS
(GWL), MAY 30-JUNE 8, 1979
Station
Alf, Wien
Illnitz, Austria
Roschnitz, Austria
Zentralstation, Frankfurt
Delft, Netherlands
Canvey, UK
WSL, .UK
Sibton, UK
Rorvik, Sweden
Langesund, Norway
R 801, Belgium
R 822, Belgium
Helsinki, Finland
GWL
Ma
30
92
161
107
50
28
45
40
56
114
57
63
HNFA
y
31
100
183
111
122
56
23
43
112
94
47
56
HNFA
1
108
189
99
48
53
58
29
53
150
90
44
44
HNFA
2
119
186
98
66
40
41
27
44
114
70
39
57
HNFA
3
119
198
63
66
65
25
77
80
61
100
50
HNFA
June
4
128
193
78
78
65
56
75
95
65
63
51
HNFA
5
99
180
70
47
44
25
61
90
71
54
HNFA
6
154
166
23
48
42
23
47
92
93
37
57
HNFA
7
160
220
42
41
30
17
43
84
90
57
BM
8
70
161
84
30
65
41
28
60
64
28
59
BM
and in Great Britain but significantly lower in Scandinavia. In cases with
cyclonic circulations and low pressure areas over Central Europe, the
concentrations could remain high in Scandinavia while they were significantly
lower in other parts of Northwestern Europe.
465
-------
At the Austrian station of Illmitz, located in a rural area 65 km southeast
of Vienna, the concentrations were high throughout the period April to September
1979. During this period, the maximum concentration exceeded 150 ppb on
90 days.
Two episodes were studied in more detail with respect to synoptic weather
situations, local meteorological conditions, and trajectory analyses, in order
to assess the origin and transport of the polluted air masses. Six other
episodes are discussed in the project report of the pilot study. In some cases,
long-range transport appeared to be more important than local and mesoscale 03
formation, while the contrary occurred on other occasions. Due to the
methodology used and the limited measurement data, it was impossible to assess
quantitatively the role of various production scales. Thus, it has not been
possible to evaluate quantitatively the influence of the various precursor
source regions on the 03 concentrations at the various receptor points.
The highest 03 concentrations were reached during weather conditions
conducive to oxidant formation, when local precursors were emitted into polluted
air masses transported from other source regions. A typical episode of this
kind occurred in England during the hot spell in June/July 1976.
466
-------
In most parts of Northwestern Europe, the maximum 03 concentrations are as
high as, and in some cases higher than, the threshold levels associated with
plant damage and health effects.
The need for further, concerted studies of photochemical oxidants in Europe
is clearly present.
RECOMMENDATIONS
Among the many aspects of photochemical air pollution in Europe that should
be investigated further are the following:
An emissions inventory of NOX and VOCs for northwestern Europe. The
emissions should be given for grid squares of approximately
100 km x 100 km. Volatile organics should be specified in terms of
chemical reactivity.
The transport of air pollutants during high pressure situations. In
these cases, air trajectories are often highly uncertain. The highest
concentrations of photochemical oxidants are, however, most likely to
occur during these weather situations. The need for more research on
this issue is thus obvious.
Large-scale photochemical transport models. These models will increase
the general understanding of and aid in predicting the effect of future
emission controls.
A consistent data base of relevant air concentrations. Consistency in
measurement and calibration methods, and in the criteria for sampling
sites is essential.
Measurements above the surface layer. Aircraft measurements are of
value in determining the horizontal and vertical extent of high
concentrations. Measurements from meteorological towers are of value in
the study of the diurnal concentration variation above the nocturnal
surface inversion layer.
467
-------
It should also be pointed out that other parts of Europe may experience
concentration levels of photochemical oxidants as high or higher than those
given in this report. Of special importance is the entire southern part of
Europe. The total precursor emissions there are smaller than those measured in
Northwestern Europe, but the weather of Southern Europe is much more sunny and
warm.
It may turn out that in all the Mediterranean countries, from Spain to
Turkey, oxidant levels exceeding internationally accepted threshold values can
occur. Studies in these regions are thus highly desirable.
The Austrian 03 data also indicate that high concentrations can occur in
parts of Central Europe. Measurements downwind of other major metropolitan
areas of Central Europe will give insight into this phenomenon.
In order to study large-scale photochemical oxidant formation and transport
in Northwestern Europe without, local influence, rural monitoring sites are
preferred. For the establishment of a minimum sampling network in northwestern
Europe, a distance of about 300 km between neighbouring stations is recommended.
Table 7 lists the number of stations that would be required in each country to
meet such a distribution.
468
-------
TABLE 7. NUMBER OF STATIONS
PROPOSED FOR A MINIMUM SAMPLING
NETWORK IN NORTHWESTERN EUROPE
Country
Austria
Belgium
Denmark
FRG
Finland
France
Ireland
Netherlands
Norway
Sweden
UK
No. of Stations
2
1
1-2
3
2
2-3
1
1
2-3
2-3
2-3
The air pollutants of interest may tentatively be grouped into priority
categories as follows:
First priority: 03 (continuous)
Second priority: Sulphate (24-h)
Visibility (at least once per day)
Nitric acid (24-h)
PAN (continuous, if possible)
Third priority: NOX (continuous)
VOC (continuous)
S02 (24-h)
The quality of the measurement data should be assured by regular
intercalibration procedures and station performance audits.
469
-------
ACKNOWLEDGMENTS
Financial support for this study was provided by the National Swedish
Environment Protection Board and the Norwegian Ministry of Environment. The
Environment Directorate of the Organization for Economic Cooperation and
Development (OECD) was helpful in establishing the necessary contacts for the
exchange of information.
A list of persons/institutions who submitted data is given below. Their
contribution is gratefully acknowledged. However, the interpretation and views
expressed in this paper are those of the authors and are not necessarily shared
by the contributors.
Austria: Ruth Baumann, Bundesstaatliche bakteriologisch-serologische
Untersuchungsanstalt, Wien.
Belgium; Jacques Bouquiaux, Institut d'Hygiene et d'Epidemiologie,
Ministere de la Sante Publique et de la Famille, Bruxelles.
FRG: Ulrich Schurath, Institut for Physikalische Chemie der Universitat
Bonn, Bonn; H-W Georgii, Institut fur Meteorologie und Geophysik der Johann
Wolfgang Goethe-Universitat, Frankfurt a.M; Werner Rudolf, Umweltbundesamt
Pilotstation Frankfurt, Frankfurt a.M.
Finland; Risto Lahdes, Helsingin kaumpungin terveysvirasto, Helsinki.
470
-------
Netherlands: Robert Guicherit, Research Institute for Environmental
Hygiene TNO, Delft.
Norway: Leif Stige, Norwegian State Pollution Control Authority (SFT),
Porsgrunn.
UK: Richard G. Derwent, Environmental and Medical Sciences Division, AERE
Harwell, Oxfordshire; Alan J. Aplig, Warren Spring Laboratory, Department of
Industry, Gunnels Wood Road, Stevenage.
BIBLIOGRAPHY
Apling, A. J., E. J. Sullivan, M. L. Williams, D. J. Ball, R. E. Bernard, R. G.
Derwent, A. E. J. Eggleton, L. Hampton, and R. E. Wallace. 1977. Ozone
concentrations in Southeast England during the summer of 1976. Nature,
269:569-573.
Atkins, D. H. F., R. A. Cox, and A. J. E. Eggleton. 1972. Photochemical ozone
and sulphuric acid aerosol formation in the atmosphere over Southern
England. Nature, 235:371-376.
Ball, D. J., and R. E. Bernard. 1978. An analysis of photochemical pollution
incidents in the greater London area with particular reference to the
summer of 1976. Atmospheric Environment, 12:1391-1401.
Becker, K. H., U. Schurath, H. W. Georgii, and M. Deimel. 1979. Untersuchungen
ueber Smogbildung, Inbesondere ueber de Ausbildung von Oxidanten als Folge
der Luftvrunreinigung in der bundesrepublik Deutschland. Umweltundesamt,
Berlin.
British Meteorological Office. 1976-79. Weather Log. Bracknell, Berkshire,
England.
Cox, R. A., R. G. Derwent, and F. J. Sandalls. Some Air Pollution Measurements
Made at Harwell, Oxfordshire, 1973-1975. UKAEA Report AERE-R 8324.
Cox, R. A., A. E. J. Eggleton, R. G. Derwent, J. E. Lovelock, and D. H. Pack.
1976. Long range transport of photochemical ozone in northwestern Europe.
Nature, 255:118-121.
471
-------
Economic Commission for Europe. 1977. The Cooperative Programme for Monitoring
and Evaluation of the Long Range Transmission of Air Pollutants in Europe.
ECE/ENV/15, Annex II, United Nations, Geneva.
Fricke, W., and W. Rudolf. 1977. Ozonkonzentrationen in Luv and Lee vond
Ballungesgebieten auf der Flugline Munchen-Rotterdam. Staub-Reinhalt.
Luft, 37:341-345.
Grennfelt, P. 1976. Ozone Episodes on the Swedish West Coast. In:
Proceedings of the International Conference on Photochemical Pollution and
Its Control. EPA-600/3-77-001, U.S. Environmental Protection Agency,
Research Triangle Park, North Carolina, pp.329-337.
Grennfelt, P. 1975. Measurement of ozone in Gothenburg, January 1972-
August 1973 and Studies of Co-Variations Between Ozone and Other Air
Pollutants. IVL-Report B 221, Swedish Water and Air Pollution Research
Institute, Gothenburg, Sweden.
Guicherit, R., editor. 1978. Photochemical Smog Formation in The Netherlands.
TNO, Delft.
Guicherit, R., and H. van Dop. 1977. Photochemical production of ozone in
Western Europe (1971-1975) and its relation to meteorology. Atmospheric
Environment, 11:145-155.
Harrison, R. M., and H. A. McCartney. 1980. Ambient air quality at a coastal
site in rural Northwest England. Atmospheric Environment, 14:233-244.
Harrison, R. M., and C. D. Holman. 1979. The contribution of middle- and
long-range transport of trophosperic photochemical ozone to pollution at a
rural site in Northwest England. Atmospheric Environment, 13:1535-1545.
Hess, P., and H. Brezowski. 1969. Katalog der Grosswetterlagen Europas.
Berichte des Deutschen Wetterdienstes, Offenbach a.M., Federal Republic of
Germany.
Killingmo, 0. H., and C. Mollergren. 1978. Matningar av ozon i Stockholm och
Nykoping under perioden 1973-1976 (in Swedish). SNV PM 1087, Statens
Naturvardsverk, Studsvik, Sweden.
Norwegian Institute for Air Research. 1978. Long Range Transport of
Photochemical Oxidants: Report from a Planning Conference on Future
Research Cooperation. NILU TN 16/78, Norwegian Institute for Air Research,
Lillestrom, Norway.
Office of the Federal Register. 1979. National primary and secondary ambient
air quality standards. Federal Register, 44:8202-8237.
Office of the Federal Register. 1971. National primary and secondary ambient
air quality standards. Federal Register, 36:8186-8201.
472
-------
Organization for Economic Cooperation and Development. 1978. Photochemical
Oxidants and Their Precursors in the Atmosphere: Effects, Formation ,
Transport and Abatement. Paris, France.
Organization for Economic Cooperation and Development. 1978. The OECD Programme
on Long-Range Transport of Air Pollutants. Measurements and Findings.
Paris, France.
Schjoldager, J., H. Dovland., P. Grennfelt, and J. Saltbones. 1981.
Photochemical Oxidants in Northwestern Europe, 1976-79. A Pilot Project.
NILU OR 19/18, Norwegian Institute for Air Research, Lillestrom, Norway.
Schjoldager, J. 1980. Ambient Ozone Measurement in Norway, 1975-1979.
presented at the 73rd Meeting of the Air Pollution Control Association,
Montreal, Canada.
Schjoldager, J. 1979. Observations of high ozone concentrations in Oslo,
Norway, during the summer of 1977. Atmospheric Environment, 13:1689-1696.
U.S. Environmental Protection Agency. 1978. Air quality control for ozone and
other photochemical oxidants. EPA-600/8-78-004, Research Triangle Park,
North Carolina.
473
-------
Emissions Inventories in Europe*
Lothar Kropp
Technischer Ueberwachungs-Verein Rheinland e.V.
Cologne, Federal Republic of Germany
INTRODUCTION
The most important objective of an air pollution control policy is the
maintenance of air quality, a goal affecting humans, animals, and the biological
and material environment. In addition to the effects of air pollutants in the
neighborhood of air pollution sources, the effects of the long-range transport
of these pollutants are also important. Sulfur compounds and nitrogen compounds
are mainly responsible for the presence of oxidant pollutants.
An important element of an air quality management system is the
establishment and implementation of clean air plans, and the basis for any
clean air plan is an emissions inventory. According to the German Clean Air
Act, clean air plans and emissions inventories must be established in heavily
polluted areas (HPAs), which must be declared as such by the German provinces.
Other European countries like The Netherlands, Belgium, Denmark and Norway are
also establishing emissions inventories and clean air plans.
*This paper has not been reviewed by the U.S. Environmental Protection Agency
and therefore does not necessarily reflect the views of the Agency, and no
official endorsement should be inferred.
474
-------
SURVEY OF CLEAN AIR PLANS AND EMISSIONS INVENTORIES IN THE FEDERAL REPUBLIC OF
GERMANY
Six of the 11 regions in the Federal Republic of Germany (FRG) have
formally declared 21 areas in Germany to be HPAs (see Table 1 and Figure 1), and
emissions inventories have been completed or are being prepared in 28 HPAs.
Data on 24 of these areas were available for this report from publications and
other sources. The total size of the 24 areas is about 9,500 km2, and the total
population is approximately 17.5 million (see Table 2). About 28% of the German
population lives in these 24 areas, which amounts to only 3.8% of the total land
mass of the FRG. The average population density in the HPAs is about
2,500 inhabitants, which is 10 times higher than the average population density
for the entire FRG.
Clean air plans for HPAs have already been established in 3 of the
11 German regions. Several of the other regions are presently establishing
clean air plans (e.g., Baden-Wuerttemberg, Bavaria, Hamburg, and Niedersachsen).
Clean air plans have already been completed in 9 of the aforementioned
28 HPAs. Comprehensive parts of clean air plans have been completed in more
than half of these areas. Table 3 summarizes those areas where emissions
inventories or clean air plans are being researched or have been completed.
475
-------
TABLE 1. HPAS IDENTIFIED IN THE FRG
Region
Bayern
Niedersachsen
Baden-Wuerttemberg
Nordrhein-
Westfalen
Hessen
Rheinland-
Pfalz
Schleswig-
Holstein
Saarland
Hamburg
Berlin
Bremen
FRG
Area
(km2)
70,500
47,400
35,800
34,100
21,100
19,800
15 , 700
2,600
750
480
400
248,630
Inhabitants
( thousands )
10,800
7,300
9,200
17,200
5,600
3,700
2,600
1,100
1,700
2,000
700
61,900
Population
Density
(Inhab./km2)
150
150
250
500
260
190
160
430
2,300
4,200
1,800
250
HPAs
Declared Number
yes 8
no (4)a
no (2)
yes 5
yes 4
yes 2
no
yes 1
no (1)
yes 1
no
21 (7)
•HPAs in parentheses not yet formally declared as such.
476
-------
Figure 1. Federal Republic of Germany.
477
-------
TABLE 2. SUMMARY OF EMISSIONS INVENTORIES IN THE FRG
•-"-""•"• n "'"- "" "'-•-^-M/-""["','• w»»'m»«m-»g«g=i <•—?** '* ii i ie«—a^f -r!SB--M. JKI"' BMSag—-——~g^gj--.-— —.-..^ ...—.^_..j^^ M-mia-.— •
Status Number
Areas formally declared as HPAs 21
Areas in which emissions inventories have been completed or are in
preparation 28
Areas in which emissions inventories have been completed or are in
preparation, which have a total area of approximately 9,500 km2
(approx. 3.8% of FRG) and a total population of 17.5 million
inhabitants (approx. 28% of FRG) 24
478
-------
TABLE 3. SURVEY OF HPAs
Date of Completion
Region
Bayern (total)
Aschaf fenburg
Augsburg
Burghausen
Erlangen-Nuremberg
Ingolstadt
Muenchen
Regensburg
Wuerzburg
Berlin (total)
Hessen (total)
Frankfurt+
Kassel
Wetzlar
Wiesbaden+
Nordrhein-Westfalen (total)
Dortmund+
Duisburg+
Duesseldorf
ESSCR+
Koeln+
Rheinland-Pfalz (total)
Ludwigshafen
Mainz+
Saarland Sarbrueckken (total)
Baden-Wuerteraberg (total)
Mannheim*
Karlsruhe*
Hamburg
Niedersachsen
Area
(km2)
2940'
740'
2301
65'
431
540"
820§
55
601
480
785
466
148
50
120
3200
712
711
356
756
649
205
111
• 94
235
319
145
174
"750
~500
Inhabitants
(thousands)
3315
210
300
25
770
170
1600
130
110
1900
1600
1000
260
47
290
6700
1200
1300
800
2000
1400
400
210
190
370
590
310
280
'1700
'850
Population
(Inhab./km2)
1100
290
1300
380
1800
320
1900
2400
1800
4000
2000
2100
1800
900
2400
2100
1700
1800
2200
2600
2200
1900
1900
2000
1600
1800
2100
1600
'2300
Emissions
Inventory
1982
1981"
1982
1981
1980
1978
1978
1975
1981
1980
1972
1977
1978
1978/79°
1980
Clean
Air
Plan
1983
1982
1981
1978
1977
1982
1980
1976
1980
1982
Braunschweig
Goslar/Oker
Hannover
Nordenham
1981
•Estimate.
"Traffic source category.
'Excludes industrial source category.
479
-------
CLEAN AIR PLANS
A clean air plan consists of the following elements: an emissions
inventory, an air pollution (impact) inventory, an effects inventory, an
analysis of causes, and a catalog of measures.
The emissions inventory is a detailed list of air pollutant emissions,
indicating type, amount, spatial and temporal distribution, and emission
conditions (source dimensions and location, waste gas volume flow, and
temperature).
The air pollution (impact) inventory includes the results of measuring the
most frequently occurring air pollutants present on a large (spatial) scale,
i.e., S02, NOX (NO and N02), CO, gaseous organic compounds, and dust
(concentration and dust fall). These data are supplemented by air pollution
impact data, which is determined from emissions inventory data via dispersion
model calculations.
The effects inventory lists effects found in the area under investigation
that are possibly precipitated air pollutants (e.g., forest damages, material
damages, epidemiological effects).
The analysis of causes demonstrates the relationship between effects, air
pollution impact, and emissions. Its aim is to determine measures to reduce air
pollution emissions.
480
-------
The catalog of measures results from the establishment of a clean air plan
and outlines those measures that are most effective in maintaining air quality
and reducing air pollution and its impact.
Emissions Inventories
Establishing the emissions inventory is a major element of a clean air
plan. In the inventory, emission data are given for three source categories:
(1) industrial installations, (2) domestic heating, trade, and small industries,
and (3) traffic (mainly road traffic).
Industrial installations include all installations subject to licensing,
such as power plants, chemical and petrochemical plants, steel works, nonferrous
metal industries, and coke ovens. The second source category, referred to as
"household," contains domestic heating facilities, fuel stations, dry cleaning
installations, and printing offices. The traffic source category includes road,
rail, water, and air traffic.
Due to the diversity of the industrial processes and sources, and the high
number of air pollutants resulting from industrial plants, emissions data for
the industrial installation source category must be investigated individually.
This can be done either by having independent experts collect the necessary data
pertaining to the different plant parts or by having the plant operator submit
emission reports.
481
-------
The emissions data for domestic heating, trade, and small industry
installations are evaluated during investigations on the type, size, and spatial
distribution of these installations. Emissions data are determined and
correlated with local distribution by using emission factors for corresponding
plant standards or standard plants. Because there are functional dependences
between the technologies used and the emissions procedures followed, this type
of assessment is sufficient.
The same approach applies to traffic sources. It is sufficient to
determine the traffic population and the corresponding driving modes. Traffic
emissions are then determined via emission factors for the different driving
modes and related to local distribution and frequency of occurrence.
For the traffic and the domestic heating, trade, and small industry
(household) categories, the emissions data are combined for line and area
sources. Emissions data for the industrial category, however, are described
mostly via point sources. The smaller emissions and the more uniformly
distributed emissions of the industry are also described as line and area
sources or even as volume sources. In emissions inventories or clean air plans,
emissions are evaluated in terms of a grid that is 1 km x 1 km.
When an emissions inventory is established, there is no a priori limitation
on the air pollutants included. Although the number of air pollution components
for the traffic and the household categories are limited, the industry category
includes many substances that can lead to air pollution impacts. The number of
these substances in the industry category may be 1,000 or more. Table 4 gives a
482
-------
rough summary of the components considered when analyzing the different source
categories.
Emissions inventories in the FRG formerly consisted of investigations
conducted by independent experts of all plants, sources, substances, and
emissions. This practice has now changed; the plant operators must now declare
all emissions data annually. Thus, an actual updating is guaranteed every year.
Updating emissions data for the traffic and the household source categories
is performed 5 yr by the local administration for the respective area. Clean
air plans have to be established or reviewed every 5 yr, too, by the regional
administration and its institutions.
The approach for investigating an emissions inventory in The Netherlands
and Belgium is very similar to that practiced in the FRG.
Status of Investigation
Presently, clean air plans with emissions inventories have been published
for the following regions in the FRG: Cologne, Duisburg, Dortmund,
Ludwigshafen, Essen, Wiesbaden, Duesseldorf, Mainz, and Wetzler. In addition,
emissions inventories are available for the following areas: Berlin (traffic
only), Saarland (traffic and domestic heating only), and the Mannheim,
Karlsruhe, Goslar/Oker, Nuremburg, and Frankfurt regions. Emissions inventories
are being prepared for Hamburg and Berlin (industrial emissions), and for the
483
-------
TABLE 4. AIR POLLUTANTS IN EMISSIONS INVENTORIES INVESTIGATED
Pollutant
S02
NOX
CO
Fluorine compounds
Chlorine compounds
Total dust/particulates
Lead/lead compounds
Soot
Organic compounds
Aldehydes
Benzopyrene
All other organic compounds
All other inorganic compounds
Industry
X
X
X
X
X
X
X
X
X
X
X
X
Source Catategory
Small Domestic
Industry Heating
X X
X X
X X
X X
X X
X X
X X
X
Traffic
X
X
X
X
X
X
All other types of dusts and
particulates
484
-------
Kassel region. Clean air plans are being prepared for the Kassel and Frankfurt
regions and may be available towards the end of 1983.
The first updated clean air plan in the FRG will be available in 1983 for
the Cologne region. It will be particularly interesting to compare the results
of the (experts) investigations for the first clean air plan with those of the
emissions declarations, which will be included in the second clean air plan for
the region.
Emissions inventories for areas outside the FRG are available for Gent and
Liege, Belgium, for five areas of The Netherlands, and for Denmark. This
listing may be incomplete as requests for emissions inventories were only made
recently to several nations.
Results with Respect to Oxidants and Their Precursors
Selected results of one emissions inventory, a general survey of data
available in the Cologne clean air plan, are given in Tables 5 through 8 and
Figures 2 through 4. The survey includes the topics of this workshop.
Table 5 presents data from the major emissions groups, particularly
organic compounds, for the three source categories. Table 6 presents the major
inorganic compounds, particularly NOX, for the three source categories. Table 7
presents the organic compounds, subdivided into HCs and halogenated HCs, for the
industry source category. Table 8 is a rough survey of halogenated HCs.
485
-------
TABLE 5. EMISSIONS DATA REPORTED IN THE COLOGNE CLEAN AIR PLAN
Emissions (thousand tons/yr)
Source Inorganic Organic Dust and
Category Compounds Compounds Particulates
Industry 298 84 25
Household 107 6 5
Traffic 134 5 0.5
Total for all
categories 539 95 30.5
TABLE 6. INORGANIC COMPOUNDS REPORTED IN THE
COLOGNE CLEAN AIR PLAN
Emissions (thousand tons/yr)
Source Other
Category NOX SOX CO Species
Industry 76 147 69 6
Household 4 18 84 1
Traffic 10 1.5 123 0
Total
for
Category
407
118
140
665
Total
for
Category
298
107
134
Total for
all categories
90
166.5
276
539
486
-------
TABLE 7. ORGANIC COMPOUNDS FROM INDUSTRIAL SOURCES
AS REPORTED IN THE COLOGNE CLEAN AIR PLAN
Compound
Hydrocarbons
C,-C4-HC
Fuel-HC (200JC)
Aroma ties
Fuel-HC (200JC)
Other compounds
Halogenated HC
Other
Total of all organic compounds
Emissions
( tons/yr )
56,127
29,780
10,771
9,414
1,970
27,990
5,725
22,265
84,177
(Z of
Total)
67
35
13
11
2
33
7
26
100
TABLE 8. HALOGENATED HCs INDUSTRIAL SOURCES
AS REPORTED IN THE COLOGNE CLEAN AIR PLAN
Compound
Vinyl chloride
Dichlorome thane
1 , 2-Dichloroethane
Trichloroethylene
Ethyl chloride
Carbon tetrachloride
Perchloroethylene
Other
Total of all HCs
Emissions
(tons/yr)
1,335
1,209
1,161
504
167
167
152
1,030
5,725
(% of
Total)
23
21
20
9
3
3
3
18
100
487
-------
10000-
[t/kmz.a]
20 km
10 km
Figure 2. Total annual emission of nitrogen oxides,
488
-------
Gesamtemission NOX (als NO2)
E = 89994 t/a = 100%
40
Industrie
KFZ
Hausbtand u. Kleingewerbe
Ul
g
M
E
0)
E
ID
V
o
v
TJ
c
ra
I
0 10 11-20 21-30 31-40 4160 61 80 81 101
TOO 120 120
Hohenklassen [m]
Figure 3. Total emissions of NOX (as N02).
489
-------
Gesamtemission organischer Case und Dam pie
E = 95191 t/a = 100%
^ 30
UJ
o
'in
£
0)
*••
a 20
«
O
w
•o
c
'5
«.*
< 10
0
n
.
^H
•
1
^^^! Industrie
gPUr KFZ
l t ( l { ,
010 1120 21-30 31-40 4160 6180 81 KM
-10O -120 120
Hohcnklassen jm]
Figure 4. Total emissions of organic compounds.
490
-------
Figure 2 is a three-dimensional display of the NOX (as N02) distribution.
The grid size is 1 km x 1 km, and the peak heights represent the total emissions
per grid. Figures 3 and 4 are displays of NOX (as N02) and organic compound
emissions, respectively, for different height categories. Although most NOX are
emitted at heights over 50 m, most organic compounds are emitted at heights
below 50 m.
The most important data with respect to this workshop are emissions data
for NOX, HCs, and halogenated HCs in the FRG and Europe. Tables 9 through 12
survey emissions in FRG for the following source categories: industry,
household, and traffic. Emissions from sources over 100 m high were excluded
where possible.
Table 9a shows a total N02 emission of about 675,000 tons per year for all
areas with established emissions inventories. About one-third (200,000 tons per
year) of these emissions occur at heights over 100 m. In comparison to the
total NOX emissions from these HPAs, the total NO? emissions for the FRG is 4 to
5 times higher (3 million tons of N02 per year). This is due to the large
amounts of emissions from traffic and domestic heating sources, as well as
emissions from large power stations outside the HPAs. Estimates of NOX
emissions over large stacks (i.e., with heights greater than 100 m) are 5 to
6 times higher than those for emissions within HPAs.
Table lOa shows the corresponding data for S02. Because S02 is less
important in oxidant formation than NOX, these data are given for comparison
only. The total S02 emissions from HPAs is approximately 1.2 million tons per
491
-------
year, one-third of the total SOj emissions for the FRG. More than two-thirds of
the emissions occur via stacks over 100 m high. The emissions data for organic
compounds are given in Table lla. About 480,000 ions of organic compounds per
year are emitted in the HPAs, only one-third to one-fourth of the total
emissions for the FRG. Only a very small part (less than 1%) is emitted via
stacks over 100 m high.
Industrial HC emissions amount to 270,000 tons per year, or 60% of the
total emissions for all organic compounds (see Table 12). Of the HCs emitted by
industry, only 8% are halogenated HCs. These figures refer solely to emissions
data for HPAs. Corresponding figures for the FRG are not available. Although
the HCs emitted at higher altitudes cannot be extracted from published data,
these values can be extracted from the base data used to establish emission
figures.
Some corresponding data for NOX» S02, and organic compound emissions in
other nations are presented in. Tables 9b through lib.
492
-------
TABLE 9a. N02 EMISSIONS IN THE FRG
Region
Bayern
Berlin
Hessen
Nordrhein-
Westfalen
Rheinland
Pfalz
Saarland
Baden-Wuert-
temberg
Total for all
Total for FRG
HPA
Nuremberg
Berlin
Frankfurt
Wetzlar
Wiesbaden
Dortmund
Duisburg
Duesseldorf
Essen
Cologne
Ludwigshafen
Mainz
Saarbrueckken
Karlsruhe
Mannheim
HPAs
, 1978
Ems s ions
Industry
14. 6
+"
(37)"
0.4
3.2
81.1
75.1
24.2
137.0
75.8
23.5
9.1
+
11.1
55.1
550
1,520
(thousand
Household
1.9
+
(4)
0.2
0.9
2.7
3.3
1.9
3.8
4.5
0.6
0.6
1.3
0.5
0.4
27
140
tons/yr )
Traffic
6.6
4.0
13.0
0.6
2.3
11.8
12.7
8.6
16.6
9.7
2.9
2.1
3.2
1.7
1.1
98
1,340
Total
for
HPA
23.1
+
(54)
1.2
6.4
95.6
91.1
34.7
157.4
90.0
27.0
11.8
(10)
13.3
56.6
675
3,000
Emissions
at
>100 m
10
+
+
(1)
45
35
8
60
24
(14)
1
+
7
37
(200)
(1,100)
"Data not available indicated by "+".
bNumbers in parentheses are estimates.
493
-------
TABLE 9b. N02 EMISSIONS IN OTHER EUROPEAN COUNTRIES
Emissions (thousand tons/yr)
HPA Industry Household Traffic
Belgium
Gent 20.4 1.0 5.3
Liege 13.8 0.8 3.5
The Netherlands
South Holland
North Holland
Utrecht
Oberi jssel
Denmark 96 66 64
West Germany 1,520 140 1,340
Total Estimates
for at
HPA >100 m
26.7 (10)a
18.0 4
105.3
62.0
45.6
20.0
226
3,000 (1,100)
'Numbers in parentheses are estimates.
494
-------
TABLE lOa. S02 EMISSIONS IN THE FRG
Region
Bayern
Berlin
Hessen
Nordrhein-
Westfalen
Rheinland-
Pfalz
Saarland
Baden-Wuert-
temberg
Total for all
Total for FRG
HPA
Nuremberg
Berlin
Frankfurt
Wetzlar
Wiesbaden
Dortmund
Duisburg
Duesseldorf
Essen
Cologne
Ludwigshafen
Mainz
Saarbrueckken
Karlsruhe
Mannheim
HPAs
, 1978
Emissions
Industry
20.8
+a
(75)b
0.7
5.9
112.3
193.4
29.1
147.5
75.8
32.8
13.1
+
60.3
63.9
1,050
.3,000
5 (thousand 1
Household
5.6
+
(10)
0.4
2.7
12.7
15.1
5.0
17.6
18.5
1.4
1.2
4.7
1.9
1.5
100
450
:ons/yr )
Traffic
0.4
0.3
0.7
0.1
0.2
1.3
1.9
0.5
1.3
1.5
0.2
0.1
0.4
0.4
0.2
10
100
Total
for
HPA
26.7
+
(86)
1.2
8.8
126.3
210.4
34.6
311.9
167.5
34.4
14.4
(10)
62.6
65.6
1,160
3,550
Emissions
at
>100 m
20
+
+
-
(1)
80
100
8
230
45
12
1
+
44
42
(600)
(2,500)
"Data not available indicated by "+".
"Numbers in parentheses are estimates.
495
-------
TABLE lOb. S02 EMISSIONS IN OTHER EUROPEAN COUNTRIES
Emissions (thousand tons/yr)
HPA
Belgium
Gent
Liege
Total for
Belgium
Denmark
West Germany
Industry
130.2
41.3
690
295
3,000
Household
5.7
3.6
103
172
450
Traffic
0.7
0.3
16
5
100
Total
for
HPA
136.6
45.2
808
472
3,550
Emissions
at
>100 m
65
25
u.a.
(2,500)
496
-------
TABLE lla. ORGANIC COMPOUNDS IN THE FRG
Region
Bayern
Berlin
Hessen
Nordrhein-
Westfalen
Rheinland-
Pfalz
Saarland
Baden-Wuert-
temberg
Sum HPAs
FRG (1978)
HPA
Nuremberg
Berlin
Frankfurt
Wetzlar
Wiesbaden
Dortmund
Duisburg
Duesseldorf
Essen
Cologne
Ludwigshafen
Mainz
Saarbrueckken
Karlsruhe
Mannheim
Emissions
Industry
6.7
+"
(18)"
0.5
9.4
4.4
150.7
2.7
29.3
84.1
12.5
2.4
+
9.4
6.9
340
480
(thousand
Household
1.1
+
(22)
0.1
1.0
4.9
12.4
3.8
9.4
6.3
1.0
0.5
2.1
0.3
0.3
65
630
tons/yr )
Traffic
5.2
4.5
9.1
0.5
2.1
10.2
9.1
6.4
13.7
4.8
2.1
1.4
3.4
2.3
1.9
77
650
Total
for
HPA
13.0
+
(49)
1.1
12.5
19.5
172.2
12.9
52.4
95.2
15.6
4.3
(6)
12.0
9.1
480
1,750
Emissions
at
>100 ra
1
+
+
-
(0.9)
(0.2)
0.5
0.1
0.1
1.4
0.1
-
+
0.1
0.1
4
(13)
"Data not available indicated by "+".
bNumbers in parentheses are estimates.
497
-------
TABLE lib. ORGANIC COMPOUNDS IN OTHER EUROPEAN COUNTRIES
Emissions (thousand tons/yr
HPA Industry Household Traffic
Belgium
Gent 6.2 0.3 1.9
Liege 5.3 0.1 1.6
Total for
Belgium 12 0.4 4
The Netherlands
South Holland
North Holland
Utrecht
Gelderland
Oberijssel
Total for The Netherlands
West Germany 480 630 650
Total Emissions
for at
HPA >100 m
8.4 (l)a
7.0 (1)
16 2
130
88.8
38.7
87.3
64.0
409
1,750 (13)
"Numbers in parentheses are estimates.
498
-------
TABLE 12. HYDROCARBONS AND HALOGENATED HYDROCARBON EMISSIONS8 IN FRG
(FIGURES IN BRACKETS ARE ESTIMATES)
Rheinland-
Pfalz
Industry
Region
Bayern
Berlin
Hessen
Nordrhein-
Westfalen
HPA
Nuremberg
Berlin
Frankfurt
Wetzlar
Wiesbaden
Dortmund
Duisburg
Duesseldorf
Essen
Cologne
HCs at
HCs >100 m
6.0 +a
++» ++
17.4 ++
0.3
3.0 +
1.8 +
145.0 +
0.5 +
19.3 +
56.1 +
Halogenated
HCs
3.1
++
3.7
0.2
1.7
0.2
1.5
0.4
2.0
5.7
Halogenated
HCs > 100 m
+
++
+!
:
Ludwigshafen
Mainz
4.8
1.5
1.8
0.2
Saarland
Saarbrueckken
Baden-Wuert
temberg
Karlsruhe
Mannheim
8.3
3.6
0.1
0.2
0.6
Total for all HPAs
270
33
"Emission in thousand tons per year.
bData not available are indicated by "+".
cData not available from publication are indicated by
499
-------
SUMMARY
Detailed emissions inventories are already available or will shortly be
available (within this or the next year) for about 3.8% of the FRG, which
includes about 28% of all German inhabitants.
During detailed investigations of the emissions inventories of the three
source categories—industrial; domestic heating, trade, and small industry, and
traffic—emissions data for more than 1,000 substances were collected.
Emissions inventories that have been published and are already available
for 15 regions indicate that:
• More than 680,000 tons of NOX are emitted per year (81% from industry,
4% from domestic heating, and 15% from traffic sources); this is about
22% of the total NOX emissions per year for the FRG.
• Nearly 0.5 million tons of organic substances are emitted per year (70%
from industry, 14% from households, and 16% from traffic sources); this
is 27% of the total organic compound emissions for the FRG.
• About 270,000 tons of HCs and about 22,000 tons of halogenated HCs are
emitted per year due to industrial emissions.
Although 37% of the nitrogen compounds are emitted at heights of over 100 m,
less than 1% of the organic compounds are emitted at such heights.
REFERENCES
Bayerisches Staatsministerium fuer Landesentwicklung und Umweltfragen. 1983.
Emissionskataster Erlangen/Fuerth/Nuernberg (Emissions Inventory—
Nuernberg).
500
-------
Dreyhaupt, F. J., et al. 1979. Handbuch zur Aufstellung von
Luftreinhalteplaenen, Koeln (Manual for the Establishing of Clean Air
Plans). Verlag Tuev Rheinland.
Duewel, L., and 0. J. Zuedorf. 1974. Emissionen luftfremder Stoffe aus
Industriebetrieben, Koeln (Emissions of Air Pollutants from Industrial
Plants). Umweltschutz Band 4, Verlag TUEV Rheinland.
Emissions Inventory—Frankfurt. 1983. Preliminary information.
Senator fuer Stadtentwicklung und Umweltschutz. 1981. Emissionskataster
Kraftfahrzeugverkehr Berlin-Innenstadt (Emissions Inventory/Traffic—
Berlin).
Lindackers, K. H., et al. 1971. Erhebung und katastermaessige Dokumentation
der Emissionen luftfremder Stoffe in die Atmosphaere—Aufbau und Auswertung
des Emissionskatasters, Koeln (Collection and Documentation of Emissions of
Air Pollutants into the Atmosphere—Establishment and Evaluation of
Emissions Inventories).
Luftforurening i Danmark, Emission og luttkvalitet. 1980. Miljo Projekter.
Hessischer Minister fuer Landesentwicklung, Umwelt, Landwirtschaft und Forsten
des Landes Hessen. 1981. Luttreinhalteplan Rhein—Main (Clean Air
Plan-Wiesbaden).
Hessischer Minister fuer Landesentwicklung, Umwelt, Landwirtschaft und Forsten
des Landes Hessen. 1982. Luftreinhalteplan Wetzlar (Clean Air
Plan—Wetzlar).
May, H., and Plassmann, E. 1973. Abgasemissionen von Kraftfahrzeugen in
Grossstaedten und industriellen Ballungsgebieten, Koeln (Pollutant
Emissions from Motor Vehicles in Metropolitan and Industrial Areas).
Umweltschutz Band 3, Verlag TUEV Rheinland.
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1972. Emissionskataster Koeln (Emissions Inventory—Cologne).
Minister fuer Arbeit, Gesundheit und Sozialordnung des Landes Baden—
Wuerttemberg. 1980. Emissionskataster Mannheim/Karlsruhe (Emissions
Inventory—Mannheim/Karlsruhe).
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein—Westfalen.
1982. Luftreinhalteplan Rheinschiene Mitte, Ducsseldorf 1982-1986 (Clean
Air Plan—Duesseldorf).
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1976. Luftreinhalteplan Rheinschiene Sued—Koeln, 1977-1981 (Clean Air
Plan—Cologne).
501
-------
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1980. Luftreinhalteplan Ruhrgebiet Mitte—Essen, 1980-1984 (Clean Air
Plan—Essen).
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1978. Luftreinhalteplan Ruhrgebiet Ost—Dortmund, 1979-1983 (Clean Air
Plan—Dortmund).
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1977. Luftreinhalteplan Ruhrgebiet West—Duisburg - Oberhausen - Muelheim,
1978-1982 (Clean Air Plan—Duisburg).
Minister fuer Arbeit, Gesundheit und Soziales des Landes Nordrhein-Westfalen.
1975. Luftverunreinigungen im Raum Duisburg—Oberhausen—Muelheim
(Emissions Inventory - Duisburg).
Minister fuer Landwirtschaft und Umwelt des Landes Hessen. 1978.
Emissionskataster Rhein-Main, Wiesbaden-Mainspitze (Emissions
Inventory—Wiesbaden).
Minister fuer Soziales, Gesundheit und Sport des Landes Rheinland-Pfalz. 1976.
Emissionskataster Frankenthal/Pfalz (Emissions Inventory—Frankenthal).
Minister fuer Soziales, Gesundheit und Sport des Landes Rheinland-Pfalz. 1977.
Emissionskataster Ludwigshafen/Rhein (Emissions Inventory—Ludwigshafen) .
Minister fuer Soziales, Gesundheit und Sport des Landes Rheinland-Pfalz. 1978.
Emissionskataster Raum Mainz (Emissions Inventory—Mainz).
Minister fuer Soziales, Gesundheit und Sport des Landes Rheinland-Pfalz. 1980.
Luftreinhalteplan Ludwigshafen/Frankenthal, 1979-1984 (Clean Air
Plan—Ludwigshafen).
Minister fuer Soziales, Gesundheit und Umwelt des Landes Rheinland-Pfalz. 1982.
Luftreinhalteplan Mainz—Budenheim (Clean Air Plan—Mainz).
Minister fuer Umwelt, Raumordnung und Bauwesen. 1978. Emissionskataster
Hausbrand/Kleingewerbe fuer die Raeume Dillingen—Saarlouis—Voelklingen—
Saarbruecken—Neunkirchen (Emissions Inventory, Domestic Heating and Small
Industry—Saarland).
Minister fuer Umwelt, Raumordnung und Bauwesen. 1978. Emissionskataster
Kraftfahrzeugverkehr fuer die Raeume Dillingen—Saarlouis—Voelklingen—
Saarbruecken—Neunkirchen (Emissions Inventory/Traffic—Saarland).
North Atlantic Treaty Organization. Committee on the Challenges of a Modern
Society. 1979. Air Pollution Emissions Inventory Systems. Document
No. 104.
502
-------
North Atlantic Treaty Organization. Committee on the Challenges of a Modern
Society. Emissions Inventory in Belgium, Canada, Federal Republic of
Germany, The Netherlands, Norway, United States of America (Appendices A
to F).
Plassmann, E., and 0. J. Zuendorf. 1977. Emissionen Luftfremder Stoffe aus
Hausbrandfeuerstaetten und Kleingewerbebetrieben. (Emissions of Air
Pollutants from Domestic Heating and Small Trade Installations).
Umweltschutz Band 5, Verlag TUEV Rheinland.
Services du Premier Ministre. Programmation de la Politique Scientifique.
1981. Enregistrement des emissions (1981). Rapport Scientifique Final
(1978-1981), No. 1 (1981).
Sources et Niveaux de la Pollution de I1Air et Impact sur 1'Environnement;
Application a la Region de Gand (Gent). Vol. A: Enregistrement des
Emissions. 1981. Rapport Scientitique Final (1978-1981), No. 18A.
Sources et niveaux de la pollution de I1air et impact sur 1'environnement:
Application a la region de Liege. Vol. A: Enregistrement des Emissions.
1981. Rapport Scientifique Final (1978-1981), No. 19A.
503
-------
APPENDIX A. GUIDELINES FOR EMISSION INVENTORY PRESENTATIONS
Data base name/source
Reasons for inventory development?
Who collects the raw data? (private
industry, national/provincial government,
etc.)
How is raw data collected?
(questionnaire, permit system, inspection,
other)
How frequent are data updated?
Are updates legally required?
List legal or confidentiality restrictions
which may prevent release of the data
Area of coverage
Coordinate system
Point source information; define a point
source
A) Raw data collected:
List stack, information
List major contributing source
categories (industries)
List types of raw data collected and
temporal resolution where appropriate
Spatial resolution
Dates of available data
B) Emission estimates:
List pollutant species
Temporal resolution of calculated
emissions
List information available for temporal
apportionment
List percentage of emissions estimated
bv following methods:
Standard emission factors with
specific plant information
Nonstandard emission factors with
specific plant information
Source test
Material balance
Other, specify
What emission factors, if any, are used?
List publication describing emission
factor development program.
Various emissions inventories in FRC.
Air quality management, clean air plans.
Independent experts bv order of lander
administrations (provincial) or plant operators.
Individual inspection and/or emissions
declaration, licensing procedures, emission
factors.
Industry, yearly; all others; every 5 yr.
Yes, legal regulations (
Restricted for use bv local authorities or lander
administrations onlv.
Total approx. 10.000 km2 (indiv. 507..50 km2).
Causs-Kruger grid (rectangular grid)
Single source with great emission, e.g., large
stacks
height, diameters, flow rate, temperature,
chemical, petrochemical, metal industry
, S02-emissions from fuel consumption and
sulfur content: 1 h.
1 m, if necessary.
Previous vear.
S02. NO. NO;. CO. HC. Pb. and some 1.000 others.
One hour.
Daily distribution of traffic, operating rate of
day, week, vear.
Very rough estimates.
20%
30%
20%
20%
10%
Traffic, domestic heating, small industry
(corresponding to standard plants).
504
-------
Are reported emissions controlled or
uncontrolled?
Are control equipment and efficiency
information available?
Describe method of estimating volatile
organic compound emissions
Area source information; define an area
source
A) Raw data collected
List major contributing source
categories
List subclasses of stationary area
mobile sources
List types of raw data collected,
spatial and temporal resolution where
appropriate
Dates of available data
B) Emission estimates:
List pollutant species
Temporal resolution of calculated
emissions
List information available for temporal
apportionment
Describe grid system or spatial
resolution
List information available for spatial
apportionment
Are published standard procedures used
for temporal and spatial
allocation and emission calculations?
If yes, list major references
Describe method of estimating volatile
organic compound emissions
Partly controlled.
For individual point sources, only.
APTI—formula, e.g., tanks breathing.
For example. 1,000 m x 1,000 m (domestic
heating, traffic if not as main street
described)
Traffic, domestic heating, small industry,
leakages of chemical/petrochemical plants.
Motorcycles, trucks, vessels, aircraft.
Type, size, and spatial distribution of
domestic heating, fuel sales, local
distribution, and frequency of traffic:
1 km, 1 h if necessary.
1970-1982, depending on area of investigation.
SO;, CO. NO. N03. HC. particulates. others
1 h i£ required.
Annual, seasonal., weekly, daily, hourly
distributions.
Longitude/Latitude in meters.
Housing location, street.
Publication 280 of
Regulation
API
505
-------
General
Comment on completeness
Comment on currentness
Summarize quality assurance program
Who is responsible for data quality?
Are source inventory data handled manually
or by computer?
Attach detailed record formats
The completeness of SO; is better than
of NO and N02 (mostly NO, as NO;). Point
sources better than area sources.
Point source data are mostly 1 yr old.
Area source up to 5 yr old.
According to updating procedure
(emissions declarations). Control and
comparison of data by lander administration.
German lander administration.
All data handled on computers.
506
-------
APPENDIX B. GUIDELINES FOR AEROMETRIC DATA PRESENTATIONS
Surface Air Quality
Data base name/source
Area of coverage
Total number of monitoring sites
Spatial distribution (attach site map)
Year of record
Check available site information:
physical location (lat-long, UTM)
geographic location (state/province/
elevation (MSL, AG)
classification (i.e., urban, rural
suburban, remote)
environment of site
descriptive information
dominating influence (i.e., industrial,
residential, mobile)
other, specify:
Parameters (attach table of measured
parameters, associated equipment
type/analvsis method and temporal
resolution)
Upper Air Quality
Data base name/source
Spatial distribution (attach standard
of flight paths)
Year/date of record
Parameters (attach table of measured
parameters, associated equipment
type/analysis method and temporal
resolution)
Spatial resolution
Surface Meteorology
Data base name/source
Area of coverage
Total number of stations
Spatial distribution (attach site map)
Site information available
Parameters measured
Time interval of measurements
Year/date of record
NOAA
20° - 55°N, 65° - 130-W
See map.
Standard airways observation
Hourlv
Aug. 1979; June, July, Aug. 1980
507
-------
Upper Air Meteorology
Data base name/source
Area of coverage
Total number and type of stations
Spatial distribution (attach site map)
Site information available
Year/date of record
Rawinsonde release
parameters measured
frequency of release
maximum level
Pibal release
parameters measured
frequency of release
maximum level
Others, specify:
I NOAA
i 20" - 55"N, 65' - 130"W
See attached map
Aug. 1979; June. JuLv. Aug. 1980
0000, 12; 0600. some; 1800. special
Standard RAOBs
12 hours - some 6 hours
500 mbar
508
-------
SESSION IV
MODEL EVALUATION
April 13, 1983
Session IV consisted of panel discussions on six selected models. These
presentations are summarized in Session V.
509
-------
SESSION V
MODEL EVALUATION
INDIVIDUAL DISCUSSION GROUPS
April 14, 1983
511
-------
STEM MODEL
P. Builtjes
Yesterday, we discussed the STEM model by Gregory Carmichael. Very briefly, it
is an Eulerian grid model used to calculate oxidants over an area of, sav,
1,000 x 1,000 km2, with the emphasis on vertical resolution. The model has
about 10 or 11 layers in the vertical direction.
As I understand it, the model has four objectives. In order of importance, the
main objective is that the model be used to provide a detailed depiction of
field studies. That is the reason the model has about 10 or 11 layers, so that
you can clearly see and follow plumes, their mixing, and their interaction with
the surroundings.
Another objective is that the model be used to determine the influence of
regional scales on global scales. That is one reason why the model has a larger
height than the other models, a height of about 8 km, with the troposphere more
or less.
The third objective is to use the model to investigate specific processes. So,
you can use it as a kind of learning model, with the purpose of incorporating
processes if they appear to be of importance in a simpler model.
Finally, you can also use the model for control stratrgv purposes. However, I
think the main objectives are the three I have just mentioned.
The STEM model requires a considerably larger computer memory than the Eulerian
SAI model or Lamb model.
Different modules of the model, have been tested, but that does not mean that the
total model is fully operational. It has been estimated that it will take 3 yr
before the model can be made fully operational, that is, if the iunding is
available to do so.
A major problem with this kind of model, as with the Lamb model, is the
incorporation of clouds, that is, the real dynamic direction inside the model
itself with clouds.
The difference between the Carmichael model and the Lamb model is that the
Carmichael model is fully Eulerian. The Lamb model has a special treatment of
interchange between vertical layers, but you can sav it is not fully Eulerian.
In our panel meeting, we also concluded that there will be a problem in the way
in which these models have to be validated, especially due to the large grid
size used in this kind of Eulerian model.
We also discussed whether a region of about a 1,000 x 1.000 km? exists in Europe
with sufficient input data, say emissions and aerometric data, to validate the
models. Although there was not complete agreement about whether such an area
512
-------
exists, my opinion is that it could be done. There are difficulties, but data
are available.
Whether it is necessary to incorporate certain phenomena into the Carmichael
model and to what level of sophistication the phenomena can be incorporated
cannot yet be decided, because no real sensitivity runs and no validations runs
have been made. So, we do not know exactly how sophisticated the treatment of
clouds should be, whether it is necessary, and whether it has any influence in a
real validation approach.
Let's summarize by considering the objectives of the workshop.
The first item concerns the state of the art of existing emissions inventories
and the plan to complete these inventories as model input data. Emissions
inventories already exist for the corridor units in the United States, which are
complete between (marks), and they are available in principle for parts of
Europe, also between (marks).
We did not discuss the second item, regarding the refinement of the best
available models.
The third item involved the merits of selected models as the bases for
developing control strategies. As I said, the Carmichael model can be used for
control strategies, but the major objectives are different.
The fourth item, the chemistry, was not discussed, and there is no
recommendation for the chemistry. There are several chemical mechanisms. Of
course, the model is modular. You can take chemistry out or leave it in if you
like.
The fifth point was to examine whether the aerometric data base, including
boundary condition data, is sufficient and, if not, to formulate a plan to
obtain the missing data. I do not know about the United States, but in the
European situation we mainly lack HC measurements, field measurements of HCs,
especially somewhere along the boundaries of your areas.
The sixth point was to determine when the model will be fully operational. As I
said, about 3 yr from now, 1986.
A. Galli: Is that when it will be a validated model or just available for use?
P. Builtjes: It will be a validated model around then.
B. Dimitrjades: I thought the question was whether the model would be
operational.
P. Builtjes; It was stated as fully operational, and we interpreted the term
"fully operational" to mean whether we could use the model to develop control
strategies and whether it could be validated.
The final point involved defining the modeling domain in terms of the available
emissions and aerometric data, the topography, and the model's capability. As
513
-------
far as we are concerned, the domain should be the northwestern part of Europe,
because most of the available input data are for that area.
DISCUSSION
E. Runca; Regarding the validation of the model, it is not clear to me what
data are necessary. Could you comment on that?
P. Builtjes; Perhaps Gregory Carmichael should comment.
G. Carmichael; Although we tried to discuss that at the end of the session
yesterday, we really did not come up with a concrete assessment of a validation
plan for such models. We have talked about the validation procedure, but I do
not have a detailed plan for that procedure.
We mentioned some of the difficulties in validating large-scale models, implying
that, with a plan, you have look at data requirements and whether there are
enough observational data to really verify field study data, and we pointed out
deficiencies in the HC measurements as one area of concern. In terms of a
detailed plan, that is included in the 3-yr period for the detailed verification
plan. Until we have a better idea of how to verify large-scale regional models,
we need that plan first and then an assessment of—now with an understanding of
your data bases, then we can make a better assessment of that. I think it would
be premature for me to speculate.
A. Galli: What do you need to make your model operational? Do you need
meteorological data, chemistry data?
G. Carmichael; Do you mean to exercise the model?
A. Galli: Right, what is implicit in the model? What do you already have
there, and what is needed?
G. Carmichael: If you are really doing an event assessment and it is a
three-dimensional model, you need good vertical resolution. So, the best test
would be under conditions where you have as much upper air data as possible, so
that you can specify the transport as well as possible. That would mean
conditions where you not only have as high a density of upper air data as
possible but also a frequency of upper air data at least four time a day for
such, and you may only have data for twice a day.
So, some of the deficiencies are that. If you select a certain time period,
there may have been field studies that have been conducted that there are data
for that time period.
A. Galli; Do you need anything else besides upper air data?
514
-------
G. Carmichael: You need land use data, but you can presumably get that, and
surface roughness and emissions data. If you select the proper area, there is
apparently adequate emissions data for HCs and NOX.
A. Galli: So the key is the upper air?
G. Carmichael: Well, that is one. You also want to specify the transport as
well as possible. This model takes the raw data through a preprocessing.
Whether it is done by our preprocesses or other people's preprocesses, you end
up with a description of the met field.
If you are really operating the model and looking at it from the standpoint of
whether you want to look at the effects of clouds, etc., there is a real
question of how you get those data, how you locate the clouds in the region, and
whether those data are available for the Europeans. This is an area where we
have ideas and other people have other ideas, but there is no good way of doing
it or no implemented way of doing it.
E. Runca; Where would you put the emphasis for your concentration—HC
measurements, oxidant measurements to distinguish between 03 and N02?
G. Carmichael: In this type of model where you are looking at many different
species, you would obviously want 03 measurements and all the species that
significantly influence 03 measurements. You would want NO, N02, and HC
measurements as much as possible. It depends on what level. If you do a total
validation, you obviously want as much data and related species data as
possible, so that you can tell whether you are predicting 03 correctly, how you
are predicting the other species related to 03, or whether you are predicting
all of the relationships in a similar biased fashion.
D. Jost: Are the HC measurements going to be used for testing the output of the
model or do you need other input in addition to the emissions?
G. Carmichael; You want it from many standpoints. First, it gives you a better
idea of boundary conditions. Wherever you put this region, you need some idea
of boundary conditions. Since you are predicting HCs, you also want to test the
model on how it does the HCs, their influence, some of the photochemistry.
G. Whitten; Your report indicates that part of the validation procedures has
already occurred in that you tested the numerical scheme for solving the
chemistry package. In my judgment, the numerical scheme does not work very
well. Is there some consideration of using in the next 3 yr alternate numerical
schemes for the chemistry-solving package?
G. Carmichael; I do not agree with your statement. In what sense is the
chemistry scheme calculation—what are you basing that observation on?
G. Whitten; The results presented in your paper indicate that, as you function
the time, the present scheme diverges from the base case. In such a case, for
application to a mesoscale over a period of several days, you want something
that is more stable over long time periods and numerical algorithms that can be
better performed. I was just wondering if that was of concern to you?
515
-------
G. Carmichael; I still do not understand what you are talking about.
G. Whitten; You had a figure in your report that showed various time steps
being used for the chemistry.
G. Carmichael: CSMP.
G. Whitten; What did you mean by CSMP?
G. Carmichael: We used a stiff integrator, in this case in the package.
G. Whitten; What is CSMP?
G. Carmichael; It is a simulation package. It's a stiff solver, so we'll take
that to be an analytical—we'll take that to be the correct solution.
If I can clarify. If you compared the integration of the transport species of
the chemical species over a 5-h period by using that technique, you would get
the types of divergence, 10% or so after 5 h. However, our transport step
involves 15-min intervals, and this chemistry is embedded in those transports
only over 15-min intervals at 0.1 rain. So, when you locate each individual time
step from those initial conditions, you have very good agreement between those
at that time scale, that resolution.
G. Whitten: My experience as a chemical modeler is that it is very important to
follow a chemical scheme over a period of several days, and to integrate over
15-min intervals even over many days. And that diverged from the STEM solving
package, I found it to be of considerable importance and of some concern.
There are other methods of solving the chemistry package that do not diverge
from the stiff package.
D. Jost: Could we come to this question again in case it would be necesary for
the further procedure after this meeting? If people feel that this question
needs to be clarified now, then we could come back it. Would you both agree?
G. Carmichael; Yes.
G. Whitten: Yes.
B. Dimitriades: I just want to bring to the conferees' attention the
possibility for confusion here. I want to go back to the comment about an
operational model and full validation. It seems to me the point is when a model
will be ready to be used by OECD in the subsequent evaluation program. By that
time, the model should be fully operational, validated, verified, or what?
Perhaps someone could define this for us. What is fully validated? What is
fully operational? What is evaluated and what is verified? This may cause some
confusion because I hear that the model should be fully validated in time to be
used by OECD in another validation program.
D. Jost; Several people are asking for the floor. Perhaps I should attempt to
answer this.
516
-------
S. Reynolds; As part of any model application at this stage, epecially for
regional models, any modeling study has to include an evaluation step before
going into a control strategy analysis, and I think it is really incumbent upon
the model users to demonstrate that the model is working adequately, whatever
that means. So, it would be helpful if there was some previous experience.
Nevertheless, it will be incumbent upon the user to provide evidence in the
specific context, in this case of whatever the OECD application would be, that
in Europe the model is operating properly. So that first and the application
subsequent.
B. Dimitriades; You mean to do this with field data, I presume? A small-scale
evaluation of the model?
S. Reynolds; Certainly some kind of evaluation. Ideally, some special
measurement study might be mounted, given that existing data have not been
collected, with the objective of evaluating or validating a model. It is
possible to use the existing data to get an idea of how well the model might
work, and that might be a first evaluation step, perhaps pointing the direction
to further field studies that might provide a more comprehensive evaluation,
perhaps a little further down the road in time.
B. Dimitriades: Perhaps Mr. Lieben could speak?
P. Lieben; Yes, I support this interpretation. What we are trying to do in
OECD is to arrive at some later stage where we can recommend to member countries
one model or several models that can be used for developing and implementing
control strategies. That is the goal.
Having reviewed the situation during these 3 days, we now have to determine what
needs to be done before we arrive at that point. The question is simple, but
the answer is not so simple. That is what I would think is a goal in order to
present to the member countries a model upon which they can rely. T would not
say it would be absolutely perfect, but it would give sufficiently reliable
answers needed for control strategies. What we have to do now is the work to be
able to arrive at that point.
D. Jost: This includes some steps for validation and evaluation, and it depends
on the state of the art of those models for which steps have already been made,
whether or not they have already been validated or evaluated. As for the
question, against which data should those evaluations be done, it is mostly
atmospheric data, but this is not necessary. Other possibilities could be given
against which such an evaluation could be done.
A. Christie; Are we perhaps oversimplifying this? In terms of the complexity
of the models being discussed, you can tune the model to fit almost any data
base that is available. So, it is fairly obvious that any kind of model
development is going to involve each of these models. Each of these models is
developed according to what will occur.
It is also fairly obvious that each of these models will have to have a separate
evaluation and data base, which is probably a great deal finer than what will be
517
-------
available for evaluating the final total model. If you are going to put in more
than one complex study in chemistry, you are probably going to need an
evaluation set for the chemistry module, an evaluation set for the advection
modules, and something that is going to evaluate whether you have cloud
processes. Then, you can start looking at an evaluation data set.
If each person talks about evaluating a model in a different way with a system
that tunes the model to what he/she is suggesting as an evaluation set, that
really is not going to be a proper evaluation at all. I think the whole
question is posed in much too general a way.
The development of a model of this kind is going to involve whole strategies of
evaluation between the cost of putting all these separate modules together.
A. Galli: I look at this from a slightly different standpoint; that is, that
Dr. Builtjes essentially said that the Carmichael model had not been validated,
that essentially no sensitivity work has been run on the model, and that it
would roughly be 3 yr before the thing would be completed for any real use for
control strategies.
Regardless of what scheme is utilized to validate test sensitivity, the time
frame that we are looking at in this whole control strategies program is 2 to
3 yr before the thing is set up, completed, and capable of giving someone
something.
If the model is not going to be ready for 3 yr, we have obviously got a problem
in the time frame that is being set up to begin with. As we look at each of the
models, I think we are going to see some inherent problems, either from a
technical standpoint of where they are in their development, not that their
development is wrong or that their purpose is wrong, but that they are off in
the time frame that we have set up, perhaps in OECD. As we go through some of
these things, that will become very obvious to us, regardless of the technical
plans that are being put forth. The bottom line is that where they are in the
development is going to be a self-limiting factor as far as their use for the
project, regardless of whether they are scientifically sound or unsound. It is
a little hard to sit down, tear down, and build up a model when it is really not
ready and it is not going to be ready in the time frame that you need it in.
D. Jost: That is right; I think this time frame is important. Nevertheless,
although it is the object and OECD could resolve and propose some strategies,
one needs to be aware of the development of these models that will be used in
the OECD project and that they and will not be the final end in development in
this field.
518
-------
SAI MODEL
H. van Dop
The next model is the SAI model. The core of the model is Eulerian. It
contains more than one layer and it has vertical resolution, parameters that are
essential for regional oxidant modeling. It is suitable for use on a regional
scale, that is, a scale of 1,000 km x 1,000 km. The grid size is limited not
only by the intrinsic model properties but also by the availability of data.
The meteorological data are in fact critical in this sense. We think that the
grid size should be as small as possible. It depends on where the regional
application is. For the United States, the size would be of little use if it
were less than 50 km, because there would no longer be any resolution of the
meteorological data.
The model is to be used in the development of control strategies. Therefore, I
think that the model should discriminate between high and low sources. I think
the SAI model fulfills this requirement.
As for the meteorology, the meteorology put into the SAI model, as with the
other Eulerian models that were presented this morning, is to be prescribed more
or less. This is an important point. If it is not prescribed or contained by
the model construction itself, the model cannot be used in an application. The
performance of a dispersion model can only be judged together with the model
input fields. I think it is essential to prescribe how meteorological data are
obtained and processed for these models. Without that, the model is incomplete.
Another point is the availability of meteorological data. The availability of
meteorological data depends on the area concerned. In the U.S. and Canada, the
density of the meteorological network is more or less critical. The data are
too sparse in most regions to .construct a detailed windfield and turbulence,
which are required for this model. In large parts of Europe, the meteorological
data are sufficiently dense in time and in space.
As for the chemistry, there was no consensus on the chemical submodel to be
used. In fact, we had the same problem with chemistry as we had with
meteorology. The modeler does not prescribe a particular system; he/she makes a
recommendation. If the user does not like the recommendation, the modeler is
willing to replace it. Like the problem of meteorological data, I think this is
important. In my opinion, a photochemical oxidant dispersion model should
contain both routines for chemistry and preparation of meteorological fields.
A few chemical models were discussed and the consensus was that there is not one
good model and one bad model. They all give different results. A study could
perhaps provide some answers as to which chemical subsystem should be
recommended for use in regional oxidant modeling.
A separate discussion developed on the individual grid scale phenomena mentioned
by Robert Lamb. Nobody knows exactly what the effects of subgrid concentration
519
-------
fluctuation will be. It presumably plays an important role, and it should be
looked into further.
As for emissions inventories, the mesh size for emissions inventories should not
be too small and should be consistent with other mesh sizes used elsewhere, for
instance 50 km x 50 km or somewhat less.
I would like to make a separate point for Europe and the U.S. In Europe, a
large emissions inventory, the OECD emissions inventory, exists. It is, in our
opinion, too coarse for regional oxidant modeling. It consists of grid cells of
127 km x 127 km. This is too coarse for photochemical modeling studies. A
large number of emissions inventories exist on smaller scales, on urban scales
or on scales far less than 1,000 km x 1,000 km, which could be integrated for
regional modeling, but that is not yet the case. In general, these small-scale
emissions inventories are of fairly good quality. They contain a lot of
components and fine resolution, and they are categorized accordingly. So, the
quality is not too bad. The problem is the integration of these inventories.
In the U.S., there is at least one data base on the right scale
(1,000 km x 1,000 km). I am not quite sure whether it contains all the
necessary components such as HCs.
Further, it was felt that a good emissions inventory should also contain
background emissions, that is biogenic and natural emissions from outside the
area.
As for the operational properties, the SAI model is operational now. It is
"validated," which means the model has been compared with data. Some
correlations and statistical comparisons have been made, resulting in a
satisfactory correlation coefficient of approximately 0.7.
Our discussion indicated that it might be a good idea, as an extra aid in
validating a model, to have a .panel of experts judge all the "ins-and-outs" of a
particular model. The panel could then come to a consensus such as: "This
model meets the present standard of knowledge and does not contain erroneous
assumptions. As far as we know, it can be applied for the specified purpose."
To have such a panel review a model might be a useful institution in the
validation process of modeling.
DISCUSSION
S. Reynolds; I would just like to expand a couple of points that you made. The
first concerns meteorological input procedures and the idea that there is a lack
of consensus regarding a procedure or specific recommendation. We have had
occasion to apply the model in the U.S., and the procedures that we used with
the available data base in the U.S. for preparing the complete set of
meteorological inputs employed in the model do exist and reside with the model.
Any time you consider applying a model in a new situation, you need to look at
520
-------
the available data and the prevailing phenomena in that area and find the
procedures that best match the data and the phenomena, the procedures that will
give an adequate representation, interpretation, of the data in light of the
phenomena that you are trying to represent.
So, on the one hand you have procedures that are with the model. On the other
hand, you wish to scrutinize those procedures in terms of any new application to
see if they are indeed adequate, to determine whether other procedures might be
better in that particular setting.
With regard to chemistry mechanisms, we could make a recommendation on how the
chemistry should be treated. There are also other investigators who are dealing
with chemistry, who have other opinions on how chemistry might be treated, just
as meteorologists have different opinions as to how certain meteorological
phenomena should be handled. So, on the one hand we have suggestions as to how
the chemistry should be treated, and these are embedded in the model. On the
other hand, we recognize that other people have their own opinions as to how
these should be handled. Chemistry packages can be interchanged in the models
or modified; they are not fixed in time.
J. Killus; To add to the comment concerning chemistry, we have specifically
designed the current chemistry module in the SAI model to be both
computationally efficient and consistent with the current knowledge of
photochemical kinetics. If you wanted to replace that module with some other
chemistry, which in fact has been done other for test purposes, emission
studies, and other applications, some small additional effort would be required
to replace the module. However, in our view such an effort would result in a
considerable increase in the computational time required by the model. The
specific design criteria of these chemistry data are computationally efficient.
So, our recommendation for the chemical kinetics is perhaps considerably
stronger than our recommendations for the various meteorological input
preparation procedures.
E. Runca: I would like to have some further clarification on the way the model
describes the vertical processes in the atmosphere.
S. Reynolds: I am not sure I understand the question.
E. Runca; I would like to know how the model is treating vertical processes in
the boundary layer, especially exchanges between the mixing layer and the layer
above the mixing layer, and how the model is treating the processes that take
place during the night.
S. Reynolds: If you recall, there are 2-1/2 layers treated. Starting from the
top, there is an inversion layer; under that, there is a mixed layer and a
parameterized surface layer near the ground. We do consider exchange between
the layers as accounting for large-scale convergence-divergence effects,
entrainment, or detrainment, as layers rise and fall and as transport removal
occurs at the surface. Within the mixed layer and the inversion layer, however,
the pollutants are assumed to be well mixed.
521
-------
H. van Pop; That is why I brought the point up here, because the interface of
these layers is meteorologically determined. Since you mentioned this
relationship, it is essential to know how the layers behave in time and space
and how they are modeled.
E. Runca; Is the behavior in time and space of these layers computed in some
standard way in the model or is it an input to the model?
S. Reynolds; Well, it is completely determined by data inputs. The model does
not assume any particular spatial or temporal behavior. It is coming from the
data.
R. Lamb; I understood you to say that the grid size of the model should be
smaller than the resolution of the meteorological layers. Is that correct?
H. van Pop; Yes.
R. Lamb; I disagree with that. In this problem and the problem with oxidants,
two things are in effect going on, chemistry and transport diffusion. If you
have a limited amount of meteorological data, you have to treat the subgrid
scale flow variations and a diffusion. The other part of the chemistry, the
chemical reaction, the rate at which the reactions are going on, depends on
whether the materials have mixed and the amount of segregation of the materials.
H. van Pop; Of course, but the chemistry is a molecular process. To describe
the mixing of contaminants down to that scale, you would need a mesh size of a
few millimeters, rather than 10 km or 20 km.
R. Lamb; No, I am saying that the resolution of the concentration model should
be comparable to the spatial variations in the emission distribution. If the
emissions are isolated in various pockets and the pockets are separated by a
distance smaller than the resolution of the meteorology, you in effect premix
all the emissions from all these sources in the model because you lump them all
into one cell. For reactive materials such as 03, if you alter the NOX and HC
concentrations in as similar fashion, you are talking about large errors in the
prediction of 03.
So, we are right back to the subgrid issue. In my judgment, the necessary
resolution of the concentration model should also depend on the spatial
variations in the emissions for the region of interest.
H. van Pop: How long would it take before it will be mixed in boxes of
20 km x 20 km, for example?
R. Lamb; Although the chemistry and the mixing are coupled to some extent,
imagine a case in which you have two isolated sources within one grid cell. The
wind is blowing in a direction that makes the two plumes parallel to each other.
They are not mixing at all. However, when you put the emissions in the model,
you mix them because you put in all the emissions. You premix them when, in
fact, they are not mixed. So, the total 03 in the two plumes is "x"
concentration of 03 in each of the two parallel plumes. If you mix the
522
-------
emissions, you get an area average concentration that is less than, probably
more, than that.
H. van Dop: This discussion is perhaps somewhat technical. However, if you
have a windfield with a resolution of that same 20 km and if you have a
variation in flow direction with height, it is easy to get a deviation from the
mean flow direction of 20° from hour to hour or from level to level.
Consequently, the flow field will be such that, due to shear effects, complete
mixing will occur within a few hours.
R. Lamb; During convection in the daytime, there is not much shear in the mixed
layer. So, plumes tend to be long and slender.
H. van Dop; That is right.
R. Lamb: Some of the data that Henry Cole showed us for the New York area
indicate that there are quite narrow plumes, even from those cities. The
contours of 03 are quite high. So, if you lumped all the emissions in that area
into one large cell, you would probably not predict 03 levels comparable to what
they saw.
H. van Dop: What would you have tried as grid size?
D. Jost; Just to stay informed of what you are discussing, is there a subgrid
mechanism in this model? The subgrid may be for chemistry and for transport,
but mainly for chemistry.
R. Lamb; It is mainly for the chemistry. I am saying that the subgrid scale
problem increases in magnitude as the grid size increases. There is no good way
of parameterizing that in general. The only way to mitigate the problem is to
make the grid size small.
S. Reynolds: I agree with Dr. Lamb regarding the resolution of emissions. If
we look at other applications for photochemical models on the urban scale,
meteorological data aloft are relatively sparse. Yet, we tend to use relatively
fine emissions grids. You might push for finer grid resolution, which would
enable the discrimination between sources a bit better. So, from intended
applications you can perhaps see a little more in detail the characteristics in
the concentration field, but one is not looking to go very fine in scale. We
are perhaps talking about factors for two choices of grid size.
F. Vukovich; In theory, I think Bob Lamb is right. In practice, however, a
large problem occurs when you take meteorological data from stations that are
100 mi apart and try to apply them on a grid spacing of 18 to 20 km. I think
Bob would be the first one to agree that you are going to produce all kinds of
errors in wind field by some kind of interpolation-extrapolation techniques in
developing a wind field that you would use in a transport model. If you look at
some of Ed Lorenz1 work on predictability, you will find that tremendous errors
occur in your prediction field an hour after you have started predicting because
you have a misguided wind field. So what you have is a problem of theory versus
practice. It also depends on how you want to use your model. If you want to
use it as a diagnostic model, in terms of performing control strategies, then a
523
-------
20-km grid is all right. However, if you are going to make real-time
predictions with your model, then you have problems. A 100-mi station variation
between wind measurments, bringing it down to 20 km, is going to produce all
kinds of errors.
J. Killus: I might amplify Bob's statement by pointing out another phenomenon
that occurs during chemical reaction. There is a time constant associated with
various chemical processes in 03 formation, which is fairly long, several hours.
So, phenomena that occur on a shorter time scale, less than several hours, are
often more or less damped out by the time the 03 begins to build up through the
day. That does not mean that subgrid scale of less than 80 km do not occur. In
our simulation with 80-km grid cells, it was quite clear in some of these cases
where the station observation diverged from the model predictions. In fact,
there were some processes smaller than 80 km operating. Our solution to that is
in fact inherent in what we designed the model for, which was to generate a
large-scale 03 flux for, among other things, specification of boundary
conditions for much smaller and finer resolved models. In that situation, if
you had your two plumes that perhaps would not mix at the smaller resolution, we
would suggest that you apply a much more detailed model in the grid. In that
case, we have a specific situation that we have again offered in the
Philadelphia example, where we used an urban model with a resolution of 5 km,
much less than what we are talking about on a regional scale, but whose boundary
conditions were prescribed by a model that had a resolution of 100 km. The
Philadelphia results indicated that those local emissions generated local
concentration patterns. However, the regional transport from, say the New York
metropolitan area, perhaps only needed to be resolved at the 100-km level.
524
-------
HOV MODEL
E. Runca
As for the Hov Lagrangian model, we did not try to explicitly answer all of the
questions posed by the workshop objectives. We developed the discussion in such
a way to keep these questions under consideration, but we tried mostly to
identify the fundamental characteristics of the model, the applicability of the
model, and further actions that could be taken in order to make the model
operational.
In terms of the characteristics of the model, there are two fundamental
assumptions, which derive mostly from the fact that the model is Lagrangian and
receptor oriented. The model assumes no lateral diffusion and it is a one-layer
model. Therefore, there is no vertical resolution, and the concentration of the
advected pollutants is considered uniform in the layer. This assumption makes
the model simple on the one hand. On the other hand, it might considerably
affect the model's results.
We were not able to quantify this effect. There was a very strong
recommendation that the model be extended to include the contribution of the
pollutants that diffuse and are advected into the above mixing layer. This
implies something that I will leave to the people responsible for the model. I
think this problem can be solved in different ways. The question is whether one
of these ways will prove to be an effective description of the real processes.
The discussion of the model's characteristics focused on the way the model
treats the chemistry. Our conclusion is that the model can be applied with any
chemistry. Indeed, the chemistry is an interchangeable component of the model.
We also discussed which chemis.try could be recommended. Since the model can
treat different chemical schemes, the model could be run with different schemes
to see what results are obtained by applying these selected schemes. However,
we recommend that a scheme be found that reduces the number of species
considered and at the same time provides realistic results. On this point,
maybe some further comments could be obtained from Gary Whitten, a panel member
who strongly supported this suggestion.
Finally, our discussion focused on the model's applicability. In this context,
we directed our attention to the temporal resolution of the model. Not too much
discussion occurred on the spatial resolution.
The model, as I said, assumes no lateral diffusion. This assumption requires
that the grid size be sufficiently large. The grid size that is now used in the
model is 150 km. We were not really able to evaluate the effect of this
assumption on the model's results.
The discussion focused on the temporal resolution, and we did not reach a firm
conclusion on this point. Some further studies should be made in order to
establish whether the model can properly simulate hourly average concentrations.
525
-------
There was some general consensus that the model may be able to provide better
results if the time, the length of the averaging time, is extended. Considering
these aspects, it is questionable at this point what conclusions can be reached
on the application of the model for the development and evaluation of control
strategies. However, some remarks can be made in this regard.
First, the model is a simple model in some way. Therefore, it can be used to
analyze different options. At this stage, we do not expect that the model can
really provide a detailed diagnosis of the impact of adopting different control
strategies. However, it can provide some insight on control strategies. One
point that should be made in this regard is that the model is receptor oriented
and is a Lagrangian type. Therefore, it is in principle more suitable for
analyzing episodes than it is for analyzing the large-scale effects of control
strategies.
To evaluate the concentration of oxidants in a given point, we have to evaluate
trajectories that have reached that point. Therefore, we can evaluate the
effect of strategies on that specific point. If we want to understand the
effect of control strategies on a region, we have to run the model for many
points in the region. We also have to run the model for different time periods.
At this point, it becomes questionable whether it would not be more appropriate
to apply an Eulerian model. The panel tended to agree that, for the analysis of
the large-scale impact of control strategies applied on a regional scale, the
Eulerian and Lagrangian models might be compared.
The final issue we discussed was what further action should be taken to progress
with this model, to make the model operational. The model is now available, at
least the core of the model. However, a detailed sensitivity analysis is
necessary to evaluate the effect of the parameterization scheme that the model
adopts on the model results, that is, the assumptions the model has adopted in
relation to the treatment of clouds and other processes relevant to the
formation of oxidants.
Some priority should also be given to a validation study. Such a study
obviously requires data, and there are not sufficient monitoring data for
oxidants all over Europe to enable a model validation on the European scale.
However, a validation study can be undertaken by utilizing existing data sets
and also by performing field experiments.
For example, as far as the chemistry is concerned, smog chamber data are now
available from many different laboratories that could be utilized to get an
understanding of the chemistry model that is more suitable for this model.
Also, in order to evaluate some of the assumptions connected with the Lagrangian
treatment of the advection of the air parcel, it seems necessary to perform some
trajectory, to perform some field experiments to characterize the Lagrangian
advection of pollutants in Europe.
As for the next step, the application of the model, suitable emissions data
should be prepared in parallel with the sensitivity analysis and validation.
Considering the resolution of the model in space, the spatial resolution of the
526
-------
emissions inventory should be approximately 100 km. Since Eastern countries are
major contributors to oxidant formation during meteorological situations
occurring in Europe, the domain of the emissions inventory should be extended to
include these Eastern countries. If that is not possible, the frequency of
these conditions and the relevance of the meteorological conditions affected by
Eastern emissions should be evaluated before a program of model implementation.
Finally, we recommend that natural emissions be considered. In relation to the
evaluation of HC emissions, we recommend that some agreement be found between
the countries in defining emission factors.
I would like to conclude with a personal comment. It seems to me that a problem
of this size requires the coordinated effort of several European institutions,
and I would like to recommend that more emphasis be placed on the validation of
the model than on the application of the model.
DISCUSSION
R. van Aalst: We pointed out that a detailed chemistry treatment of HCs is
probably not quite suited to the detailed HC data in Europe. So, we pointed out
the lack of that as well. And about that one of the main reasons for reducing
the complexity of HCs.
A. Eliassen: Regarding Dr. Runca's presentation, we pointed out, first, that we
have no lateral diffusion, depending on travel time obviously. We have the
first, instantaneous diffusion, the grid size, but there are certain problems
with Lagrangian models that are more or less unavoidable. One of these problems
is that it is very difficult to have lateral diffusion, depending on the travel
time and the nonlinear chemistry. If you think about it, you will find out that
this is more or less impossible. So, this is a property of the approach that we
are taking, this should not be concealed in these discussions.
On the other hand, one might argue that to go to a grid size smaller than 150 km
is questionable if you account for two things. One is the availability of the
emissions data and the quality they have impressed in the meteorology. The
other is the uncertainty of the trajectory calculations. If you follow a
trajectory backwards for 4 days, the uncertainty is considerably larger than for
150 km.
The possibility of making the model suitable for calculating hourly
concentrations, for example, is rather small, because you need rather
sophisticated meteorology to calculate hourly concentrations. One would like to
have wind shear together with, for example, the diffusion for such effects. The
possibility of doing that with this approach is not good.
If you use a very short time scale, you have to calculate the degree of
trajectory that arrives very often at the receptor point. The computer time
required then approaches that of an Eulerian model. This also happens if you
use very long trajectories, because you do very nearly the same calculations
527
-------
many times following nearby trajectories. Then, the advantages of the
Lagrangian approach disappear.
As for the advantages in this calculation, you can have almost any chemistry
scheme that you can think of, and it can be run for quite a long time. We can
cover a much larger area than we are running at the present.
My last comment is on the need to have an exchange between the boundary layer
and, say, the free troposphere. This is also difficult because these have
different velocities on the boundary layer, and you get into trouble. It is not
impossible to solve, but it is rather cumbersome. It is much easier if you can
treat the troposphere as a very large reservoir with constant concentrations.
Then you can perhaps do some sensitive things.
E. Runca; I do not find any contradiction with what has been said, unless I
misunderstood your comment.
A. Venkatram: I would like to follow up your comment on validation. You said
we should do validation. We really do not know how to validate the model, do
we? It is very subjective at this stage. Maybe we should have some discussion
on the procedure or something of that sort. My personal feeling is that the
experts agree that you do not have to rely on statistics to determine whether
the model is good. We have been talking about a lot of models, but we have
forgotten that a lot of experience with oxidant modeling has been acquired over
the past 10 yr. Is there any hard evidence to suggest that these models do not
work? If they do not, have we identified the physical mechanisms that give them
errors?
E. Runca; The answers to your questions really require a discussion from the
audience. However, as my last recommendation, which is a personal
recommendation, I really believe that derives from the presentation and the
discussion we have had during the previous days. Models appear to be in the
research stage rather than in .the application stage. To progress further, the
coordination of different groups of researchers is required in such a wav that a
clearer strategy for the next step of model development and application can be
identified. This means, in my view, validating which types of modeling
approaches, according to various operational needs, should be applied.
G. Whitten; I would like to clarify a couple of points on things that I have
recommended in the chemistry. As for the chemical approach, you can take either
a broad approach, where you follow many primary compounds, or a more in-depth
approach, where you follow fewer compounds but where you follow them further
through their oxidation steps. The present chemical mechanism in the Hov model
is a broad approach. It follows very many primary species, but the chemical
steps are very short and some are left out. There are approaches where you have
fewer species but greater depth to them.
In evaluation procedures, a chemical mechanism can be validated out of the
atmosphere in a data base that exists in smog chambers, and this is a
recommendation that you pointed out. I think that another word in place of
"validation" could be used at this point for the atmospheric models. Perhaps
you could say that it is time for a "dress rehearsal," to go out and use the
528
-------
models as they now exist with the data base that now exists, to find out what is
missing. These models would not be used at the present time where you would
expect any regulations or control strategies that were performed to be
corrected. However, we need to find out how well the models can perform and, it'
they are not performing, what is missing.
J. Bottenheim: I strongly agree with what Venkatram says and with what Gary
says too, but I suspect that if we start to validate, every modeler will
accommodate his own model in his own way. In particular, if a validation shows
that the modelers results do not match his validation set, he will come up with
fudge factors for which there is more or less a physical reasoning and the
modeler will thereby improve the outcome of his model. So, if we get into this
validation business, it is very important to have a panel or an independent
review committee to assure that the validation is a realistic validation and not
a modeler just propping up his own model.
Secondly, I would like to ask Venkatram which models over the last 10 yr worked,
because I think that almost all oxidant models are urban models.
A. Venkatram; We, first, haven't decided by what means. My understanding is
that we know enough about the chemistry, that there is no substantial difference
between the chemical modules. That is one important component of the oxidant
model.
The second point is we have been making a lot of points about subgrid-scale
chemistry and other phenomena. Is there any evidence to suggest that these will
really screw up the results? Hard evidence?
E. Runca: Maybe I should explain myself. When I say that the validation
program should be a coordinated effort, I am implying that the validation should
not be just an exercise of running the model and comparing the results with some
measured data. It is just a problem of research coordinated between different
institutes with different expertise to assess which processes are important and
what we can say in relation to some of the issues mentioned so far. I really
think that this should have a strong priority in Europe before moving to the
implementation and application phase.
D. Jost: I too got the impression that it is not as easy as it may be inferred
from your comment. Even the different chemistry models are so different that
you may come up with very different results with such models. Therefore, we are
not at the stage where we can take what is available and rely on it.
G. Whitten; One problem that exists with the chemistry modules right now is
that, in many cases, a given data set can be simulated with several different
chemistry modules. However, if you then apply control strategies using
different chemistry modules, they predict different control strategies.
The research that is now underway is investigating why one chemical mechanism
gives different control strategies, and we are at the stage of arguing about
parts of the chemical mechanism that lead to different control strategies. So,
the fact that you may validate on a given data set is one thing, but when you
529
-------
apply the control strategies that these models will be used for, you produce a
new problem.
J. Bottenheim; One case in point is trying to validate a chemical mechanism
through smoke chambers. Everybody knows that smoke chambers have problems, and
different modelers use different chemical mechanisms, different fudge factors,
to take into account what the smoke chamber does to the chemistry. And all the
chemical models work pretty well! But, have all the chemical models really been
tested from the same smoke chambers with the same fudge factor?
J. Killus: I want to explain what Gary was describing and disagree with Jan
Bottonheim somewhat. There are definitely certain chemical mechanisms which,
when applied to smog chambers or certain sets of smog chamber data, do not
replicate the smog chamber data, no matter what fudge factor you use. We have
various examples of that, and that is part of what Gary was saying when we try
to locate what specific features of kinetic mechanisms are responsible for these
differences in control strategies.
Part of the problem is that kinetic mechanisms have been in a state of intensive
development for the past 10 to 15 yr. The lead time necessary for publication
of these mechanisms is such that, by the time the mechanism is published, it is
more or less obsolete. Mechanisms that are available to people who do not
actually design and develop them are usually 4 or 5 yr out of date and may
contain certain known errors in fundamental chemistry that have been corrected
in more recent mechanisms, errors that might produce some differences in control
strategies.
We have found that the differences in control strategies are in fact much less
than one would expect from the differences in fundamental chemistry, that is,
mechanisms that have even fairly large errors can still often predict similar
control strategies within a factor of 20% or 25% of other mechanisms. That 25%
difference in HC control can, of course, mean many hundreds of millions of
dollars, which is not an irresponsible issue. However, the difference is
similar or much smaller to the differences that are going to occur in
atmospheric modeling methodology.
In agreement with Venkatram, I would say that most of the mechanisms of recent
vintage are reasonable. Properly applied in the atmosphere, they will give more
or less proper results. The difficulty is proper application. The various
assumptions that one must make in applying these for emission inventories, which
may be improperly speciated or may not be speciated at all, will overwhelm the
differences in kinetic mechanisms.
My own feeling is that more of a difference exists in kinetic mechanisms in
their application state than in the actual fundamental chemistry. Some
mechanisms are much easier to use. Some of the parameters associated with these
mechanisms are much easier to specify, and that is perhaps what Gary was talking
about.
530
-------
LAMB/NOVAK MODEL
F.B. Smith
Panel III discussed Bob Lamb's and Joan Novak's U.S. EPA model. To summarize
the basic nature of the model, it is an Eulerian-type model with a grid length
of about 18 km. There might be a good case for reducing this grid length to
about 9 km, half of the present size, in order to develop a model that can be
used for some regional-scale and also urban-scale modeling.
In the vertical, there are four layers. The model does include a surface layer,
in which all of the interesting initial chemistry occurs. Mean vertical motions
are incorporated in the model, and these are derived from the horizontal
divergences of the windfield. There is also vertical and horizontal diffusion.
The horizontal diffusion takes into account the stability and depth of the
mixing layer in the usual way.
Advection depends upon the measured winds at the surrounding meteorological
stations. As Bob described, this is one area where there is a certain amount of
uncertainty, i.e., various trajectories can be derived according to the
different ways in which you interpolate between the measured winds and according
to the forms of advection.
We might want to discuss afterwards whether you might do better by incorporating
some sort of mesoscale meteorological model, into this whole system. Of course,
it would enlarge the model considerably; nevertheless, it might give you a
better feeling for the advecting winds.
A variable mixing depth is incorporated. Additionally, there is an actual
treatment for convective cloud. Cloud plays a very important role. If it is
sitting on the top of the mixing layer, various pollutants within the mixing
layer can be advected into the cloud, undergo chemistry there, and come out
again into the mixing layer. It can thus be a very important part of the whole
system.
The model also recognizes the effects of the surface terrain. The terrain has
an effect by producing slow vertical velocities through the divergences of the
wind field. It also affects the dry deposition of the various substances.
Because the model is directed towards rather extreme situations in which
precipitation does not likely occur, there is presently no provision for
representing the effects of wet removal by washout and rainout.
As Bob mentioned earlier, there are problems in representing the chemistry going
on within the individual plumes on a subgrid scale. Roughly 80% of the computer
time is involved in working out the chemical processes, and only 20% is involved
in the meteorology, the advection, etc.
As for the background to the model, which we discussed in our panel, the U.S.
has decided upon certain air quality standards for oxidants, certain maximum 1-h
531
-------
values that should not be exceeded. They have also developed possible
strategies for emission control. The aim is to find a model, such as this one,
that can determine by 1987 which of the various control strategies is the optium
one to apply. As we have already seen, the model involves the "best available"
physics and chemistry as it is presently known. Its modular form is very
developed; as our understanding of various processes improves, we can change the
various modules.
One very important aspect of the model is that it predicts the concentration
fields, recognizing the uncertainties in various processes, the uncertainties in
determining wind field, boundary layer height, etc. Because of these
uncertainties, a statistical aspect is introduced into the model. Thus, the
outcome is not just a single predicted concentration but a range of
concentrations with certain probabilities associated with them.
This is a very important, interesting, and novel aspect, which will enable us to
look at the model's sensitivity to possible errors. In addition, it will enable
us to decide the likely accuracy and certainty with which you can make
predictions and the possible certainty with which control strategies can be
applied. I think this is a very interesting and useful aspect of the model.
As to the time schedule, the model has been coded and is essentially ready to
run. There will be some computer runs this summer. The emissions inventory is
being developed under contract and should be available at the end of this year.
When that is available, Bob's team can run the model to see how it operates in
practice.
Next year, the model will be run under the same conditions as those that
occurred during two monthly field experiments in 1979 and 1980, when a great
deal of surface data were collected by balloon and by aircraft. The model will
be tested for these two monthly periods, and certain developments will obviously
result from this test. Having attained the second-generation stage for the
model, various proposed strategies will then be investigated in the following
2 yr.
The panel next discussed the appropriateness of the Lamb/Novak model for
application to OECD Europe. The first thing that came out of this discussion
was that Bob's team really does not have the capacity to revise the model or
apply the model to the European scene. To do so, some team in Europe would
presumably have to take the model over, refine it or revise it, find its wind
fields, etc.
In doing this, the team would be faced with the problem of emissions data. As
we have heard from Anton Eliassen and others, the emission situation in Europe
is not as good as it is in the United States, which brings us to the problem of
boundaries. One aspect of applying the model in the United States is that most
of the emissions, certainly as far as the Eastern United States is concerned,
are concentrated in the northeastern corridor. As a result, it is possible to
choose boundaries, as Bob has done, sufficiently far away from the main emission
areas so that, by using the measured surface concentrations in a relatively
simple way, you can estimate the influxes across the boundary without incurring
any great errors in the whole process. The same is not necessarily true in the
532
-------
European situation where the industrial belt runs from Central U.K. across
Northern France, Belgium, Holland, through the Ruhr, to East Germany, and
Poland, going east-west and covering a very large area. Consequently, there are
no really convenient boundaries to choose in Europe in contrast to the situation
in the U.S.
Also, the model is a very complex model. It is not intended for daily
operational use. It is really intended for testing and developing strategies.
This brings up several questions. What are we trying to do in Europe? What are
our objectives? Are they the same as in the U.S.? Are we going to consider
certain strategies for limiting oxidant levels or will there be other
objectives? If there are other objectives, this will affect the sort of model
that we will need to apply. If we are going to use the model in Europe, we will
really need an institute in Europe, either an existing institute or a newly
created one where a lot of effort could be devoted to a model of this kind, not
only to the model itself but also to the whole backup behind the model—what are
the objectives, how do you define them, what are the objectives based on, and so
on. So, I recommend that OECD consider setting up an institute for this sort of
work.
As an alternative, if we do not want to use a model of this kind and of this
complexity in Europe—if we want to develop our own simpler model—a model of
this kind could at least be used as a standard by which to test a simpler model.
Maybe we could take up the offer Basil made at the beginning of this meeting.
Maybe somebody from this European institute could come along and test the model
side-by-side with the selected U.S. complex model.
Finally, coming back to the point of what our aims in Europe, we have to
identify the actual hazards that we are particularly concerned with in Europe.
Are they hazards to health, to vegetation? What are the time scales involved?
Is it a question of an hourly oxidant level, a daily oxidant level, or an annual
level? Having decided upon the levels and time scales, which model is the most
appropriate for this sort of investigation?
DISCUSSION
S. Reynolds: When you state that the model provides an estimate of the
uncertainty, is that a direct calculation or the result of performing many
simulations in a sensitivity mode?
F. Smith; As I understand, it involves repeated running of the model using
different assumptions for some of the factors in the model. Is that right, Bob?
R. Lamb: It is primarily associated with the specification of flow fields,
which gets back to our earlier discussion about the resolution of meteorological
data. It is a rather complicated thing to describe in a few words.
Essentially, the flow fields cannot be described uniquely. There is a set of
possible flows, that is, a set of flows that are possible within the constraints
533
-------
of all the observations that we have in space and time and for physical
transport. Within all of these constraints, there are many possible flows.
We can assign probability for each of these flows on the basis of empirical
spectral data that have been seen for the flow field. In essence, the
uncertainty exists mainly in looking at the nonuniqueness of the flow field.
The uncertainty is associated with that and not so much the uncertainty due to
errors in emissions or errors in known factors. There are separate problems.
A. Christie: How much confidence do you have in the kind of data used to
indicate the degree of uncertainty, the probability of having any particular
flow distribution field? You have a spectrum of possibilities.
R. Lamb: Right.
A. Christie: Now, you say that you are going to do this from an analysis of
information on the flow fields. Do you feel that there is enough information at
the present time to carry out this analysis and that you would have the same
confidence in the flow analysis across North America as a whole, or would these
flow patterns incorporate mesoscale phenomena and terrain effects? Do the data
exist over the model domain to adequately define the probability of the flows at
all locations?
R. Lamb: With any given, the uncertainties in the flow would be very large,
even if there is only one source of meterological data, only one station.
Theoretically, if you are given a set of observations in space and time and if
you are given the physical principles that are used in the mesoscale flow
model—momentum, energy, mass conservations—you can delineate a set of flow
fields to whatever resolution you wish. There is a set of flow fields, and each
field is consistent with that information. Each of them obeys continuity,
momentum conservation, all the observations you have. Within that set of
possible flows, you can then take information such as the climatological data
that are used to get optimal interpolation formulas where you find weighting
functions that minimize in some climatological sense mean square error. We can
use climatological data or historical meteorological data as the empirical basis
on which to assign probability to the flows in this set. Some of these flows
may have no probability or vanishingly small probabilities on the basis of what
we have observed from the atmosphere in each individual region over long periods
of time. In our case, the empirical data base would then be used over about a
2-day period, where we have observations and space and time. We review these
empirical data to assign probabilities to all the flows that are possible during
that period, given the physical constraints.
In theory, we have worked this out and we are trying to implement it. We have
made some progress, but we have quite a way to go before we can look at just how
large the uncertainties are.
G. Whitten: It is not clear to me why some of the uncertainties that need to be
investigated, that are being built into this model, cannot be investigated with
existing models right now. For instance, uncertainties in the flow field could
be, and their effect could be tested now with other models to see how important
that would be when the Lamb model is ready.
534
-------
Another thing is the microscale mixing effects of parallel plumes. If you use
an existing model with a very fine mesh size in an area, you could make
hypothetical parallel fields and investigate the effects with existing models.
If they are found to be extemely important, they would have to be built into the
model when it is ready.
I do not see the necessity of waiting for this model to be ready and then tested
when a lot of these tests could be done at the present time.
D. Jost; As I understand it, there are some available models that are in some
respects very detailed. One would be able to check the sensitivity relative to
simulations with those models.
R. Lamb: It is not quite as simple as that. In testing the uncertainty due to
atmospheric flow, you can certainly take any model, drive it with different flow
fields, and look at the differences you get in the prediction. The problem is
defining a set of flow fields that is consistent with all you know. Unless you
are testing with the proper set of flow fields, the results you are getting will
not be applicable.
As to the effects of the grid resolution, the subgrid, could be tested to some
extent. However, tests for flow field uncertainty, which is in our case a major
consideration, will have to wait until we have developed a way of defining flow
fields and assigning them probabilities.
G. Whitten: The sensitivity to the flow fields could be tested. They might be
incorrect flow fields, but you could determine the sensitivity to them.
R. Lamb: Yes, but you may be using such an extravagant range of flow fields
that your results are very pessimistic, whereas in reality the situation is not
that bad. Or, you can take a flow field and not disturb them. There is an
entire range of variations of flow fields that you can put in and get a very
wide range of answers. If you can define that set of flow fields in advance,
then, yes, you can do that.
E. Runca; I am not questioning the reality; I am only commenting. I think that
this group made an important point, the need for a European institute for a
study of this type. I support that idea. We now recognize that many
environmental problems in Europe are not local, that they are problems for the
whole of Europe. It might be worthwhile to devote more attention to the
possibility of creating an institute to deal with these problems.
A. Venkatram; I would like to follow up on Bob's remarks about specifying the
velocity for looking at the sensitivity to concentration predictions to the
unresolved components of velocity. We are doing some preliminary studies on
that.
Instead of spending too much time on that, I would like to point out that if you
want to calculate concentration variance, you have to worry about not only this
so-called inherent variance, but also the variance associated with the model
inputs. You cannot predict what the meteorology will be next year; therefore,
your control strategy must account for that variance, which of course means that
535
-------
the total variance is going to overwhelm things. That is going to pale into
insignificance. Regulators have to start thinking in terms of probabilities; in
fact, it is an educational process.
S. Reynolds; Dr. Lamb, what is your schedule for doing the flow field
uncertainty work?
R. Lamb: The project is running somewhat parallel to the model development, and
we are hoping to do part of it within the next year. The main problem is this:
One often talks about using a mesoscale flow model to drive the concentration
models. In such cases, you incur a great amount of error as you initialize the
flow field model with the meteorology at some moment and as you use the
prediction of that model to go into the future, because the predictions of a
mesoscale model do not arbitrarily remain acceptable far into the future. In
our case, we are doing a simulation of a historical event. It is a so-called
"worse-case meteorology" in the past in which you have observations of what
happened for a time interval of 2 days.
In our approach, we have taken the equations and in effect put in all the
observations of time and space. We are now defining a set of functions that
satisfies all of those observations and all of those equations. Within that
set, we can then assign probabilities. This is a very complicated problem. It
requires working in high dimensional spaces, and it is a very time-consuming
problem for the computer. However, we are proceeding with it because we think
it is worth exploring. If it turns out to be impractical, we will just have to
face that.
A. Eliassen; It seems to me that you do not really know what sort of
probabilities of the front flow fields you will have in the future, say, if you
are going to construct control strategies. You will get certain situations and
you will assume certain probabilities in those situations according to the flow
field. What is going to happen next year and what sort of probabilities will
exist at that point is very difficult to say. The degree of the control
strategy will greatly depend on the flow functions. Then, I would be very
surprised if this is at all really necessary to consider.
R. Lamb: There is some misunderstanding about what I mean by probability. We
can discuss it later.
H. van Pop; I would like to remark on what Runca said a few minutes ago. We
are talking about regional oxidant modeling in the United States and in Europe.
In Europe, no place exists to carry out such studies, and I would like to hear
the participants' comments on that topic, because it is now clear to me that
only in the United States is there a platform for such studies. That is an
important conclusion if it is true.
D. Jost; I think it was presented here in a different way. If we want to apply
this model in Europe with all the SAT submodels, it needs a home in Europe that
is not yet available.
E. Runca; The question is a little bit more general than the way you are now
discussing it. I think we have reached a point where we recognize that in
536
-------
Europe there are many problems—not only this oxidant problem—which are of
interest to Europe and which in many aspects concern the European environment,
problems that are essentially multidisciplinary and require the coordination of
different groups of experts for their solution.
If we consider, for example, the long-range transport of sulfur, we realize that
the problem is extremely complex, because the development of a model that
describes this problem has to take into account so many factors, which really
requires the contribution of many groups of experts. It requires the synthesis
of the findings of these groups. So, there is really a need for a platform in
this sense, not only in coordinating, but also in taking these results and doing
additional research on these results in order to mesh them and create the
possibility for the analysis of results and their application to the evaluation
and development of control strategies in Europe.
My statement may be naive in terms of a model for consideration, but it seems
that a practical problem to be solved, in order to progress in Europe in dealing
with these environmental problems, is to find a home for these types of studies.
Sooner or later, we will have to consider the creation of an environmental
institute in Europe.
D. Jost; I did not want to get into this; I only wanted to explain what the
group meant. We could not propose an European environmental institute. The
group's proposal was directed toward the handling of these very models.
A. Galli; Another thing is perhaps not just limited to Bob's model, the Lamb
model. I am not sure that it does not exist if they choose to use the SAI model
or the Carmichael model.
All of these American models really do not have a home in Europe. Bob's simply
came to the front for use over there, but none of them really does.
S. Zwerver; I do not think this is the place to discuss it. We are discussing
technical questions of models. This is a problem for European countries. We
discuss with each other in international grous, like OECD or. I think that's
more of a platform to be decided by an international cooperation, the Europe
International Institute, as this here. I think it is good to say that in Europe
there is something missing in the technical coordination.
D. Jost; I wasn't going to decide here on the kind of cooperation in Europe,
but there is the recommendation for further coordination.
P. Lieben: Just a very short comment here. OECD is quite ready to assist in
developments and recommendations on which strategies should be adopted and so
on, but in the European context a lot of emissions are impacting the situation
in Eastern European countries. We have our normal boundaries in OECD. It is a
question of coordination not only among European countries, but also among
different international platforms that exist on the European scene.
D. Jost; That is right. One more question, and we will come to the next
presentation.
537
-------
Unidentified Speaker; Did your group discuss computer resource requirements for
a typical scenario, a control strategy scenario, using the Lamb model?
F. Smith: We did not actually discuss this in detail. Bob, would you like to
comment on that? You actually know.
R. Lamb: We have done some 24-h simulations with the NEDS data, the old
emissions data. For one 24-h simulation, the requirement was about 10 h of
UNIVAC time. As to how that compares with some other system, perhaps Joan
knows.
Unidentified Speaker; I am concerned about the probabilistic aspects of the
wind field model. That multiples it?
R. Lamb: Exactly, it is multiplied by a factor of 10.
J. Novak; In terms of the application of the flow fields, we intend to use
another computer, which we already have. With that computer, there is a
difference of about a factor of 10 in terms of performance, which puts the CPU
time for a typical 24-h simulation to around 1 h. This at least becomes
affordable in terms of multiple runs, say, the 50 to 100 runs that may be
required for a scenario. It is expensive, but you have to decide whether the
results you are getting are worth the price.
Unidentified Speaker; Do these model runs involve the preprocessing of the
data, the meteorological data base, or are they based on the assumption that
preprocessing has already been done?
J. Novak: These figures involve only the model time, not preprocessing time.
They are not comparable in terms of preprocessing. Possibly, the flow field
generation would be the most time consuming. The others could be several orders
of magnitude lower.
R. Lamb: If you consider the regional problem in all its complexity, this
problem is inherently one that requires a large amount of computer resources. I
mean, it is in the level of the climate modeling work, the large-scale global
simulations. We are out of the realm of the old plume model calculation where
we speak about seconds of CPU time. This problem has moved up many orders of
magnitude into a whole new range of problems if you treat it in this complexity.
If you decide that you do not want to pay the price of that and want to do
something simpler, then you can do that. The question then becomes: Are the
numbers you are getting really reliable enough to base strategy decisions on?
538
-------
U.K./ADOM MODEL
A. Eliassen
Panel IV discussed two models, the U.K. model, which was developed at Harwell,
and the Acid Deposition and Oxidant model (ADOM), which was presented by P.K.
Misra. The U.K. model is very complex as far as the chemistry is concerned—
40 emitted species and 300 reactions approximately. The meteorology is
considerably simpler. It involves instantaneous mixing into a box with the
dimensions of 450 km x 360 x 1,300 m in the vertical. It is operational, and it
has been used to compare predictions with measurements.
The ADOM model is a multilevel Eulerian model; the North American version has a
127-km horizontal resolution. It has dry and wet chemistry and deposition and
nine classes of HCs. It is modular in structure and it is comparable to the
Carmichael model. The ADOM model should be operational in 1986, when it will be
tested against historical data.
Our discussions concerned the uses, possible uses, and limitations of these two
models.
We will present the U.K. model first. In its present version, it does not
handle transport at all. Therefore, it cannot be used to develop oxidant
control strategies for OECD Europe.
Its complex chemistry might be used as a reference for less sophisticated
schemes. However, we felt that it would be necessary before that to test the
model against smog chamber data.
In addition, there are plans to further develop the model into a moving cell
model. That is a natural extension of this modeling approach, and it would make
the model more or less similar to the Norwegian model presented. If this
development is done, its complex chemistry would then require an emissions
inventory of the 40 emitted species over Europe, and we are far from actually
having those data. Also, the work involved in developing the moving cell model
would require approximately one person-year.
As to the ADOM model, it is being constructed specifically as a tool for
developing control strategies. The model is perhaps tilted slightly more
towards acid rain than oxidants, but that may not represent a great problem.
The meteorological input data for this Eulerian model are available in principle
for both North America and Europe. The data are more or less sufficient,
especially if we cooperate with the National Weather Service and related
institutions.
The emissions data are better for North America. In Europe, the situation
exists that I mentioned earlier. The quality of the available data is very
variable from country to country.
539
-------
The ADOM model is planned for use in both regions, so it may be of interest to
know that, regardless of what OECD decides, the model will more or less be used
in both regions anyway. So we could compare, of course, if OECD is choosing
another model.
Since the model is still under development, we did not really have a detailed
discussion on it. Instead, the group discussed so-called complex models.
To summarize, the simple models are useful for long-term averages, and they
often do quite well on those things, for example annual averages. The problem
arises when you want to use them for control strategies. Some of their
parameterizations are such that it is doubtful whether they respond correctly to
assumed emission variations. Thus, even if they calculate averages very well,
they are sometimes suspect for use in control strategies.
The complex models are more or less unavoidable. If you are interested in
short-term concentrations such as hourly concentrations, the shorter the
averaging time, the more complex the model required. Hopefully, these complex
models have more correct responses to assumed emission variations. Not being a
chemist, I wonder if the chemists' overview of their complex schemes is really
correct.
DISCUSSION
S. Reynolds: Could you clarify what you had in mind as a simple model versus a
complex model?
A. Eliassen; We more or less regarded the ADOM model as a complex model. The
U.K. model is a complex model.as far as the chemistry is concerned and a simple
model as far as the meteorology is concerned. The same characterization applies
perhaps to the Norwegian model.
G. Whitten: I would like to reiterate my comment on the chemistry of the U.K.
model, since it is very similar to the Hov model. Namely, it appears to be very
complex to a nonchemist. As a chemist, it is far too simplistic to me in that
it treats many compounds as primary compounds. The chemistry on each compound
is treated rather simplistically, so it lacks in-depth treatment. I would be
skeptical of its ability to handle long-term effects, because the chemistry of
each individual compound is not treated in much depth.
A. Eliassen: Perhaps another chemist could respond to that.
0. Hov: I am not a chemist, but I think that your comment applies more to our
Norwegian model than to the U.K. model, since the chemistry in our model as a
transport model is much more simplified than that in the U.K. model. I think
that the U.K. model's chemistry was up to date with the current literature at
the time of its construction, which was in 1978.
540
-------
J. Novak: I want to make a general observation. Several of the presentations
on the first day made reference to an interest in acid deposition modeling. It
might be interesting to do in terms of the ADOM model's ability to accomplish
both the acid deposition problem and the oxidant problem.
In terms of the success in this OECD effort, each representative has to carry
back to his/her country some kind of argument for funding this effort, some
argument that might give extra weight to obtaining support for funding and
commitment, because there is an attempt to solve two problems, especially the
acidification problem, which seems to be of prime concern in many countries.
A. Eliassen: It would surprise me if OECD once again initiated the program on
acid rain.
D. Jost: Perhaps not OECD, but the countries where the problem exists.
N. Laulainen: I'm Nels Laulainen from EPA, the Pacific Northwest Lab. I think
OECD is aware of this. There has been a proposal to compare acid rain
models—long-range transport models that deal with acid rain—in the future.
The U.S. position is that they would like to do this; however, the U.S. would
like to have such an effort wait until the completion of this oxidant
comparison. For one thing, we can learn something about it. For another,
rather extensive model development is occurring in the U.S. now, namely, the
development of a new Eulerian acid deposition model. We would like to have this
included in such a comparison, and it is not ready at this time.
In summary, we are interested in a comparison with acid rain models, but the
comparison would best wait until this is completed and until current development
efforts are completed.
C7. Whit ten: I support Ms. Novak's ADOM approach, combining the acid deposition
and oxidant. As a chemist, my understanding is that the acid deposition problem
depends very strongly on oxidant chemistry, and there have been a lot of
approaches to the acid deposition problem independent of the oxidant problem.
The two are much more closely related than approaches have been in the past. I
wanted to add my support to that.
N. Laulainen: The work that is being done in developing acid deposition models
does not include the type of models that you may be thinking of, the type that
tries to just treat sulfur chemistry by itself. The model that is being
developed in the acid rain program is quite a comprehensive model that includes
oxidant chemistry as well.
G. Whitten; Then that makes it an oxidant model as well.
N. Laulainen: It is, but the model is not ready at this time for comparison,
and there are some other models that are being developed that are not ready. We
would like to delay until they are ready for comparison.
G. Carmichael: I would just like to point out that this is the approach that we
have taken in the development of our model. We are looking at both the oxidants
as a part of the regional-scale cycle of trace gases in the atmosphere.
541
-------
A. Galli: As Dr. Eliassen indicated, the Misra model is going to proceed with
or without the support of OECD or the Control Strategies Program, and it is
going to be available for comparison. I do not see getting the acid rain
situation directly tied in with the control strategies project at the present
time, since the acid rain project as a whole is getting tied up in a number of
international organizations and is getting coverage there—even though, as far
as acid rain is concerned, OECD has perhaps been tied into an eva]nation of some
of the models.
If we stick to the models before us and the one that is going to proceed with or
without our blessings, we need to look at those besides the Misra model that may
or may not proceed or receive support. We had a considerable discussion about
the acid rain situation in our last AMPG meeting and decided to defer a lot of
this, even though oxidant work and acid rain work are very similar in modeling,
chemistry, and in a number of other areas such as emission inventories. We may
be getting far afield. I can appreciate people wanting to get heavily involved
in acid rain, but there are too many people involved in it right now. It is a
buzz word now rather than a rationality.
J. Killus; I would like to offer this observation. It is a fairly natural and
easy thing at a meeting like this—where you have many technical people
approaching a problem, where all of us have an opinion about the various
approaches and most of us have fairly strong personalities and strong
opinions—to slip into a competitive stance. You know, this mode] is better
than that or let's use this approach rather than that approach, but as this
conference has proceeded, most of that competitiveness is, at least in mv mind,
slipping away. I find nothing exclusionary in these various models or modeling
approaches.
There is nothing mutually exclusive about, say the Lamb model, the SAI model,
the Hov model, or being able to apply it in Europe. We will use whatever
emissions we can get. We will use whatever meteorological characterization we
can get, and that will be reflected in the model results.
There is nothing that says that the emissions grid used for the SAI model cannot
be used with the Lamb model. We will all be certainly requesting a 20-km
emissions grid if you can supply it to us. So, I would just say that I see no
reason why we cannot proceed now, and I think that that would probably be a good
idea.
R. van Aalst: I would like to make a very trivial comment. I liked a model
with simplistic meteorology and simplistic chemistry. If you are going to model
the long-range transport of oxidants in Europe and if you start out in a
diagnostic way, which would very much be the way to start in Europe, I would
advise the use of extremely simple chemistry in terms of modeling oxidant
production and oxidant destruction, and just give it a diurnal cycle or
something like that and try to relate that to the amount of emissions in a grid.
It is just as simple as that to see what is going on and to analyze the results
in terms of such simple parameters. Probably nobody proposes such an approach
because everybody can do it. 1 would like to urge that such an approach be
taken.
542
-------
SESSION VI
CONCLUSIONS AND RECOMMENDATIONS
April 14, 1983
543
-------
CONCLUSIONS AND RECOMMENDATIONS
P. Lieben, Chairman
We are now arriving at the end of the workshop and trying to put everything
together. During the next 2-1/2 h, I would like for us to consider what OECD
should do after we have reviewed the situation. I would like for us to consider
the objective to be achieved, the subjects to be commented on, and the facts
that are not known.
The objective is for OECD to recommend to member governments within the near
future—not 10 yr from now, but within 2 or 3 yr—regional model(s) that they
can use to develop and implement control strategies for chemical oxidants.
I have stated this in very simple terms. I am not an expert, and I am not
familiar with all the details that are involved in the chemistry or the
meteorology. The question is: What should be done to meet this objective?
First, we will certainly have to assess the effects of oxidants. Do they affect
human beings, crops, vegetation, or materials? This question certainly has a
bearing on the control strategies that will be chosen by member governments. In
Europe, most member governments have not yet decided which control strategies to
adopt. They may have some idea, but it is not fixed.
Second, we will have to look at the models and make some kind of selection.
This does not imply that some models are good and others are not. We will have
to select one, two, or three models that appear to be the best candidates for
meeting the objective of developing control strategies. We will have to
concentrate on these models, and this will be part of the work plan that we will
try to develop. We will certainly have to look at the capabilities of the
models, the time frame for availability, the necessary input data, the costs,
and possibly other factors. These factors are not meant to be restrictive; they
are simply pointers to direct the discussion.
The third item is the development of emissions inventories, especially for
Europe where they are largely missing. Emissions inventories will certainly
have to include NOX; they will also have to include HCs, and we will have to get
some ideas about the species to track.
One very important point is the grid resolution for such inventories. It will
have to fit with the model(s) that are likely to be selected for control
strategy purposes. We have heard that getting such inventories is a major
effort. I would not like, after 2 yr of effort, to finish with an emissions
inventory that is done in such a grid resolution that it cannot be used with the
model we finally select.
I think the domain covered by the models that are likely to be selected is also
important and will certainly require 03 and meteorological data. It is probably
less important than the emissions inventories.
544
-------
So, this is a very rough, tentative agenda for developing a work plan. I see
these three activities going on in parallel, but converging at some point. Tn
other words, there will be some point where we are ready with the three
different elements, where we can put everything together and make a
recommendation that constitutes our objective.
Do you agree with structuring the discussion in this fashion this afternoon? I
do not think that we will achieve a definite answer on everything now, but we
can perhaps get from the meeting as clear an indication as possible of the major
elements that have to go into such a work plan.
I should also give you some idea of what is coming in the future. We have a
commitment to prepare a work plan for the special session of the OECD Air
Management Policy Group in mid-June. We have about half a month to write the
plan and to present it at the meeting for discussion with those countries that
are willing to go ahead with it, countries that are willing to put forth the
necessary effort.
This will be done in mid-June; then we will have the regular meeting. We will
have to review the plan in terms of the willingness of member countries to go
ahead with it, according to the possibilities they have. In September, we will
have the regular meeting of the Air Management Policy Group to finally adopt a
plan for implementation.
Are there any questions about the proposed structure for the discussion and the
outline for a work plan?
G. Whitten: To add a slightly different complexion to the discussion, we can
also consider as part of the plan whether you want to concentrate on a single
model, a single type of model, or a battery of models. Our recent experience is
that, if you put together an inventory on a grid scale, you can use a grid
model, a trajectory model, and a box model from that inventory. Using all three
models at once is cost-effective and time-effective in that you learn much about
the chemistries, about the adequacy of the specjation and the amounts there,
just by running a simple box model. It tells you things very rapidly. So, I am
suggesting that a plan of the types of models to use all at once be part of the
plan.
P. Lieben: That is a very relevant point. I was expecting some discussion on
my point that the selection of models for further work within OECD is
understood, that models will progress and that people will go ahead with
developing models, etc., but that we have to select for OECD purposes what will
suit best. That is certainly something to look al.
Unidentified Speaker: If we are going to get into assessing the various models
that have been presented, it may be good to reach an agreement here as to what
such a model needs to have inside of it, what it needs to predict, and what time
scale it needs in order to qualify as a strategy-development oxidant model. It
may be good to determine by what standard a particular model does or does not
qualify before you even get into the various models. It seems to me that the
various routes may have different criteria within their discussion as to what is
an oxidant model for strategy development.
545
-------
P. Lieben: Yes, I think this will be part of the discussion covered under the
second item. What I am trying to do now is establish these three points as
areas for discussion and take them one after the other. When we come to the
discussion about models, your comment is certainly valid.
E. Runca: Can you specify better "assessment effects of oxidants"? What will
this imply? Is it just a review of what has been done so far?
P. Lieben: This leads us to the first point, which I would like Lars Lindau to
introduce.
L. Lindau: As has been said many times during this workshop, the time averages
are very important. We ask that for the models and also for the discussion of
strategies. In talking about health effects, we have 1-h averages. We have
both a recommendation from WHO and we have the U.S. EPA standards. Thus, I do
not think there is much need for a discussion of the normal time averages when
we are talking about health effects.
As for the effects on crops, vegetation, forests, etc., we have short-time
averages. To my knowledge, there is also information about the effects over
longer time periods for both crops and forests. I have divided these up. Of
course, it could be different when you are talking about vegetation. Whereas,
when you are talking about forests, you have to talk about the needle life, and
that means not just the (summer, half-year); it could mean a couple of years.
Needle life is 4 yr.
In a draft proposal made by Swedish scientists about a half a year ago on
criteria for determining the effects of chemical oxidants on vegetation, a 1-h
average, an 8-h average, a 1-mo average, and a 1/2-yr average were used.
Galli proposed and I am also proposing that we have to assess the effects at the
beginning of this project for OECD. OECD and the OECD countries must have a
base of knowledge, so they will know which effects they are looking for. This
base could be developed in several different ways. Experts could get together
to assess the existing literature, or it could be done on a bigger scale. I
have not thought the issue over very thoroughly, but there is a lot of
information on the effects of chemicals in the literature. If you use
scientists from a couple of countries, they could come up with a conclusion, and
we could get some information that would be very useful for future work.
P. Lieben: Are there any comments on the issue of effect? This is an important
point in establishing priorities and developing control strategies. You have to
know what you are supposed to protect and how. Are there any comments on that
particular point?
E. Runca: Is there any plan to undertake some projects in Europe to quantify in
some way the effects of oxidants on crops, forests, etc.?
P. Lieben; I am not aware of any plan like that. The issue before us now, an
assessment of effects, will certainly involve looking at what has been
published, what has been put together, and the studies that will have to be made
in time to extract a sense of the rest.
546
-------
In addition to that, will it be necessary to conduct some special studies? I do
not know. We do not have any plan to do so in OECD. We are just looking at
which plan we should have.
E. Runca: Should we try to make some proposals now?
P. Lieben: I would like to repeat what I said at the beginning. I am not
trying to fix anything, but to get some ideas from the meeting of the elements
that we can take into account when developing this work plan.
J. Novak: One thing that would need to be determined is how you are going to
evaluate the control strategies, and this has something to do with the effects
of 03. If you run these models with different emission scenarios and different
control strategies, then a choice has to be made as to which is the best and the
most cost-effective model. In order to make that decision, you have to weigh
the levels of effectiveness you have with the different control strategies
against the cost of implementing them and the costs to humans, vegetation,
crops, whatever.
So, it seems that some type of economic analysis would be worthwhile in which a
functional relationship is developed between 03 levels and crop damage or
economic loss—however you wanted to define it in terms of the effects of 03.
Thus, given a function of economic loss due to 03 concentration and the cost of
controlling that 03, you can determine, at least on a more scientific basic,
what the optimum control strategies might be in terms of cost and effect.
P. Leiben: Thank you. Are there comments on that point?
B. Thompson; A number of studies were done in the United States on the economic
effects of oxidants. Boyce Thompson Institute did several of these studies.
They found a strong synergism with other pollutants. So, it is not just
oxidants at work; it is oxidants plus the 02. If you look further, it gets even
more complicated than that.
Beyond the first order of effects, it could therefore be a question of ever
increasing complexity, and it could end up in a situation in which the acid rain
problems are being talked about now. It might be because some pollutant causes
some metal ion to mobilize and whatnot. It could turn out to be a Pandora's
box.
P. Lieben; It may be.
D. Balsillie: There is a network across the United States called the National
Crop Loss Assessment Network, which is systematically trying to find out what
the economic loss is on crops across the U.S. It turns out that the crop loss
is very substantial. As Joan Novak pointed out, there is nowhere near the
amount of crop loss in terms of costs, is nowhere near the billions of dollars
it would cost to abate the problem. In place of this, the U.S. raised its
standard from 80 ppb to a 120 ppb, knowing that they would have to accept that
level of damage.
547
-------
In Ontario, we are losing about 15 to 20 million dollars worth of crops a year,
based on the 1980 averages. However, 15 to 20 million dollars does not come
close to the amount of money we would have to spend to reduce the VOC emissions
from our two major areas, the Windsor corridor and the Golden Horseshoe around
Toronto, Ontario. So, there is going to have to be something more than just the
vegetation losses put together. In other words, I think you are going to have
to come up with a health effect as well as a vegetation effect, because I do not
think that we are going to be able to justify the cost of abatement based on
vegetation damage.
A number of people in Europe are working on the effect of 03 on plants, and they
are quite knowledgeable in this area. They have been in the United States on
several occasions to take part in meetings such as this, so they are aware of
what is happening on both sides of the Atlantic. If you will look into it, you
will find that such information is available for Europe. Whether it is actually
documented in dollars and cents or some other currency, I do not know.
P. Lieben: As to the assessment of effects on crops, that was never the
intention. I think the assessment covers a number of effects in order to judge
which one(s) we will try to reduce, and this will determine which control
strategies are adopted.
J. Schjoldager; I think the question of plant damage is important. Along with
that and as early as possible, we should include in any plan a strong
recommendation that countries carry out measurement programs to find out what
the concentration levels of oxidants actually are.
Knowing what a slow process it really is from your plan to carry out
measurements until you have a program and a good program operational. It is
important to urge both countries and research institutions to put this into
operation as early as possible.
P. Lieben: Thank you. Are there any more comments on this issue? I know that
is not the point that we intended to discuss. I do not think we have to prolong
it.
P. Grennfelt: Before we proceed with our discussion of the models, I want to
point out that it is essential that the models will present (sounds) that fit
into the work to estimate damage.
P. Lieben; I am not sure I understand.
P. Grennfelt: There are many different ways in which biologists use data to
estimate effects. We do not know exactly what type of data they really want to
get in estimating the effects. I think it is necessary to really get the models
that the biologists define what type of data they need so they can put together
what would allow this analysis factor to reach abatement. The models discussed
until now have very much focused on control strategies without considering the
issue of effect.
548
-------
P. Lieben; Okay, you think at some stage biologists should be brought together
with analysts and control people to have them discuss and agree on a way to
provide it.
E. Runca: I would like to emphasize this point. These three items—the
effects, the model, and the emissions inventory—and the measurement of other
data should be the components of a problem. However, these three components
should not proceed independently. There should be an overall framework, which
is continuously verified. Information from these three components could then be
used for the evaluation of control strategies.
S. Zwerver: I get the feeling we are moving to conclusions concerning other
fields and fields that we have not discussed in the last two days. We were
discussing models.
We have received a suggestion about cost-benefit analysis and effects, and we
have just discussed effects as an influence on the models to be selected. So,
are we going to select a model for a long average, or are we going to select a
model to (produce values)?
I should warn for all these other discussion because they are very complex, and
I don't think we have the expertise there to—
P. Lieben; I was going to move on to the second point. If you remember, it
concerns the preselection of models or advice about which models are the best
candidates to focus on for the next 2 to 3 yr in the OECD program, in view of
our objective. I will ask Dieter Jost to introduce that subject, and we will
have comments afterwards.
D. Jost: After the discussion this morning, perhaps more aspects more technical
and scientific than the models that have been presented here. I think now I
(prescribe to some of our discussion panel) the (apparent) recommendation based
on this technical and scientific discussion as a result of these discussions and
the report of the chairman of this panel.
I would like to describe some of the more administrative aspects of the
discussion, which could be a basis for further discussion within OECD. I have
taken the models that were discussed and considered some questions on this. I
have already tried to answer part of these and I would like to ask you to
correct this. You may have additional items that could be taken as the basis
for later decisions.
This should include the Hov model, Lamb model, Misra model, SAI model, and
Carmichael model. These models are operational to some time, which was
mentioned this morning. I did not find out when the Hov model will be
operational?
549
-------
0. Hov; The model has been applied, as published in the Journal of Applied
Meteorology, for an episode at the Norwegian southeast coast, and a number of
sensitivity studies have been carried out. So, it is operational.
D. Jost: I think this is true. It may have been 1982 or 1983, but it does not
matter
Unidentified Speaker; Is the Carmichael model pretty much running right now?
G. Carmichael: As pointed out, when we discussed the term "operational," we
were talking about operational as a verified model to field study. We can
compute numbers at this point in time, and it has all the pieces. We plan to
apply it to some field study data next year. So, it depends on how you look at
the term "operational."
B. Dimitriades; Perhaps we should be asking in what year the model will be
turned over to OECD.
D. Jost: By operational, I meant available for people who are not the authors
of the models. Should the date be moved to 1986?
G. Carmichael: Yes, 1985-1986. We plan for a users' manual to be available
then.
R. van Aalst; I strongly recommend that the Hov model be viewed as a two-layer
structure. I do not know whether this should be included in the definition of
the operation or condition needed for OECD. Could someone comment on that?
D. Jost: Perhaps I could answer. This takes us to the last point, what we can
do with these models. As it now is available, the model is a one-layer model.
It could, nevertheless, be used for some tests. This would be the model as it
was available already in 1982, without the recommended changes.
Perhaps we could come back to this third point. You can use it for strategies,
which is the purpose of this project.
This morning I got the impression that the model would be used for studies in
which you want to analyze air quality situations or special effects on oxidant
formation. Perhaps this may be changed due to the input data needed for the
model. Let's come back to this line a little later.
There have been some estimates on the effort needed for running the models.
This was estimated for the Lamb model, but I do not remember the numbers. As I
am not a computer specialist, I would like to ask which unit should be taken.
H. van Pop: Make three classes: low, medium, and high.
D. Jost: If I remember correctly, the effort is high for the Lamb model. What
is it with respect to the Hov model?
0. Hov: That depends on how many changes are required.
550
-------
D. Jost: Does that mean high, as it is now?
0. Hov: As it is now, it is low.
D. Jost: Is it medium or high for the Misra model?
P. Misra: High.
D. Jost: The modeler will not be high, but the input. I do not know whether
"high" has a different inference in your language. The SAI model?
G. Whitten: Medium.
D. Jost: And the Carmichael model, high. As to evaluation, this is no longer
planned since an evaluation has already been done.
A. Christie: However, it has not been evaluated in terms of model resolution, a
criticism made this morning.
D. Jost: The model has been evaluated as far as it is available, so the
evaluations that you are mentioning are not possible.
P. Lieben: Maybe Mr. Hov has a comment on that.
0. Hov: If the model is to be evaluated according to a set of recommendations
made at this meeting, that has not been done of course. The only model
evaluation has been a comparison with a measured period of time at one site in
southern Norway.
E. Runca: In the final discussion, there was some agreement that studies to
evaluate the model are needed. So far, it has been mostly a qualitative
analysis comparing with not only one single station. (in terms of describing)
the trends of the process, I don't think it can be concluded that the evaluation
is completed. I think it needs some further (management). I cannot evaluate
how long this will take because (I am not around the model).
0. Hov: One might say that the sensitivity of the model tolerates realistic
changes in model input parameters like deposition velocities, emission field,
transport direction, wind speed, as it has been performed for this one episode,
but the need clearly exists to perform more sensitivity studies, such as a study
on the effect on the very simple way horizontal diffusion is included as well as
vertical mixing. There should be additional sensitivity studies to acquire a
feeling for the sensitivity of the model towards realistic changes in the model
process.
D. Jost; We will take note of this.
0. Hov: As to plans for evaluation, it might be sufficient to consider some
sensitivity studies that might be done this year or in 1984.
D. Jost: How about the other models?
551
-------
P. Misra: Some evaluation will have been done for the ADOM model by 1986.
D. Jost: The SAI model?
S. Reynolds: We have conducted evaluations on portions of the model. I say
"portions" because we have conducted these studies in the context of SO2 and
sulfate, but that uses transport. So, the physical process is part of the
oxidant model. A study has been done and a report of it will be appearing in
Atmospheric Environment. Also, we presented some results at this meeting for an
8-day application that was done in the Northeastern United States. This
application is part of an evaluation going on at this time.
D. Jost: Thank you. Is it being done for SOX?
G. Whitten; For oxidants as well. It was presented at this meeting.
D. Jost; Then I misunderstood.
G. Whitten: Yes, the oxidant results were presented here; the sulfate results
are being published.
D. Jost: So the evaluation for your model will be done in 1984.
G. Carmichael: We will be doing some calculations of field studies in 1984, but
the entire verification will be operational in 1986.
D. Jost: This would be understandable from those numbers. With respect to
input data, I think the emissions data will be handled later on, so we will not
look at them at this moment. We will limit ourselves to the meteorological
data. Perhaps we can indicate for the European case what routine meteorological
data will be sufficient for application of the model.
0. Hov: With respect to input data, I would like to suggest that OECD recommend
that a unified approach be taken when the emissions inventories are established,
so that the inventories from the various countries can be compared.
D. Jost: I wold like to say that the emissions inventories will be handled
separately in addition to this table, and we should keep this in mind.
Therefore, I will ask that we restrict our discussion at this time to the
meteorological data.
D. Jost: For the Hov model, it is sufficient. This cannot be answered for the
Lamb model, as this model may handle each data set that is provided.
P. Misra: My answer is yes, but I would like Anton to comment on that.
A. Eliassen: It requires some work, but it is sufficient.
D. Jost: Yes, it will require some work to apply European meteorological data
to the Lamb model as well. That is, meteorological input data. As you have
11 layers in the SAI model, can these layers be taken from European
meteorological balloon measurements?
552
-------
G. Carmichael: Yes, from the upper air. You have the same problems with some
of the other models. However, the more frequency you have, the better off you
are.
R. van Aalst: Does this exclude the cloud data?
G. Carmichael: I am not sure about the cloud data. They are harder to obtain.
D. Jost; Do you have available routine WHO measurements from cloud data?
G. Carmichael: In principle, yes.
D. Jost: Then, this will not be different from the United States?
A. Christie: A great deal of preprocessing will be required to get the cloud
data into the scale we are discussing here in order to be able to compute the
input data from the available data, regardless of whether the data are from
North America or Europe.
D. Jost: With respect to the output data, I would like to ask for time scale,
how the average data will average and also with respect to the model scales that
results from the models as they exist now. This will be given for the Hov model
(as it is a trajectory model).
0. Hov: An appropriate scale for the output is not an hourly average but 6- or
8-h averages or more that that. The longer the time period we average, the
better the agreement I would think. Also, I would say that as a Lagrangian
model, it is most practical when a limited number of receptor points are of
interest. If you want to compute an Eulerian field with a Lagrangian model,
that becomes very expensive.
D. Jost: A limited number of receptors, whatever the limit is.
0. Hov: Much less than the number of grid points, which is 39 times 37.
D. Jost: With respect to the Lamb model, the Misra model, and the SAI model, I
assume that it will be 1 h?
G. Whitten: We have a graphics output that is continuous.
D. Jost: I think we could take this as 1 h—continuous and 1 h.
A. Christie: I think what you're talking about, there is a minimum time
resolution. (inaudible comment) ...daily averages or weekly averages. In other
words, an analysis of the effect you are going to get may be based on either a
short-term effect or an integrated effect.
D. Jost; That is right, but I do not think that we can decide or recommend this
right now. Obviously, you may come to a decision later on—I think 24 h would
be sufficient—and obviously the quantification of different models will change.
There will be models that will be more suitable for 1 h than other ones. There
is, for example, 24 h where all these models might be equal, sufficient.
553
-------
Unidentified Speaker: It might be worthwhile to at least quantify these input
data a little more, at least according to low, medium, or high input
requirements or according to the degree of sophistication. In other words, does
the wind field model needed to drive it represent the state of the art? If you
want to put in things like liquid water profiles and really do a good job on
clouds, then I would call that a fairly sophisticated method. So, I think we
ought to go at least one step further on the input data.
D. Jost; Are you saying that we should split up this question on the
meteorological data?
Unidentified Speaker; Not just the meteorological data. It affects other
things as well as the geophysical data, the degree of sophistication in your
emissions files and so forth.
Basically, for every data set that you are inputting, there are various degrees
of sophistication for which these models were either originally developed or
various degrees of sophistication that would be consistent with the degree of
approximation inherent within them. So at this point, I would not want to go
beyond much more than putting low, medium, or high numbers on them in terms of
the effort it would take to develop the input field. However, it might be
worthwhile to have more than just what is there.
D. Jost: Low, medium, or high concerning all the input data, with the exception
of the emissions inventories?
Unidentified Speaker: You might want to quantify it in terms of the man-years
that would be required to actually compare the input, to adequately apply that
model.
D. Jost: Starting from routinely available data, which effort is necessary to
prepare the data in such a way that they are suitable for the model?
Unidentified Speaker; That is just one suggestion.
D. Jost: There seems to be agreement on it. To make this task easier, I should
perhaps start with the Lamb model since I was a member of that panel discussion.
In the discussion, I got the impression that the preparational work represents a
high effort in terms of what we are discussing here. That is, you are starting
from continuously available meteorological data, what you continuously know,
topography, and so on.
R. Lamb: As far as the meteorological data are concerned, you can put in as
much or as little data as you want You are limited by putting in one number or
one observation for every grid cell, or you can put in a guess or assume a
constant flow field. As for the way we do it in this simulation, the
uncertainty and the error bounds on the calculation will depend on how much
information you put in. So, you put in a lot or a little and you get out, in
principle, reliable numbers or less reliable numbers.
D. Jost: This is why I was at first a bit afraid to put in more details as I
saw. I guess that the complicated models are also simplified and then you have
554
-------
simplified. In the same sense, you may also simplify the input data and then it
is no longer a simple or a complex model, but it is still possible.
H. van Pop: In constructing the complicated model, you have in mind a certain
amount of meteorology. So, you need to say that you could put it as simple as
you want it, but I do not think you really want that. You have in mind a
certain sophistication of your chemistry too.
R. Lamb: That is true. On the one hand, having a lot of sophistication can
allow you to handle the chemistry properly. You may need it for that. That is
partly why this model has the resolution and the structure it does, to treat the
chemistry, which is independent of the meteorology. So, for the flow fields
that go into it, you would ideally like to have the maximum utility involved in
putting as much meteorological information into it as you can get. For that
reason, you want to have as much, so that would require a great amount of data,
more than exists.
D. Jost; One could perhaps approach it from a (third). One could say that this
model is meaningful for this model to use (a high effort) with meteorological
data (where there could be models). There is no sense in making such a high
effort, and then such a thing could be put here as high. It is possible and it
is meaningful to spend much effect in preparing the input data for this model.
R. Lamb; I would agree.
0. Hov: I would like to comment on the meteorological input to the first model.
As it is now, it operates on the same data base as is used in the normal weather
forecast procedures. The trajectories are computed at this level in Oslo and
uses the meteorological data along the trajectories that come from the weather
forecast data base. So, I would say that the data base is fairly extensive, but
that it is a very simple job to get those data to ihe model. So that's a low,
work effort drops.
P. Misra: My first reaction would be a high effort, but again I would like to
have Anton comment on how quickly he can get the data.
D. Jost: Do you object to this?
A. Eliassen: No.
D. Jost: What is you comment on this?
S. Reynolds: I think medium to high, depending on your definition. Less than
some, more than others.
J. Killus: Since the Carmichael model- has in fact many more layers and much
more detail, the degree of input that you would expect is high-plus.
H. van Pop: Are we finished with input data?
P. Jost: I assume.
555
-------
H. van Pop: I would like to go back to another point, which was discussed
rather fast, I think, That is the output data. The 1-h time resolution I
object to. Maybe it is true, but it is not thought out well.
P. Jost: Your point is right, but when I considered the 1 h, I thought that the
people that are working with this model are meaningful to it, that they do not
output 1-h averages knowing that there cannot be any difference between them.
H. van Pop: If you have a grid spacing of 150 km, it would be very useful to
depict your data every hour. Every 3 h may be more meaningful.
P. Jost: This takes us to the discussion we had this morning on how to
integrate reactions and all that happens within the grids. Perhaps we should
keep this in mind, as I do not think that I will be able to assess this right
now.
H. van Pop: Could we ask each modeler to indicate whether his model is supposed
to give useful information every hour?
P. Jost; I assumed you would phrase the question a little differently. I do
not think that any of the authors will say "No."
E. Runca: In describing the output, we should also consider the spatial scale
of every model.
D. Jost: Yes, this could be added. The spatial scale of the output from the
model is used right now.
As you mentioned during your discussion this morning, it is not too meaningful
to discriminate hours—to distinguish scale hours and something like 100 km.
A. Eliassen: The emissions data are for 150 km.
P. Jost: I would like to put these numbers here in addition. The times—the
scale for the output for your model?
R. Lamb: What is the number now?
P. Jost; Do you mean the local scale for the output of the model?
R. Lamb: The grid scale?
P. Jost: Yes.
R. Lamb; It is in latitude/longitude coordinates, but it is about 18 km.
D. Jost; 18 km.
R. Lamb: 18.5 km.
Unidentified Speaker: It is 50 km on the European scale.
556
-------
D. Jost: Why is it different for the European and U.S. scales?
P. Builtjes: Small country.
D. Jost: I know that everything in this country is larger; that is why they are
expressed in miles instead of kilometers. The SAT model?
J. Killus: The SAI model has been exercised both at the 20-km level and at the
80-km level.
D. Jost: 20 to 80 km.
G. Whitten: The S02/sulfate that we are doing is at 80 km and the Seabreeze
that I showed is designed at 8 km.
A. Eliassen; As to the spatial scales you just wrote down, noone can convince
me that you can calculate oxidants for Europe or anywhere on a 20-km scale,
because the wind fields and the other meteorological information are far too
uncertain. Noone can convince me that you can say exactly with what probability
you can do this.
Now, I would like to ask OECD what it really wants here. Do you want to
calculate oxidants on a 20-km or a 50-km scale for Europe, or do you want to
prevent the occurrence of high concentration of oxidants? Whether these high
concentrations hit one city or a neighboring city, does that really matter? On
an international scale, if the plume hits Belgium or The Netherlands, docs that
really matter?
In Europe, you have high emissions of different species coming out in the air,
and these mix over a number of days. If you have a puff that is emitted from
one source and it does not exactly hit the source 50 km down in that direction,
it will hit another source 10 degrees off in another direction. Whatever you
do, you more or less get oxidant production. So, it seems to me completely
unnecessary to do this on such a fine scale, as some people here are proposing.
You might also argue that this could result in a suggestion to reduce sources
over certain regions or whatever and so on. This would have consequences for
the weather next year. What sort of situations will you have then? The
development of the weather is so uncertain that all sorts of cases can occur.
This therefore limits the degree of accuracy that can be of interest at all when
you apply models for oxidant control strategies.
Having listened to the discussion on all these very complex models, I have to
say this. I have defended complex modeling at this meeting over the previous
few days. Now, I resume my usual position of supporting the more simple models.
Thank you.
P. Lieben: I think maybe the question here, I, not being an expert, feel that
if the accuracy is 20 km, it can also be run for 50-55 km.
A. Eliassen: Of course. Any model is—
557
-------
P. Lieben: The other way is not common, but provided you can get 5- or 10-km
accuracy, you can also get 50- or 55-km accuracy. Is this right?
A. Eliassen: Yes, that is right.
P. Lieben: Again, just to comment on your point about these figures, it will be
useful, once the OECD member countries have better defined what they would like
to use as control strategies and, again, I say it is not for the group here to
do it and to tell them what they should do that will help the countries to
decide what to do. Once they have more ideas about the kind of control
strategies they will try to achieve, this will probably help in looking at the
final selection of a model. If they have to do it on a grid of 20 km x 20 km or
higher but if they have to do it, then again that's information.
G. Whit ten: As a chemist, I think that these models can be used to provide an
accurate estimate of the effectiveness that an HC control strategy might have
over an NOX control strategy—an estimate of which would be more effective. It
is true that we cannot predict the meteorology for future years, so we do not
know exactly where or how big the 0$ will be. However, with a typical
meteorological field, a typical emissions inventory, and a good model, we can
predict the order of magnitude of the response from a control strategy in terms
of HC versus NOX and in terms of mobile sources versus stationary sources. This
is what the models are being used for at the urban scale and they can also be
extended to the mesoscale.
Unidentified Speaker: I just wanted to point out a few things that went into my
comment about 50 km. Almost all of these models can be applied on any grid size
you choose. As for what went into that 50 km, we looked at a map of Europe and
decided we wanted to put the boundaries in fairly clean areas to the extent
possible. So, that means that you run out the western boundary to include
England, the northern one to include southern Scandinavian countries, etc. With
the eastern boundary, you are in rough shape, but you can only do what you can
do. So you pick an approximate scale for a model and consider how many grid
points you can afford to run with. Now, 30 and 40 grid points in one direction
(X and Y) are the kinds of numbers that people run with. If you say 50 or 60 in
one direction, that is an extremely ambitious model; if you say 100, the model
might not run on present day computers. So, even though they sound very
arbitrary, that is the way in which these number get back down to the problem.
J. Bottenheim: If you are going to control HCs, there will be some consequences
for the acid rain problem. I know you do not want to talk about acid rain, but
the people who do research in acid rain are certainly interested in what you
decide here for oxidant control, because it influences their decisions as well.
So, you are right, Gary. However, there are consequences in other fields. I am
not sure if you should only stick to your own little niche of oxidants.
S. Zwerver; I do not think this is a question of policies or abatement
strategies. I cannot imagine that the same abatement strategies or policies
will work for all countries. What is one country to do for abatement or the
other. Here, it is just a question of the model itself; it has nothing to do
with policy. I think there are two different points, strategies, but the
558
-------
question here has to do with the quality of the model. So if we stick to the
issue of quality, whether the model needs a 1 x 1 square, we can accept that.
D. Jost: Is there any further discussion?
H. van Pop: The spatial resolution in the model is up to the modeler. The
modeler has to decide which spatial resolution is needed for his/her numerical
schemes, for his/her chemistry. Nobody cares if it is 1 m or a 100 km. The
question is: What spatial scales do we want to have numbers for control
strategies. Do we want every 10 km? Do we want to test every 10 km in our area
or do we want 100-km values?
D. Jost: As Zwerver mentioned, there are several reasons for those scales. One
reason is inherent in the model. The other involves local-scale concerns as
effects. Or, due to a particular air quality situation, you may know that
typical oxidant situations need a local resolution of 1 km, 10 km, or 50 km. By
different aspects, we mean which scale you want to use to abate air pollution.
This is, I think, quite different from this.
H. van Pop; It is in part a technical question because, you have to ask the same
question (as to the time required of the grid), at what distance do you expect a
significantly different value from the source point.
D. Jost: I think this is given by this 10, 50, 80 km. I assume that the
numbers that I mentioned with respect to the time scale are meaningful. Perhaps
you normally should take half of the scale and look for differences in order
to—It should not be the same scale, but half the scale. That is, there should
be real differences when these numbers are doubled. But, are you proposing
further items of a more administrative nature that should be taken?
A. Venkatram: Perhaps this is a very naive question, but what is a control
strategy? What are the types of questions thai comprise a control strategy? Do
you have to pose the questions before you can discuss the utility of the models
or answer those questions? What is a typical control strategy? Do you need to
be specific about it so that you can match the model's answers to the questions
you pose, especially with some of the models aspiring to predict probabilities.
I think it is quite critical to ask these questions.
B. Luebkert: We are not trying to propose one typical control strategy, but we
all asking whether you can use these models altogether to develop control
strategies and we are evaluating different ones against each other to see what
effects they have on the oxidant picture on a regional scale.
A. Venkatram: Do you have to ask the model a certain number of qxiestions before
it can give you an answer?
B. Luebkert; I thought we were trying to do that.
P. Jost: Didn't your question in this direction deal with control? For
example, a typical strategy would be to decrease the NOX emissions from high
stacks and in the same strategy decrease the HC emissions from the motor
traffic.
559
-------
A. Venkatram; Qualitative.
D. Jost: Those could be enough to—
A. Venkatram: Is that a qualitative statement? People are looking for
something like a 30% reduction in HCs to lead to a 30% reduction, for example.
However, those percentages mean nothing unless you specify time scales, spatial
scales, and probabilities.
A. Galli; That is true, but it is a lot easier for the U.S. to define those
kinds of things than it is for the European countries. The U.S. has a standard
that is based on 1-h time periods. We also have a standard of 0.12, which means
that we are aiming for a percentage reduction to meet a particular number. The
European countries do not have a corresponding level or regulatory basis with
which to work. Therefore, it is a little harder for them to define that kind of
information here when they have not defined it for themselves, for Europe or for
a country within Europe. We more or less have defined that in the United
States, at least for 03, and our primary and secondary standards are identical,
right now at 0.12.
A. Venkatram: If you work on the numbers, you can reduce the HCs by 20% and you
might see an increase in 03 next year because of the meteorology. This has to
be posed in terms of probability. What is the probability of a decrease below a
certain level?
D. Jost; In my opinion, such a result would help to say it is possible to do
something for oxidant air pollution, and there are several possibilities to do
this. Yes, I agree with you. Nevertheless, I think that it is necessary to
base all of the very simple measures on more complicated scientific studies.
P. Lieben: What is your conclusion from the discussion regarding the best
candidates for future focus by OECD?
D. Jost: I cannot give you a clear answer. To select one model or several
models, we need some more basic information on the possibilities which we will
have as it was discussed several times on the needs. To put it in a more
qualitative way, I think it would be worthwhile to do some simple tests, first
to run a test with one of the not-so-simple models in order to check several
emission scenarios and then to check the model, or a third model that has been
used, against a more complicated one, which would really take into account
almost all the known chemical and physical effects. Although I cannot nominate
one very simple model to do the whole business, I propose such a stepwise
procedure, including testing, developing strategies, and checking the model
sensitivity for areas in which the model is going to be used.
P. Lieben: Perhaps somebody in the audience would respond to the question.
From the discussions we have had these 3 days, are there best candidates for the
future by OECD?
J. Schjoldager; The comparison was very interesting, but from an applications
point of view in Europe, a few things are missing. One important thing was
brought up by Smith in terms of a home in Europe. These models really have a
560
-------
home in Europe. As far as I am concerned, three of them are now available and a
fourth is underway. It is very important to consider which of these models has
a home in Europe.
Unfortunately, the models you already have in Europe are quite different—the
U.K. model, the Hov model, and the SAI model. Then, there is the Misra model,
which will definitely have a home in Europe.
Given that and given Gary Whitten's comment, which is that we should not really
restrict the discussion to one model, with which I fully agree, we should allow
for a group of models to be used. I think these two things together early
indicate that we should really not try at all the various models we have at home
in Europe. We should more or less continue the discussion here along the lines
that we should prepare the necessary data bases so that all of these models can
be continued by the people involved with them. Thank you.
P. Lieben: Thank you for that very useful comment. It is time to go on to the
third item, which may be the most difficult point, that is emissions
inventories. I would like to ask Dr. Galli to introduce the subject for
discussion.
A. Galli; Emissions inventories can perhaps be looked at in four different
ways. The first one brings up a problem with what Joergen said. He talked
about models finding a home. The first question that occurs with emissions
inventories is: What compounds do you measure? You have several choices. The
U.K. model and some of the other models require 40 specific compounds. If you
are going to do an emissions inventory in Europe and if you are going to upgrade
your inventory, you will have to look at all 40 compounds; otherwise, such a
model will be useless to you. Furthermore, the more compounds you are looking
at, the more expensive your emissions inventory will be, and the commitment on
the part of the country will have to be a lot higher. While I understand the
rationale, you immediately run into a problem gathering the data base for your
emissions inventory, which illustrates the first question right here on your
list: Exactly what types of compounds or materials are you going to measure?
Are you going to measure for the 40 compounds that are identified in the brief
compound, or are you going to limit it to a more reasonable number, which may be
somewhere between 1 and 10?
I have indicated some likely candidates, and I am asking for suggestions and
comments. I have suggested NOX and HCs. In the case of HCs, what species are
you talking about? I have also included 03. The relationship between 03 and
acid rain has been discussed here, and it cannot be ignored. So, I have to
include SOX. What else? What about the other 36 compounds that are mentioned
in some of the other models? Do you really want to talk about that from a
European standpoint when, in essence, the place where your data bases are really
going to need to be updated, and most of you have identified that as a problem
in Europe. Do you want to do it for 40 compounds?
J. Bottenheim; Are you talking about just field measurements or source
inventories?
561
-------
A. Galli: I am generally talking about emissions inventories from sources, not
ambient monitoring.
J. Bottenheim: If you are talking about inventories, it might not be as
disastrous as you picture it.
A. Galli: I do not know; I am asking the question. T do not live in Europe; I
live in North America.
J. Bottenheim; Neither do I, but apparently it has been done £or the U.K. with
some success. Maybe they can say something more about it.
K. Brice: Surely the question here about whether you speciate the HCs does not
come down to what the model can handle. If you parameterize the chemistry, you
still have to have as much information as possible about the HC composition. I
think everybody would agree with that. For emissions, there are enough data
available in some cases to be able to break down the HCs into more than just
five or six categories. The more emissions you have, the better. In the field,
measuring all of the HCs is of course more of a job. You are dealing with a lot
of concentrations. It is a more extensive field project. So, I would argue
that the more information you can get on the HCs, the better your
parameterization of any chemical scheme.
A. Galli: I guess I was hearing from the group in general that the HC data base
was the weakest of the European data base. If what you are saying is true, you
really have a bigger problem than what you are describing, because what you are
saying you need the most is your weakest area.
P. Builtjes: I said the HC field data are missing.
J. Killus: The more experience you have with emissions inventories, the more
possible it is to make assumptions concerning the various speciation that
occurs. Given an emissions inventory that is split into source-emission
categories without precisely defining what HC properties are in those sources,
it is still possible to make intelligent guesses as to what those new
speciations are.
It you have mobile sources, for example, it is very difficult to go to a local
gas station, run a few drops of gasoline through a gas chromatograph, and find
out more or less what the unburned gasoline should look like. That generally
applies across the board. If you then make a certain number of ambient
speciation characterizations, you can compare the emissions inventory speciation
to the ambient speciation.
For U.S. inventories in the early days, inventory speciation and ambient
speciation did not correlate well at all. In more recent times, they have
correlated quite well. In a recent urban study for Philadelphia, there was an
extremely good match between the ambient speciation and the emissions inventory
we were given. For that, we were profoundly grateful.
A. Gal1i; Are you saying that we should be looking at both an ambient ifcventpry
as well as an emissions inventory for specific sources?
562
-------
J. Killus: Ambient measurement for quality assurance.
A. Galli: Any advice on the size of the list of the compounds?
R. van Aalst; If you are going to include acid rain components, I strongly
suggest that you include ammonia. Our estimates for The Netherlands include a
contribution of about 35% of ammonia compounds to acid deposition.
A. Galli: I am not sure whether acid rain compounds should be included.
Personally, I have tried to make a clear distinction between the oxidant-type
things and the acid rain, realizing that there is some relationship and that you
cannot draw a distinct line between them. How about some suggestions of other
compounds we have looked at?
B. Luekert: I would like to see the discussion separated into two different
areas, first a discussion on emissions inventory and later one on ambient data.
There is a lot of confusion about this, and I would like to make this
distinction. If we first focus on the emissions data, I think the problem in
Europe is that you have to agree on the methods to develop a common inventory.
This is exactly what we are looking for—some suggestions, a basis that the
countries come up with an inventory that fits whatever model will be selected or
that fits different models.
However, it has to be compatible. There is no need for one country to use
different emission factors than another country. When you do, you introduce
mistakes into you model that your can predict right in the beginning. This is
where we are looking for some recommendations.
S. Reynolds; You might want to consider CO.
G. Whitten; Carbon monoxide can also play a quality assurance role in
validating a model. The chemisty of CO is slow; therefore, following CO clouds
is a way of testing meteorology and testing against the emissions inventory.
G. Whitten: Our experience with emissions inventories to date suggest that a
common problem is carbonyl compounds, aldehydes especially. Many of the
chemical mechanisms respond to these types of compounds, so we are talking about
oxygenates as opposed to pure HCs.
Our problem was that we asked for HCs. People were putting out oxygenated HCs
and saying that these were not HCs. So, they did not list them. However, the
oxegenated HCs still reacted in the atmosphere. That is why we are now using
the terminology "VOCs and reactive organics." These have a particularly strong
reactivity in the early morning when the sun comes up. If they are missing from
the inventory, the model does not perform correctly.
L. Lindau: In response to Barbara Luebkert's comments about the emissions
inventories in Europe, I agree that we have to develop them in a similar way for
each country. However, the emission factors will be different. A refinery in
one country is different from a refinery in another country. The emissions
factors will be different and will have to be considered in some
563
-------
source-by-source way. The way of doing it, the principles, etc., have to be
discussed.
H. van Pop: Should particulate matter be included?
R. van Aalst: Acid rain substances are not included; SOg should be removed as
well. We agree on that.
J. Bottenheim: Just like 03, yes?
A. Galli: What are you taking off?
J. Bottenheim: Take off 03 and SOX if you do not want to talk about acid rain,
03 at any rate.
G. Whitten; We do not really know of any 03 emissions.
B. Luebkert: To continue the discussion, we could consider some of the
countries that have undertaken a joint, cooperative effort, such as the
Dutch-German experience. The people here who have had experience with that may
want to point out the problems they have run into with the inventories, the
differences that exist between these countries. These experiences could be a
learning process so that OECD would not repeat the same mistakes, but take it
from there and avoid mistakes in the beginning.
D. Jost: There has not been much experience of this type up to now. We had a
lot of time lag in preparing our emissions inventories just for the Dutch-German
effort. One of the reasons was that we needed to know at a very early stage of
the project which emissions data, what accuracy, what time and scale, and what
resolution were needed. It took a lot of time to prepare this information.
This had to be discussed with a lot of political people and already for this
discussion one needs to know what the need of the single data one has. There is
no possibility for change later on, as one has to follow the entire procedure
all over again.
J. Killus: The SOX should probably be placed back on the list. I can at least
come up with a rationale to include it in photochemical episodes, for not
looking at it specifically in terms of acid rain. After all, we tested the
transport sections of our regional model using the S02 and sulfate portions of
that, especially the S02/sulfate reaction that is a moderately good tracer for
photochemial reactivity and for regional transport.
I am not really familiar with the various reasons for wanting to separate the
two issues of acid deposition and regional oxidant modeling. However, for
transport reasons and in order to test for transport calculations, I believe we
should include acid rain.
A. Galli: You have to include acid rain only from the standpoint that it tends
to be a buzz word within a number of countries for getting additional research
support. The group here has more or less steered clear of acid rain, because it
is being taken up in a number of other foreign and domestic forums. In these
564
-------
forums, the discussion is focusing directly on acid ran and only peripherally on
oxidants and their effects upon acid rain.
It is not a hard and fast thing; it is something that helps to sell research
programs within the individual countries. As for specifically talking about
acid rain here, there are enough other forums that are bigger from both a
technical and a political standpoint, forums where that is being taken up now,
without adding this meeting as another one.
J. Killus: I am not talking about acid rain at all though. In fact, the
regional oxidant episodes are not generally rainfall episodes. However, there
are reasons for wanting to have S02 and sulfate in the transport model in order
to get the test to do long-range transport. It is good validation situation and
unless there is some reason to exclude it, I think it should be included.
P. Lieben: Maybe the question is a bit ridiculous, since the OECD countries are
engaged anyway in making a detailed emissions survey of SOX. So, it will be
included anyway with this project or another projet. We will match the truth
with the same grid square. However, there may be specifc reasons for including
SOX in an oxidant poject. That is where we need advice.
A. Galli; We are not trying to specifically exclude anything. I am saying that
I do not want the acid rain, acid deposition, or whatever term you are using
today, to be the driving force of this group, because there are many other areas
where it is the driving force.
S. Reynolds: I might just comment on the experience that I perceived in the
Dutch photochemical modeling effort and their assembling of an emissions
inventory. They were able to develop a gridded inventory in a fairly
expeditious fashion, given the data bases they had assembled there, including
the derivation of the kinds of HC speciation that the photochemical model
required. That was about a four-to-six lumped-species inventory. They had
quite detailed estimates of the HC composition for various sources in their
country.
A. Galli: I am not sure I know where you are heading.
R. van Aalst: It is probably a detail, but I might add that in The Netherlands
at the moment we are concentrating a little bit on emissions in water as well.
It may be a point that you forget when you are trying to get emission
inventories; you may just concentrate on air emissions, although organic
species, especially volatile organic species, tend to come from water. It is a
detail that could be significant.
J. Bottenheim: I wanted to make another comment on other European situations.
You say you are already developing an SOX inventory; what other inventories are
you developing? If you are already obtaining one for NOX, why should you
consider doing it in this group as well, and similarly for HCs or CO. In other
words, it is nice to say that you want to stay with the oxidants, but if you are
developing these inventories for the acid rain program, even if somebody else
does it, it is rather superfluous to consider starting this effort again.
565
-------
P. Lieben: No, the emissions survey for NOX and HCs has not been fixed so far.
We are trying to get some ideas about the grid square to be used and so on, from
this particular meeting.
B. Luebkert: I would like to point to one application, especially the
Dutch-German effort in that experience. You said that you had no problems in
getting a gridded inventory. Was that only on a grid base or was that also with
respect to source categories?
I think OECD is talking about eventually making some recommendations with
respect to policy for control strategies. If so, then it is of no use to have
only a gridded inventory that is needed by source categories. This is where
there may be a source problem. If you come across any of those, I would like
for them to be discussed.
P. Builtjes: It is available for all kinds of source categories. No problem at
all.
S. Zwerver: I could not follow the discussion completely. As to the question
of categories, it is, in principle, a very difficult problem to solve. In
Europe and in OECD, we also have examples of miscategorized inventories for NOX
and HCs. For the Dutch situation and I suppose for the German situation, we
were able to categorize in a very detailed way.
As I understand it, the problems now under discussion are the emissions
inventories, the long-term average emissions inventories, and the oxidant models
that will be applied, especially to short time periods (e.g., days).
Another problem arises: What are the emissions, the meteorology, etc., for that
specific date? We have heard from the United States; they have very complicated
inventories to collect data directed especially to the application of these
short-time-period models.
I think that will become in Europe to gather information for certain periods,
because that would mean a high degree of organization and so on, but I do not
think that—
A. Galli: That is not my understanding of the discussions that have occurred
over the past 3 or 4 days. My understanding is that it would be difficult to
come up with these inventories. Maybe I am naive because I have not been there,
but I believe there is an inventory problem in Europe. However, I do not
believe that it is of such a level that these data could not be gathered in a
1-, 2-, or 3-yr time frame, the same time frame in which a lot of these models
could be updated and tested in something like this.
S. Zwerver: It depends on what you mean by emissions inventories. For Europe,
it is possible to relate, in a very simple way, energy use, energy consumption,
etc., in order to make an emissions inventory, but would that inventory be
sufficient? Do we need more detailed emissions inventories, directed to special
periods.
566
-------
P. Builtjes: In principle, the data are available. It is in time. But what
you want to know is the traffic. So there are traffic grids specified by day.
This information is available. You can even have specific episodes in mind.
You can even try to look up specific situations. So you have to check, for
example, where the large emitters—when they are working.
You can just by inquiring look to power plants to check whether specific things
happenend, like an enormous strike in the petroleum industry or something, but
that is not difficult. You have to organize it. It will take some time to do.
You would really have to check whether there are strikes, you know. But that's
not a difficult procedure.
J. Novak: One other area that was briefly mentioned up in these discussions was
biogenic species. Right now, there is nothing on our list for biogenic HCs. In
my opinion, you were talking about all anthropogenic sources. So, I would at
least question whether this is important, whether it is something that Europeans
would like to consider in their inventories and whether the chemistries will
handle them exclusively or just lump them in with the other anthropogenic HCs.
P. Builtjes: As far as I know, the chemistry does not explicitly handle it. So
the HCs coming out of these sources are categorized in the same way as the other
emissions. But emissions factors, whether they are accurate or not (is not a
question). But emissions are available, so you know by land use this 2-km2 grid
gives emission of so many—because it's grassland or because it's urbanized,
etc.
J. Novak: Are you saying that biogenic emissions inventories could be
constructed?
P. Builtjes: They are available. They are existing on the list, on the file.
Because the land use is on the file.
J. Novak; In biomass factors?
P. Builtjes; Yes.
G. Whitten: What can be done explicitly in the chemistry is not always clear.
If you handle them specifically in the emissions inventory and you bring them in
and out of the emissions inventory, you get a differential effect in the results
from your model of the biogenic emissions. Certain types of chemistry would
need to be treated explicitly; other types could be treated in a lump. It would
not make any difference.
A. Galli: Perhaps we should move on a bit. The next thing we need to discuss
is the grid resolution. To a certain extent, the grid resolution we should be
looking at was covered by Dieter in the previous discussion. In the various
models, the resolution varied anywhere from—what did you say, eight-and-a-half,
Greg?
G. Carmichael: Yes, that was for Seabreeze.
567
-------
A. Galli: I think it was 80 km in one case. We have heard one argument that
the 20-km grid resolution is too small and rather meaningless. You need
something a little bit larger, but I guess if you have a grid resolution at the
smallest level (20 km), you can always do a resolution at 40, 50, or 60 km
because the basic information is available. Do we really need to discuss grid
resolution further?
S. Reynolds: It would perhaps be helpful if an inventory were put together in
the future on a somewhat finer resolution than might seem appropriate at the
moment, so that the inventory might be available at a subsequent time. I am
thinking of something in the range of 24 km to 50 km. A hundred kilometers
might be a little gross for some subsequent time.
G. Whitten; If you were going to use an urban model on a mesoscale, you would
want a concentrated emissions inventory for that area.
P. Grennfelt; I would like to point out that there is one cell for Southern
Europe. If this is too large, why not divide it so we get four squares? That
would make it a little easier to compare the sulfur emissions. To construct a
new grid system would be difficult in many countries to understand why you are
using a new system.
P. Lieben: For example, your proposal would be 60 km to 65 km. So, if we put
four squares together, we have one square of the existing grid.
P._ Grennfelt: Yes.
A. Eliassen: It seems to me that the main reason for going to a very fine grid
is related to chemistry.
If you want to go just a short distance downwind from an industrial area, say
12 h, you might say that it is important to have very good spatial resolution
for the emissions in that area. The further downwind you go, perhaps the less
important the emissions resolution is. So, it is difficult to say how important
this is, and it is difficult to discuss it here without having good data to base
your arguments on.
I suggest that you perform calculations with a very coarse emissions grid, say
150 km, then a very fine grid, and that you qualify them by your calculations,
with a realistic presentation of the different species in the fine grid. Then,
compare the results. That would be a very interesting guiding experiment and it
would perhaps tell OECD what resolution is really necessary in this case.
P. Lieben; That idea is interesting, but it would just postpone any decision
about the grid size to use.
A. Eliassen: Just because there is no basis on which to decide that now?
P. Lieben: Well, I was thinking that we could have some ideas, some proposals
from the meeting, I am sorry to come back to this question, but I think it is a
very important point.
568
-------
I have heard during these 3 days that in any mode] development project,
especially, the emissions data are costing the most. Once we have the emissions
data, once we have put forth the effort to get it, we will not do it again for
another grid size resolution for a long period oL time. So, I think it is
important to get an idea of which size we should propose.
J. Bottenheim: I am not sure if the size is really relevant if you start making
an inventory. If you make an inventory, you try to figure out where the point
sources are and you establish coordinates for them. You go out to where the
area sources are and you establish coordinates for them. Then, it is up to the
user to divide them and add them up in a certain grid size, and I think the
major effort goes into determining where these sources are on the map and
putting coordinates on them. Whether you next add them up on a 20-km grid or on
a 100-km grid is up to the user, and it is presumably a relatively simple effort
compared to just coming up with where those sources are, with the coordinates.
In other words, I think this grid discussion is rather irrelevant.
D. Jost: I agree. If you are establishing an emissions inventory and if you
are going beyond a scale of several kilometers, say 5 or 10 km, there is not
much difference in the cost between using the 10-km grid scale or the 100-km
grid scale, since you will have to rely on the same data.
On the other hand, it is true that we do not have much data here before us.
However, the data we have seen are more or less of a typical scale for the 03
concentration, somewhere in the range of 20 to 50 km. The whole plume has a
scale of 90 km or something like this.
B. Luebkert: I am not sure that it is very easy to go from a large grid to a
smaller grid. It is easy to go from a smaller grid to a larger one, but if you
agree on something like a 100-km2 resolution and later find that you need a
20-km resolution, a major effort will be required to take into account area
sources. Certainly you have point sources, coordinates on the map, but area
sources are calculated from land use patterns, etc. So, you would need to do
that exercise over again. For an area like Europe, that is quite costly.
J. Bottenheim: You are basically suggesting that area sources over 100 km not
be done. In my opinion, if you do that, you do not take too large an area
anyway. You still need—for areas sources, you would divide it up on smaller
units anyway. If you do not want to use it, then you don't have to add them up.
A. Galli: I do not think at this point that we (or OECD) have really made any
decision based on which way to go. OECD is gathering opinions, information on
which to make recommendations to the member countries, who may or may not
approve the suggestions given to them that are based on the conversations coming
out of this workshop. So, your comments are well taken and there are other
people here who have supported these comments, but they have to be considered in
the context of the total discussion we are having here. They may be accepted or
they may be rejected down the line, depending upon the opinions of the different
countries, but I do not think a decision is being made here at all.
569
-------
F. Smith: I think we do have to decide upon a scale. We do have to recognize
Anton's point when we are dealing with nonlinear chemistry, that no absolute
scale can be chosen that solves the problem.
A. Galli: Right.
F. Smith: When we are dealing with the interplay of different chemical species,
the only scale that really matters is almost a molecular scale. Of course,
there is no way we can get down to that. So, at some stage we have to
parameterize what is going on at the molecular level. The former
parameterization may in fact be a function of the actual scale that we choose.
Recognizing that, the most logical scale to impact here is perhaps related to
the size and the variation of the sources.
A. Galli; Any other questions? Let's move on to the next problem—the domain,
in other words the boundaries for your region. As I have perceived from prior
discussions, this has been more of a problem to the European community and than
it has been to North America, including the U.S. and Canada.
I have heard two things mentioned in term of the European community: (1) the
boundary problems between Eastern and Wester bloc nations and the inability, to
date, to really obtain emissions and source information from the Eastern bloc
and (2) the identification of an emissions inventory for a specific area,
whether that area is some subset of a country or a country plus a little bit of
another one.
We need to hear comments on how the European area might be handled or how an
emissions inventory might be handled for the European-type situation. Also, we
need to hear suggestions for problems on the North American continent.
P. Lieben: This has to be looked at both in connection with the models we have
considered and in connection with the models that may finally be retained for
use, since a model might perform better than the others for a limited part of
Europe than it does for the whole system.
D. Jost: As mentioned several times, we have the Dutch-German cooperative
effort in which modeling of this oxidant problem has already been studied. As
we expect to measure relatively high 03 concentrations in the southern part of
our country, it would be necessary for us to include in this OECD project larger
parts of our country, larger parts in this Dutch-German effort. Obviously, the
Dutch-German effort and the experience we acquired could be used as a basis for
determining what needs to be done within the OECD project.
P. Lieben: Any comments on that?
A. Galli: Let's go on to the last point, meteorology. You are talking about
two types of observations, surface observations right at the land and
observations of the upper atmosphere. It has been suggested that we include
wind speed, wind direction, temperature, humidity, etc., the things that you
routinely look at. In the upper atmosphere, you are looking at inversion
layers, anything that most modelers would not ordinarily consider as part of
their meteorological information.
570
-------
L. Kropp: I have a question regarding an issue that has already been discussed
but I missed the answer. Did we agree to a certain time resolution for the
emissions inventories. Should it be 1 h, 1 day, 1 yr?
A. Galli; I do not think that we made any real decision.
L. Kropp; I think it is an issue that must be resolved before establishing—
A. Galli: I agree, and once this meeting is over, OECD will be putting together
some recommendations based on the comments and information that have been
presented here. The only comment that I made on the subject was that it is
little bit easier for the U.S. to define the time period, since it has a
regulation that is essentially based on 1 h. The European countries really do
not have regulations based on a particular time frame. So, with the U.S. it is
1 h. It is defined; it is known. With the foreign countries, it is not that
way. Many of them do not have a time period.
J. Bottenheim: 1 disagree with you. This is precisely the answer that should
come out of this workshop. Some of the models will need source inventories on
an hourly basis if they are ever going to produce hourly answers.
A. Galli; I am not saying it should not come out.
P. Builtjes: For traffic, it is on an hourly basis.
J. Bottenheim: If you want to run a model and get hourly answers to validate
your model, you need hourly inventories.
P. Builtjes: Sure.
J. Bottonheim: That is a major problem that has been found in all of these
models. They try to predict 1-hour, 6-h, or 1-day values and use yearly source
inventories. What does that mean? It means that most of the uncertainty comes
from the source inventories. If OECD is just going to recommend source
inventories on a yearly basis, there will be certain consequences for whatever
model you are going to use or validate with. This is a very important point
that has to be discussed here.
A. Galli: I do not have a problem with that, and I do not mind it being
discussed here. I did not say it was unimportant. I said that this forum is
not making the final decision.
P. Lieben: That is a very important point, and I agree. It must be looked at
in relation to the model to be used. The emissions inventory has to match the
model requirements. If they do not match, you can do nothing.
A. Galli: I could not agree with you more; that lias to be decided.
D. Jost: I agree that we need such emissions inventories, but there is some
misunderstanding here. None of the known emissions inventories is in a
real-time 1-h resolution that processes timed hourly values. That is not too
571
-------
hard, but getting real-time 1-h emissions inventories is impossible and not
really necessary.
Also, this very fine emissions inventory from the Northeast Corridor produces
hourly values, but they are not actual hourly values. They are processed based
on some assumptions on the emissions. This would be possible for Europe, too.
G. Whitten; I am not quite sure I followed the discussion. If you are talking
about meteorology, you have a 100-km grid size. The meteorology with that mass
of air cannot do a lot in 1 h; it cannot move very far. So, there is less of a
constraint on the time scale of the meteorology from that standpoint. You have
this huge air mass and there is no way that it is going to move very far in 1 h.
However, the sun can go up and come down a lot in 1 h. There are certain parts
of the model that you do have resolution for.
A. Christie; It seems to me that, by using 15-min time steps, we are talking
about insulation changes that may have a fairly massive emission (of every
hour). You average that over a 1-h or a 3-h period and you say it has not moved
very far, but it may have a significant effect in any Eulerian model. While you
may not need that kind of resolution when it comes to using the model for
strategy evaluation, you will probably need it to carry out an evaluation on the
model, at least for short time periods. We have already discussed the
evaluation of these models, and it seems that we cannot carry out an adequate
evaluation without that kind of resolution over at least a limited period.
J. Novak: I was just going to make a couple of comments again about the
emissions resolution. I think it is a good idea for OECD, as part of its
recommendations, to list those specific parameters that it might want to include
in relation to the emissions inventory. As Peter was saying, most inventories
in the U.S. are collected on an annual basis, on a county-wide basis for area
sources.
The types of statistics are exactly that, about different fuel uses or whatever
it might be. I think it would be worthwhile to go through that exercise in
Europe, looking at the kinds of statistics that are available to do temporal
distributions and sending somebody out to look for that type of information.
Also, it would be worthwhile to try to characterize the differences between the
countries. It was mentioned before in terms of the different emissions factors,
the different refineries, and other things in terms of the fuel types that are
used. The differences that could occur need to be specified on a country basis.
I think you will frequently find that these are uniform and that you can isolate
those that are not uniform and handle them in specific ways. So, an important
part of actually producing your emissions is looking for this additional
information and finding out exactly what is available.
P. Lieben: Thank you. Are there any other comments on the emissions inventory
input data?
G. Whitten; I might make one final comment from the chemist's standpoint and
cite some experiments that were done at the North Carolina Outdoor Smog Chamber
on behalf of the Australian Government. They were quite concerned with a
572
-------
spectrum of solvents that seemed to have different HC composition. So, several
smog chamber experiments were conducted with a wide range of solvent
compositions. The result was that the reactivity of the solvents was all about
equal, and they all produced about the same rate and the same amount of 03 for a
given weight of HC.
The point is that, even though there might be a different brand of gasoline in
one country than in another, a different car spectrum, or things like that, you
see changes amongst a large amount of HCs. The result to date is that, if you
have a wide spectrum of HCs, once you have one spectrum, they all seem to react
very similarly. We have not yet seen too many important differences. So, this
kind of thing becomes unimportant, the fine structure and the reactivity
spectrum.
P. Lieben: Thank you. Are there any other comments on this point? It seems we
have more or less exhausted the possibilities.
Well, I would like to thank you very much for all these comments. I would like
to turn the meeting over to our general chairman.
D. Jost: As we are nearing the end of this meeting, I would like to thank Basil
Dimitriades and our colleagues from the United States, especially the U.S.
Environmental Protection Agency, for inviting us, for making possible this
meeting, and for providing the scientific background to help us come a little
bit closer to the definition of our project within OECD. Again, thank you very
much.
B. Dimitriades: You are quite welcome, and thank you very much.
I would just like to say a few things. First, when the decision was made in
November to hold this conference in April, I thought it would be a hopeless
undertaking and that the time we had to prepare was not nearly enough to tackle
this truly complex subject, but I am delighted that the discussions went so
well. They were well thought out and informative. When all of this information
is digested, I feel that there will be some very concrete and useful conclusions
to be turned over to the OECD.
I would like to thank the speakers and the experts, who were invited to come
here and help us on such a short notice. They graciously accepted, and they did
an outstanding job. A lot of credit also goes to the other participants. Their
critical comments in the discussions made the conference as successful as it
turned out to be.
Speaking for myself and the other members of the EPA staff from ERG and from
Washington, we are delighted for this experience.
I want to say once again that we seriously invite one or two modelers from OECD
to spend a year or so with us in our laboratory, working with us in the regional
modeling area. Please give it some thought, and we can talk about this again.
573
-------
PARTICIPANTS
575
-------
David Balsillie
Ontario Ministry of the Environment
Toronto, Ontario, Canada
Susana Cerquiglini-Monteriolo
Institute del Territorio
Rome, Italy
George Bergeles
National Technical University
Athens. Greece
Jason Ching
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Francis S. Binkowski
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
George P. Christich
U.S. Environmental Protection Agency
Washington, DC (USA)
John C. Bosch, Jr.
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
A.D. Christie
Environment Canada
Downsview, Ontario, Canada
Jan W. Bottenheim
Environment Canada
Downsview, Ontario, Canada
Terry Clark
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Thomas N. Braverman
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
John F. Clarke
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Kenneth A. Brice
AERE Harwell
Oxfordshire, United Kingdom
Julia Clones
Embassy of Greece
Washington, DC (USA)
Joseph J. Bufalini
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Henry Cole
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Peter Builtjes
MT-TNO, Dept. of Fluid Mechanics
The Netherlands
Thomas Dann
Environmental Protection Service
Ottawa, Ontario, Canada
Gregory R. Carmichael
University of Iowa
Iowa City, IA (USA)
Basil Dimitriades
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
576
-------
Marcia Dodge
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Dieter Jost
Umweltbundesamt
Berlin, West Germany
Anton Eliassen
Norwegian Meteorological Institute
Oslo, Norway
William Keith
U.S. Environmental Protection Agency
Washington, DC (USA)
Alfred H. Ellison
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
James Killus
Systems Applications, Inc.
San Rafael, CA (USA)
Jack Fishman
NASA-Langley Research Center
Hampton, VA (USA)
Lothar Kropp
Technischer Ueberwachungsverein Koeln
Koeln, West Germany
Gary J. Foley
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Kenneth Ladd
U.S. Environmental Protection Agency
Washington, DC (USA)
Alfred Galli
U.S. Environmental Protection Agency
Washington, DC (USA)
Robert G. Lamb
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Bruce Gay
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Nels Laulainen
Battelle-Pacific Northwest Laboratory
Richland, WA (USA)
Gerald L. Gipson
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Pierre Lieben
Organisation de Cooperation et de
Developpement Economiques
Paris, France
Peringe Grennfelt
Swedish Environmental
Research Institute
Goeteberg, Sweden
Lars Lindau
National Swedish Environment
Protection Board
Solna, Sweden
Oeystein Hov
Norwegian Institute of Air Research
Lillestrom, Norway
Barbara Luebkert
Organisation de Cooperation
et de Developpement Economiques
Paris, France
577
-------
Eija Lumme
Finnish Meteorological Institute
Helsinki, Finland
Steven D. Reynolds
Systems Applications, Inc.
San Rafael, CA (USA)
Fred Lurmann
Environmental Research
and Technology, Inc.
Westlake Village, California (USA)
Elidoro Runca
International Institute
for Applied Systems Analysis
Laxenburg, Austria
Charles 0. Mann
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Kenneth L. Schere
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Jerome Mersch
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Francis A. Schiermeier
U.S. Environmental Protection Agency
Research Triangle, Park, NC (USA)
Edwin Meyer
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Joergen Schjoldager
Norwegian Institute for Air Research
Lillestroem, Norway
P.K. Misra
Ontario Ministry of Environment
Toronto, Ontario, Canada
M.T. Scholtz
MEP Company
Toronto, Ontario, Canada
Joan Novak
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
A. Sheffield
Environment Canada
Ottawa, Ontario, Canada
Deran Pashayan
U.S. Environmental Protection Agency
Washington, DC (USA)
Lou Shenfield
Ontario Ministry of Environment
Toronto, Ontario, Canada
William T. Pennell
U.S. Environmental Protection Agency
Washington, DC (USA)
Jack Shreffler
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Norman C. Possiel
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Francis B. Smith
Meteorology Office
Berkshire, United Kingdom
578
-------
James Southerland
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Gary Whitten
Systems Applications, Inc.
San Rafael, CA (USA)
Leslie L. Spiller
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
Robert J. Yamartino
Environmental Research
and Technology, Inc.
Concord, MA (USA)
Jacob G. Summers
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
S. Zwerver
Ministry of Housing, Physical Planning
and the Environment
Leidschendam, The Netherlands
Joseph Tikvart
U.S. Environmental Protection Agency
Research Triangle Park, NC (USA)
R.M. van Aalst
MT-TNO, Division for Technology
Delft, The Netherlands
Han van Dop
Koninklijk Nederlands Meteorologisch Institut
De Bilt, The Netherlands
Frank Vena
Federal Canadian Government
Ottawa-Hull, Canada
Akula Venkatram
ERT, Inc.
Concord, MA (USA)
Fred Vukovich
Research Triangle Institute
Research Triangle Park, NC (USA)
Boris Weisman
MEP Company
Research Triangle Park, NC
(USA)
579
U.S. GOVERNMENT PRINTMOOfnCE:'»« -7S9-OU/86U
------- |