Environmental Protection
Agency
uttiee ot Hesearoh and
Development
Washington DC 20460
EPA/6Q0/R-99/030
March 1999
4>EPA
Science Algorithms of the
EPA Models-3 Community
Multiscale Air Quality
(CMAQ) Modeling System
\
-------
United States Office of Research and Development EPA/600/R*99/030
Environmental Protection Washington, DC 20460 March 1999
Agency
SCIENCE ALGORITHMS OF THE EPA MODELS-J
COMMUNITY MULTISCALE Am QUALITY (CMAQ)
MODELING SYSTEM
Edited by:
D. W. BYUN* and J. K. S. CHING*
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of
Commerce
Wy Printed on Recycled Paper
-------
EPA/600/R-99/030
DISCLAIMER
The information in this document has been funded wholly or in part by the United States
Environmental Protection Agency. It has been subjected to the Agency's peer and administrative
review, and has been approved for publication as an EPA. document. Mention of trade names or
commercial products does not constitute endorsement nor recommendation for use.
-------
EPA/600/R-99/030
FOREWORD
The Models-3 Community Multiscale Air Quality (CMAQ) modeling system has been developed
under the leadership of the Atmospheric Modeling Division of the EPA National Exposure
Research Laboratory in Research Triangle Park, NC. This new generation of modeling software
was under development for seven years and was made available in June 1998 without charge for
use by air quality regulators, policy makers, industry, and scientists to address multiscale,
multi-pollutant air quality concerns.
Models-3/CMAQ has a unique framework and science design that enables scientists and
regulators to build their own modeling system to suit their needs. Users can access pre-installed
modeling systems provided by the EPA or can incorporate their own modeling systems to work
within the existing framework software.
This direct user involvement is key to the concept of a community modeling and analysis
system. This approach to model development, application, and analysis leverages the
community's complementary talents and resources to set new standards for rapid incorporation of
better science into air quality model applications. The resulting comprehensive system forms the
foundation upon which the community, including governments, industry, academia, and other
stakeholders, can collaborate in the examination of issues and the subsequent development of
strategies that meet society challenges of environmental protection.
The release of Models-3/CMAQ is one of the many steps which we hope will unite the
community under the common goal of advancing our knowledge and abilities to tackle critical
problems of the future in far more effective ways than have been attempted in the past.
Scientifically sound modeling systems, developed and supported by the community, are one
method of achieving this goal.
The June 1998 release of the Models-3/CMAQ computer code was accompanied by a User
Manual [EPA-600/R-98/069(b)] to serve as a reference on how to use the software system. This
Science Document is the counterpart to the User Manual in that it presents the peer reviewed
scientific bases for the Models-3/CMAQ modeling systems. This document also includes
components such as interface processors, process analysis routines, and the present and planned
evaluation program.
F. A. Schiermeier
March 1999
-------
EPA/600/R-99/030
CONTENTS
FOREWORD v
ACRONYMS xvii
ACKNOWLEDGMENTS xxi
EXECUTIVE SUMMARY ES-1
1. INTRODUCTION TO THE MODELS-3 FRAMEWORK AND THE
COMMUNITY MULTISCALE AIR QUALITY MODEL (CMAQ)
(J. ChingandD. Byun)
Abstract 1-1
1.0 Introduction to the Models-3 Framework and the Community Multiscale Air
Quality Model (CMAQ) 1-2
1.1 The Models-3 Emissions, Meteorology, and the CMAQ Modeling Systems .. 1-4
1.2 CMAQ Interface Processors 1-6
1.3 The CMAQ Chemical Transport Model (CCTM) , 1-8
1.4 Analysis of CMAQ Output 1-10
1.4.1 Process Analysis (Chapter 16) ....' 1-10
1.4.2 Aggregation (Chapter 17) 1-11
1.5 Management of CMAQ Science Information Objects and Codes in Models-3 1-11
1.5.1 Program Control Processors (Chapter 15) 1-11
1.5.2 CMAQ Code Integration (Chapter 18) 1-12
1.6 Post Release Studies and Near-Future Plans 1-12
1.6.1 CMAQ Evaluation Study 1-12
1.6.2 Testing Operational Configurations 1-12
1.6.3 Extensions and Science Additions 1-13
1.7 Opportunities and Encouragement for Long Term Extensions and Science
Community Involvement 1-13
1.8 References 1-15
2. MODELS-3 ARCHITECTURE: A UNIFYING FRAMEWORK FOR
ENVIRONMENTAL MODELING AND ASSESSMENT
(J. Novak and S, Leduc)
Abstract 2-1
2.0 Models-3 Architecture: A Unifying Framework for Environmental Modeling and
Assessment 2-2
2,1 Introduction , 2-2
2.2 Overview of the Models-3 Framework 2-2
vii
-------
EPA/600/R-99/030
2,2.1 Dataset Manager ;.... , ; 2-2
2,2,2 Program Manager , ,.,.,, 2-3
2.2.3 Study Planner ,...,.. 2-3
2.2.4 Strategy Manager .: , 2-4
2.2.5 Tool Manager , , , 2-4
2.2.6 Science Manager , 2-6
2.2.7 Model Builder 2-6
2.2.8 Source Code Manager 2-7
2.2.9 Framework Administrator , 2-7
2.3 Models-3 System Architecture ,.,, 2-7
2.4. Schedule and Future Plans ... 2-10
2.5 References , 2-10
3. DEVELOPING METEOROLOGICAL FIELDS
(T. Otte)
Abstract 3-1
3.0 Developing Meteorological Fields 3-2
3.1 Credits and Disclaimers for Use of MM5 ....'.. 3-2
3.2 Meteorology Model Pre-Processing 3-2
3.2.1 Defining the Simulation Domain (TERRAIN) 3-2
3.2.2 Processing the Meteorological Background Fields (DATAGRID) 3-3
3.2.3 Objective Analysis (RAWINS) 3-3
3.2.4 Setting the Initial and Boundary Conditions (INTERP) 3-4
3.3 The Meteorology Model (MM5) 3-4
3.3.1 Brief History 3-4
3.3.2 Horizontal and Vertical Grid 3-5
3.3.3 Prognostic Equations .,..,.,. 3-5
3.3.4 Model Physics ;..... 3-6
3.3.5 Nesting ....... ....... 3-10
3.3.6 Four-Dimensional Data Assimilation 3-10
3.4 Meteorology Model Post-Processing 3-11
3.5 Changes to the MM5 System's Software for Models-3 3-11
3,6 References 3-12
4. EMISSION SUBSYSTEM
(B. Benjey, J. Godowitch, and G, Gipson)
Abstract 4-1
4.0 Emission Subsystem 4-2
4.1 Emission Inventory Processors 4-2
4.1.1 Discussion 4-2
4.1,2 General MEPPS Structure 4-11
4,2 The MEPPS Emission Processing System 4-12
4.2.1 The Inventory Data Analyzer (IDA) 4-14
viii
-------
EPA/600/R-99/030
4.2.2 The MEPPS Input Processor (INPRO) 4-16
4.2.3 Processing Procedure '...,' 4-17
4.2.4 Modeled Emission Data ., 4-33
4.2.5 Chemical Speciation of Emission Data 4-78
4.2.6 Output Processor (OUTPRO) ... ; 4-90
4.3 Models-3 Emission Projection Processor (MEPRO) 4-91
4.4 Emission Processing Interface ...:.,.... 4-94
4.4.1 Overview of Key Features of ECIP 4-94
4.4.2 Plume Rise of Point Source Emissions 4-96
4.4.3 Method for the Treatment of Initial Vertical Plume Spread 4-99
4.4.4 Vertical Allocation of Plume Emissions ....................... 4-100
4.4.5 Generation of 3-D Emissions •... 4-100
4.5 Data Requirements 4-100
4.6 Plans for Improvement .'.. 4-102
4.7 References 4-103
5. FUNDAMENTALS OF ONE-ATMOSPHERE DYNAMICS FOR MULTISCALE
AIM QUALITY MODELING
(D. Byun)
Abstract , 5-1
5.0 Fundamentals of One-Atmosphere Dynamics for Multiscale Air Quality
Modeling . 5-2
5.1 Governing Equations and Approximations for the Atmosphere 5-2
5.1.1 Governing Equations in a Generalized Curvilinear Coordinate System 5-3
5.1.2 Assumptions of Atmospheric Dynamics ..5-6
5.2 Choice of Vertical Coordinate System for Air Quality Modeling 5-11
5.3 Coupling of Meteorology and Air Quality 5-18
5.3.1 Meteorological Data for Air Quality Modeling ».-...„-. 5-18
5.3.2 Off-line and On-line Modeling Paradigms 5-18
5.4 Mass Conservation ,....„. 5-22
5.4.1 Mass Consistency in Meteorological Data .-. 5-22
5.4.2 Techniques for Mass Conservation in Air Quality Models 5-23
5.4.3 Temporal Interpolation of Meteorological Data 5-25
5.5 Conclusion 5-26
5.6 References .................,.......;........:;.;....... 5-27
Appendix 5 A. Tensor Primer and Derivation of the Continuity Equation in a
Generalized Curvilinear Coordinate System 5-30
IX
-------
EPA/600/R-99/030
6. GOVERNING EQUATIONS AND COMPUTATIONAL STRUCTURE OF THE
COMMUNITY MULTISCALE AIR QUALITY (CMAQ) CHEMICAL
TRANSPORT MODEL
(D. Byun, J. Young, andM.T. Odman)
Abstract , 6-1
6.0 Governing Equations And Computational Structure of The Community
Multiscale Air Quality (CMAQ) Chemical Transport Model 6-2
6.1 Derivation of the Atmospheric Diffusion Equation 6-3
6.2 Representation of Science Processes in CMAQ Modeling System 6-9
6.2.1 Supporting Models and Interface Processors 6-9
6.2.2 Modularity Concept of CMAQ 6-10
6.2.3 Description of Science Processes 6-15
6.3 Equivalent Model Formulations for Different Vertical Coordinates 6-26
6.4 Nesting Techniques '.... 6-28
6.5 Summary 6-31
6.6 References 6-31
Appendix 6A. Concentration Units Used for Air Quality Studies 6-34
7. NUMERICAL TRANSPORT ALGORITHMS FOR THE COMMUNITY
MULTISCALE AIR QUALITY (CMAQ) CHEMICAL TRANSPORT MODEL
IN GENERALIZED COORDINATES
(D. Byun, J. Young, J. Pleim, M.T. Odman, andK. Alapaty)
Abstract 7-1
7.0 Numerical Transport Algorithms for the Community Multiscale Air Quality
(CMAQ) Chemical Transport Model in Generalized Coordinates 7-2
7.1 Numerical Advection Algorithms , 7-3
7.1.1 Conservation Form Equation for Advection 7-4
7.1.2 Classification of Advection Schemes 7-6
7.1.3 Description of Advection Schemes in CCTM 7-6
7.1.4 Treatment of Boundary Conditions 7-11
7.1.5 Test of Algorithms with Idealized Linear Horizontal Flow Fields 7-12
7.1.6 Vertical Advection 7-15
7.1.7 Adjustment of Mass Conservation Error 7-16
7.2 Vertical Mixing Algorithms 7-18
7.2.1 Closure Problem 7-19
7.2.2 Computing Vertical Mixing with the Eddy Diffusion Formulation:
K-Theory 7-21
7.2.3 Flux Form Representation of Vertical Mixing 7-31
7.3 Horizontal Mixing Algorithms 7-43
7.4 Conclusions 7-45
7.5 References 7-46
Appendix 7A Numerical Solvers for Diffusion Equations 7-51
-------
EPA/60WR-99W30
8. GAS-PHASE CHEMISTRY
(G. Gipson andJ. Young)
Abstract ...,....,..,...,... 8-1
8.0 Gas-Phase Chemistry 1 8-2
8,1 Background 8-3
8.2 Chemical Mechanisms in the CMAQ System 8-4
8.2.1 CB4 Mechanism 8-5
8.2.2 . RADM2 Mechanism '-. 8-6
8.2.3 SAPRC-97 Mechanism 8-7
8.2.4 Extended Mechanisms ..,.....,,..... 8-8
8.2.5 Changing or Adding Mechanisms in CMAQ 8-11
8.3 Reaction Kinetics 8-12
8.3.1 Reaction Rates 8-12
8.3.2 Rate Constant Expressions 8-13
8.4 Mathematical Modeling 8-15
8.4.1 Governing Equations 8-16
8.4.2 SMVGEAR 8-18
8.4.3 QSSA Solver 8-22
8.4.4 Summary , 8-25
8.5 References 8-25
Appendix 8A Chemical Mechanisms Included in the CMAQ System 8-29
9. PLUME-IN-GRID TREATMENT OF MAJOR POINT SOURCE
EMISSIONS
(N. Gillani andJ. Godowitch)
Abstract 9-1
9.0 Plume-in-Grid Treatment of Major Point Source Emissions 9-2
9.1 Introduction .............'. 9-2
9.2 Overview of the Conceptual Framework of the Plume-in-Grid Treatment .. 9-4
9.3 Formulation of the Plume-in-Grid Modeling Components 9-5
9.3.1 Description of the Plume Dynamics Model 9-5
9.3.2 Formulation of the Plume-in-Grid Module : 9-12
9.4 Summary 9-28
9.5 References 9-28
XI
-------
EPA/600/R-99/030
10. AEROSOLS IN MODELS-3 CMAQ
(F. Binkowski)
Abstract , 10-1
10.0 Aerosols in Models-3 CMAQ 10-2
10.1 Aerosol Dynamics 10-4
10.1.1 Modal Definitions 10-4
10.1.2 New Particle Production by Nucleation 10-5
10.1.3 Primary Emissions 10-6
10.1.4 Numerical Solvers 10-7
10.1.5 Mode Merging by Renaming 10-10
10.2 Aerosol Dry Deposition 10-11
10.3 Cloud Processing of Aerosols 10-11
10.4 Aerosol Chemistry 10-14
10.5 Visibility 10-15
10.6 Summary 10-17
10.7 References 10-17
11. CLOUD DYNAMICS AND CHEMISTRY
(S. Roselle and F. Binkowski)
Abstract 11-1.
11.0 Cloud Dynamics and Chemistry 11-2
11.1 Background , 11-2
11.2 Model Description 11-2
11.2.1 Subgrid Convective Cloud Scheme 11-3
11.2.2 Resolved Cloud Scheme 11-7
11.3 Conclusions 11-8
11.4 References 11-9
12. METEOROLOGY-CHEMISTRY INTERFACE PROCESSOR (MCIP) FOR
MODELS-3 COMMUNITY MULTISCAL1 AIR QUALITY (CMAQ)
MODELING SYSTEM
(D. Byun, J. Pleim, R. Tang, and A. Bourgeois)
Abstract 12-1
12.0 Meteorology-Chemistry Interface Processor (MCIP) For Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System 12-2
12.1 Introduction 12-2
12.1.1 MCIP Functions 12-4
12.1.2 MCIP's Data Dependency 12-6
12.1.3 Computational Structure 12-8
12.2 Data Types, Coordinates, and Grids 12-10
12.2.1 Meteorological Data Types 12-10
12.2.2 Coordinates 12-14
12.2.3 Modification of Grid Structure 12-18
xii
-------
EPA/600/R-99/030
12.3 Estimation of Physical Parameters , 12-23
12.3.1 PEL Parameters 12-23
12.3.2 Dry Deposition Velocities 12-37
12.3.3 Cloud Parameters and Solar Radiation 12-49
12.4 Meteorological Data for CCTM with Generalized Coordinate System 12-52
12.4.1 Thermodynamic Variables: Pressure, Density and Entropy 12-52
12.4.2 Vertical Jacobian and Layer Height 12-54
12.4.3 Contravariant Velocity Components 12-57
12.4.4 Mass Consistent Temporal Interpolation of Meteorological
Parameters «.. 12-60
12.4.5 Optional Conversion of Nonhydrostatic Data to Hydrostatic
Meteorological Data for MM5 . 12-61
12.5 Operation of MCIP 12-63
12.5.1 MCIP Modules 12-63
12.5.2 Building MCIP 12-63
12.5.3 Executing MCIP 12-65
12.5.4 Defining Grid and Domain for MCIP . 12-68
12.5.5 Extension of MCIP for Other Meteorological Models 12-72
12.6 Concluding Remarks . . .»....,.. 12-73
12.7 References . 12-74
Appendix 12A MCIP Output Data 12-80
Appendix 12B Examples of Nest Domain Definitions for CMAQ system 12-83
Appendix 12C Sample MCIP Configuration File 12-85
Appendix 12D Sample MCIP Run Script 12-87
13. THE INITIAL CONCENTRATION AND BOUNDARY CONDITION
PROCESSORS
(G, Gipson)
Abstract ; ...,; 13-1
13.0 The Initial Concentration and Boundary Condition Processors 13-2
13.1 Introduction .„ 13-2
13.2 Overview of the ICON and BCON Processors 13-2
13.3 Input Sources ,, 13-3
13.3.1 Time Invariant Concentration Profiles 13-3
13.3.2 CCTM Concentration files 13-5
13.3.3 Tracer Species ...•.,.,.» .* 13-5
13.4 Spatial Interpolation 13-7
13.4.1 Horizontal Interpolation 13-7
13.4.2 Vertical Interpolation 13-8
13.5 ICON/BCON Species Processing 13-9
13.6 Mechanism Conversions 13-11
13.7 ICON/BCON Applications I:.... 13-12
13.8 References 13-13
xiii
-------
EPA/600/R-99/030
14. PHOTOLYSIS RATES FOR CMAQ
(S. Rosette, K. Schere, J. Pleim, A, Hanna)
Abstract 14-1
14.0 Photolysis Rates for CMAQ 14-2
14.1 Background 14-2
14.2 Preprocessor JPROC: Calculate Clear-sky Photolysis Rate Table 14-3
14.3 Subroutine PHOT: Table Interpolation and Cloud Attenuation 14-4
14.4 Conclusions 14-5
14.5 References 14-6
15. PROGRAM CONTROL PROCESSING IN MODELS-3
(J. Young)
Abstract 15-1
15.0 Program Control Processing in Models-3 15-2
15.1 Domain Configuration 15-3
15.2 Input/Output Applications Programming Interface 15-4
15.3 Other CCTM Configuration Control 15-4
15.3.1 CCTM Process Analysis 15-4
15.3.2 CCTM Fixed Data 15-5
15.4 Generalized Chemistry 15-6
15.4.1 Design 15-6
15.4.2 Operation 15-7
15.4.3 Supported Reaction Types 15-9
15.4.4 Mechanism Parsing Rules 15-10
15.4.5 Chemical Species Include Files 15-10
15.6 Conclusion 15-16
15.7 References 15-16
16. PROCESS ANALYSIS
(G. Gipson)
Abstract 16-1
16.0 Process Analysis 16-2
16.1 Integrated Process Rate Analysis 16-3
16.1.1 Computation of Integrated Process Rates 16-3
16.1.2 Example IPR Analyses 16-5
16.1.3 Implementation of IPR Analysis in the CMAQ System 16-6
16.1.4 Use of the PACP to set up an IPR Analysis 16-7
16.2 Integrated Reaction Rate Analysis 16-9
16.2.1 Computation of Integrated Reaction Rates 16-10
16.2.2 Example IRR Analyses 16-11
16.2.3 Implementation of IRR Analysis the CMAQ system 16-13
16.2.4 Use of the PACP to set up an IRR Analysis 16-14
xiv
-------
EPA/6QG/H-99/Q3Q
16.3 Conclusion , , 16-19
16.4 References 16-20
17. AN AGGREGATION AND EPISODE SELECTION SCHEME DESIGNED TO
SUPPORT MODELS-3 CMAQ
(R. Cohn, B. Eder, andS. Leduc)
Abstract 17-1
17.0 An Aggregation and Episode Selection Scheme Designed to Support Models-3
CMAQ 17-2
17.1 Introduction 17-2
17.1.1 Background 17-2
17.1.2 Objectives 17-3
17.2 Summary of the Approach 17-4
17.2.1 Basic Elements of the Methodology .....», 17-4
17.2.2 Rationale, Scope, and Limitations 17-5
17.2.3 Strategy 17-6
17.3 Cluster Analysis of Wind Fields 17-8
17.3.1 Description of Wind Data , 17-8
17.3.2 Basic Cluster Analysis Technique ,... 17-8
17.3.3 Illustration of Cluster Analysis Results 17-9
17.4 Evaluation of Alternative Aggregation Approaches 17-11
17.4.1 Description of Alternative Approaches 17-11
17.4.2 Description of Meteorological Data 17-12
17.4.3 Analysis Methods 17-12
17.4.4 Results 17-13
17.5 Refinement of the Sampling Approach 17-21
17.5.1 Determination of Appropriate Numbers of Strata and Events 17-22
17.5.2 Selection of Stratified Sample of Events ; 17-27
17.6 Application and Evaluation 17-29
17.6.1 Application of the Aggregation Procedure 17-29
17.6.2 Evaluation 17-31
17.7 Summary and Discussion ....; 17-32
17.8 References 17-33
18. INTEGRATION OF SCIENCE CODES INTO MODELS-3
(J. Young)
Abstract 18-1
18.0 Integration of Science Codes into Models-3 18-2
18.1 Introduction 18-2
18.2 Classes and Modules 18-4
18.2.1 Operational Design 18-4
18.2.2 CGRID 18-5
18.2.3 Class Driver 18-6
xv
-------
EPA/600/R-99/030
18.2.4 Synchronization Time Step 18-7
18.3 Input/Output Applications Programming Interface 18-8
18.4 Code Configuration Management 18-10
18.4.1 The Need 18-10
18.4.2 The Tool 18-11
18.4.3 The Repository 18-12
18.5 How a Model is Constructed , 18-14
18.5.1 Object Oriented Concepts 18-14
18.5.2 Global Name Table Data 18-14
18.5.3 Build Template 18-15
18.6 How a Model is Executed 18-18
18.7 Using the Models-3 Framework 18-18
18.8 Conformant Code 18-19
18.8.1 Thin Interface 18-19
18.8.2 Coding Rules 18-20
18.8.3 Science Process Code Template 18-21
18.8.4 Robustness and Computational Efficiency 18-24
18.9 Conclusion 18-25
18.10 References 18-25
xvi
-------
EPA/600/R-99/030
ADEOS
ADOM
AQM
ASD
AVS
BCs
BCON
BDF
BEIS
BEIS2
HELD
EOT
CAAA-90
CB-IV (or CB-4)
CBWQM
CCM2
CCTM
GEM
CFL
CG
CMAQ
CORBA
CPU
CTM
CVS
ECIP
EMPRO
EPA
FAST
FDDA
FG
FIPS
FSL
GCIP
GEMAP
HSPF
I/O API
ICs -
ICON
IDA
IEEE
ACRONYMS
Advanced Earth Observing Satellite
Acid Deposition and Oxidant Model
Air Quality Model
Accurate Space Derivative advection scheme
Advanced Visualization System .
Boundary Conditions
Boundary Conditions processor
Backward Differentiation Formulae
Biogenic Emissions Inventory System
Biogenic Emissions Inventory System -. version 2
Biogenic Emissions Landuse Database
Bott Scheme for advection
Clean Air Act Amendments of 1990
Carbon Bond-IV chemical mechanism
Chesapeake Bay Water Quality Model
Community Climate Model Version 2
CMAQ Chemical Transport Model processor
Continuous Emission Monitor
Courant-Friedrich-Levy condition
Coarser Grid
Community Multiscale Air Quality model
Common Object Request Broker Architecture
Central Processing Unit of a computer
Chemical Transport Model
Concurrent Versions System
Emissions-Chemistry Interface Processor
Emissions Processor
Environmental Protection Agency
Flow Analysis Software Toolkit
Four-Dimensional Data-Assimilation
Finer Grid
Federal Identification Protocol System
Forecast Systems Laboratory
GEWAX Continental Scale International Project
Geocoded Emission Modeling and Projection
Hydrologic Simulation Program FORTRAN
Input/Output Applications Programming Interface
Initial Concentration
Initial Conditions processor
Inventory Data Analyzer
Institute of Electrical and Electronics Engineers, Inc.
TfVll
-------
EPA/600/R-99/030
INPRO
IPR
IRR
JPROC
LRPM
LUPROC
MAPS/RUG
MCIP
MEPPS
MEPRO
MEPSE
MEPSEs
MM5
MPS
MRF
NAAQS
NARSTO-NE
NASA
NCAR
NCEP
NOAA
NOX
NQS
NSSDC
03
ODE
OODBMS
OUTPRO
PACP
PAN
PARTS
PAVE
PEL
PCP
PDM
PinG
PM
PM10
PM2.5
PPM
PSU
QSSA
RADM
MEPPS Input Processor
Integrated Process Rates
Integrated Reaction Rates
Photolysis Rate Processor
LaGrangian Reaction Plume Module
Landuse processor for MCIP
Mesoscale Analysis and Prediction System/Rapid Update Cycle
Meteorology-Chemistry Interface processor
Models-3 Emissions Processing and Projection System
Models-3 Emissions Projection Processor
Major Elevated Point Source Emitter
Major Elevated Point Source Emitters
Fifth Generation Mesoscale Model
Multiple Projection System
Medium Range Forecast
National Ambient Air Quality Standards
North American Research Strategy for Tropospheric Ozone - Northeast
National Aeronautics and Space Administration
National Center for Atmospheric Research
National Centers for Environmental Prediction
National Oceanic and Atmospheric Administration
Oxides of Nitrogen
Network Queuing Service
National Satellite Service Data Center
Ozone
Ordinary Differential Equations
Object Oriented Data Base Management System
MEPPS Output Processor
Process Anaylsis Control Program
Peroxy Acetyl Nitrate
Mobile Source Particulate Model 5
Package for Analysis and Visualization of Environmental Data
Planetary Boundary Layer
Program Control Processor
Plume Dynamics Model
Plume-in-Grid
Particulate Matter
Particulate Matter less than 10 /
-------
EPA/6QQ/R.-99fG3Q
RADM2
RCS
RELMAP
RMSE
ROG
ROM
RPM
SAPRC
SAQM
SAS
SCC
sees
SIP
SMO
SMOKE
SMVGEAR
TOG
SOX
STEM
SVOC
TKE
TOMS
UAM
UCAR
UTC
VMT
voc
WRP
XDR
YAM
Regional Acid Deposition Model Version 2
Revision Control System
Regional Lagrangian Modeling of Air Pollution
Root Mean Square Error
Reactive Organic Gases
Regional Oxidant Model
Regional Particulate Model
State Air Pollution Research Center chemical mechanisms
SARMAP Air Quality Model
Statistical Analysis System
Source Classification Code
Source Code Control System
State Implementation Plan
Smolarkiewicz advection scheme
Sparse Matrix Operator Kernel Emissions system
Sparse Matrix Vectorized Gear algorithm
Total Organic Gases
Oxides of Sulfur
Sulfur Transport and Emissions Model
Semi-Volatile Organic Compounds
Turbulent Kinetic Energy
Total Ozone Mapping Spectrometer
Urban Airshed Model
University Corporation for Atmospheric Research
Universal Time Coordinate
Vehicle Miles Traveled
Volatile Organic Compound
Weather Research and Forecasting
external Data Representation
Yamartino-Blackrnan Cubic Scheme for advection
xix
-------
-------
EPA/600/R-99/030
ACKNOWLEDGMENTS
The development of the science components of the Models-3 Community Multiscale Air Quality
(CMAQ) system represents a major undertaking by a large team of dedicated atmospheric
scientists and a relatively long term effort beginning in the early 1990s, and covers a wide area of
subject material as embodied in each of the contributing chapters. Most of the EPA Models-3
CMAQ Science Team are from the Atmospheric Modeling Division (AMD) in the EPA Office of
Research And Development's National Exposure Research Laboratory (NERL). With several
exceptions the contributing authors from this Division are on detail from the Air Resources
Laboratory of the National Oceanic and Atmospheric Administration (NOAA). The NOAA-
AMD effort is supported through Interagency Agreement (DW13937252) with the EPA. The
authors of each of the chapters of this document would like to collectively express their gratitude
to the other members of the NOAA Division, In particular we acknowledge the encouragement
and support of the Director of the Atmospheric Modeling Division, Francis Schiermeier. We are
indebted to Dr. Robin Dennis for his vision and role in ensuring the implementation of holistic-
one-atmosphere approach into CMAQ. We also recognize the technical help from our colleagues
Tom Pierce, Steven Howard, Alfreida Torian, and Gary Walter.
We also want to recognize the contributions, helpful discussions, and assistance of many members
of the science community, either through formal agreements or through peer interest. For
example, the work was supported by several cooperative agreements including:
« Atmospheric Modeling Research-Scientific and Computational (CR-822066: MCNC),
Principal Investigator (PI): Kenneth Galluppi Ed Bilicki, Steve Fine, Alison Eyth, and
Rohit Mathur;
* Research on Computational Framework in Generic Grids, Adaptive Grids, and Subgrid
Treatment of Air Quality Simulation (CR-822053: MCNC), PI: Talat Odman; with R.K.
Srivastava and D.S. McRae, North Carolina State University);
• Transport Algorithms for Air Quality Simulation Models (CR-822059, MCNC), PI: Talat
Odman;
• Emissions Modeling Research with High-Performance Computing (CR-822074: MCNC),
PI: Carlie Coats;
• Advanced Modeling of Meteorology in Support of Air Quality Models (CR-822628:
(MCNC), PI: Aijun Xiu, Kiran Alapaty, John N. McHenry, and Adel F. Hanna; with
Nelson L. Seaman and Aijun Deng, Pennsylvania State University; and John S. Kain,
National Severe Storms Laboratory);
• A Flexible and Efficient Methodology for Modeling Aerosol for Air Quality Models
(CR-823634: MCNC), PI: Uma Shankar, Mark Read and Atanas Trayanov; with Anthony
Wexler, University of Delaware; and John H. Seinfeld, California Institute of Technology;
• Develop Methods for Technology Transfer of Advanced Regional/Urban Air Quality
Models (CR-822080: NCSU), PI: Alan Schula; and
« Plume-in-Grid development for a Multiscale Air Quality Modeling System (IAG
DW64937190, Tennessee Valley Authority (TVA)), Pi's: Robert E. Imhoff; with Noor
Gillani, Arastoo Biazar (now at Monash University, Australia), and Yu-Ling Lu,
xxi
-------
EPA/600/R-99/030
University of Alabama, Huntsville).
The following scientists also contributed under the formal contracts listed: Ruen Tang, Chris
Maxwell, and Hao Jin of the Technical Support Group, Dyntel Corp. (General Services
Administration IAG DW47937823); Tod Plessel, and Yan Ching Zhang of the Visualization
Laboratory, Lockheed Martin (EPA Contract 68-W7-0055); and Nick Moghari, Joe Susick, and
Dave Tivel of Science Applications International Corporation (SAIC) (EPA Contract 68-W1-
0055).
Significant contributions were made by: visiting scientists, including Sang-Mi Lee of the Seoul
National University, Korea, Chong Bum Lee of the KangWeon University, Korea, and Seiji
Sugata of the National Institute for Environmental Studies, Japan; UCAR's post doctoral fellows
Yonghong Li, Qingyuan Song and Shoba Kondragunta; and Dr. Ingmar Akermann of Ford
Research in Aachen, Germany.
It is also a pleasure to thank the colleagues who engaged us in many seminal technical discussions
on various aspects of the CMAQ system. We recognize Professors Harvey Jeffries, University of
North Carolina at Chapel Hill; Ted Russell, Georgia Institute of Technology; Nelson Seaman and
Jack Kain, Pennsylvania State University; Dick McNider, University of Alabama, Sonia
Kreidenweis, Colorado State University, Panos Georgeopolus of Rutgers University, and Itsushi
Uno of Kyushu University, Japan. Many of our EPA colleagues provided stimulating discussion
or reviews of our effort including Gail Tonnesen, Deborah Luecken, Carey Jang, John S. Invin
(NOAA) and Ed Edney.
We are indebted to the following peer reviewers, who gave unselfishly of their time and whose
comments and suggestions were extremely valuable in improving the scientific aspects of the
report. Listed in alphabetical order, the reviewers are followed by the chapter(s) they reviewed:
Jeff Brook, Atmospheric Environment Service, Canada (Ch. 17); David Chock, Ford Research
Laboratory (Ch. 2 and 7); Henry Hogo, Southern California Air Quality Modeling Division (Ch.
8 and 16); Sasha Madronich, National Center for Atmospheric Research (Ch. 14); Paulette
Middleton, RAND Environmental Science & Policy Center (Ch. 10); Ted Russell, Georgia
Institute of Technology (Ch. 18); Nelson Seaman, Pennsylvania State University (Ch. 3);
Christian Seigneur, Atmospheric and Environmental Research, Inc. (Ch. 10 and 11); Trevor
Sholtz, ORTECH Corporation (Ch. 4); Saffet Tanrikulu, California Environmental Protection
Agency (Ch. 9 and 12); Robert Yarmartino, Earth Tech (Ch. 5 and 6); and Zion Wang (Ch. 13,
15, and 16).
Finally, the editors note with great appreciation, the efforts of Brian Eder, who facilitated many of
the activities and tasks involved in the preparation and completion of this document. We thank
the Raleigh, North Carolina, staff of SAIC, especially Alice Gilliland (now with NOAA-AMD)
and Andrea Verykoukis for their critical contributions in the area of technical editing, assuring a
high level of consistency and quality in all the chapters of this manuscript. We are also grateful to
Alice Gilliland for her invaluable work on the Executive Summary.
xxii
-------
EPA/600/R-99/OSO
EXECUTIVE SUMMARY
THIS SCIENCE DOCUMENT PRESENTS THE PROCESSORS AND ALGORITHMS THAT
EMBODY THE INITIAL RELEASE OF THE MODELS-3 COMMUNITY MULTISCALE
AIR QUALITY (CMAQ) MODELING SYSTEM. CMAQ is A MULTIPLE POLLUTANT
MODEL THAT CONTAINS NEW SCIENTIFIC APPROACHES TO AIR QUALITY
MODELING, WHICH REPRESENT THE CURRENT STATE OF SCIENCE. This CMAQ
Science Document is a living document that will be updated as the state of the science progresses.
The CMAQ Science Document provides a basis and point of reference for the state of the science
captured in the June 1998 initial release of Models-3. Current and future efforts to improve the
Models-3 modeling system(s) will depart from the scientific reference points presented in this
document.
Models-3 is a flexible software system that provides a user-interface framework for CMAQ air
quality modeling applications and tools for analysis, management of model input/output, and
visualization of data. The Models-3 framework relies on two modeling systems to provide the
meteorological and emissions data needed for air quality modeling. With this data, the Models-3
CMAQ modeling system can be used for urban and regional scale air quality simulation of
tropospheric ozone, acid deposition, visibility, and particulate matter (PM25 and PM,0). The
meteorological and emissions modeling systems that are provided with the current release of
Models-3 will be described in this document. However, CMAQ is designed as an open system
where alternative models can be used to generate the data.
This CMAQ Science Document contains chapters that address specific scientific and technical
issues involved in the development and application of the Models-3 CMAQ modeling system.
The principal researchers for each model component or function authored the coinciding chapter
in this document. They serve as the points of contact for scientific questions regarding their
CMAQ air quality model components. For instructions on using the Models-3 framework and
using the MM5, MEPPS, and CMAQ modeling systems, refer to the Models-3 User Manual
(EPA/6()0/R-98/069b, National Exposure Research Laboratory, Research Triangle Park, NC) and
Tutorial (EPA/600/R-98/069c, National Exposure Research Laboratory, Research Triangle Park,
NC). ' •
An overview of the MEPPS emissions, MM5 meteorology, and CMAQ air quality modeling
systems is provided in Chapter 1. Chapter 2 then introduces the Models-3 framework and
structure and explains how the framework's user-interface is used with the MEPPS and CMAQ
modeling systems. More detailed discussions on the modeling systems are then discussed
separately in the following chapters. The amount of detail and the length of these discussions
vary depending on whether this information has already been provided elsewhere. Some chapters
provide a synopsis of the scientific components and refer to previously published material on the
subject, while other chapters provide extensive detail on new scientific techniques that are not
currently described in other publications.
-------
BPA/600/R-99/030
MM5 Meteorological data are essential for many processes simulated in the CMAQ chemical
transport model including transport, chemistry, and cloud processes. The Fifth-Generation Penn
State/NCAR Mesoscale Model (MM5) is the only meteorology model compatible with the initial
release of Models-3. MM5 is a complex, state-of-the-scienee community model, and it is
maintained by NCAR. MM5 is well documented by its primary developers in technical notes and
refereed journal articles. Chapter 3 briefly describes the scientific aspects of MM5, including grid
definitions, model physics, nesting, and four dimensional data assimilation. These descriptions
generally direct the user to more complete documentation about particular aspects of MM5. To
promote the flexibility of CMAQ, additional meteorology models will be compatible with Models-
3 future releases.
MEPPS Chapter 4 provides a description of the Models-3 Emission Processing and
Projection System (MEPPS) structure, its scientific approach, and the assumptions used in
modeling and processing emission data in the Models-3 framework. The chapter also discusses
data flow and quality control used with emission inventory and meteorological input data for
MEPPS. The description of the main Emission Processor addresses the basis of spatial and
temporal allocation procedures, and the methods and assumptions used in modeling mobile and
biogenic emissions and in the "lumping" of individual chemical species are also presented. This
chapter also explains the procedures used by the Models-3 Emission Projection Processor to
estimate emission data for use in modeling future air quality scenarios.
THE CMAQ CHEMICAL TRANSPORT MODEL (CCTM)
Fundamentals of One-Atmosphere Dynamics for Multiscale Air Quality Modeling
Chapter 5 provides information essential to the proper use of meteorological data in air
qualify modeling systems. The chapter introduces a robust and folly compressible set of
governing equations for the atmosphere, which provides an integral view of atmospheric
modeling. The limitations of several simplifying assumptions on atmospheric dynamics are
presented, as are concepts of on-line and off-line coupling of meteorological and air
quality models. In addition, this chapter describes a procedure for conserving the mixing
ratio of trace species even in the case of meteorological data that are not mass consistent.
In summary, Chapter 5 attempts to bridge the information gap between dynamic
meteorologists and air quality modelers by highlighting the implication of using different
meteorological coordinates and dynamic assumptions for air quality simulations.
Governing Equations and Computational Structure In Chapter 6, the governing
diffusion equation is derived in a generalized coordinate system, which is suitable for
multiscale atmospheric applications. CMAQ's use of generalized coordinates for its
governing equations provides the flexibility to span multiple scales and to incorporate
meteorological data on different coordinates. The CMAQ system's modularity concepts
and fractional time-step formulation, and CCTM's key science processes are described.
Chapter 6 also presents the dynamic formulations of several popular Eulerian air quality
ES-2
-------
EPA/600/R-99/030
models as emulated by the governing diffusion equations in the generalized coordinate
system,
Numerical Transport Algorithms The transport processes in the atmosphere primarily
consist of advection and diffusion. In Chapter 7, CMAQ's numerical algorithms for
advection and vertical and horizontal diffusion are discussed. To provide the CMAQ
system with multiscale capability, the transport processes, both advection and diffusion,
are formulated in conservation (i.e., flux) forms for the generalized coordinate system.
Therefore, CMAQ's numerical transport algorithms will function under a wide variety of
dynamical situations and concentration distribution characteristics. Users are encouraged
to experiment with their own algorithms to test different numerical schemes for air quality
simulations.
Gas Phase Chemistry Chapter 8 examines the way gas-phase chemistry is treated
in CMAQ. The CMAQ system currently includes two base chemical mechanisms,
RADM2 and CB4, while the incorporation of a third, the SAPRC97 mechanism, is
planned for the future. Chapter 8 describes each of these chemical mechanisms as well as
the manner in which the first two are linked to the aqueous chemistry and aerosol
formation processes. The chapter also discusses procedures for entering new chemical
mechanisms in the CMAQ system, the representation of reaction kinetics, the numerical
modeling of gas-phase chemistry, and the two numerical solvers included in CMAQ,
SMVGEAR and a variant of the QSSA method.
PIume-in-Grid Chapter 9 introduces the plume-in-grid (PinG) technique developed
for CMAQ. PinG is designed to treat more realistically the dynamic and chemical
processes impacting selected major point source pollutant plumes in CMAQ. The Plume
Dynamics Model (PDM) simulates plume rise, horizontal and vertical plume growth, and
transport of each plume section during the subgricl scale phase. The PinG module
simulates the relevant physical and chemical processes during a subgrid scale phase. This
technique is in contrast to the traditional Eulerian grid modeling method of instantly
mixing the point source emissions into an entire grid cell volume. Chapter 9 describes the
technical approach and model formulation of the relevant processes, and discusses the
capabilities and limitations of the initial version of the PinG approach.
The Aerosol Module One of CMAQ's key strengths is that it is a multi-pollutant model
that fully addresses the criteria pollutants PM and ozone. Chapter 10 discusses the
aerosol module of CMAQ, which is designed to be an efficient and economical depiction
of aerosol dynamics in the atmosphere. This chapter discusses the techniques for
distributing particulates in three modes: coagulation, particle growth by the addition of
new mass, and particle formation. The aerosol module considers both PM2 5 and PM10 and
includes estimates of the primary emissions of elemental and organic carbon, dust, and
other species not further specified. Secondary species considered are sulfate, nitrate,
ammonium, water and organic from precursors of anthropogenic and biogenic origin.
ES-3
-------
EPA/600/R-99/030
Cloud Chemistry and Dynamics Chapter 11 discusses the role and functions of
clouds in CMAQ. Clouds are involved in aqueous chemical reactions, vertical mixing of
pollutants, and removal of aerosols by wet deposition, all of which affect the concentration
of air pollutants. CMAQ's cloud module performs several functions related to cloud
physics and chemistry, and it models three types of clouds: sub-grid convective
precipitating clouds, sub-grid non-precipitating clouds, and grid-resolved clouds. The
cloud module vertically redistributes pollutants for the sub-grid clouds, calculates in-cloud
and precipitation scavenging, performs aqueous chemistry, and accumulates wet
deposition amounts.
CMAQ INTERFACE PROCESSORS PREPARE INPUT DATA FROM SOURCES
INCLUDING THE EMISSION AND METEOROLOGICAL MODELING SYSTEMS FOR USE
IN THE CMAQ CTM. EACH OF THESE PROCESSORS, EACH OF WHICH HAVE
SPECIFIC FUNCTIONS, ARE DESCRIBED IN THE CHAPTERS MENTIONED BELOW. The
interface processors that handle input data from the emissions and meteorological models are
essential because CMAQ is an open system in which meteorological and emissions data are
calculated separately (i.e., "off-line"), rather than during the chemical transport model simulation.
These interface processors also add extra quality control, so that inconsistencies between input
data and the CCTM are minimized.
ECIP In addition to describing the Models-3 MEPPS emission modeling system, Chapter
4 discusses the Emission-Chemistry Interface Processor (ECIP). ECIP serves as the key
link between the MEPPS system and CCTM. ECIP's primary function is to generate
hourly 3-D emission data files for CCTM from the individual emission file types produced
by the MEPPS. The key inputs for ECIP are the area emissions file, the stack parameter
and emission files for the point sources generated in MEPPS, and a set of meteorological
data files generated by the Meteorology-Chemistry Interface Processor (MCIP) for the
CCTM domain. All major point sources are subject to plume-rise and initial vertical
dispersion processes before being allocated to a particular vertical model layer.
MCIP Chapter 12 describes MCIP, which links meteorological models, such as MM5,
with the CCTM system to provide the complete set of meteorological data needed for air
quality simulation. To support CCTM's multiscale generalized coordinate
implementation, MCIP provides appropriate dynamic meteorological parameters to allow
mass-consistent air quality computations. MCIP deals with issues related to data format
translation, conversion of parameter units, diagnostic estimations of parameters not
provided, extraction of data for appropriate window domains, and reconstruction of
meteorological data on different grid and layer structures. MCIP also relies on the
Landuse Processor (LUPROC) to provide landuse and vegetation information to define
surface characteristics to compute dry deposition and other PBL parameters. LUPROC
extracts information about landuse in the CMAQ domain from a landuse database and
converts it into the fractional landuse data used in MCIP.
ES-4
-------
EPA/600/R-99/030
Initial and Boundary Conditions Initial conditions provide a simulation's starting
point, while boundary conditions define influences from outside the domain.
Chapter 13 describes the two interface processors that generate the concentration fields
for the initial and boundary conditions for CCTM. The chapter describes how the initial
condition (ICON) and boundary condition (BCON) processors can be used to generate
the concentration fields from either predefined default vertical profiles or from other
CMAQ simulation results when model nesting is being performed. This chapter also
discusses generating initial and boundary concentrations for special tracer species and
procedures for horizontal and vertical interpolation, and conversions between chemical
mechanisms.
Photolysis Rate Processor Many chemical reactions in the atmosphere are initiated by
the photodissociation of numerous trace gases, including NO2,03, and HCHO, Chapter
14 describes the photolysis rate processor (JPROC) that produces the photolysis rates
used in the CMAQ chemical transport simulation. JPROC predicts photolysis rates for
various altitudes, latitudes, and zenith angles. Currently, the radiative transfer algorithm
assumes clear-sky conditions (i.e., no clouds present), and CCTM then attenuates for
cloudiness.
As described above, each of the CMAQ interface processors incorporate raw data for CMAQ and
perform functions such as calculating parameters and interpolating or converting data. The
functions of the interface processors also include capabilities to handle raw data with various
resolutions or measurement units. Raw input data is currently specified in the source code for
JPROC, LUPROC, ICON, and BCON; however, the interface processors in future releases of
CMAQ will be modified to handle a more generalized set of input data.
PROGRAM CONTROL PROCESSORS, A SET OF PROGRAMS EMBEDDED m THE
MODELS-3 FRAMEWORK, HANDLE SCIENCE INFORMATION OBJECTS SUCH AS
GRID AND LAYER SPECIFICATIONS, CHEMICAL MECHANISMS, AND MODEL
CONFIGURATIONS FOR REPEATED USE ACROSS SEVERAL PROCESS COMPONENTS
OF CMAQ.
Program Control Processing (PCP)Chapter 15 explains how PCP is used within
Models-3 to set up internal arrays, map species names, define global parameters, and
establish linkages among processors in the Models-3 CMAQ system. Specifications
needed for the CCTM simulation (e.g., grid and coordinate conditions and chemical
species names) are entered into the Models-3 system once by the graphical user interfaces,
and an object-oriented database accessible by all model components is established. PCP
utilizes this information in the object database and automatically generates the required
global FORTRAN include files. As a part of PCP, Models-3 CMAQ system employs a
generalized chemistry mechanism processor (MP), also called the "mechanism reader." It
greatly simplifies the task of implementing or altering gas-phase chemistry mechanisms
ES-5
-------
EPA/600/R-99/030
and provides the capability of easily and safely using different mechanisms in the CMAQ
system.
Integration of Science Codes into Models-3 One of the major objectives of the
Models-3 project was to develop a flexible, comprehensive air quality modeling system
with a modular coding structure that allows easy replacement of science process
components. Chapter 18 describes the modularity concepts, code management method,
and integration schemes of CMAQ science code with the Models-3 framework. The
CMAQ FORTRAN code was integrated into the Models-3 framework with the following
set of design, coding, and implementation standards: (1) modularity to allow easy
exchange of science process solvers, (2) a standard subroutine interface at the module
level, (3) restriction of coding practices, (4) the Models-3 I/O API
(http://www.iceis.mcnc.org/EDSS/ioapi/index.html/), which contains standardized file I/O
functions and a modeler-friendly interface built on top of self-describing netCDF
(http://www.unidata.ucar.edu/packages/netcdf/) files that are portable across most Unix
platforms.
MODELS-3 ALSO PROVIDES ANALYSIS ROUTINES FOR USE WITH CMAQ OUTPUT,
WHICH CAN BE USED TO PROVIDE PROCESS ANALYSIS RESULTS AND STATISTICAL
AGGREGATION TECHNIQUES.
Process Analysis Chapter 16 describes the implementation of process analysis
techniques in the CMAQ modeling system. These techniques can be used in CMAQ to
provide insights into how model predictions are obtained, which is particularly useful
when modeling nonlinear systems like atmospheric photochemistry. Two techniques are
available in the CMAQ system, integrated process rate (IPR) analysis and integrated
reaction rate (IRR) analysis. IPR analysis can be used to determine the relative
contributions of individual physical and chemical processes, and IRR analysis can help
elucidate important chemical pathways and identify key chemical characteristics.
Aggregation Chapter 17 discusses a statistical procedure called aggregation that is
applied to CMAQ's outputs in order to derive the seasonal and annual estimates required
by assessment studies. Assessment studies require CMAQ-based distributional estimates
of ozone, acidic deposition, and PM2 5, as well as visibility, on seasonal and annual time
frames. Unfortunately, it is not financially feasible to execute CMAQ over such extended
time periods. Therefore, in practice CMAQ must be executed for a finite number of
episodes or "events," which are selected to represent a variety of meteorological classes.
The aggregation technique is used to incorporate these episode simulations into annual
and seasonal estimates.
THE MODEL-3 CMAQ MODELING SYSTEM is BEING FORMALLY EVALUATED TO
ASSESS THE PERFORMANCE OF CMAQ'S NEW DEVELOPMENTS IN AIR QUALITY
ES-6
-------
EPA/600/R-99/030
MODELING. THE EVALUATION WILL PROVIDE THE BASIS FOR UNDERSTANDING
THE STRENGTHS OR WEAKNESSES OF THE CURRENT STATE-OF-SCBENCE IN
CMAQ. With an evaluation of CMAQ simulations of 36, 12, and 4 km grid resolution,
CMAQ's performance can be evaluated on both the regional and urban scales. This evaluation
will include an initial comparison of relative performance against the RADM model and diagnostic
evaluation against databases from regional studies such as the 1995 Southern Oxidant Study
conducted in the vicinity of Nashville, TN and the 1995 NARSTO-NE study.
CMAQ can be configured for a wide range of applications, from scientific studies to regulatory
applications. While the scientific community can take advantage of CMAQ's ability to create
alternative applications for research and development purposes, regulatory applications depend
upon a standardized, evaluated form of CMAQ for regulatory applications. The CMAQ
evaluation program will provide the scientific benchmark needed for this.
FUTURE EXTENSIONS OF CMAQ INCLUDE NEAR-TERM EFFORTS TO PROVIDE A
NEW CHEMICAL MECHANISM AND EMISSION MODELING SYSTEM.
The SAPRC-97 gas phase mechanism will soon be incorporated into CMAQ, in
addition to the current CB-IV and RADM2 mechanisms available. The SAPRC
mechanism will be incorporated with a fixed subset of the approximately 100 organic
species contained in the semi-explicit version of the SAPRC mechanism.
The Sparse Matrix Operator Kernel Emissions modeling system (SMOKE)
(http://envpro.ncsc.org/products/) will also be incorporated into CMAQ in the near-term.
The SMOKE model formulates emissions modeling in terms of sparse matrix operations
which require considerably less time to perform than current systems.
WE ENCOURAGE THE FULL PARTICIPATION AND INVOLVEMENT OF THE
SCIENTIFIC AND MODELING COMMUNITIES IN THE GROWTH AND USE OF MODELS-
3 CMAQ. THERE ARE MANY IDEAS AND PLANS FOR FUTURE DEVELOPMENTS OF
CMAQ, INCLUDING TOXIC POLLUTANT MODELING AND LINKAGES TO OTHER
MODELS.
Modeling atmospheric toxic pollutants A key opportunity for CMAQ is developing the
capability to model toxic pollutants. Models of airborne toxic pollutants are essential for
human exposure and risk assessments. They can also be used to assess the exchange of
toxic compounds between the atmosphere and sensitive ecosystems. With the ability to
simulate toxic pollutant processes in addition to the current photochemical oxidants and
particulates, it is planned to transport the CMAQ model to a finer than urban scale to link
with human exposure models.
ES-7
-------
EPA/600/R-99/030
New linkages with global models It is hoped that information from the urban and
regional CMAQ applications and from global modeling applications can be bridged.
CMAQ output, produced using state of the science techniques, can be used to benchmark
or examine the parametric basis of process formulations in global models. In addition,
global model output can be used to improve or enhance the initial and boundary conditions
for regional and urban scale CMAQ simulations.
Modeling ecosystems Efforts to combine environmental modeling techniques to
encompass an entire ecosystem is needed to address issues including (a) nutrient cycling
through the atmosphere, water bodies, and soil and (b) acidic wet and dry deposition into
sensitive ecosystems, including critical load analyses. With this ecosystem modeling
approach, air quality issues can be studied in combination with other aspects of
environmental health.
This is the Executive Summary of Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
ES-8
-------
EPA/600/R-99/030
Chapter 1
INTRODUCTION TO THE MODELS-3 FRAMEWORK AND
THE COMMUNITY MULTISCALE AIR QUALITY MODEL (CMAQ)
Jason Ching* and Daewon Byun**
Atmospheric Modeling Division .
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Models-3, a flexible software framework, and its Community Multiscale Air Quality (CMAQ)
modeling system form a powerful third generation air quality modeling and assessment tool
designed to support air quality modeling applications ranging from regulatory issues to science
inquiries on atmospheric science processes. The CMAQ system can address tropospheric ozone,
acid deposition, visibility, fine particulate and other air pollutant issues in the context of "one"
atmosphere perspective where complex interactions between atmospheric pollutants and regional
and urban scales are confronted. This CMAQ Science Document contains chapters that address
specific scientific and technical issues involved in the development and application of Models-
3/CMAQ system; collectively, it provides the scientific basis and point of reference for the state of
the science captured in the June 1998 initial release of the CMAQ. This chapter provides an
overview and context of each contributing chapter to the CMAQ system.
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Jason Ching, MD-80, Research Triangle Park, NC 27711.
E-mail: Ching.Jason@epamail.epa.gov
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
1.0 INTRODUCTION TO THE MODELS-3 FRAMEWORK AND THE
COMMUNITY MULTISCALE AIR QUALITY MODEL (CMAQ)
Air quality simulation models are important tools for regulatory, policy, and environmental
research communities. In the United States, the Clean Air Act provides a societal mandate to
assess and manage air pollution levels to protect human health and the environment. The U.S.
Environmental Protection Agency (USEPA) has established National Ambient Air Quality
Standards (NAAQS), requiring the development of effective emissions control strategies for such
pollutants as ozone, particulate matter, and nitrogen species. National and regional policies are
needed for reducing and managing the amount and type of emissions that cause acid, nutrient and
toxic pollutant deposition to ecosystems at risk and for enhancing the visual quality of the
environment. Air quality models are used to develop emission control strategies to achieve these
objectives. Optimal control strategies should be both environmentally protective and cost
effective. Up to now, air quality model paradigms typically addressed individual pollutant issues
separately. However, it is becoming increasingly evident that when pollutant issues are treated in
isolation, the resulting control strategies may solve one set of problems but may lead to
unexpected aggravation of other related pollutant issues. Pollutants in the atmosphere are subject
to myriad transport processes and transformation pathways that control their composition and
levels. Also, pollutant concentration fields are sensitive to the type and history of the atmospheric
mixtures of different chemical compounds. Thus, modeled abatement strategies of pollutant
precursors, such as volatile organic compounds (VOC) and NOX, to reduce ozone levels may
under a variety of conditions, cause an exacerbation of other air pollutants such as particulate
matter or issues of acidic deposition.
The development of comprehensive air quality models started in the late seventies. The Urban
Airshed Model (UAM) (Morris and Meyers, 1990) followed by the Regional Oxidant Model
(ROM) (Lamb, 1983a, 1983b) provided Eulerian-based models for ozone, the former for urban
and the latter for regional scale. Strategies for State Implementation Plans (SIPs) used ROM to
provide boundary conditions for UAM simulations. Attention to acid deposition issues were
addressed in the eighties with the development and evaluation of regional acid deposition models
such as the Regional Acid Deposition Model (RADM) (Chang et al., 1987; Chang et al., 1990),
the Acid Deposition and Oxidant Model (ADOM) (Venkatram et al., 1988), and the Sulfur
Transport and Emissions Model (STEM) (Carmichael and Peters, 1984a, 1984b; Carmichael et
al., 1991). Other major modeling systems included the Regional Lagrangian Modeling of Air
Pollution model (RELMAP) (Eder et al., 1986), a Lagrangian framework system, and semi-
empirical and statistical models. The genre of models of this period were designed to address
specific air pollution issues such as ozone or acid deposition and to be applied under relatively
prescriptive implementation guidance strategies. Thus, flexibility to deal with other issues, such
as particulate matter or toxics, was very limited. Further, With the passage of the Clean Air Act
Amendments of 1990 (CAAA-90), a wide range of additional issues were identified including
visibility, fine and coarse particles, indirect exposure to toxic pollutants such as heavy metals,
semi-volatile organic species, and nutrient deposition to water bodies. The direct response
approach is to modify, adapt or extend current models to handle more complex implementations
and issues but is both cumbersome and limiting.
1-2
-------
EPA/600/R-99/030
Seeking a more strategic approach to handle the increased modeling requirements of the CAAA-
90, more comprehensive modeling approaches appear to be needed. With projections for the
increasing rapid pace of the development of computational capabilities at the start of the nineties,
the opportunity arose for a strategic review of modeling approaches leading to design of a system
that would both meet and keep pace with the increasing requirements on ah- quality modeling, of
incorporating advances hi state-of-science descriptions of atmospheric processes, as well as
eliminate impediments of the current genre of models. The scope of such a system must be able to
process great and diverse information from complicated emissions mixtures and complex
distributions of sources, to modeling the complexities of atmospheric processes that transport and
transform these mixtures in a dynamic environment that operates on a large range of time scales
covering minutes to days and weeks. The corresponding spatial scales are commensurately large,
ranging from local to continental scales. On these temporal and spatial scales, emissions from
chemical manufacturing and other industrial activities, power generation, transportation, and
waste treatment activities contribute to a variety of air pollution issues including visibility, ozone,
particulate matter (PM), and acid, nutrient and toxic deposition. The residence times of pollutants
in the atmosphere can extend to multiple days, therefore transport must be considered on at least
a regional scale. NAAQS requirements and other goals for a cleaner environment vary over a
range of time scales, from peak hourly to annual averages. These challenges suggest that more
comprehensive approaches to air quality modeling are needed, and that assessments and pollution
mitigation are achieved more successfully when the problems are viewed in a "one atmosphere"
context that considers multiple pollutant issues. Further discussion of the needs for the third-
generation air quality modeling system can be found hi Dennis et al. (1996) and Dennis (1998).
To meet both the challenges posed by the CAAA-90 and the need to address the complex
relationships between pollutants, the USEPA embarked upon the Models-3 project and developed
the Community Multiscale Air Quality (CMAQ) system, an advanced air quality modeling system
that addressed air quality from this "one atmosphere" multi-pollutant perspective. Based on its
conceptual design, the high performance computational Models-3 framework serves to manage
and orchestrate air quality simulations, using the CMAQ modeling system. The Models-3
framework is an advanced computational platform that provides a sophisticated and powerful
modeling environment for science and regulatory communities. The framework provides tools
used to develop and analyze emission control options, integrate related science components into a
state-of-the-art quality modeling system, and apply graphical and analytical tools for facilitating
model applications and evaluation. Descriptions of the Models-3 architecture are provided hi
Chapter 2 and in Novak et al. (1998). CMAQ is a multi-pollutant, multiscale ah* quality model
that contains state-of-science techniques for simulating all atmospheric and land processes that
affect the transport, transformation, and deposition of atmospheric pollutants and/or their
precursors on both regional and urban scales. It is designed as a science-based modeling tool for
handling all the major pollutant issues (including photochemical oxidants, particulate matter,
acidic, and nutrient deposition) holistically.
More than six years of investment and commitment .from Federal staff and from scientists and
model developers from the environmental and information communities were expended to
develop the Models-3 framework and the CMAQ air quality modeling system. Models-3 CMAQ
1-3
-------
EPA/600/R-99/030
was released to the public in June 1998. The science and model engineering concepts and
progress of the project have been described in two peer-reviewed AWMA Transaction papers
(Byun et al., 1995a; Coats et at, 1995) and others (Byun et al.} 1995b, 1996, and 1998a; Ching et
al., 1995). This release version of the Model-3 software is supported by the following documents
under the overall heading "Third Generation Air Quality Modeling System":
• System Installation and Operations Manual. EPA/600/R-98/069a, National Exposure
Research Laboratory, EPA, Research Triangle Park, NC
• Users Manual. EPA/600/R-98/069b, National Exposure Research Laboratory, Research
Triangle Park, NC
• Tutorial. EPA/600/R-98/069c, National Exposure Research Laboratory, EPA, Research
Triangle Park, NC
« Science Concepts of the Third Generation Air Quality Models: Project Report. In
preparation. Edited by D.W. Byun and A. Hanna. National Exposure Research
Laboratory, EPA, Research Triangle Park, NC
» Science Algorithms of the EPA Models-3 Community Multiscale Air Quality (CMAQ)
Modeling System, Edited by D.W. Byun and J.K.S. Ching. National Exposure Research
Laboratory, EPA, Research Triangle Park, NC (this document)
The project report "Science Concepts of the Third Generation Air Quality Models" summarizes
the basic atmospheric science concepts and mathematical principles pertinent to the development
of the third generation air quality modeling. This document discusses the key scientific features
and options incorporated into the Models-3 CMAQ modeling system.
1.1 The Models-3 Emissions, Meteorology, and the CMAQ Modeling Systems
The structure of the Models-3/CMAQ system is shown in Figure 1-1. Orchestrated through the
Models-3 system framework, the Community Multiscale Air Quality (CMAQ) modeling system
incorporates output fields from emissions and meteorological modeling systems and several other
data source through special interface processors into the CMAQ Chemical Transport Model
(CCTM). CCTM then performs chemical transport modeling for multiple pollutants on multiple
scales. With this structure, CMAQ retains a flexibility to substitute other emissions processing
systems and meteorological models. One of the main objectives of this project was to provide an
air quality modeling system with a "one atmosphere" modeling capability based mainly on the
"first principles" description of the atmospheric system. CMAQ contains state-of-science
parameterizations of atmospheric processes affecting transport, transformation, and deposition of
such pollutants as ozone, particulate matter, airborne toxics, and acidic and nutrient pollutant
species. With science in a continuing state of advancement and review, the modeling structure of
CMAQ is designed to integrate and to test future formulations in an efficient manner, without
requiring the development of a completely new modeling system. Contents of the CMAQ in the
1-4
-------
EPA/600/R-99/030
June 1998 release version of Models-3 are summarized in Ching et al. (1998) and Byun et al.
(1998b).
Currently, the Models-3 Emission Projection and Processing System (MEPPS) produces the
emissions and the Fifth Generation Perm State University/ National Center for Atmospheric
Research Mesoscale Model (MM5) provides the meteorological fields needed for the CCTM.
They are considered to meet the present application needs for diverse air pollution problems in
urban and regional scales. However, given the CMAQ paradigm, and other considerations, the
emissions processing and meteorological modeling systems can be replaced with alternative
processors.
Each of these three modeling systems are described briefly below, where associated chapters of
this document are highlighted to provide directions to more in-depth discussions of these topics:
• The PSU/NCAR MM5 meteorological modeling system (Grell et al., 1994) generates the
meteorological fields for CMAQ. MM5 is a complex, state-of-the-science community
model, which is maintained by NCAR, MM5 is well-documented by its primary
developers in technical notes and referenced journal articles. Chapter 3 briefly describes
the scientific aspects of MM5, including grid definitions, model physics, nesting and four-
dimensional data assimilation,
* The MEPPS emission modeling system is based on the Geocoded Emission Modeling and
Projection System (GEMAP) (Wilkinson et al., 1994) now known as the Emission
Modeling System-95 (EMS-95). MEPPS processes emission inventory data, performs
future projections (including control scenarios), and pre-processes data for use in the
CMAQ model (Chapter 4). It provides speciated emissions consistent with CB-IV or
RADM2 chemistry mechanisms.
* The CMAO chemical transport modeling system (CCTM) is then used to perform model
simulations for multiple pollutants and multiple scales with these input data (Chapters 6,
7,8,9,10 and 11). The fundamental concepts used for the one-atmosphere dynamic
modeling is described in Chapter 5. The techniques used for the management of
CMAQ's source code are discussed in Chapter 18.
The CMAQ modeling system also includes interface processors that process input data
for the emission and meteorological modeling systems, and other processors that calculate
photolysis rates, and develop initial and boundary conditions (Chapters 4,12,13, and
14). CMAQ also has an internal program control processor which is discussed in
Chapter 15.
Using the analysis routines provided hi Models-3, the CMAQ output can be processed to
provide process analysis information (Chapter 16) and/or analyzed further to provide
aggregated statistical information (Chapter 17).
1-5
-------
EPA/600/R-99/030
An important design requirements for CMAQ is that it addresses multiple scales and pollutants,
which requires that governing equations and computational algorithms among the different
systems should be consistent and compatible across the multiple scales. However, modeling
assumptions used in various modeling systems may not be valid across all scales. For example,
the atmospheric dynamics description in a meteorological model may have been optimized for
application of certain scale or limited range of scales (e.g., global vs mesoscale vs complex terrain
to urban). It is incumbent upon the user community to ensure the model component formulations
are applicable to the range of scales upon which CMAQ is applied. The current version of MM5
and the CCTM is designed for regional to urban scales. Furthermore, when using nesting
procedures to scale down from regional to urban scales and for avoiding feedback between the
scales, one way nesting is recommended. In addition to the challenges of creating a multiscale air
quality model, CMAQ's multi-pollutant capability cannot be achieved if the emissions modeling
system does not provide appropriate precursor or pollutant emissions to the chemical transport
model (CTM). The development of Models-3 and CMAQ overcome these hurdles by providing
the flexibility to modify specific requirements (e.g., chemical mechanisms, model inputs, etc.), a
generic coordinate system that ensures consistency across spatial scales, and user interfaces that
can integrate alternative emissions or meteorological modeling systems,
1.2 CMAQ Interface Processors
The CMAQ modeling system includes interface processors to incorporate the outputs of the
meteorology and emissions processors and to prepare the requisite input information for initial
and boundary conditions and photolysis rates to the CCTM. Figure 1-1 illustrates the relationship
and purpose of each of the CMAQ processors (and requisite interfaces) and their relation to the
chemical transport modeling system. The arrows show the flow of data through the modeling
system. Two additional functional features of the CMAQ system are included, one for process
analysis, which is primarily for model diagnostic analyses, and a second one that is an aggregation
methodology for estimating longer term averaged fields. Each of these processors is described
briefly below, and the associated chapter numbers are also listed to note where detailed
discussions can be found on the topics.
1-6
-------
EPA/600/R-99/030
Fundamentals of \
Dynamics for CMAQ '"••
(Chapter 5)
Program Control
Processing
(Chapter 15)
CMAQ Chemical Transport Model (CTM)
Figure 1-1. Emissions and Meteorological modeling systems and the CMAQ
Chemical Transport Model and Interface Processor
The Emission-Chemistry Interface Processor (ECIP) translates data from the MEPPS
emission model for use in the CCTM. ECIP generates hourly three-dimensional emission
data for CMAQ from the separate source type files produced by MEPPS, which include
mobile, area, and point sources (Chapter 4). ECIP calculates the plume rise and initial
vertical plume spread of point source emissions to determine the vertical level(s) of
CCTM into which point source emissions should be introduced. Since meteorological
conditions affect both point source plume rise and biogenic emissions, meteorological data
from MCIP is also used in ECIP.
The Meteorology-Chemistry Interface Processor (MCIP) translates and processes model
outputs from the meteorology model for the CCTM (Chapter 12). MCIP interpolates the
meteorological data if needed, converts between coordinate systems, computes cloud
parameters, and computes surface and planetary boundary layer (PBL) parameters for the
CCTM. MCIP uses landuse information from the landuse processor (LUPRDC) to
calculate the PBL and surface parameters.
Initial Conditions and Boundary Conditions (ICON and BCON) provide concentration
fields for individual chemical species for the beginning of a simulation and for the grids
surrounding the modeling domain, respectively. The ICON and BCON processors
(Chapter 13) use data provided from previous three-dimensional model simulations or
1-7
-------
EPA/600/R-99/030
from clean-troposphere vertical profiles. Both the vertical profiles and modeled
concentration fields have a specific chemical mechanisms associated with them, which are
a function of how these files were originally generated.
• The photolysis processor CJPROO calculates temporally varying photolysis rates
(Chapter 14). JPROC requires vertical ozone profiles, temperature profiles, a profile of
the aerosol number density, and the earth's surface albedo to produce the photolysis rates
for the CCTM. JPROC uses this information in radiative transfer models to calculate the
actinic flux needed for calculating photolysis rates. JPROC generates a lookup table of
photo-dissociation reaction rates.
Each of these CMAQ interface processors incorporates raw data into CMAQ and performs
functions such as calculating parameters and interpolating or converting data. Raw input data is
currently specified in the source code for JPROC, LUPROC, ICON, and BCON. However, the
interface processors in future releases of CMAQ will be modified to handle a more generalized set
of raw input data, so that alternative data sets with varying resolutions or measurement units can
be used.
1.3 The CMAQ Chemical Transport Model (CCTM)
The CCTM simulates the relevant and major atmospheric chemistry, transport and deposition
processes involved throughout the modeling domains. The fundamental basis for CMAQ's one-
atmosphere dynamics modeling is discussed in Chapter 5. Governing equations and model
structure, including definitions of CMAQ's science process modules, are discussed in Chapter 6.
The science options available to the user include the gas phase chemistry mechanisms, RADM2
and CB-IV, a set of numerical solvers for the mechanisms, options for horizontal and vertical
advection schemes, algorithms for fine and coarse particulate matter predictions, photolysis rates,
and a plume-m-grid approach. Through the Models-3 framework, CMAQ simulations can be
developed using these different options without modifying source code. A general overview of
these science process options is provided below along with a reference to the chapter(s) of this
document where more scientific detail can be found.
• Advection and Diffusion (Chapter 7): Several advection methods are implemented in the
CMAQ; these include a scheme by Bott (1989), a piecewise parabolic method (PPM)
(Collela and Woodward, 1984), and the Yamartino-Blaekman cubic algorithm. Options
for computing subgrid vertical transport include eddy diffusion, and the Asymmetric
Convective Model (ACM) (Pleim and Chang, 1992) applicable to convective conditions.
Horizontal diffusion is modeled using a constant eddy diffusion coefficient. Numerical
methods differ in the handling of advection of concentration fields.
• Gas Phase Chemistry (Chapter 8): CMAQ includes both the RADM2 and CB4 gas-phase
chemical mechanisms. The CMAQ version of CB4 includes the most recent
representation of isoprene chemistry and two additional variants of the RADM2
mechanism also contain the newer isoprene chemistry at two levels of detail. In addition,
1-8
-------
EPA/600/R-99/030
CMAQ provides the capability to edit these mechanisms or to import a completely new
mechanism by means of a generalized chemical mechanism processor. CMAQ also
accounts for the formation of secondary aerosols and the reactions of pollutants in the
aqueous phase, and aqueous reactions are simulated by means of the aqueous chemical
mechanism incorporated in RADM. All CMAQ gas-phase mechanisms are linked to these
processes to provide the capability to simulate multi-phase interactions.
Two chemistry solvers are available — the Sparse Matrix Vectorized Gear (SMVGEAR)
algorithm developed by Jacobson and Turco (1994) and the Quasi-Steady State
Approximation (QSSA) method used in the Regional Oxidant Model. SMVGEAR is
generally recognized as the more accurate of the two, but it is much slower than QSSA on
non-vector computers.
Plume-in-Grid (PinG) Modeling (Chapter 9): CMAQ includes algorithms to treat subgrid
scale physical and chemical processes impacting pollutant species in plumes released from
selected Major Elevated Point Source Emitters (MEPSEs). The PinG modules simulate
plume rise and growth, and the relevant dynamic and chemical reaction processes of
subgrid plumes. PinG can be used for the simulations at 36 km and 12 km resolutions,
PinG is not invoked at 4 km resolutions and the MEPSE emissions are directly released
into the CTM 3-D grid cells.
Particle Modeling and Visibility (Chapter 10): One of the major advancements in CMAQ
is the modeling of fine and coarse mode particles, with the use of the fine particle model
described in Binkowski and Shankar (1995). CMAQ predicts hourly gridded
concentrations of fine particle mass whose size is equal to or less that 2.5 microns in
diameter (PM 2 s), speciated to sulfate, nitrate, ammonium, organics and aerosol water.
Secondary sulfate is produced when hydroxyl radicals react with sulfur dioxide to produce
sulfuric acid that either condenses to existing particles or nucleates to form new particles.
CMAQ model output includes number densities for both fine and coarse modes. The
modeling of aerosols in CMAQ also provides the capability to handle visibility, which is
another CMAQ output. In another potential application, CMAQ can provide the basis for
modeling the atmospheric transport and deposition of semi-volatile organic compounds
(SVOC) with parameterizations for their rates of condensation to and/or volatilization
from the modeled particles.
Cloud processes (Chapter 11 and 12): Proper descriptions of clouds are essential in air
quality modeling due to their critical role in atmospheric pollutant transport and chemistry
processes. Clouds have both direct and indirect effects on the pollutant concentrations:
they directly modify concentrations via aqueous chemical reactions, vertical mixing, and
wet deposition removal processes, and they indirectly affect concentrations by altering
radiative transmittances which affect photolysis rates and biogenic fluxes. CMAQ models
deep convective clouds (Walcek and Taylor, 1986) and shallow clouds using the
algorithms as implemented in RADM (Dennis et al., 1993) for 36 and 12 km resolutions.
At 4 km resolution, the clouds are generally resolved, and explicit type cloud dominates.
1-9
-------
EPA/600/R-99/030
* Photolysis Rates (Chapter 14): The photochemistry of air pollutants is initiated by
photodissociation of smog precursors, which are driven by solar radiation. The amount of
solar radiation is dependent on sun angle (time of day), season, latitude, and land surface
characteristics, and is greatly affected by atmospheric scatterers and absorbers.
Photolytic rates are also wavelength- and temperature-dependent. Within CCTM,
temporally resolved 3-D gridded photolysis rates are interpolated from a lookup table
generated by JPROC processor and corrected for cloud coverage.
1.4 Analysis of CMAQ Output
Air quality modeling simulations arise from modeling of complex atmospheric processes. It is
important to assure and to understand the model results. Sensitivity tests are needed to detect
problems in model formulations and to determine if the model is credible for assessing emission
control strategies. A very powerful sensitivity analysis tool called process analysis is provided
with CMAQ. Also, an aggregation technique is provided with CMAQ. Aggregation can be used
to estimate seasonal or annual concentrations for pollutants from CMAQ simulations which are
usually performed for shorter time periods due to time and computational limitations.
1.4.1 Process Analysis (Chapter 16)
Sensitivity analyses are needed to detect errors and uncertainties introduced into a model by the
parameterization schemes and the input data. Results must also be analyzed to ensure that
realistic values are obtained for the right reasons rather than through compensating errors among
the science processes. Process Analysis techniques quantify the contributions of individual
physical and chemical atmospheric processes to the overall change in a pollutant's concentration,
revealing the relative importance of each process. Process analysis is particularly useful for
understanding the effects from model or input changes. CMAQ provides the capability to
perform process analyses using two different pieces of information: Integrated Process Rates
(IPRs) and Integrated Reaction Rates (IRRs).
• The IPRs are obtained during a model simulation by computing the change in
concentration of each species caused by physical processes (e.g., advection, diffusion,
emissions), chemical reaction, aerosol production, and aqueous chemistry. Values provide
only the net effect of each process. IPRs are particularly useful for identifying
unexpectedly low or high process contributions which could be indicative of model errors.
The IRR analysis involves the details of the chemical transformations. For gas-phase
chemistry, the CCTM has been designed to compute not only the concentration of each
species, but also the integral of the individual chemical reaction rates. IRR analyses have
typically been used to understand the reasons for differences in model predictions obtained
with different chemical mechanisms.
1-10
-------
EPA/600/R-99/030
1.4.2 Aggregation (Chapter 17) .
In support of studies mandated by the CAAA-90, CMAQ can be used to estimate deposition and
air pollutant concentrations associated with specific levels of emissions. Assessment studies
require estimates of ozone, acidic deposition, particulate matter as well as visibility, on seasonal
and annual time frames. A statistical procedure called the "aggregation" has been developed and
is provided for the CMAQ to derive the required seasonal and annual estimates. This
methodology is an efficient technique and can be used instead of executing multiple CMAQ model
runs for the intended period of averaging.
A typical CMAQ simulation provides hourly air quality fields for regional to urban scales for
multi-day episodes, typically up to five days in duration. The new PM2 5 standard includes an
annual average value, so utilization of CMAQ for PM2 5 will require the use aggregation
techniques in order to estimate annual average PM2 5 values, One such technique, initially
developed for RADM wet deposition applications, was recently modified and successfully applied
to PM25 by Eder and LeDuc (1996). The approach utilizes visibility data as a surrogate for PM25,
and it will be applied to CMAQ on a continental scale (i.e., contiguous United States, southern
Canada, and northern Mexico). Future efforts will be needed to validate this approach when a
network of PM2 5 samplers is deployed; also, aggregation approaches for mesoscale domains will
need to be developed perhaps utilizing the method by Eder and LeDuc (1994).
1.5 Management of CMAQ Science Information Objects and Codes in Models-3
The CMAQ source code is managed through the Models-3 framework to make the CMAQ
modeling system more efficient and easier to use by applying a program control processor,
management and integration of source code, and implementing a modularity concept. These
techniques also help users customize CMAQ for their own modeling applications without source
code modifications. .
1.5.1 Program Control Processors (Chapter 15)
Certain science information, such as grid and layer definitions and dimensions, chemical
mechanisms, species list, model configurations, and episode (case) specification, is used
repeatedly across the several process components in Models-3 CMAQ modeling system.
Program control processors are a set of programs embedded in the Models-3 framework to
handle these science information components and their codes. Program Control Processing (PCP)
refers to setting up internal arrays, mappings of species names in the input processors, defining
global parameters, and establishing linkages among processors in the CMAQ system. PCP allows
users to define globally shared information on model components, and it uses that information to
generate the global FORTRAN include files required for building a model in CMAQ.
1-11
-------
EPA/600/R-99/030
1.5.2 CMAQ Code Integration (Chapter 18)
CMAQ's modularity facilitates efficient coordination of development work and management of
the science codes. Chapter 18 describes the modularity concepts, code management method, and
integration schemes of CMAQ science code with the Models-3 framework. Integrating the
CMAQ code into the Models-3 framework is achieved by following a set of design, coding, and
implementation standards that include:
• A standard subroutine interface at the module level
* The restriction of coding practices to avoid practices that can conceal data dependencies,
hinder maintenance and foster hidden bugs
* The Models-3 Input/Output Applications Programming Interface (I/O API)
(http://www.iceis.mcnc.org/EDSS/ioapi/index.html/), which contains standardized file I/O
functions. The I/O API is an interface built on top of self-describing netCDF
(http://www.unidata.ucar.edu/packages/netcdf/) files that are portable across most Unix
platforms.
1.6 Post Release Studies and Near-Future Plans
1.6.1 CMAQ Evaluation Study
It is important to conduct extensive evaluation of the CMAQ. Subsequent to the initial release of
the Models-3/CMAQ, the development team is engaged in a substantial program of evaluation.
The scope of the effort includes analyses of the performance and veracity of each individual
process module as well as the integrated air quality system. Findings from this evaluation can be
incorporated into future releases of the CMAQ modeling system. The degree and rigor of this
evaluation provides the basis for understanding the strengths or weaknesses of the current state-
of-science in CMAQ. The evaluation of the initial release version of CMAQ is underway for three
nested grids with 36, 12, and 4 km grid resolutions. With these results, CMAQ's performance
can be evaluated on both the regional and urban scales. This model evaluation activity for CMAQ
will be staged with the initial efforts to show relative performance against the RADM model,
which has undergone extensive model evaluation efforts. Diagnostic evaluation will continue
using databases from different regional studies such as the 1995 Southern Oxidant Study
conducted in the vicinity of Nashville, TN and the 1995 NARSTO-NE study.
1.6.2 Testing Operational Configurations
CMAQ can be configured for a wide range of applications from science studies and investigations
to regulatory applications. While the scientific community can take advantage of the CMAQ open
system and flexibility to create alternative applications of CMAQ for research and development
purposes, regulatory applications depend upon a standardized, evaluated form of CMAQ. The
CMAQ evaluation program will provide the scientific benchmark needed for this. As science
advances in CMAQ, future configurations of a more operational nature can also be periodically
1-12
-------
EPA/600/R-99/030
re-benched as appropriate. As understanding of atmospheric processes improves, it is a natural
tendency for models to become more complex and have increased computational demands.
Efforts are underway to improve model computational efficiencies to compensate for this.
1.6.3 Extensions and Science Additions
Two major extensions are planned for the CMAQ modeling
• A version of the SAPRC-97 gas phase mechanism will be incorporated into CMAQ, in
addition to the current CB-IV and RADM2 mechanisms available. The initial
configuration of S APRC will be in a fixed parameter mode, with a preset number of
organic species. Another possible future implementation of S APRC will allow the user to
select from about 100 organic surrogate species in the semi-explicit SAPRC mechanism to
construct a user-defined, smaller SAPRC mechanism. Another approach for representing
gas-phase chemistry in CMAQ is also being developed. It will use a limited number
reactive entities termed "morphecules" to include in the chemical reactions while using a
much larger number of chemical species (singly or lumped) called allomorphs to provide
extra chemical detail. This latter approach provides a means for including a much more
detailed representation of atmospheric chemistry than conventional chemical mechanisms
without significantly increasing computational burden, albeit at the expense of additional
computer memory.
• A new emissions processor will be implemented which is called the Sparse Matrix
Operator Kernel Emissions modeling system (SMOKE)
(http://envpro.ncsc.org/products/). The linear operations used in emission processing can
be represented as multiplications by matrices. Since most entries in these matrices are
zeros, the SMOKE model formulates emissions modeling in terms of sparse matrix
operations that require considerably less time to perform than current systems. Efficiency
is enhanced even further when considering variations in emissions projections from base
. case scenarios by temporally modeling once per episode, calculating gridding matrices
only once per grid, and calculating speciation matrices only once per chemical mechanism.
Development and testing of several science options is underway for incorporation into future
releases of the CMAQ. These include an advanced surface-PBL linked system (Pleim and Xiu,
1995), optional meteorological processors such as the Regional Atmospheric Modeling System,
RAMS, and an advanced 4-D Photolysis Rates Processor.
1.7 Opportunities and Encouragement for Long Term Extensions and Science
Community Involvement
The Models-3/CMAQ concept is based on an open system design, we encourage the full
participation and involvement of the scientific and modeling communities in the growth and use
of Models-3 CMAQ. As described in this document, the Models-3 CMAQ system has flexibility
for incorporating scientific and modeling advances into CMAQ processors, for testing of
1-13
-------
EPA/600/R-99/030
alternative modeling techniques for science processes, and for extending its current capability to
handle multimedia environmental issues. Additionally, the community of users should be vigilant
in performing evaluations against improved databases and measurement technology to assess the
realism of model performance and to measure the strengths and weaknesses of the current state of
the-science as presented hi the CMAQ modeling system. Some suggestions for extensions and
community involvement are provided below, but certainly not limited to:
(1) Modeling airborne and deposition of atmospheric toxic pollutants: A key opportunity for
CMAQ is the development of a modeling capability for toxic pollutants into the CMAQ chemical
transport modeling system. Models of airborne toxic pollutant provide an important tool for
understanding the transport and chemical pathways that are concerns to human exposure
assessments and for risk assessments and its management. It also provides a powerful means to
assess the exchange of toxic compounds between the atmosphere and sensitive ecosystems.
However, developing such models is challenging. Air toxics arise from a wide variety of sources,
which may have a wide range of chemical lifetimes and reactivity. Consequently, their
lexicological impacts as well as their time-space distributions may be highly variable. These
complex mixtures of reactive compounds can exhibit wide range and variability in physical
properties and exist at various gas, liquid or particulate ambient metastates. Modeling paradigms
might evolve from introducing gas-particulate partitioning of the semi-volatile species (Ching et
aL, 1997) to more fundamental modeling with detailed chemical mechanisms.
(2) Development of modeling capability to link with human exposure models: With the ability to
simulate toxic pollutant processes in addition to the current photochemical oxidants and
particulates, it is planned to transport the CMAQ model to a finer than urban scale to link with
human exposure models. International efforts to determine and understand the etiology of
adverse human effects of air pollutants, especially those that or associated with fine particles is
underway. Modeled concentrations of air pollutant constituents at neighborhood scales when
coupled with the limited numbers of sampling provide a powerful basis for driving human
exposure models. With this necessary data, human exposure models provide a basis for the
causality studies and for risk assessment research. Models have not yet been developed to predict
the spatial and temporal distributions of the various causal pollutant classes under current
consideration at the neighborhood scale. The downscaling requirements represent a great
challenge to the scientific basis in current meteorological processors.
(3) Advanced data assimilation capability: As air quality modeling efforts extend to finer
horizontal resolutions and time scales, it becomes increasingly important to make use of higher
frequency asynoptic meteorological data such as satellite, radar wind profilers and NEXRAD.
There is a need to develop and extend data assimilation techniques to these meteorological data
sources. Improved methods for introducing cloud information into the CMAQ should also be
investigated. Such fields have significant impact on chemical photolysis rates, the atmospheric and
surface energy budgets, the stability and dispersive power of the boundary layer, and aqueous
chemistry.
1-14
-------
EPA/600/R-99/030
(4) Development of an air quality predictive mode: With further development of the emissions
and meteorological modeling systems, plans include developing a predictive capability in the
CMAQ modeling system. CMAQ is currently developed in retrospective mode for performing
assessments and for implementation of the National Ambient Air Quality Standards (NAAQS). In
principle, CMAQ may be extended to a forecast mode in order to provide air quality advisories
for pollutants such as ozone and visibility and for operational support for issues such as smoke
and PM/haze from prescribed fires. These capabilities will require the development and testing of
day-specific emissions inventories and a real time data processing system for the meteorological
mode.
(5) Up and downscale links to global scale models: New linkages with other areas of modeling,
including scale and media, are envisioned. It is hoped that information from the urban and
regional CMAQ applications and from global modeling applications can be bridged. Because
CMAQ offers the state of science to simulate atmospheric process as realistically as possible at
time scales commensurate with reality, CMAQ output can be used to benchmark or examine the
parametric basis of process formulations in global models. From a downscale perspective, global
model output can be used to improve or enhance the initial and boundary conditions in CMAQ
regional and urban scale simulations.
(6) Ecosystem modeling: Efforts to combine environmental modeling techniques to encompass
an entire ecosystem is needed to address issues including: (a) nutrient cycle modeling, which
includes pathways through the atmosphere, water bodies,, and soil; and (b) acidic wet and dry
deposition into sensitive ecosystems, including critical load analyses. With this ecosystem
modeling approach, air quality issues can be studied in combination with other aspects of
environmental health. For example, nitrogen deposition can cause adverse nutrient loadings to
ecosystems that can result in a reduction of water quality due to adverse biological responses.
Further, toxic deposition can lead to adverse indirect human exposure from bioaccumulation
through the food chain.
1.8 References
Binkowski, F.S., and U. Shankar, 1995: The Regional Paniculate Model: Part I. Models
description and preliminary results. J.Geophys. Res., 100(D12): 26191-26209.
Bott, A., 1989: A positive definite advection scheme obtained by nonlinear renormalization of the
advective fluxes, Mon. Wea. Rev. Ill: 1006-1015.
Byun D. W., A. Hanna, C. J. Coats, and D. Hwang, 1995a: Models-3 air quality model prototype
science and computational concept development. Transactions of Air & Waste Management
Association Specialty Conference on Regional Photochemical Measurement and Modeling
Studies, Nov. 8-12, San Diego, CA. 1993, 197-212.
1-15
-------
EPA/600/R-99/030
Byun, D. W., C. J. Coats, D. Hwang, S. Fine, T. Odman, A. Hanna and K. J. Galluppi, 1995b:
Prototyping and implementation of multiscale air quality models for high performance computing.
Mission Earth Symposium, Phoenix, AZ, April 9-13, 1993. 527-532.
Byun, D.W., D. Dabdub, S. Fine, A. F. Hanna, R. Mathur, M. T. Odman, A. Russell, E. J.
Segall, J. H. Seinfeld, P. Steenkiste, and J. Young, 1996: Emerging Air Quality Modeling
Technologies for High Performance Computing and Communication Environments, Air Pollution
Modeling and Its Application XI, ed. S.E. Gryning and F. Schiermeier. 491-502.
Byun, D.W., J.K.S. Ching, J. Novak, and J. Young, 1997: Development and Implementation of
the EPAis Models-3 Initial Operating Version: Community Multi-scale Air Quality (CMAQ)
Model, 1998a: Twenty-Second NATO/CCMS International Technical Meeting on Air Pollution
Modelling and Its Application, 2-6 June, 1997. Air Pollution Modeling and Its Application XII,
ed. S.E. Gryning and N. Chaumerliac, Plenum Publishing Coorp. 357-368.
Byun, D.W., J. Young., G. Gipson., J. Godowitch., F. Binkowsk, S. Roselle, B. Benjey, J. Pleim,
J.K.S. Ching., J. Novak, C. Coats, T. Odman, A. Hanna, K. Alapaty, R. Mathur, J. McHenry, U.
Shankar, S. Fine, A. Xiu, and C. Jang, 1998b: Description of the Models-3 Community Multiscale
Air Quality (CMAQ) model. Proceedings of the American Meteorological Society 78th Annual
Meeting, Phoenix, AZ, Jan. 11-16, 1998: 264-268.
Carmichael G.R. and L.K. Peters, 1984a: An Eulerian transport/transformation/removal model for
SO2 and sulfate-I. Model development, Atmos. Environ. 18: 937-952.
Carmichael G.R. and L.K. Peters, 1984b: An Eulerian transport/transformation/removal model for
SO2 and sulfate-II. Model calculation of SOX transport in the Eastern United States, Atmos.
Environ. 20: 173-188.
Carmichael G.R., L.K. Peters, and R.D. Say lor, 1991: The STEM-II regional scale acid
deposition and photochemical oxidant model-I. An overview of model development and
applications, Atmos. Environ. 25A: 2077-2090.
Chang, J.S., P.B. Middleton, W.R. Stockwell, C.J. Walcek, J.E. Pleim, H.H. Landsford, S.
Madronich, F.S. Binkowski, N.L. Seaman, and D.R. Stauffer, 1990: "The Regional Acid
Deposition Model and Engineering Model." NAPAP SOS/T Report 4, in National Acid
Precipitation Assessment Program: State of Science and Technology, Volume 1. National Acid
Precipitation Assessment Program, 722 Jacksn Place, N.W., Washington, D.C., September 1990.
Chang, J.S., R.A. Brest, I.S.A. Isaksen, S. Madronich, P. Middleton, W.R. Stockwell, and C.J.
Walcek, 1987: A three-dimensional Eulerian acid deposition model: Physical concepts and
formulation." J. Geophy. Res. 92: 14681-14700.
Ching, J.K.S., D.W. Byun, A. Hanna, T. Odman, R. Mathur, C. Jang, J. McHenry, K. Galluppi,
1995: Design requirements for multiscale air quality models. Mission Earth Symposium, Phoenix,
AZ, April 9-13 p532-538.
1-16
-------
EPA/600/R-99/030
Ching, J.K.S., D.W. Byun, J. Young, F. Binkowsk., J. Pleim, S. Roselle, J. Godowitch, W.
Benjey., and G. Gipson, 1998: Science features in Models-3 Community multiscale air quality
system. Proceedings of the American Meteorological Society 78th Annual Meeting, Phoenix, AZ,
Jan. 11-16,1998:269-273
Ching, J.K.S., Francis S. Binkowski, and O. Russell Bullock, Jr., 1997: Deposition of semivolatile
toxic air pollutants to the Great Lakes: A regional modeling approach. In " Atmospheric
Deposition of Contaminants to the Great Lakes and Coastal Waters, Ed by Joel Baker, SETAC
Press, pp 293-304.
Coats, C, J., A.H. Hanna, D. Hwang, and D.W. Byun, 1995: Model engineering concepts for air
quality models in an integrated environmental modeling system. Transactions of Air & Waste
Management Association Specialty Conference on Regional Photochemical Measurement and
Modeling Studies, Nov. 8-12, San Deigo, CA. 1993,213-223.
Colella, P., and P. L. Woodward, 1984: The Piecewise Parabolic Method (PPM) for gas-
dynamical simulations, J. Comput. Phys. 54: 174-201.
Dennis, R, 1998: The Environmental Protection Agency's third generation air quality modeling
system: An overall perspective. Proceedings of the American Meteorological Society 78th Annual
Meeting, Phoenix, AZ, Jan. 11-16,1998: 255-258.
Dennis, R.L., D.W. Byun, J.H. Novak, KJ. Galluppi, C.J. Coats, and M.A. Vouk, 1996: The
next generation of integrated air quality modeling: EPA's Models-3. Atmospheric Environment,
30, No, 12, 1925-1938.
Eder, B.K., D.H. Coventry, C; Bollinger and T.L. Clark: RELMAP, 1986: A Regional
Lagrangian Model of Air Pollution User's Guide. U.S. Environmental Protection Agency Report
EPA/600/8-86/013, Research Triangle park, NC 146 pp.
Eder, B.K., J.M. Davis and P. Bloomfeld, 1994: An automatic classification scheme designed to
better elucidate the dependence of ozone on meteorology. J. Appl. Meteor, 33: 1182-1199.
Eder, B.K. and S. LeDuc, 1996: Can selected RADM simulations be aggregated to estimate
annual concentrations of fine particulate matter? Reprint from the 11th Annual International
Symposium on the Measurement of Toxic and Related Pollutants, May 7-10, 1996, RTF, NC
732-739. ...
Gillani, N.V., A. Biazar, Y. Wu, J. Godowitch, J. Ching, R. Imhoff, 1998: The Plume-in-Grid
treatment of major elevated point source emissions in Models-3. 10th Joint Conference on
Applications of Air Pollution Meteorology with AWMA, January 11-16, 1998, Phoenix, AZ,
Amer. Meteorol. Soc., Boston, MA.
Grell, G.A., J. Dudhia, and D.R. Stauffer, 1993: A description of the Fifth-Generation Perm
State/NCAR Mesoscale Model(MMS). NCAR Technical Note, NCAR/TN-398+IA
1-17
-------
EPA/6QO/R-99/030
Jaeobson M. and R.P. Turco, 1994: SMVGEAR: A Sparse-Matrix, vectorized Gear code for
atmospheric models. Atmos. Environ, 28: 273-284.
Jeffries, H. E. and S. Tonnesen, 1994: A comparison of two photochemical reaction mechanisms
using mass balance and process analysis, Atmos. Environ, 28(18): 2991-3003.
Lamb, R.G., 1983a: A Regional Scale (1000 km) Model of Photochemical Air Pollution, Part 1:
Theoretical Formulation. EPA-600/3-83-035, U.S. Environmental Protection Agency, Research
Triangle Park, NC.
Lamb, R.G., 1983b: A Regional Scale (1000 km) Model of Photochemical Air Pollution, Part 2:
Input Processor Network Design. EPA-600/3-84-085, U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Morris, R.E. and T.C. Meyers, 1990: User's Guide for the Urban Airshed Model, Volume 1:
User's Manual for UAM(CB-IV). EPA-4 5 0/4-90-007A. U.S. Environmental Protection Agency,
Research Triangle Park, NC.
Novak, J., J. Young, D.W. Byun, C. Coats, G. Walter, W. Benjey, G. Gipson, S. LeDuc, 1998:
Models-3; A unifying framework for environmental modeling and assessments. Preprint Volume,
10th Joint AMS and A&WMA Conference on the Applications of Air Pollution Meteorology,
Phoenix, AZ, Jan 11-16,1998: 259-263.
Pleim J.E., and J.S. Chang, 1992: A non-local closure model in the convective boundary layer.
Atm Environ., 26A: 965-981.
Pleim, J.E., and A. Xiu, 1995: Development and testing of a surface flux and planetary boundary
layer model for application in mesoscale models, J. Appl. Meteor., 34:16-32.
Wilkinson, J.G., C.F. Loomis, D.E. McNally, R.A, Emigh, and T.W. Tesche, 1994: Technical
Formulation Document: SARMAP/LMOS Emissions Modeling System (EMS-95), Final Report,
Lake Michigan Air Directors Consortium and the California Air Resources Board, AG-90/TS26
and AG-90/TS27, 120pp.
Venkatram, A.K. and P.K. Karamchandani, 1998: The ADOMII Scavenging Module, ENSR
Consulting and Engineering Report 0780-004-205. Camarillo, CA.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. 1C. S.
Ching, 1999.
1-18
-------
EPA/600/R-99/030
Chapter 2
MODELS-3 ARCHITECTURE: A UNIFYING FRAMEWORK FOR
ENVIRONMENTAL MODELING AND ASSESSMENT
Joan Novak* and Sharon Leduc"
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The Models-3 framework and Community Multiscale Air Quality (CMAQ) model were designed
and built to function together. The CMAQ science is documented in the other chapters and this
chapter is intended to provide science and technical details of the Models-3 framework which
integrates the CMAQ. The Models-3 User Manual is available and useful for those who want
operational details beyond the "big picture" provided in Chapter 2.
This chapter present the various components of the Models-3 framework (Dataset Manager,
Program Manager, Science Manager, Study Planner, Strategy Manager, Model Builder, Source
Code Manager, Framework Administrator, and Tool Manager). The Tool Manager provides
access to third party applications for emissions processing, visualization, and analysis. The
Framework Administrator provides the system administrator access to system lists and
administrative functions in Models-3. Each of the other components is provided for developing,
executing, and managing CMAQ applications. .
The computer architecture, the client-server design, the object-oriented data base management
system, and the graphical user interface are also addressed in this chapter, along with the design
features of Models-3.
"On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Joan Novak, MD-80, Research Triangle Park, NC 27711. Email:
novak.joan@epamail.epa.gov
"On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
2.0 MODELS-3 ARCHITECTURE: A UNIFYING FRAMEWORK FOR
ENVIRONMENTAL MODELING AND ASSESSMENT
2.1 Introduction
Models-3 is a flexible software system designed to simplify the development and use of
environmental assessment and decision support tools for a wide range of applications from
regulatory and policy analysis to understanding the interactions of atmospheric chemistry and
physics. The initial version of Models-3 contains a Community Multiscale Air Quality (CMAQ)
(Byun, et al., 1997, and other chapters of this document) modeling system with capabilities for
urban to regional scale air quality simulation of tropospheric ozone, acid deposition, visibility and
fine particulate. The Models-3 framework provides an interface between the user and operational
models, between the scientist and developing models, and between ever changing hardware and
software platforms.
The concept of an integrated modeling and analysis framework (Novak, et al., 1995; Dennis, et
al., 1996) was formulated in response to 1) problems and duplication of effort involved in
maintaining the separate data pre-processing and post-processing software systems required for
each model used to analyze different air quality pollutants or scales, 2) the difficulties States and
industry encountered using more complex modeling systems, and 3) the large expense required to
modify existing models to incorporate scientific advancements or to adapt the models to new
domains. Based on technology advances in the National High Performance Computing and
Communications program, Models-3 was designed to overcome these limitations and to serve as a
community framework for continual advancement and use of environmental assessment tools.
Capabilities and design features of the Models-3 framework provide more flexibility in data
handling, modeling, visualization and analysis.
2.2 Overview of the Models-3 Framework
The Models-3 framework contains components that assist 1) the model user with air quality
modeling studies and analysis of results, and 2) the model developer with creating, testing, and
performing comparative analysis of new versions of air quality models. The major design goal is
to simplify and integrate the development and use of complex environmental models, beginning
with air quality and deposition models. The function of each framework component is described
below.
2.2.1 Dataset Manager
Dataset Manager provides the user with the capability to register files for use with the modeling
and analysis programs within the Models-3 framework. The registration process involves entering
the location of the dataset (full path name) and metadata (information about the data such as
spatial - temporal extent and resolution, source of data, time convention, units, etc.) into the
Models-3 data base. Models-3 follows the Federal Geospatial Metadata Standard for metadata
2-2
-------
EPA/600/R-9W030
content. The datasets may be located on any network-connected computer system known to the
Models-3 system installed at the user's site. Once a dataset is registered, the user can search for
the dataset based on its metadata information, file type, etc.
Dataset Manager allows the user to view the details of the selected dataset to ensure the correct
one has been selected for use with an application. Dataset registration eliminates the need for the
user to type the entire path name each time the dataset is used. Instead the user can highlight the
dataset from a list of candidates that satisfy the search criteria specified by the user. Models-3
will automatically move selected data to the host where it is needed for a model execution.
Dataset Manager also provides standard capabilities such as deleting, copying, archiving, and
restoring files and metadata.
2.2.2 Program Manager
Program Manager allows the user to register, update, and search for executable programs and/or
scripts to make them available for use in defining studies within the Study Planner component.
During program registration, the user enters characteristics of the program into the framework
including descriptive information on program function, input requirements, output specifications,
runtime environment variables, target architecture and operating system. Once the program or
script is registered, this executable can be used in Study Planner to sequence a series of
executions which may depend on previous executions for input data. The user can access and
execute programs that are not registered. However, registering programs ensures that all
mandatory inputs have been specified and automatic naming and registration of output files is
activated, to facilitate tracking output from numerous program executions. Recommended model
configurations for standard domains are preregistered in the system, eliminating the need for the
typical user to deal with the details of program registration.
2.2.3 Study Planner
Study Planner allows the user to define a study and control the execution of its associated
processes. A study is a collection of plans and properties that are necessary to describe and
perform one or more environmental modeling analyses. A plan is a collection of information
defining dataset and program interdependencies and the sequence of execution of one or more
programs. Study Planner gathers much of its information from the Program Manager and Dataset
Manager registration data. The relationship between a program (node) and its required and
optional datasets (links) is user-defined through the process of constructing and annotating a
graphical diagram with simple drag-and-click mouse operations. Once a plan is constructed and its
graphical diagram fully annotated with desired input datasets and options, the plan can be
executed. User specified program options are entered by editing program environment variables.
Studies and associated plans are named entities that are saved in the system data base. Therefore,
a typical user can start with an existing study plan provided by the model developer and simply
change the dataset annotations by selecting, through a file browser, appropriate datasets needed
2-3
-------
EPA/600/R-99/030
for execution. The Study Planner provides capabilities to create new studies, copy and modify
existing studies, and delete existing studies.
2.2.4 Strategy Manager
Strategy Manager provides the capability to estimate point-, area-, and mobile-source emissions
for future years and to determine the relative effectiveness of specified control scenarios. The user
may adjust pollutant growth factors and emissions control factors to perform "what if analyses
for entire EPA regions, states, counties or for user-defined study areas. By applying estimated
yearly emission growth factors from the Emissions Growth and Assessment System, control
efficiency, rule effectiveness, and rule penetration factors to the EPA 1990 base year emissions
inventory, the Strategy Manager provides estimates of future year (1991 - 2010) emissions for
carbon monoxide, nitrogen oxide, particulate matter up to 10 microns, sulfur dioxide, and volatile
organic compounds. Strategy Manager is based on EPA's Multiple Projection System. An input
data processor is included to process the National Emissions Trend (NET) inventory (URL
http://www.epa.gov/oar/oaqps/) data format.
2.2.5 Tool Manager
Tool Manager provides access to third-party applications (tools) that are registered with the
Models-3 framework. Third-party applications that are currently available include Models-3
Emission Processing and Projection System (MEPPS), VisSD, Text Editor, Package for Analysis
and Visualization of Environmental Data (PAVE), SAS®, Arc/Info®, IBM Visualization Data
Explorer™, DXDriver, and VisDriver. Tool Manager allows the user to add tools to the system
that will help with the user's work. A file converter is included to convert between ASCII, SAS,
and I/O API files.
2.2.5.1 Emissions Processing
The majority of the emissions processing needed to support air quality model applications can be
prepared through minor modifications to pre-specified emissions plans in the Study Planner.
However, for those users who need total flexibility in specifying details for emissions preparation,
the Models-3 Emissions and Projection System (MEPPS) (Benjey and Moghari, 1996) is available
for interactive processing of emissions. MEPPS provides capabilities to input and perform quality
control on emission inventory data, and reformat and subset data for the user-specified modeling
domain. The main emissions processing is performed by a significantly revised version of the
Geocoded Emission Modeling and Projection (GEMAP) (Wilkinson and Emigh, 1994) system
which requires the user to have Arc/Info® and SAS® licenses for operation. This Emissions
Processor (EMPRO) transforms the annual county based emissions inventories into spatial and
temporal resolutions that are consistent with the target model application, i.e. typically hourly
gridded emissions for a selected domain. The processor also estimates mobile and biogenic
emissions and performs chemical speciation consistent with the chemical mechanism in the target
model. Mobile emission factors are calculated using Mobile 5a; biogenic emissions are calculated
2-4
-------
EPA/600/R-99/030
using the Biogenic Emissions Inventory System (BEIS2). Finally, according to a user-specified
criteria, an output processor merges the emissions files to provide the 2-D gridded files and a
point source stack emissions file required for the model execution. The user is offered numerous
choices for interactively changing parameters and reviewing results in all of the emissions
components. The global science specifications created by the user in Science Manager are
accessible from within MEPPS to ensure consistency when desired.
2.2.5.2 Visualization and Analysis Tools
Several freely available visualization tools are accessible from the Models-3 framework as well as
commercial visualization and statistics packages, such as IBM Visualization Data Explorer™ and
the SAS. A graphical user interface component, VisDriver, provides an interface for finding and
selecting Models-3 data files on local or remote hosts and launches visualization applications with
VisSD or PAVE. VisDriver performs requisite data conversions from the internal Models-3 data
format to the data format required by these third party visualization packages. VisDriver also
provides a data export capability to such formats as Advanced Visualization System (AVS™),
Flow Analysis Software Toolkit (FAST™), and an ASCII spreadsheet. On-line help is available to
assist the user with these visualization tools
(http://www.epa.gov/asmdnerl/models3/vistutor/user/user-guide-00.htm/).
PAVE
The Package for Analysis and Visualization for Environmental data (PAVE), developed at the
North Carolina Supercomputing Center of MCNC, allows the user to visualize 2-D multi-variate,
gridded environmental data with smooth tile plots,3-D mesh plots, time series line and bar plots,
vertical cross sections, wind vectors, and scatter plots. More information can be found at
http://www.iceis.mcnc.org/EDSS/pave_doc/Pave.html.
VisSD
VisSD, a public domain package developed at the University of Wisconsin Space Science and
Engineering Center, provides interactive visualization of large five-dimensional gridded datasets -
the data are real numbers at each point of a grid that spans three space dimensions, one time
dimension, and a dimension for enumerating multiple physical variables. VisSD provides
isosurfaces, contour line slices, colored slices, volume rendering of data in 3D, and rotation and
animation of the 3D image in real time. There is also a feature for wind/trajectory tracing. More
information can be found at http://www.ssec.wisc.edu/~billh/vis.html.
IBM Visualization Data Explorer
IBM Visualization Data Explorer™, a commercial visualization package, has been used to create
custom visualizations to support diagnostic evaluation of the air quality models with visualization
of multiple/nested grids, terrain following grids, grid cell time aggregate statistics such as
2-5
-------
EPA/600/R-99/030
maximum ozone concentration and hours of non-compliance, comparison of measured aircraft
data with model predictions along the path of flight, and detailed chemical analysis of the
integrated reaction rates predicted over space and time in the air quality models. Other standard
visualization capabilities for scalar and vector quantities, 2-D and 3-D graphs have also been
integrated through the DXDriver interface.
2.2.6 Science Manager
Critical model specifications typically found hardwired in air quality model codes, such as details
of the horizontal coordinate system, specification of horizontal grid dimensions, vertical layers,
chemical mechanism specification (reactions and rate constants), etc. are treated as globally
shared information hi the Models-3 framework. Use of this global information enables more
consistency throughout the system (i.e. emission, meteorological, and chemistry-transport
components). Detailed specifications for these key science components are entered by the user
only once using Science Manager graphical user interfaces. These specifications are saved as
named entities in an object-oriented database accessible by all model components. A typical user
would access and modify previously defined Science Manager components to define a new
modeling domain.
The functionality of Science Manager, however, is targeted to the model developer to facilitate
experimentation with new model components. For example, extensive research is done on
improving photochemical mechanism specifications. One component of Science Manager enables
the model developer to either edit an existing chemical mechanism or import a new set of
chemical reactions, specify chemical species, molecular weights, etc. Both the Regional Acid
Deposition Model, Version 2 (RADM-2) and the Carbon Bond-4 mechanisms are available in
the Models-3/CMAQ framework and the State Air Pollution Research Center (SAPRC) chemical
mechanism is being modified to fit within this paradigm to facilitate comparative studies with
these chemical mechanisms. As long as the chemical species in any modified mechanisms are
present in source emission profiles associated with Source Classification Codes (SCC), the
chemical mechanism specification will propagate to the emission processing subsystem to enable
generation of emission species consistent with the newly defined chemical mechanism.
2.2.7 Model Builder
The Models-3 framework facilitates the interchange of science components within a model,
customization of the chemistry mechanism, and changes in the modeling domain, horizontal grid
resolution, and vertical layering without the need for reprogramming. Modifying an existing
model or building a new model is a two-step process. First, the model developer must use
Science Manager to specify the horizontal coordinate system, horizontal grid dimensions, vertical
layers, chemical mechanism, and select science modules for each class (i.e. advection, diffusion,
cloud process, etc.) of science process to be included in the new model. Module selections are
contained in a named configuration file. After the needed science components are specified, the
2-6
-------
EPA/600/H-99/030
model developer uses Model Builder to select from available science'specifications and
configuration files to create a single custom model executable.
2.2.8 Source Code Manager
CMAQ source code is distributed, with the Models-3 framework to provide the user flexibility to
create a model for a user specified domain or to allow a user to build a new executable with a
selected chemistry mechanism. Model Builder uses Source Code Manager to access appropriate
source code to transform the user's selections into a working executable. Source Code Manager
allows a user to retrieve a version of a source code file; modify it, and return it to the code
archive after the change has been tested. Source code should only be modified by knowledgeable
model developers. Source Code Manager is based on Concurrent Versions System
(CVS)/Revision Control System (RCS) public domain software for code configuration
management and tracks the history and version numbers for all source code changes. These
version numbers become part of the history information associated with each output file.
2.2.9 Framework Administrator
The Framework Administrator subsystem menu allows the Models-3 system administrator to
manipulate system lists and to access administrative functions of other Models-3 subsystems.
System lists include users, hosts, device types, site IDs, file format types, compiler names,
operating system names, platform names, and time zone names. Through this subsystem, only the
authorized system administrator can set access controls for users, add or delete entries from the
system lists, and maintain the integrity of controlled files.
2.3 Models-3 System Architecture
The Models-3 framework is designed as a three-tier client-server architecture with an object-
oriented data base management system (OODBMS) ,ObjectStore™, to store persistent data (i.e.
studies, horizontal grid definitions, metadata containing pointers to physical dataset locations ,
etc.). When the user brings up the Models-3 graphical user interface on user's desktop system, the
user becomes a "client" requesting services from a "server". Several Models-3 servers act as
"clients" requesting services from other Models-3 "servers". The client and server can be on the
same machine or on separate machines giving the system scalability by either adding more client
workstations with only slight impact on performance, or moving to faster and larger server
machines. Each of the subsystem functions described in Section 2.2 is implemented as a server
and can be can run on different machines. The Models-3 servers were initially developed on a
Sun workstation using the Solaris 2.5.1 operating system (Models-3 Version 2.1). Models-3 is
being ported to a Sun Solaris 2.6 operating system (Models-3 Version 2.2). The next server to be
released for Models-3 will be SGI (late 1998). Models-3 Version 3 (summer 1999) will allow
server to be a PC with a Windows NT® operating system.
2-7
-------
EPA/600/R-99/030
This client-server architecture in conjunction with commercial cross-platform communications
interface - ORBIX™ by lona Technologies, Ltd. which is compliant with the Common Object
Request Broker Architecture (CORBA) standard - and an OODBMS containing pointers to
datasets distributed over the network enables transparent use of multiple computing platforms and
access to data across the network. Figure 2-1 illustrates a possible configuration for multiple
installations for Models-3 software. Each independent site must have at least one server to
support the local OODBMS and Models-3 sessions. Each site can run autonomously. A two-
server configuration is recommended at each installation: one to support emissions processing
activities, and the second to support other client-server and model computations. Of course,
additional servers will improve performance for larger applications and the current version of
Models-3 also supports batch submissions of model executions to a supercomputer with job status
reporting back to the framework through the standard Network Queuing Service (NQS). In the
future, we envision maintaining a master database to foster sharing of data and to minimize
duplication of effort.
Figure 2-1. Seamless Computing and Data Management
from PC to Scalable Parallel Computer
2-8
-------
EPA/600/R-99/030
To adapt to ever changing hardware and software, the framework uses a layered interface design
that isolates critical system components, thus minimizing the impact of hardware and software
upgrades. Each layer, as seen in Figure 2, is connected to its adjacent layer(s) by a well specified
interface thus easing the task of replacement if a more advanced component becomes available.
Another key design choice is the use of an input/output application programming interface (I/O
API) for the data access layer. This concept is implemented by using a standard FORTRAN or C
callable library for all input/output in the air quality model codes. Thus I/O efficiencies can be
improved by replacing this centralized library or new data types can be handled by adding new
routines to the library. -
The I/O API is layered on top of the netCDF standard data file format developed at the National
Center for Atmospheric Research. This data format uses the underlying external Data ,
Representation (XDR), which is IEEE (Institute of Electrical and Electronics Engineers, Inc.)
compliant and, therefore, enables cross-platform transfer of data without conversion. The
Models-3 implementation conforms to the netCDF specification, but sets additional conventions
for information stored in the file header. NetCDF is a self documenting file format that contains
complete specifications (parameter names, units, etc) for the file contents. All netCDF records can
be accessed via direct access methods to minimize data access times. When registering
conforming programs, the Models-3 framework automatically writes execution history
information to the file header record of each output file and to the OODBMS metadata
associated with the output file. The history information, which establishes traceability of output
datasets to the model that generates them, contains the date/time of model execution, the version
numbers of the science process modules contained in the model executable, Science Manager
specification names, host computing environment, and other pertinent information. Science
Manager's global shared data objects and the netCDF format are the key aspects of building "plug
compatible" science modules that conform to Models-3 guidelines.
USER
INTERFACE
USER INTERFACE
Management Layer:
Data Manager
Program Manager
Study Manager
Strategy Manager
Tool Manager
Science Manager
Model Builder
Source Code Manager
Framework Adm inistrator
Environment Layer: OS, System "Personality"
Computational Layer: Programs: models, analysis, visualization,
Data Access Layer:
I/O Application Programming Interface
Data Structure/Representation: netCDF, XDR
Data Storage: File systems & databases
Physical Device Layer: Disks, networks, printers, machines
Figure 2-2. Architectural Layering: Flexibility for Future Change
2-9
-------
EPA/600/R-99/030
2.4. Schedule and Future Plans
The Models-3 framework provides a unifying foundation for continual community evolvement of
environmental modeling and assessment tools with possible extensions beyond the current air
quality implementation. The use of the Models-3 framework for multimedia exposure and risk
assessment is possible. A simple integration of the Hydrologic Simulation Program FORTRAN
(HSPF) and the Chesapeake Bay Water Quality Model (CBWQM) in the Models-3 framework
explored multimedia model linkage capabilities. Maintaining Models-3 framework to use
emerging computing capabilities and transferring that capability to science researchers and
environmental decision makers will continue to be the focus of the Models-3 effort.
2.5 References
Benjey, W.G., and N.M. Moghari. Functionality of an integrated emission preprocessing system
for air quality modeling: the Models-3 emission processor. In The Emissions Inventory:
Programs & Progress. VIP-56. Proceedings of a Specialty Conference, Research Triangle Park,
NC, October 11-13, 1995. US Environmental Protection Agency, Research Triangle Park, NC,
and Air & Waste Management Association, Pittsburgh, 463-471 (1996).
Byun, D.3 J. Young, J. Gipson, J. Godowitch, F. Binkowski, S. Roselle, B. Benjey, J. Pleim, J.
Ching, J. Novak, C. Coats, T. Odman, A. Hanna, K. Alapaty, R. Mathur, J. McHenry, U.
Shankar, S. Fine, A. Xiu, and C. Jang. Description of the Models-3 Community Multiscale Air
Quality (CMAQ) model. Proceedings of the American Meterological Society 78th Annual
Meeting, January 11-16,1998, Phoenix, AZ.
Dennis, R.L., D.W. Byun, J.H. Novak, K.L. Galluppi, C.J. Coats, M.A. Vouk. The Next
Generation of Integrated Air Quality Modeling: EPA's Models-3. Atmospheric Environment, Vol.
3Q.No.12, pp. 1925 - 1938, (1996).
Emission Inventory Improvement Program. URL http://www.epa.gov/oar/oaqps/eiip/
Novak, J.H., R.L. Dennis, D.W. Byun, J.E. Pleim, K.J. Galluppi, C.J. Coats, S. Chall, M.A.
Vouk. EPA Third-generation Air Quality Modeling System, Volume 1: Concept. EPA
600/R95/084, National Exposure Research Laboratory, Research Triangle Park, NC, 55 pp.
(1995).
Wilkinson, J.G., and Emigh, R.A. The Geocoded Emissions Modeling and Projections System
(GEMAP): Advanced Training Workshop. Prepared for Environmental Protection Agency,
Office of Research and Development, by Alpine Geophysics, Boulder, CO (1994).
2-10
-------
EPA/600/R-99/030
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
2-11
-------
-------
EPA/600/R-99/030
Chapter 3
DEVELOPING METEOROLOGICAL FIELDS
Tanya L. Otte*
Atmospheric Modeling Envision
National Exposure Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Meteorological data are important in many of the processes simulated in the Community
Multiscale Air Quality (CMAQ) model included in the first release of the Models-3 framework.
The meteorology model that has been selected and evaluated with CMAQ is the Fifth-Generation
Pennsylvania State University/National Center for Atmospheric Research (NCAR) Mesoscale
Model (MM5). All software in the Perm. State/NCAR mesoscale modeling system has been
dedicated to the public domain and is presently used for both research and operational purposes
at many organizations worldwide. Improvements to MM5 within the meteorological community
are ongoing, and these enhancements should be evaluated for their applicability to air quality
modeling. Other meteorological models are being considered for compatibility with CMAQ but
are not provided with the initial release. The application of MM5 with CMAQ is described in
Chapter 3,
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department, of Commerce.
Corresponding author address: Tanya L. Otte, MD-80, Research Triangle Park, NC 27711. Email:
tlotte@hpcc.epa.gov
-------
EPA/600/R-99/030
3.0 DEVELOPING METEOROLOGICAL FIELDS
3.1 Credits and Disclaimers for Use of MM5
The Fifth-Generation Pennsylvania State University/National Center for Atmospheric Research
(NCAR) Mesoscale Model (MM5) was developed in cooperation with Perm State and the
University Corporation for Atmospheric Research (UCAR). Perm State and UCAR make no
proprietary claims, either statutory or otherwise, to this version and release of MM5, and they
consider MM5 to be in the public domain for use by any person or entity without any fee or
charge. Perm State and UCAR shall be credited in any publications that result from the use of
MM5. The names "Perm State" and "UCAR" shall not be used or referenced in any advertising
or publicity that endorses or promotes any products or commercial entity associated with or
using MM5, or any derivative works thereof, without the written authorization of UCAR and
Penn State.
MM5 is provided to the public by Penn State and UCAR on an "as is" basis, and any
warranties, either express or implied, including but not limited to implied warranties of non-
infringement, originality, merchantability, and fitness for a particular purpose, are disclaimed.
Neither UCAR nor Penn State will be obligated to provide the user with any support, consulting,
training, or assistance of any kind regarding the use, operation, and performance of MM5, nor
will they be obligated to provide the user with any updates, revisions, new versions, error
corrections, or "bug" fixes. In no event will UCAR and Penn State be liable for any damages,
whatsoever, whether direct, indirect, consequential, or special, which may result from an action in
contract, negligence, or other claim that arises out of or about the access, use, or performance of
MM5, including infringement actions.
Data that are obtained from NCAR (e.g., global analyses and observations) are prepared and
maintained by the Data Support Section, Scientific Computing Division, NCAR. NCAR is
operated by UCAR and is sponsored by the National Science Foundation. Any published
research that uses data from the NCAR archive shall credit the maintainers of the archive. In
addition, the original source(s) of data shall be acknowledged.
3.2 Meteorology Model Pre-Processing
Before the meteorology model can be run, several smaller ("pre-processing") programs must be
run to set up the domain for the simulation and to generate a set of initial and boundary
conditions for the meteorology model. The pre-processing programs are briefly described in this
section.
3-2
-------
1PA/600/R-99/030
3.2.1 Defining the Simulation Domain (TERRAIN)
Domains for the meteorology simulations are defined by several primary parameters: number of
grid points in each horizontal dimension, grid spacing, center latitude, center longitude, map
projection (Mercator, Lambert conformal, or polar stereographic), and number of "nested"
domains and their horizontal dimensions. (Some other user-specific parameters are also defined
based on the primary parameters, as required.) These parameters are processed by a program
that is executed only when a new domain location is required. This program, TERRAIN, makes
use of high-resolution global terrain and land use data sets to create "static files" for the domain.
The static files currently include values for each grid point for terrain height and land use
specification (e.g., deciduous forest, desert, water). Future versions of TERRAIN may include
additional diurnal, seasonal, and location-specific information. The TERRAIN program is
thoroughly described by Guo and Chen (1994).
3.2.2 Processing the Meteorological Background Fields (DATAGRID)
After the simulation domain has been established, the program DATAGRID is run to process the
meteorological background fields. DATAGRID generates first-guess fields for the model
simulation by horizontally interpolating a larger-scale data set (global or regional coverage) to the
simulation domain. DATAGRID interpolates the background fields to the simulation domain for
times throughout the simulation period; these files are used ultimately to generate lateral
boundary conditions for the coarse-domain simulations. (Nested domains obtain lateral boundary
conditions from the coarse domain.) DATAGRID also processes the sea-surface temperature and
snow files by interpolating the analyses to the simulation domain. Lastly, DATAGRID calculates
map-scale factors and Coriolis parameters at each grid point to be used by MM5. The
DATAGRID program is thoroughly described by Manning and Haagenson (1992).
3.2.3 Objective Analysis (RAWINS)
The program RAWINS performs an objective analysis by blending the first-guess fields generated
by DATAGRID with upper-air and surface observations. There are four objective analysis
techniques available in RAWINS. The following descriptions are taken largely from Dudhia et al.
(1998).
• The Cressman scheme assigns a circular radius of influence to each observation. The
weights associated with the observations decrease radially from the observation to the
radius of influence. The first-guess (meteorological background) field at each grid point is
adjusted by taking into account all observations that influence that grid point. (See
Cressman, 1959.)
3-3
-------
EPA/600/R-99/030
• The Ellipse scheme alters the Cressman circles for analyses of wind and relative
humidity by elongating the circles along the wind flow. The eccentricity of the ellipses
increases as wind speed increases. This scheme reduces to the Cressman scheme under
light wind conditions. (See Benjamin and Seaman, 1985.)
• The Banana scheme alters the Cressman circles for analyses of wind and relative
humidity by elongating the circles in the direction of the flow and curving the influence
region along streamlines. The resulting influence region is banana-shaped. This scheme
reduces to the ellipse scheme under straight-flow conditions,.and it further reduces to the
Cressman scheme under light wind conditions. (See Benjamin and Seaman, 1985.)
• The Multiquadric Interpolation scheme uses hyperboloid radial basis functions to
perform objective analysis. This scheme takes much more time and memory than the
Cressman-based schemes. This scheme must be used with caution in data-sparse regions
and when more than 25% of the domain is water-covered. (See Nuss and Titley, 1994.)
In addition, RA WINS performs some data quality control checks and buddy checks with user-
defined thresholds. The objective analysis is performed for a user-defined number of pressure
surfaces (generally mandatory levels plus some additional surfaces). RAWINSis also used to
prepare the analyses for the analysis nudging in the simulation model. The RA WINS program is
thoroughly described by Manning and Haagenson (1992).
3.2.4 Setting the Initial and Boundary Conditions (INTERP)
The "standard" INTERP program sets the initial and boundary conditions for the meteorology
simulation. The analyses from RAWINS are interpolated to MM5's staggered grid configuration
and from their native vertical coordinate (pressure) to MM5's vertical coordinate (a terrain-
following, pressure-based "sigma" coordinate). In addition, the state variables are converted as
necessary, e.g., relative humidity to specific humidity. The analyses from one time (generally the
first time analyzed by RAWINS) are interpolated by INTERP to provide MM5's initial
conditions, while analyses from all times are interpolated to generate MM5's lateral boundary
conditions.
INTERP can also be used for "one-way nesting," where the MM5 output from one model run is
interpolated to provide initial and boundary conditions for the nested run. This is a "non-
standard" use of the INTERP program, but it is commonly used to support air quality modeling.
3-4
-------
EPA/600/R-99/030
3.3 The Meteorology Model (MM5)
The following subsections include brief summaries of the science and options that are available in
MM5. Thorough descriptions of the standard MM5 options are found in Grell et al. (1994) and
Dudhia et al. (1998). The source code for an early release of MM5 is documented in Haagenson
et al. (1994). The Models-3 package does not include NCAR's most current release of MM5.
MM5 user options have been tailored and expanded in the Models~3 release to support air
quality modeling. Some of these options have not yet been accepted into the official version that
is maintained'by NCAR. Caution should be exercised-when deviating, jrom the version of MM5
that is included with Models-3. Refer to Section 3.5 for a summary of changes to MM5 for
CMAQ.
3.3.1 Brief History
MM5 is the Fifth-Generation Perm State/NCAR Mesoscale Model. It has evolved from the
model used by Anthes in the early 1970s, described by Anthes and Warner (1978). The
improvements in MM5 over the previous version (MM4) include the option for non-hydrostatic
physics, as well as more sophisticated explicit moisture, boundary layer processes, radiation,
convective parameterization, among other improvements. Dudhia (1993) documented the major
changes from MM4 to MM5, Starting with MM5 Version 2 (MM5v2), the software has been
restructured to run on various hardware platforms in addition to the Cray, with emphasis placed
on workstation-based MM5 simulations. The Cray version of MM5 provided on the Models-3
installation tape has been tested on the EPA's Cray C90. A workstation-based version of MM5
may be used with subsequent releases of Models-3.
3.3.2 Horizontal and Vertical Grid
The coordinate system for MM5 is (x, y, sigma-p). The x and y dimensions are a regular lattice
of equally spaced points (delta-x = delta-y = horizontal grid spacing, hi kilometers) forming rows
and columns. Sigma is a terrain-following vertical coordinate that is a function of the pressure at
the point on the grid (in hydrostatic runs) or the reference state pressure (in non-hydrostatic
runs), the surface pressure at the grid point, and the pressure at the top of the model. Sigma
varies from 1 at the surface to 0 at the top of the model. The influence of the terrain on the sigma
structure diminishes with height, such that the sigma surfaces near the top of the model are nearly
parallel.
The horizontal grid in MM5 has an Arakawa-B staggering of the velocity vectors with respect to
the scalars (Arakawa and Lamb, 1977). The momentum variables (u- and v-components of wind
and the Coriolis force) are on "dot" points, while all other variables (e.g., mass and moisture
variables) are on "cross" points. The dot points form the regular lattice for the simulation
3-5
-------
EPA/600/R-99/030
domain, while the cross points are offset by 0.5 grid point in both the x and y directions. Note
that the interpolation of the variables to the staggered grid is done automatically within the
INTERP program.
3.3.3 Prognostic Equations
MM5 is based on primitive physical equations of momentum, thermodynamics, and moisture.
The state variables are temperature, specific humidity, grid-relative wind components, and
pressure. In the prognostic equations, the state variables are mass-weighted with a modified
surface pressure. MM5 can be run as either a hydrostatic or non-hydrostatic model. In the
hydrostatic model, the state variables are explicitly forecast. In the non-hydrostatic model
(Dudhia, 1993), pressure, temperature, and density are defined in terms of a reference state and
perturbations from the reference state. MM5 is not mass-conserving in the non-hydrostatic
mode. The vertical (sigma) coordinate is defined as a function of pressure. The model's
prognostic equations are thoroughly discussed in Grell et al. (1994) and Dudhia et al. (1998).
3.3.3.1 Time Differencing
The hydrostatic and non-hydrostatic versions of MM5 use different time differencing schemes to
filter the fast waves from the prognostic solutions in the model. In the non-hydrostatic model, a
semi-implicit scheme based on Klemp and Wilhelmson (1978) is used to control the acoustic
waves in the model solution. In the hydrostatic model, a split-explicit scheme based on Madala
(1981) is used to control gravity waves in the model solution. The time differencing in MM5 is
discussed at length in Grell et al. (1994) and Dudhia et al. (1998).
3.3.3.2 Lateral Boundary Conditions
There are five options for lateral boundary conditions in MM5: fixed, relaxation, time dependent,
time and inflow/outflow dependent, and sponge. The lateral boundaries in MM5 consist of
either the outer five grid points (relaxation and sponge options) or the outer grid point (all other
options) on the horizontal perimeter of the simulation domain. (The outer four grid points are
used for boundary conditions for "cross" point variables for the relaxation and sponge options.)
The lateral boundary conditions for the coarse domain are derived from the background fields
processed in DATAGRID and INTERP, When the one-way nest option is selected, the lateral
boundary conditions for nested domains are interpolated from the simulation on the parent
domain.
3-6
-------
1PA/600/R-99/030
3.3.4 Model Physics
Several model physics options in MM5 are briefly noted below. The model physics options are
further discussed and compared in Dudhia et al. (1998). The descriptions of the model physics
options are largely taken from Dudhia et al (1998), and other pertinent references are noted.
3.3.4.1 Radiation
There are five atmospheric radiation cooling schemes available in MM5.
• The "None" option applies no mean radiative tendency to the atmospheric temperature.
This scheme is unrealistic for long-term simulations.
• The Simple Cooling scheme sets the atmospheric cooling rate strictly as a function of
temperature. There is no cloud interaction or diurnal cycle.
• The Surface Radiation scheme is used with the "none" and "simple cooling" schemes.
This scheme includes a diumally varying shortwave and longwave flux at the surface for
use in the ground energy budget. These fluxes are calculated based on atmospheric
column-integrated water vapor and low/middle/Mgh cloud fraction estimated from relative
humidity.
• The Dudhia Longwave and Shortwave Radiation scheme is sophisticated enough to
account for longwave and shortwave interactions with explicit cloud and clean air. This
scheme includes surface radiation fluxes and atmospheric temperature tendencies. This
scheme requires longer CPU time, but not much memory. (See Dudhia, 1989.)
• The CCM2 Radiation scheme includes multiple spectral bands in shortwave and
longwave, but the clouds are treated simply as functions of relative humidity. This
scheme is suitable for larger grid scales and probably more accurate for long time
integration (e.g., climate modeling). It also provides radiative fluxes at the surface. (See
Hack etal., 1993.)
3.3.4.2 Convective Parameterization
There are currently six convective parameterization schemes in MM5. There is also the option
for no convective parameterization and an independent option for shallow convection. The
convective parameterization schemes have been designed for use at various simulation scales, and
they are not entirely interchangeable. For example, each scheme uses different assumptions for
convective coverage on the sub-grid-scale and for the convective trigger function. The convective
parameterization schemes also differ peatly in CPU usage and memory requirements.
3-7
-------
EPA/600/R-99/030
The Anthes-Kuo scheme is based on moisture convergence and is mostly applicable to
larger grid scales (i.e., greater than 30 km). This scheme tends to produce more convective
rainfall and less resolved-scale precipitation. This scheme uses a specified heating profile
where moistening is dependent on relative humidity. (See Anthes, 1977.)
The Fritseh-Chappel scheme is based on relaxation to a profile due to updraft,
downdraft, and subsidence region properties. The convective mass flux removes 50% of
the available buoyant energy in the relaxation time. There is a fixed entrainment rate.
This scheme is suitable for 20-30 km scales due to the single cloud assumption and local
subsidence. (See Fritsch and Chappel, 1980.)
The Arakawa-Schubert scheme is a multi-cloud scheme that is otherwise similar to the
Grell scheme (described below). This scheme is based on a cloud population, and it
allows for entrainment into updrafts and the existence of downdrafts. This scheme is
suitable for larger grid scales (i.e., greater than 30 km). This scheme can be
computationally expensive compared to the other available schemes. (See Arakawa and
Schubert, 1974.)
The Kain-Fritsch scheme is similar to the Fritseh-Chappel scheme, but it uses a
sophisticated cloud-mixing scheme to determine entrainment and detrainment. This
scheme also removes all available buoyant energy hi the relaxation time. (See Kain and
Fritsch, 1990,1993.)
The Betts-MiUer scheme is based on relaxation adjustment to a reference post-
convective thermodynamic profile over a given period. This scheme is suitable for scales
larger than 30 km. However, there is no explicit downdraft, so this scheme may not be
suitable for severe convection. (See Belts, 1986; Betts and Miller, 1986,1993; and Janjic,
1994.)
The Grell scheme is based on rate of destabilization or quasi-equilibrium. This is a
single-cloud scheme with updraft and downdraft fluxes and compensating motion that
determines the heating and moistening profiles. This scheme is useful for smaller grid
scales (e.g., 10-30 km), and it tends to allow a balance between the resolved scale rainfall
and the convective rainfall. (See Grell et al., 1991; and Grell, 1993.)
The "no convective parameterization" option (e.g., explicitly resolved convection on
the grid scale) is also available. This option is generally used for simulations on domains
with horizontal grid spacing smaller than 10 km.
3-8
-------
EPA/600/R-99/030
• The Shallow Convection scheme is an independent option that handles non-
precipitating clouds that are assumed to be uniform and to have strong entrainment, a.
small radius, and no downdrafts. This scheme is based on the Grell and Arakawa-
Schubert schemes. There is also an equilibrium assumption between cloud strength and
sub-grid boundary layer forcing.
3.3.4.3 Planetary Boundary Layer Processes
Four planetary boundary layer (PBL) parameterization schemes are available in MM5. These
parameterizations are most different in the turbulent closure assumptions that are used. The
PBL parameterization schemes also differ greatly hi CPU usage.
• The Bulk Formula scheme is suitable for coarse vertical resolution in the PBL (i.e.,
greater than 250 m vertical grid sizes). This scheme includes two stability regimes.
• The Blackadar scheme is suitable for "high-resolution" PBL (e.g., five layers in the
lowest kilometer and a surface layer less than 100 m thick). This scheme has four
stability regimes; three stable and neutral regimes are handled with a first-order closure,
while a free convective layer is treated with a non-local closure. (See Blackadar, 1979.)
• The Burk-Thompson scheme is suitable for coarse and high-resolution PBL. This
scheme explicitly predicts turbulent kinetic energy for use in vertical mixing, based on a
1,5-order closure derived from the Mellor-Yamada formulas. (See Burk and Thompson,
1989.)
• The Medium Range Forecast (MRF) model scheme is suitable for high-resolution
PBL. This scheme is computationally efficient. It is based on a Troen-Mahrt
representation of the counter-gradient term and a first-order eddy diffusivity (K) profile
in the well-mixed PBL. This scheme has been taken from the National Centers for
Environmental Prediction's (NCEP's) MRF model. (See Hong and Pan, 1996.)
3.3.4.4 Surface Layer Processes
The surface layer processes with the Blackadar and MRF PBL schemes have been parameterized
with fluxes of momentum, sensible heat, and latent heat, following Zhang and Anthes (1982).
The energy balance equation is used to predict the changes in ground temperature using a single
slab and a fixed-temperature substrate. The slab temperature is based on an energy budget, and
the depth is assumed to represent the depth of the diurnal temperature variation (~10-20 cm).
The 13 land use categories are used to seasonally define the physical properties at each grid point
(e.g., albedo, available moisture, emissivity, roughness length, and thermal inertia).
3-9
-------
EPA/600/R-99/030
A five-layer soil temperature model (Dudhia, 1996) is also available as an option in MM5. In
this model, the soil temperature is predicted at layers of approximate depths of 1,2,4, 8, and
16 cm, with a fixed substrate below using a vertical diffusion equation. This scheme vertically
resolves the diurnal temperature variation, allowing for more rapid response of surface
temperature. This model can only be used in conjunction with the Blackadar and MRF PBL
schemes.
In a subsequent release of CMAQ, the Pleim-Xiu land-surface scheme (Pleim and Xiu, 1995)
may be included in MM5. The Pleim-Xiu scheme has been developed to address surface
processes that can significantly impact air quality modeling, including evapotranspiration and soil
moisture. Notable in the Pleim-Xiu scheme is the more careful treatment of the surface
characteristics (particularly vegetation parameters) that are currently assigned to grid points
based on land use specification, as well as a more detailed land use and soil type classification
database.
3.3.4.5 Resolvable-Scale Microphysies Schemes
There are six resolvable-scale (explicit grid-scale) microphysics schemes in MM5. There is also
an option for a "dry" model run. The microphysics schemes have been designed with varying
degrees of complexity for different applications of the model. In addition, there are new
prognostic output variables that are generated by the more sophisticated schemes. These
microphysics schemes also differ greatly in CPU usage and memory requirements.
* The Dry scheme has no moisture prediction. Water vapor is set to zero. This scheme is
generally used for sensitivity studies.
• The Stable Precipitation scheme generates non-convective precipitation. Large-scale
saturation is removed and rained out immediately. There is no evaporation of rain or
explicit cloud prediction.
* The Warm Rain scheme uses microphysical processes for explicit predictions of cloud
and rainwater fields. This scheme does not consider ice phase processes. (See Hsie and
Anthes, 1984.)
* The Simple Ice scheme adds ice phase processes to the warm rain scheme without
adding memory. This scheme does not have supercooled water, and snow is immediately
melted below the freezing level. (See Dudhia, 1989.)
« The Mixed-Phase scheme adds supercooled water to the simple ice scheme, and slow
melting of snow is allowed. Additional memory was added to accommodate the ice and
3-10
-------
EPA/600/R-99/030
snow. This scheme does not include graupel or riming processes. (See Reisner et al.,
1993,1998.)
• The Mixed-Phase with Graupel scheme adds graupel and ice number concentration
prediction equations to the mixed-phase scheme. This scheme is suitable for cloud-
resolving scales. (See Reisner et al., 1998.)
« The NASA/Goddard Microphysics scheme explicitly predicts ice, snow, graupel, and
hail. This scheme is suitable for cloud-resolving scales. (See Tao and Simpson, 1993.)
3.3.5 Nesting
MM5 can simulate nested domains of finer resolutions within the primary simulation domain. In
MM5, the software is configured to enable up to nine nests (ten domains) within a particular run.
However, due to current hardware resources, the state of the science, numerical stability, and
practicality, the number of domains hi a simulation is generally limited to four or fewer.
Nesting can be accomplished by either a "one-way" or a "two-way" method. In one-way
nesting, the coarse-resolution domain simulation is run independently of the nest. The coarse
domain can then provide the initial and boundary conditions for its nest. In one-way nesting,
each domain can be defined with independent terrain fields, and there are no feedbacks to the
coarse domain from its nest. Note that the simulated meteorology at the same grid point in the
coarse-resolution domain is likely to be different (if only slightly) from the nest in a one-way
nest simulation.
Two-way interactive nesting (Zhang et al., 1986; Smolarkiewicz and Grell, 1992) allows, for
feedback to occur between the coarse-resolution domain and the nest throughout the simulation.
The two domains are run simultaneously to enable this feedback, and terrain in the overlapping
regions must be compatible to avoid mass inconsistencies and generation of numerical noise. The
TERRAIN program automatically defines the terrain compatibility when the user specifies the
two-way nesting interaction. When two-way nesting is used, the portion of the coarse-resolution
domain that is simulated in the nest may reflect too much smaller-scale detail from the nest to be
useful for the CMAQ simulations.
The nesting ratio between domains in MM5 is generally 3:1. (Some other mesoscale meteorology
models allow for user-defined nest ratios.) For example, if the coarse domain is a 36-km
resolution domain, its nest will be a 12-km resolution domain. This is strictly true for a two-way
nest, but is largely held as a standard for the one-way nests in MM5. The nest ratio restricts the
number of grid points in each dimension of the nest domains to a multiple of 3, plus 1.
3-11
-------
EPA/600/R-99/030
3.3.6 Four-Dimensional Data Assimilation
The four-dimensional data assimilation (FDDA) scheme included in MM5 Is based on
Newtonian relaxation or "nudging". Nudging is a continuous form of FDD A where artificial
(non-physical) forcing functions are added to the model's prognostic equations to nudge the
solutions toward either a verifying analysis or toward observations. The artificial forcing terms
are scaled by a nudging coefficient that is selected so that the nudging term will not dominate the
prognostic equations. The nudging terms tend to be one order of magnitude smaller than the
dominant terms in the prognostic equations and represent the inverse of the e-folding time of the
phenomena captured by the observations.
There are two types of nudging in MM5: analysis nudging and observation nudging ("obs
nudging"). Analysis nudging gently forces the model solution toward gridded fields. Analysis
nudging can make use of three-dimensional analyses and some surface analyses. Analysis
nudging is generally used for scales where synoptic and mesoalpha forcing are dominant. Obs
nudging gently forces the model solution toward individual observations, with the influence of the
observations spread in space and time. Obs nudging is better suited for assimilating high
frequency, asynoptic data that may not otherwise be included in an analysis.
Nudging in MM5 is extensively discussed in Stauffer and Seaman (1990,1994) and Stauffer et al.
(1991), The data assimilation is generally used throughout the MM5 simulation period for air
quality simulations. Three-dimensional analyses of wind, temperature, and moisture are
assimilated, and only surface analyses of wind are assimilated, following Stauffer et al. (1991).
3.4 Meteorology Model Post-Processing
Since the output variables that are generated by an MM5 simulation are not always useful in
their raw form, those variables must be converted into fields that are required by the chemistry
and emission models. The conversion of MM5 output to useful fields for the other Models-3
programs is accomplished in the Meteorology Chemistry Interface Processor (MCIP), which is
discussed in Chapter 12. MCIP computes variables that are useful to the subsequent models in
Models-3, and it creates an output file written with the I/O API libraries in netCDF format that
is standard in the Models-3/CMAQ modeling system.
3.5 Changes to the MM5 System's Software for Models-3
The version of MM5 included with the initial release of Models-3 is MM5 Version 2, Release 6.
The system is complete with "bug fixes" included through October 1997. Only selected "bug
fixes" and upgrades have been included beyond October 1997. (Including all of the changes to
MM5 would have jeopardized testing and evaluation of CMAQ, which uses the MM5 output.)
3-12
-------
EPA/600/R-99/030
The version of MM5 in the Models-3 release includes most of the science options in NCAR's
current version. The omission of upgrades should not result in substandard or corrupted
meteorological simulations.
The following is a summary of EPA-initiated changes that were made to the NGAR release of
MM5 and its supporting software to support air quality modeling with CMAQ:
All Programs: ,
• Standardized the radius of the earth as 6370.997 km to be consistent with chemistry and
emission models.
TERRAIN:
• Set terrain height over ocean to zero when using the 1 -minute terrain database.
• Improved representation of urban areas along coasts.
INTERP:
• Increased loop indices and parameter sizes to accommodate 120-hour simulation.
MM5:
• Modified script and source code to enable analysis nudging on a one-way nested
simulation.
3.6 References
Anthes, R. A., 1977: A cumulus parameterization scheme utilizing a one-dimensional cloud
model. Man. Wea. Rev,, 106,270-286.
Anthes, R. A., and T. T. Warner, 1978: Development of .hydrodynamic models suitable for air
pollution and other mesometeorological studies. Man. Wea. Rev,, 106,1045-1078.
Arakawa, A.5 and V. R. Lamb, 1977: Computational design of the basic dynamical process of the
UCLA general circulation model. Methods in Computational Physics, 17,173-265.
• i'
Arakawa, A., and W. H. Schubert, 1974: Interaction of a cumulus cloud ensemble with the large
scale environment. Part I. J.Atmos. Sci., 31,674-701.
Benjamin, S. G.» and N. L. Seaman, 1985: A simple scheme for objective analysis in curved flow.
Man. Wea. Rev., 113,1184-1198.
3-13
-------
EPA/600/R-99/030
Betts, A. K., 1986: A new convective adjustment scheme: Part I: Observational and theoretical
basis. Quart. J. Roy. Meteor. Sac., 112,677-692.
Betts, A. K., and M. J. Miller, 1986: A new convective adjustment scheme. Part II: Single
column tests using GATE wave, BOMEX, ATEX and Arctic air-mass data sets. Quart. J. Roy.
Meteor. Soc., 112,693-709.
Betts, A. K., and M. J. Miller, 1993: The Betts-Miller scheme. The representation of cumulus
convection in numerical models. K. A. Emanuel and D. J. Raymond, Eds., Amer. Meteor. Soc.,
246 pp.
Blackadar, A. K., 1979: High resolution models of the planetary boundary layer. Advances in
Environmental Science and Engineering, 1, No. 1, Pfafflin and Ziegler, Eds., Gordon and Briech
Sci. PubL, New York, 50-85.
Burk, S. D., and W. T. Thompson, 1989: A vertically nested regional numerical prediction model
with second-order closure physics. Man. Wea. Rev., 117,2305-2324.
Cressman, G. P., 1959: An operational objective analysis system. Mon. Wea. Rev,, 87,
367-374.
Dudhia, J., 1989: Numerical study of convection observed during the winter monsoon experiment
using a mesoscale two-dimensional model. J. Atmos. Sci., 46,3077—3107.
Dudhia, J., 1993: A nonhydrostatic version of the Perm State/NCAR mesoscale model:
Validation tests and simulation of an Atlantic cyclone and cold front. Mon. Wea. Rev,, 121,
1493-1513.
Dudhia, J., 1996: A multi-layer soil temperature model for MM5. Preprints, The Sixth
PSU/NCAR Mesoscale Model Users' Workshop, Boulder, CO, Natl. Ctr. Atmos. Res.
Dudhia, J., D. Gill, Y.-R. Quo, D. Hansen, K. Manning, and W. Wang, 1998: PSU/NCAR
mesoscale modeling system tutorial class notes (MM5 modeling system version 2). [Available
from the National Center for Atmospheric Research, P. O. Box 3000, Boulder, CO 80307.]
Fritsch, J. M., and C. F. Chappel, 1980: Numerical prediction of convectively driven mesoscale
pressure systems. Part I: Convective parameterization. J. Atmos. Sci., 37,1722—1733.
Grell, G. A., Y.-H. Kuo, and R. Pasch, 1991: Semi-prognostic tests of cumulus parameterization
schemes in the middle latitudes. Mon. Wea. Rev., 119, 5—31.
3-14
-------
EPA/600/R-99/030
Grell, G. A., 1993: Prognostic evaluation of assumptions used by cumulus parameterizations.
Man. Wea. Rev., 121, 764-787.
Grell, G. A., J. Dudhia, and D. R. Stauffer, 1994: A description of the fifth-generation Penn
State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 117 pp.
[Available from the National Center for Atmospheric Research, P. O. Box 3000, Boulder, CO
80307.]
Guo, Y.-R., and S. Chen, 1994: Terrain and land use for the fifth-generation Penn State/NCAR
mesoscale modeling system (MM5): Program TERRAIN. NCAR Tech. Note, NCAR/TN-
397+1 A, 119 pp. [Available from the National Center for Atmospheric Research, P. O. Box
3000, Boulder, CO 80307.]
Haagenson, P., J. Dudhia, D. Stauffer, and G. Grell, 1994: The Penn State/NCAR mesoscale
model (MM5) source code documentation. NCAR Tech. Note, NCAR/TN-392+STR, 200 pp.
[Available from the National Center for Atmospheric Research, P. O. Box 3000, Boulder, CO
80307.]
Hack, J. J., B. A. Boville, B. P. Briegleb, J. T. Kiehl, P. J. Rasch, and D. L. Williamson, 1993:
Description of the NCAR community climate model (CCM2). NCAR Tech. Note, NCAR/TN-
382+STR, 120 pp. [Available from the National Center for Atmospheric Research, P. O. Box
3000, Boulder, CO 80307.]
Hong, S.-Y., and H.-L. Pan, 1996: Nonlocal boundary layer vertical diffusion in a medium-range
forecast model. Mon. Wea. Rev., 124, 2322-2339.
Hsie, E.-Y., and R. A. Anthes, 1984: Simulations of frontogenesis hi a moist atmosphere using
alternative parameterizations of condensation and precipitation. J. Atmos. Sci., 41, 2701—2716.
Janjic, Z. I., 1994: The step-mountain eta coordinate model: Further development of the
convection, viscous sublayer, and turbulent closure schemes. Mon. Wea. Rev., 122, 927-945.
Kain, J. S., and J. M. Fritsch, 1990: A one-dimensional entraining/detraining plume model.
J. Atmos. Sci, 47, 2784-2802.
Kain, J. S., and J. M. Fritsch, 1993: Convective parameterization for mesoscale models: The
Kain-Fritsch scheme. The representation of cumulus in mesoscale models. K. A. Emanuel and
D. J. Raymond, Eds., Amer. Meteor. Soc., 246 pp.
Klemp, J. B., and R. B. Wilhelmson, 1978: Simulations of three-dimensional convective storm
dynamics. J. Atmos. Sci., 35, 1070-1096
3-15
-------
EPA/600/R-99/030
Madala, R. V., 1981: Finite-difference techniques for vectorized fluid dynamics calculations.
Edited by D. L. Book, Springer Verlag, New York.
Manning, K., and P. Haagenson, 1992: Data ingest and objective analysis for the PSU/NCAR
modeling system: Programs DATAGRID and RAWINS. NCAR Tech. Note,
NCAR/TN-396+IA, 209 pp. [Available from the National Center for Atmospheric Research, P.
O. Box 3000, Boulder, CO 80307.]
Nuss, W. A., and.D. W. Titley, 1994: Use of multiquadric interpolation in meteorological
objective analysis. Mon. Wea. Rev., 122, 1611-1631.
Pleim, J. E., and A. Xiu, 1995: Development and testing of a surface flux and planetary boundary
layer model for application in mesoscale models. J. Appl. Meteor., 34,16-32.
Reisner, J., R. T. Bruintjes, and R. J. Rasmussen, 1993: Preliminary comparisons between MM5
NCAR/Penn State model generated icing forecasts and observations. Preprints, Fifth Intl. Conf.
on Aviation Weather Systems, Vienna, VA, Amer. Meteor. Soc., 65-69.
Reisner, J., R. J. Rasmussen, and R. T. Bruintjes, 1998: Explicit forecasting of supercooled liquid
water in winter storms using the MM5 mesoscale model. Quart. J. Roy. Meteor. Soc., 124B,
1071-1107.
Smolarkiewicz, P. K., and G. A. Grell, 1992: A class of monotone interpolation schemes.
J. Comp. Phys., 101,431^40.
Stauffer, D. R., and N. L. Seaman, 1990: Use of four-dimensional data assimilation in a limited-
area mesoscale model. Parti: Experiments with synoptic-scale data. Mon. Wea. Rev., 118,
1250-1277.
Stauffer, D. R., N. L. Seaman, and F. S. Binkowski, 1991: Use of four-dimensional data
assimilation in a limited-area mesoscale model. Part II: Effects of data assimilation within the
planetary boundary layer. Mon. Wea. Rev., 119, 734—754.
Stauffer, D. R., and N. L. Seaman, 1994: Multiscale four-dimensional data assimilation. J. Appl.
Meteor., 33, 416-434.
Tao, W.-K., and J. Simpson, 1993: Goddard cumulus ensemble model. Part I: Model
description. Terr., Atmos., and Oc. Sci., 4, 35-72.
Zhang, D.-L., and R. A. Anthes, 1982: A high-resolution model of the planetary boundary layer -
- Sensitivity tests and comparisons with SESAME-79 data. J. Appl. Meteor., 21, 1594-1609.
3-16
-------
EPA/600/R-99/030
Zhang, D.-L., H.-R. Chang, N. L. Seaman, T. T. Warner, and J. M. Fritsch, 1986: A two-way
interactive nesting procedure with variable terrain resolution. Mon. Wea. Rev., 114,1330-1339.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
MultiscaleAir Quality (CMA0 Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
3-17
-------
-------
EPA/600/R-99/030
Chapter 4
EMISSION SUBSYSTEM
William G. Benjey* and James M. Godowitch"
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Gerald L. Gipsoii
Human Exposure and Atmospheric Sciences Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Chapter 4 provides a description of the Models-3 Emission Processing and Projection System
(MEPPS) structure, scientific approach and the assumptions used in modeling and processing
emission data in the Models-3 framework. The description includes emission data entry through
the Inventory Data Analyzer (IDA) import and quality control checks; and the data flow and
quality control used in loading emission inventory and meteorology data in the MEPPS Input
Processor, The description of the main Emission Processor addresses the basis of spatial and
temporal allocation procedures. The scientific models and assumptions used in modeling hourly
mobile source and biogenic emissions are explained (Biogenic Emission Inventory System 2 and
Mobile 5a, respectively). The rationale and assumptions are described which are used in the
allocation and grouping of individual chemical species into "lumped species" in preparation for the
lumped species chemical transformation mechanisms contained in the Community Multiscale Air
Quality (CMAQ) model. The chapter also describes the procedures used by the Models-3
Emission Projection Processor to estimate emission data for use in modeling future air quality
scenarios. Finally, the quality control report and output file options contained in the Output
Processor are described.
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Bill Benjey, MD-80. Research Triangle Park, NC 27711. E-mail:
benjey@hpcc.epa.gov
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
4.0 EMISSION SUBSYSTEM
The rationale for the Models-3 air quality emission processor is rooted in the need to estimate,
organize, and process emission inventory data for regulatory and scientific analysis and modeling.
Historically, air quality emission processing methods have been developed in an ad hoc manner,
with each procedure specifically tuned to meet the need at hand, reflecting the various data
information sources and estimation methodologies, as well as the uses of the data. The Models-3
air quality emission processor consolidates and simplifies the estimation, data handling, and
linkage of the resulting data to air quality models. This section describes the purpose, origins, and
scientific bases of the Models-3 Emission Projection and Processing System (MEPPS). Material
is presented in the general sequence of emission processing. System requirements, system design
,and detailed user information for emission data processing are not addressed in this Volume.
These items are described in Volume 7, Volume 8, and Volume 9B, respectively of the Models-3
documentation set. For example, processor names, environment variables, and flow diagrams are
given in Chapter 6 of the Models-3 Volume 9B: User Manual.
4.1 Emission Inventory Processors
In the Models-3 framework, MEPPS imports, quality controls, and processes emission inventory
data for either direct regulatory analysis or input to a chemical transport model. The MEPPS also
models hourly biogenic emissions and mobile source emissions. The MEPPS is non-conforming
within the framework because it is not an object-oriented component, and because it does not use
the NetCDF Input/Output Applications Program Interface (I/O API) format internally.
Processing is accomplished using a combination of FORTRAN, SAS®, and ARC/INFO®
programs. The emission processor results are translated into the NetCDF I/O API format by the
MEPPS Output Processor.
4.1.1 Discussion
The MEPPS builds on lessons from and functionalities of previous software developed for
processing emission inventories. The capabilities and design decisions in MEPPS are placed in the
context of emission inventory developments.
4.1.1.1 The Role of Emission Inventory Processing for Chemical Transport Modeling
Historically, there have been many air quality emission data bases which were compiled and used
for the purposes of regulatory or scientific assessment of emissions, including spatial and temporal
patterns and trends. As air quality chemical transport models (CTMs) were developed, each had
its own specific format for input data, based either on the data structure used in the model and/or
the format of the available data. Initially, CTMs were relatively simple and applied to one or a
few point sources, such as emissions from an industrial stack. As CTMs for urban and regional
scales - such as the Urban Airshed Model (Morris et al.,1992) - became more sophisticated, the
level of detail needed for input emission data became increasingly more comprehensive and
4-2
-------
detailed. It became necessary to include as complete an inventory as possible for the area of
interest, including emissions from all diffuse as well as point sources at as great a degree of spatial
and temporal resolution as possible. As the spatial and temporal resolution of the models has
increased, so has the demand for detailed emission input data. Because of resource limitations on
data gathering, the emission inventory information needed often substantially exceed the reported
or observed emission data available. The result is development of a wide range of emission
estimation and modeling techniques.
Emission inventory data are available from various sources, often state and local air pollution
control agencies. The data are commonly compiled into annual emission inventories for specific
areas to be analyzed and/or modeled. The spatial extent of an emission inventory may vary from
plant-specific emission data to data for an entire country or more. Recent regional examples
include the National Acid Precipitation Assessment Program 1985 National inventory (Saeger et
al.,1989), the 1990 EPA Interim National Emission Inventory (U.S. EPA, 1993; U.S. EPA,
1994), and the 1990 Ozone Transport Assessment Group inventory (Ozone Transport
Assessment Group, 1997). Most emission inventories are organized into four traditional general
groups of emission types:
• Point sources, which are emission sources attributable to discrete emission points, usually
a stack. The data include pollutant, source category code, stack parameters (height,
diameter, exit velocity, temperature, flow rate), emissions, location coordinates, fuel, etc.
• Area sources, which are emission sources attributable to diffuse sources or areas, such as
agricultural fields, large open mining operations, forests, or a combination of many point
sources which are too small and numerous to account for individually (e.g., residences).
Area source inventories are typically by county and include pollutant, source category
code, emissions, location, coordinates, etc.
• Biogenic sources, which are often natural emissions from vegetation, soils, and lightning.
Biogenie emissions are dependent on temperature, solar radiation, and land cover type.
They are usually modeled hourly for specific days and locations.
• Mobile sources, which are emissions from vehicular traffic on roadways, aircraft, trains,
shipping, and off-road mobile equipment. Mobile source emissions are dependent on the
ambient temperature, road type, vehicle type and age, miles traveled, etc. These emissions
are generally modeled for specific day and meteorology scenarios either by county or road
segment (link-node data).
The techniques used to estimate emissions in compiling an emission inventory are based on
extrapolation of limited direct measurements for point sources, and application of limited
measurements or estimates to spatial surrogate data for area sources. Emissions which are
dependent on environmental conditions (e.g., biogenic and mobile source emissions) are modeled
to generate either portions of emission inventories or hourly data for direct use in air quality
4-3
-------
EPA/600/R-99/030
modeling. The procedures used to process emission inventory data and to model and process
mobile and biogenic source emission data are described in more detail in the following sections.
Point source and area source emission inventory data are usually included in annual emission
inventories. Annual mobile and biogenic emissions are included if the inventory was intended for
assessing annual totals or trends of emissions. If the inventory was prepared for modeling, only
the annual "Vehicle Miles Traveled" by county or road segment and vegetation land cover may be
provided. In order to accomplish episodic air quality modeling, it is necessary to model the hourly
mobile and biogenic emissions for the episode-specific meteorological conditions.
The typical procedures used to prepare annual emission inventories for use in a CTM require
temporal, spatial, and pollutant species allocation of the data. This is accomplished in a sequential
manner. For regional modeling, initial "raw" emission inventory data files are very large, often
several megabytes in size. The data files are subjected to a variety of data quality checks,
depending upon the methodological sophistication and computing and time resources available.
Typically, visual inspections of mapped locations, value range checks, cross-checks of sums, and
routine computer checks for blank fields and valid data types are performed.
In order to reduce the size of emission data files, the data are often speciated first, depending
upon the pollutants involved, then temporally allocated to hourly data, and then spatially allocated
or "gridded" to a spatial domain with gridded cells of the resolution required by the CTM. At each
step some of the detailed information, such as source category code and geographic coordinates,
are dropped to reduce the file size. Hourly mobile source and biogenic emission data are modeled
using the appropriate hourly meteorological data, and merged with point and area source data
prior to speciation.
It is necessary to rerun the processing sequence in the event of an error, or for each new
meteorological or day-specific scenario. Projections to future years require application of source-
category-specific economic growth factors to a base year inventory to produce a projected
annual inventory. The projected inventory is then processed through the entire emission
processing sequence. Preparation of detailed regional emission inventory data for regional
modeling using this traditional approach may take weeks or month.
4.1.1.2 Function and Place of the Emission Processor in the Models-3 Framework
The Models-3 framework is designed to contain conforming object-oriented modules that pass
data in the NetCDF I/O API format, although non-conforming modules are accommodated at the
cost of reduced functionality. The emission processing system is intended to ultimately be
conforming. The initial version of MEPPS was derived from existing software and consequently
is not a conforming object-oriented program in Models-3. However, it is integrated into the
Models-3 framework and does take instructions and information from the system and provides
emission data output in NetCDF IO/API format. Figure 4-1 provides a functional view of the
Models-3 system. Processing of emission data is often a part of iterative air quality strategy
4-4
-------
EPA/600/R-99/030
building, directly or indirectly, though the effect of emissions on air quality modeling results.
Figure 4-2 is a simplified illustration of the location of MEPPS and other non-conforming
components within the Models-3 framework. The main aspects of a study (including the principal
emission processing options) are graphically defined in the high-level Models-3 Study Planner,
and the resulting information is automatically passed to MEPPS as well as other portions of the
system. Specifically, a study name is defined by the user by copying, modifying, and renaming an
existing "template" emission study. For example, a tutorial emission study could be used as a
template to define a new study. The new study is associated with input data sources, MEPPS
processing and modeling modules and their primary options by annotating links and nodes in the
specified emission study and plans. The pollutants of interest, the geographic domain, map
projection, grid spatial resolution, chemical speciation mechanism, temporal resolution, and dates
of interest (referred to as the "case") are defined in the Models-3 Science Manager and passed to
the rest of the system including MEPPS, This approach allows the user to make primary
specifications once, rather than separately to each of the system components and thus ensures
consistency. Study Planner and Science Manager also aid in using the system to run multiple
emission, projected emission, and emission control scenarios without frequent respecification of
parameters. From a functional view, MEPPS generates emission data that reflect user-defined
studies and cases, whether for regulatory analysis of emission data and/or for input to CTM runs
and comparison with CTM results.
Main Control Subsystem
(Study Planner)
Control
Strategy
Building
Model
Building
Analysis
Functionality
1
[Data
[ Models
[
Tools
Building
Blocks
Figure 4-1 Functional View of the Models-3 System
4-5
-------
EPA/600/R-99/030
Meteorology
(MM5)
Data How
Principal Non-framework Components
\>
Models-3 Interfaces
Analysis Tools
System Instructions
Figure 4-2 The location of MEPPS within the Models-3 Framework
The emission processing function of MEPPS requires that it be linked to the Models-3
framework, and to other processing, modeling, and analysis tools through the framework.
MEPPS outputs data hi the NetCDF I/O API format in order to link MEPPS to the Models-3
framework and other processing, modeling, and analysis modules. Figure 4-3 illustrates that
information (grid, chemical mechanism, and case) user-specified in Models-3 Science Manager for
a study is passed to other modules, including MEPPS. However, it is necessary to access the
Parameter Window of MEPPS through the Tools Manager, to specific study name, grid, source
(data source name), and case, and then return to the Study Planner to run the emission study.
Direct access to MEPPS is through the Tools Manager and the emission projection and control
functions are in Strategy Manager. New emission inventory data are imported and quality
controlled through the Models-3 File Converter and its subsidiary Inventory Data Analyzer
(IDA). The data are then imported directly into MEPPS through its internal Input Processor
(INPRO). Meteorological information needed to estimate biogenic and mobile source emissions
are provided to MEPPS from a meteorological model or dataset in the Models-3 system through
4-6
-------
EPA/600/R-99/030
the Meteorology-Chemistry Interface Processor (MCIP). For regional-scale modeling, the
current default meteorological model is Mesoscale Model 5 (MM5). The MEPPS processes
emission data into speciated, spatially and temporally allocated hourly emission data for use in a
Models-3 conforming CTM, The data are output as NetCDF I/O API files to the Emission-
Chemistry Interface Processor (ECIP) which adjusts the format to that needed by the CTM.
Emission data for large elevated point sources (large stack emissions) are also provided as I/O
API files to the Plume Dynamics Module, and the plume rise algorithm.
Too
i
(
lc A^Tfl fid OrPT _f
la ivlU-llil^CI "^,
V >^
f File
(Converter*
s _,. f *
1
fScier
[Mana
V
(IDA*
f V *x^
MEPPS J
Study 3
\
3lanner
^
f MM5
ice
gftr
[MCIP
~»J F
QP
\
CMAQ
T
,„
4.1.1.3
"Where: MM5 is Mesoscale Model 5
CMAQ is Community Multiscale Air Quality Model
MEPPS is Models-3 Emission Processing and Projection System
MCIP is Meteorology-Chemistry Interface Processor
ECIP is Emissions-Chemistry Interface Processor
"IDA i s Inventory Data Analyzer
"Tile Converter is generic to the Models-3 system
*TDA and File Converter are optional (used for new data)
Figure 4-3 Relationship of Principal Models-3 Framework Components
Rationale for the Basis of the Initial Version of MEPPS
In order to assure that an acceptable and functional emission inventory processor would be
present in the initial release of Models-3 given time and resources available, it was decided that
the emission processing module for the initial version would be based on software and
methodologies available as of 1994, More advanced approaches being researched and developed
then and now will be incorporated as soon as possible. The decision factors used in selecting
4-7
-------
EPA/600/R-99/Q30
from existing emission processors were availability, cost, features (capabilities), hardware and
software requirements, and operating characteristics. Table 4-1 lists the emission processing
systems considered with respect to the principal evaluation factors.
Based on the evaluation, the Geocoded Emission Modeling and Projection System (GEMAP;
Wilkinson et al.,1994), now known as the Emission Modeling System-95 (EMS-95), was selected
as the basis of the initial emission processing system, although it is non-conforming (not in I/O
API Net CDF) in the Models-3 system. The GEMAP was selected because it was readily
available in the public domain; it was state-of-the-art at the time; there was no licensing cost
(exclusive of GEMAP's internal use of S AS® and Arc/Info®); its code was modular and flexible;
and it contained a geographic information system to perform spatial allocation of emission
inventory data.
Other systems considered included the Emission Processing System (EPS), The Flexible Regional
Emission Data System (FREDS), and the Sparse Matrix Operator Kernel Emission System
(SMOKE). The EPS is used to process emission inventory data for the Urban Airshed Model
(UAM; U.S. EPA, 1992). A recent adaptation of EPS by Environment Canada for regional use is
called the Canadian Emission Processing System Version 1 (CEPS1.0) (Moran et al., 1998). Like
GEMAP, EPS would be non-conforming code in the Models-3 framework, and would require
design and development of interfaces with the Models-3 system and the addition of many features.
However, EPS does not include flexible GIS-based gridding capability of GEMAP and was
originally designed for urban scale air quality modeling as opposed to the multiscale (local to
regional and national) air quality modeling capabilities of Models-3.
The FREDS has been used for regional air quality modeling during the past ten years in
conjunction with the Regional Oxidant Model and Regional Acid Deposition Model (Modica et
al., 1989). The FREDS code is run on a main-frame computer, and is "hard-wired" and difficult
to change for different scenarios or grids. Consequently it lacks the modularity and flexibility
needed to operate in the Model-3 framework.
Design and prototyping of SMOKE was just beginning when the design of the initial version of
Models-3 was determined (Coats, 1995). The SMOKE, which is now being used in conjunction
with UAM, could be modified to be fully compliant with the object-oriented Models-3 system.
Additional analysis and quality control tools are being designed for SMOKE, and initial work has
begun to adapt it to the Models-3 framework.
4-8
-------
EPA/600/R-99/030
Table 4-1. Emission Processor Selection Factors for the Initial Public Release of Models-3 .
f^CltoifS:?-'j;-;: '!';••.:•'••.•.• :":"*
'-'-'"•;.' - ' • . • ••' - "• • ''-..-'.'••'. • , ,
'.•'^ •.:'.: '•'• .:'•• '."':'- ••'<:. \ : : '', '•: • '! '••• - •-
Acquisition Cost of Source
Code1
Availability
Degree of Development
Relative Flexibility for
Modification to Models-3
Hardware Requirements
History (Operating
Experience)
Software Requirements
Spatial Allocation Capability
: ll^$siM;Pj?6'0e$siiig:. ^
^'Sy^m||ip^;'R:!.':1^^
'•..I .:•': '•'• ' -''•:-. '.: : : .'- '••'' ':• '
.."::;.$ i :!;;.. - -': A -.-. '.. : . .J:^: - '.; \,: . ,
None
Public domain
Completed and in use
Changes required receding of
modules to allow revised
interfaces and external control
- spatial allocation would be
added
Work station with UNIX
operating system
Established system used by
EPA, state, and local agencies '
FORTRAN and SAS*
Spatial allocation grids must
be manually coded for each
allocation scenario.
I? JexiSK 'Rfepttnal \ . :-f : " ; -
:]|iijfesioni:]^[didleMng- ,
^teiti:(ERE9S):; •;;:;-:-
None
Public domain
Completed and in use
Changes required substantial
receding - many patches -
relatively inflexible -spatial
allocation would be added
IBM Mainframe system
Established system used by
U.S. EPA in regional
modeling programs
FORTRAN and SAS*
Spatial allocation grids must
be manually coded for each
allocation scenario.
IGjBbib^d, "JiiBiission ''. ::^-
"Modelrag-aiMlJl ;•. ;••:';';• *
;:lp¥pjectioiii;-S^i5ftejnri: / . .--:
None
Public domain
Initial version being tested
Changes required receding of.
modules to allow revised
interfaces and external control
- spatial allocation already
present
Work station with UNIX
operation system
New system, used only with
test data in California
SAS* and ARC/INFO®
licenses.
ARC/INFO* allows flexible
user-defined grid domains,
spatial resolution and data
overlays
•4paiRp;Mitrix";V:"- r .': ]
v6]pwlte*:;ICerMJ::;' '-. t
' lmiss&(SBIO]KB;); , • -: ;
:^st(pa::::;;;y::i;lH:- :i\!; ;i
N/A
N/A
Design and prototype
beginning
N/A .
Unix-based machines,
specifics not established
N/A '
Unix-based, specifics .not well
established .-.
Unknown
4-9
-------
EPA/600/R-99/030
Since its selection, GEMAP/EMS-95 has been substantially modified and incorporated into
MEPPS. Because of the changes, the GEMAP portion of MEPPS has been renamed the
Emission Processor (EMPRO) module. Revisions were made to allow the software to be more
generic and efficient in its operation, and new features were added. In general, new capabilities
•were written in FORTRAN and placed in modules outside of GEMAP. This was done to
minimize rewriting of code in future when the GEMAP-based portion of the system is replaced.
MEPPS must include several basic functions. There are also many enhancements, detailed
descriptions for which may be found in Volumes 7, 8, and 9 (System Requirements, System
Design, and User Manual, respectively) of the Models-3 documentation set. The experience and
design suggestions of personnel familiar with processing large emission inventories were used as
important guidance in deciding which features to include in MEPPS. The design and
implementation of MEPPS emphasized ease of use and efficiency within the overall design of the
Models-3 system. The following paragraphs present a brief description of the principal functions
included in MEPPS.
• Annual point and area source emission data from inventories are subset to the spatial
domain of interest, spatially allocated to a grid, and temporally allocated. Area source
emissions are spatially allocated using surrogate spatial coverages.
• Spatial allocation of each general type of air quality data (point, etc.) to grid cells is done
by the GIS using a grid defined by the Science Manager module of Models-3.
• Modeled estimates of mobile source emissions are prepared on an hourly basis for periods
of interest, usually several days. These estimates account for meteorological conditions
using data from a meteorology model (such as MM5) that has been passed through the
Models-3 meteorology-chemistry interface processor (MCIP). Hourly, grid-cell specific
emission factors for different vehicle and roadway classifications are prepared with MM5
using vehicle miles traveled (VMT) data from an emission inventory.
• Modeled hourly estimates of biogenic source emissions are prepared on an hourly basis for
periods of interest, using the Biogenic Emission Inventory System Version 2 (BEIS-2) and
meteorology data from MCIP and land use coverages (see Section 4.2.4.1).
* Temporal allocation of emission data to hourly data for the period of interest is
accomplished using source-type specific defaults or user-selected temporal allocation
profiles.
» Disaggregation of gridded, temporally-allocated emission data of groups of chemical
species to specific species (chemical speciation) is completed according to the users choice
of chemical speciation mechanism (currently Carbon Bond 4 (CB-4) and RADM 2.0).
4-10
-------
EPA/600/R-99/030
• Projection of emission inventory data to future years from a base-year inventory (1990)
and application of controls is performed by the Models-3 Emission Projection Processor
(MEPRO). Projected emission inventories are used to iteratively evaluate different
emission scenarios caused by economic or emission control changes. Projected emission
inventories are each processed by MEPPS for air quality modeling in the same way as the
original, or "base" inventory.
• Merging of spatially and temporally-allocated, speciated files for point, area, mobile, and
bjogenic emission data into one emission data output file, and translation into NetCDF
I/O API format are performed for use in the Models-3 system, including the chemical-
transport model. Summary and quality control reports on the output data are also
produced.
• User-defined point-source emission data are extracted and prepared for use in vertically-
layered (three-dimensional) emission files to be used with the plume dynamics module of
the Models-3 system.
4.1.2 General MEPPS Structure
The MEPPS is normally used as an integral part of the Models-3 framework from high-level
menu-driven screens and pick-lists.
MEPPS in the Models-3 Framework
A user may define a study using the Study Planner within Models-3. The Study Planner specifies
the name and description of a study, and defines data sources, models, and the relationship
between them using a graphical interface. If a study includes processing of emission data, the
sequence of processing operations, source file addresses and many options for processing are
defined by the Study Planner and passed to MEPPS annotated to the study name. Specifications
for spatial allocation grids, and definition of specific temporal cases are defined in the Science
Manager. Existing studies, grids, work space directories, speciation mechanisms, source
directories, and computer hosts must be selected from the MEPPS parameter window located
through the MEPPS icon in the Tools Manager prior to running under the Study Planner. The
MEPPS may also be directly accessed through the Tools Manager and run interactively via its
SAS® interface windows. During direct access interactive operation, some features may be run
individually in the background using an "interactive" batch mode selection, or a series of
processes may be run together as a concatenated batch run.
The MEPPS shares an import File Converter with the rest of the Models-3 framework. The File
Convenor will import any ASCII, SAS®, or NetCDF I/O API file of format known to the user.
SAS® data sets are used within EMPRO. The data are imported, converted, and subjected to basic
quality control checks, including missing or out-of-range values. The data are then put through
the Inventory Data Analyzer (IDA) for quality control and analysis specific to emission files,
4-11
-------
EPA/600/R-99/030
including Inventories (in one of four EPA formats), temporal allocation factors, control factors,
vehicle miles traveled, and continuous emission monitoring (CEM) data. The data then go to the
MEPPS Input Processor (INPRO) for final quality control and loading into EMPRO.
The Internal Components of MEPPS
The emission processor (MEPPS) includes several basic components (Figure 4-4): an emission
data Input Processor (INPRO), a main Emission Processor (EMPRO), an Output Processor
(OUTPRO), and a Models-3 Emission Projection Processor (MEPRO). The File Converter and
IDA are not intrinsic parts of MEPPS. They are used to import, quality control, and convert the
formats and units of data files of known format to the formats that are used in the Models-3
system, including emission inventories and related emission data files. The INPRO imports
emission inventory data from IDA and meteorology files from MCIP, and prepares them for use in
EMPRO. The EMPRO spatially and temporally allocates point and area source emission data to
hourly gridded data. It is also used to model biogenic and mobile-source emission data using
meteorological data generated by MM5 and processed by MCIP, as well as spatially and
temporally allocates the data. The EMPRO then allocates (groups) the chemical species from the
gridded temporally allocated emission files according to the user's selection of a chemical
speciation mechanism. The speciated files then go to OUTPRO where they are merged (a two-
dimensional file). Quality control is performed and summary reports are prepared in accordance
with the user's choices. The user also may divide point sources into categories to separate very
large sources (major-elevated point sources or MEPSE) and large sources (major sources) into
separate output files, which may be placed in a three-dimensional emission output files to be
allocated to vertical layers for CMAQ. The user defines what is MEPSE or major using a
combination of pollutant-specific emission values and/or stack parameters (see Section 4.2.6).
Remaining point sources are merged into the two-dimensional emission output file along with
area, mobile, and biogenic source emission data. Output files are in NetCDF I/O API format. The
MEPRO module projects emission inventory data to future years while applying controls. The
projected emission data can be read by EMPRO for further processing if modeling of a projected
case is desired
4.2 The MEPPS Emission Processing System
Although MEPPS is a significant advance in emission inventory data processing, it is continually
being improved. The following sections describe the scientific rationale for the 1998 (first public
release) version of the system.
This section describes the basis for emission processing procedures in MEPPS. For models
developed independently of the Models-3 framework and incorporated into MEPPS,
documentation for those models is referenced. MEPPS was developed to meet emission
processing needs specified in Models-3 Volume 7: Design Requirements of the Models-3
documentation set, as amended. The design requirements for emission inventory processing are
based on the emission data input requirements of chemical transport models, and on the needs of
4-12
-------
MEPRO
(fsJcccinn
r
I Meteorology Data
1 Point Sources
1 Area Sources
I Mobile Source Date
( mmo 1
g*
apeQaie
I Visualize
f Models
f Grid
-L
>f RADM-2
K--Q -..I 1
CD— IV 1
J ^ •
H Generic
4 Pokt
*j Area
r , • -
r- -.
s* MoMe
EMPRO
1 Point Sources
J Area Sources
[Mobile Source Date
,„ T.l Prni
"1 Emiss'
-»f Ret
s»^ Ke]
en Data
3orts
— sH Gmphs
( MEPRQ 1
OUTPRO
Reports
1 EpisesdeFile
r
I SCO Reports 1
f
Stack Files
I Tempoealized Files I
I Speckted Files
f OUTPRO 1
m
Os
O
o
t
V£>
Figure 4-4 Functional Components of MEPPS
-------
EPA/600/R-99/030
those who will use MEPPS to process emission data for both regulatory and scientific emission
analysis as well as CMAQ data input. Two guiding principles were followed in the design and
implementation of MEPPS. First, substantive changes to the existing GEMAP system were
minimized because of a long-term goal to replace GEMAP with a more flexible, less costly (in
terms of software licenses), and faster processor. Changes focused on improving the processing
efficiency and correcting basic inadequacies of GEMAP. The second principle was to place new
features and new emission processing developments outside of GEMAP to the extent possible, but
within MEPPS. The resulting structure will make it easier to replace the GEMAP-based portion
of MEPPS in the future.
MEPPS is usually run from the Models-3 Study Planner, although it can be run independently in
interactive mode. Consequently, MEPPS is configured to run in a "batch" mode (from Study
Planner for multi-day cases), and to accept Study Planner specifications for emission data
processing and location of data sources. Some MEPPS processors that may be ran when
accessing MEPPS through the Tools Manager (such as file, grid, and case creation, deletion, or
editing), will not function when MEPPS is accessed through the Study Planner. This provision
avoids conflicts when these processes are specified from the Models-3 framework under Science
Manager. The main components of MEPPS are each discussed in the following sections.
4.2,1 The Inventory Data Analyzer (IDA)
The IDA is an important enhancement to Models-3. It operates as an adjunct to the generic File
Converter, which is described in Section 10.8 of the Models-3 User Manual (Volume 9b of the
Models-3 documentation set). The File Converter is fully-compliant component of Models-3,
applicable to the entire system, and provides file format and unit conversion for files imported into
or exported from Models-3, including emission related files. The formats currently used in
Models-3 and supported by the File Converter are NetCDF I/O API (gridded data), ASCII, and
SAS®.
The IDA imports files from the File Converter and provides quality control and analysis tools
specialized for reading, comparing, editing, and analyses of emission data . Quality control reports
are generated including the results of the checks. Emission inventories that are in terms of annual,
daily, or hourly emissions can be imported. The functionality of IDA is fully described in Section
6.2 of the User Manual. Specific quality control functions include the following:
« Range checks. Ranges accepted by IDA were taken from the range of values known to be
correct in inventories currently in the system (1985, 1990, 1995). English units are used
because most U. S, inventories are submitted with those units. Values outside of the
validity ranges are considered incorrect. The validity ranges currently used in IDA for
point-source stacks are:
4-14
-------
: EPA/60D/R-99/03D
Diameter >0.5 and < 50 ft.
Diameter < 0.2 * Height
Height >1.0and50 and <2,000°F
Velocity >1 and < 1000 ft/sec
Flow rate >1 and < 30,000,000 ftVsec and/or
Flow rate must not be more than 10 percent different from flow
rate calculated from Velocity and Diameter.
Sign checks. The system reports suspect signs based on the knowledge that particular
parameters are always of a given sign (e.g., stack height are positive).
Missing data checks. The system automatically fills missing data with Source Category
Code-specific default data generated from the inventory being processed.
Incorrect data checks. Data are corrected automatically when they can be computed
based on specific mathematical relations between data elements (such as exit velocity for
stack parameters).
Completeness cross-checks are performed between the plant, stack, process, and emission
hierarchy levels of point source inventories. Each level should be associated with data at
the next lower and higher level. For example process data should be subsidiary to
corresponding stack information, and superior to associated emission data. Specific
quality control checks are provided for each level of point, area, and mobile source
inventories.
Location checks, which are applied to determine that the specified political unit (e.g.,
county) is valid in each case, and that latitude and longitude coordinates (for point
sources) are not reversed or missing. Political unit identifiers are checked against a list
within IDA. When point-source latitude and longitude coordinates are missing, they are
currently assigned to the pseudo-center of the appropriate county. The missing values are
usually associated with smaller point sources. This assignment procedure can result in
over-representation at the county center. Work is planned to better distribute point
sources reported without geographic coordinates.
Some emission inventories do not include sulfate (SO4) emission data, which are important
in modeling of particulate matter. The IDA can approximate SO4 emission data by
multiplying the source-category specific ratios of SO4 to sulfur dioxide (SO2) taken from
the 1985 National Acid Precipitation Assessment Program (NAPAP) inventory, with SO2
values in another inventory. The NAPAP inventory is considered to contain the best
separate national modeling estimates of both SO2 and SO4 emissions at this time. If an
inventory contains SO4 data, this approximation feature is not used. The SO4 estimator is
optional in IDA. However, it is applied automatically in INPRO if no SO4 values are
4-15
-------
EPA/600/R-99/030
present. With this method, SO4 estimates were supplied for the 1990 and 1995 U.S.
inventories included with the initial Models-3 release.
• Occasionally an emission inventory contains an error in particulate matter emission values
such that the fine fraction (PM25) emissions exceed the value of the coarser fraction
emissions (PMIO) of which they are a part. In the absence of more specific data, or
information that the coarser particulate emission (PM10) are in error, IDA automatically
computes a new PM2 5 value using the 1990 national inventory ratio of PM25 to PMIO ,
which is expressed as PM25 = .2411 PMI0.
• Point source stack emission parameters are essential to plume rise calculations in
modeling. However, the stack parameters often have erroneous or missing values in
emission inventories. In addition to range checks, IDA addresses the problem in two
ways:
(1) IDA examines point source files for consistency between stack flow rate,
velocity, and diameter. The following relationship is used to correct erroneous
velocity and flow values:
F = V* 0.785398 * D2 (4-1)
where:
F is flow (cubic feet/second)
V is velocity (feet/second)
D is stack diameter (feet)
(2) In the event that all or most stack parameters are missing for a given point
source, IDA supplies default values which are computed by SCC from other point
sources in the same emission inventory.
4.2.2 The MEPPS Input Processor (INPRO)
The INPRO includes the MEPPS processors used to establish an emission study, the grid
directory structure, and time-specific cases. These items are usually defined using the Models-3
Science Manager, specified through the Study Planner in the Models-3 framework, and passed to
INPRO, unless MEPPS is being operated through the Tools Manager.
Import and Processing of Emission Inventories
The INPRO serves as the principal data access point for MEPPS. Two primary sets of data
manipulations are accomplished in INPRO to prepare the data for further study-specific
processing in MEPPS:
4-16
-------
EPA/600/R-99/030
• The emission inventory data are subsetted and imported for the grid specified in the main
MEPPS window and (if necessary) created in Science Manager. The emission data are
subset from the user-selected emission inventory for the kinds of emission sources (point,
area, mobile) specified by the user. The INPRO accepts inventories from IDA in the NET
or IDA inventory formats (see Section 10.8 of the User Manual). The user must be sure
that any aggregated (not hourly) data in the inventory for biogenic and mobile source
emissions (other than vehicle miles traveled) are deleted from the inventory before use.
Otherwise, MEPPS will double-count these emissions because hourly biogenic and mobile
source emissions are modeled during processing.
• The INPRO imports meteorological data specific to the spatial domain and temporal case
defined in Science Manager and selected in the MEPPS main window, provided that the
meteorological data have been previously generated. The data come from the Mesoscale
Meteorology Model 5 (MM5), via MCIP. The MCIP converts MM5 output to NetCDF
IO/API files containing information needed by the rest of the Models-3 system (see
Section 7.3 of this volume and Section 2.3.1 of the User Manual). In MEPPS,
meteorology data are needed to model hourly biogenic and mobile source emissions
(Sections 4.2.4.1 and 4.2.4.2). The MEPPS uses four of the files provided by MCIP,
named MET_CRO_2D_ xx, MET_CRO_3D_ xx, GRID_DOT_2D_ xx, and
MET_DOT_3D_ xx. The terms "CRO" and "DOT" refer to the cross and dot points of
grids, respectively, and the suffixes "xx" are study specific identifiers. Refer to Chapter 12
for the details of the grid system definitions.
ROG-to-TOG Adjustment
Emission inventories may reflect different assumptions or methodologies with respect to volatile
organic compound (VOC) gases. They may be reported as VOC, reactive organic gases (ROG)
or total organic gases (TOG). It is typical to require TOG for air quality modeling, whereas the
VOCs comprising organic gases in many regional inventories are reported as ROG. This occurs
because some emission measurement techniques do not capture all of the discrete hydrocarbon
compounds in the emission stream. Since the hydrocarbon speciation profiles are based on total
organic compounds, the measured value of the hydrocarbon must be adjusted to account for the
missing hydrocarbon components for the emission measurements that fail to capture the total
organic emission stream. This adjustment is included as an option when importing an inventory
using INPRO. The default is "yes" - perform the adjustment. More details are given in the
speciation discussion in Section 4.2.5.
4.2.3 Processing Procedure
A detailed description of the processing procedure is given in Chapter 6 of the Models-3 Volume
9B: User Manual. Briefly, EMPRO receives point- and area-source emission data from INPRO
for use in biogenic- and mobile-source emission modeling. The general sequence of processing
for all of the emission data is (1) spatial allocation, (2) temporal allocation, and (3) chemical
4-17
-------
EPA/60Q/R-99/030
speciation. There are some variations in processing between the general types (point, area,
biogenic, mobile) of emission data because of the spatial and temporal resolution available and
method of estimation. The gridded, temporally allocated and speciated data are passed to
OUTPRO where they are merged into output files in the NetCDF I/O API format.
4.2.3.1 Spatial Allocation of Emission Data
The emission inventory data are usually provided by discrete point sources or by political units
(typically counties in the United States). In order to prepare the data for use in air quality models
and for analysis by grid-oriented visualization tools such as PAVE, the data must be spatially
allocated (gridded) on a map projection and rectangular grid. The map coordinate system, grid
position and spatial resolution are defined from windows in the Models-3 framework Science
Manager (Sections 7.1 and 7.2 of the Models-3 Volume 9B: User Manual), although the actual
grid creation is performed by the ARC/Info program. The MEPPS is capable of using Lambert
Conformal, Mercator, Albers, and Universal Transverse Mercator (UTM) map projections; and
several datums including perfect sphere, NAD83, and Clarke. However, in order to maintain
consistency with gridded MM5 meteorology data, a spherical datum must be used. The radius
used for Models-3 applications is 6370.997 kilometers.
Aside from the spherical datum restriction, the gridding processor allows development of a variety
of emission grid systems which are used in different aspects of air quality modeling studies,
including emission processing and in CMAQ. The gridding processor is written in the ARC®
Macro Language (AML®) of the Arc/Info® geographic information system (GIS). Grids are used
in MEPPS to spatially disaggregate or aggregate emission estimates, using land use/land cover
data, and area source spatial surrogate data. Surrogate data are needed to spatially distribute
emission data because the exact location of emission sources totaled for a county, for example,
are seldom known. For example, census tract population data are of higher spatial resolution than
county boundaries, and therefore population distribution could be assumed to be directly
proportional to the spatial distribution of dry cleaning establishments. Land use cover data,
detailed road network maps, and many other kinds of spatial data may be used as spatial
surrogates to locate different source types of emissions, if GIS coverages of these data are
available. At this time MEPPS is provided with GIS coverages for political boundaries and land
cover for North America, and detailed population data and road networks for the United States.
The primary input to the gridding processor consists of user-supplied data that are entered into
windows or chosen from a pick-list in the Coordinate System and Grid windows of the Models-3
Science Manager. These data define the grid projection system, the origin of the grid, cell size,
etc., which is then generated by Arc/Info® through MEPPS. The primary output of the gridding
processor for MEPPS is an ARC® coverage which consists of the intersection of the political
boundaries and the fixed grid.
Use of the Arc/Info® GIS as the key component in the gridding processor allows substantial
flexibility in the definition and use of grids and surrogate data. It also allows MEPPS to
4-18
-------
EPA/600/R-99/030
incorporate visualization methods for quality control and for managing and querying of the spatial
emission data.
To generate the emission modeling grid structure, a political boundaries (state and county
boundaries) Arc/Info® coverage named "counties" must already exist in the exact map projection
that is intended for the emission modeling grid structure. Accordingly the counties coverage
provided with MEPPS is in the Lambert Conformal projection. If inconsistent projections are
used, the gridding processor produces unpredictable results. The processor produces three files
that are used elsewhere in EMPRO:
• A running history of emission modeling grid structure changes.
• A list of the geographic location of each grid cell in the emissions modeling grid structure.
• The Arc/Info® coverage of the emission modeling grid structure.
The gridding processor supports the following subset of Arc/Info® map projections:
• UTM (Universal Transverse Mercator)
• Lambert (Lambert Conformal Conic)
• Albers (Albers Conic Equal Area)
* Geographic (actual latitude and longitude coordinates rather than a projection)
* State (State Plane Coordinate System)
For more information about the Arc/Info18 map projections, refer to the following:
• ARC/INFO® User's Guide: Map Projections & Coordinate Management, Concepts and
Procedures.
• ARC/INFO® Command References: ARC® Command References, Commands J-Z,
The UTM map projection is a specialized application of the Transverse Mercator projection. It is
limited to use between the latitudes of 84° North and 80° South, and divides the earth into 60
longitudinally-defined zones (UTM zones 1-60) with each zone spanning 6 °of longitude. UTM
zone 1 starts at 180° West longitude and ends at 174° West longitude. UTM zone 2 starts at
174° West longitude and ends at 168° West longitude, and so on around the globe. Shape, area,
direction, and distance errors are all minimized if a study area is within a single zone; however,
error increases rapidly as study areas cross UTM zone boundaries.
The Lambert map projection is a projection of the earth onto a cone intersecting the earth along
two parallels called standard parallels (for example 30° North and 60° North). The projection is
good for large scale (continental or smaller), middle latitude (between 45 ° North and 45 ° South)
study areas. Shapes are maintained on a small scale (state or smaller), and large shapes
(countries, continents) are minimally distorted. Area and distance are maintained near the
standard parallels, but area is reduced between the standard parallels and increased beyond the
standard parallels; that, is, the study area is accurately represented between the standard parallels.
4-19
-------
EPA/600/R-99/030
Direction is accurate within the entire study area. CMAQ is currently tested and used with a
Lambert conformal projection.
The Albers map projection is suitable for study domains where the east-west extent is greater than
the north-south extent. Shape is minimally distorted near the standard parallels as long as the
study domain's east-west extent is greater than the study domain's north-south extent. Area is
accurate within the study domain. Direction is accurate only at the standard parallels and
minimally distorted between the standard parallels. If the study area is in the middle latitudes (45°
North to 45° South), distances are minimally distorted at and between the standard parallels. If
the study area falls outside of the middle latitudes, distance error increases rapidly as the poles are
approached.
The Geographic system is not a true map projection; instead it is a global reference system. It is
supported as a map projection because geographic positioning (latitude and longitude) is the most
widely used method of map location. The origin of the geographic global reference system is at
0° longitude (Meridian of Greenwich) and 0° latitude (equator). Directions in the northeast
quadrant are measured in positive longitude degrees by positive latitude degrees. Directions in
the northwest quadrant are measured in negative longitude degrees by positive latitude degrees.
Directions in the southwest quadrant are measured in negative longitude degrees by negative
latitude degrees. Directions in the southeast quadrant are measured in positive longitude degrees
by negative latitude degrees.
The State system is also not a true map projection; instead it is a specialized coordinate system for
the United States, Puerto Rico, and the United States Virgin Islands. The State system divides
the aforementioned areas into a total of 120 zones of varying sizes. The map projection is
inherent in each zone. The three map projections that are used in the State system are the
Lambert Conic Conformal (for zones with where the east-west extent is greater than the north-
south extent), Transverse Mercator (for zones where the north-south extent is greater than the
east-west extent), and Oblique Mercator (used only for the Alaska Panhandle). Direction,
distance, area, and shape errors vary with the map projection inherent in each zone; however, the
zones have been designed so that error is reduced or eliminated as long as the study area falls
entirely within a state plan zone.
The spatial allocation process grid description file ($EMS_GRD/grd_desc.in) references
numerous data items that are used by the Grid Definition Model to prepare the emissions
modeling grid structure. Table 4-2 lists the run description file data items that affect processing in
the Grid Definition Model.
4-20
-------
Table 4-2. Data Items That Affect the Gridding Processor
EPA/600/R-99/030
Variable Name
gridid
griddesc
celMoc
utmzn
utmxorig
utmyorig
cellsizx
cellsizy
xcells
ycells
zcells
projectn
Menu
Identifier
Grid Id:
Description:
Cell Location:
(S,E)
UTM zone:
(S,E)
Origin X
direction: (S,E)
Origin Y
direction: (S,E)
Cell size X
direction: (S,E)
Cell size Y
direction: (S,E)
Number of
cells X
direction: (S,E)
Number of
cells Y
direction (S,E)
Number of
cells Z
direction: (S)
Name: (S,E)
Type
C
C
C
N
N
N
.N
N
N
N
N
C
Example
grid a
Tutorial 36km
Resolution
Grid
SOUTHWEST
10
-108000
-1080000
36000
36000
21
21
10
LAMBERT
.Description
directory name and grid identifier
descriptive text for the emissions
modeling grid
location on/in the grid cell from
which to make measurements
(NORTH, NORTHEAST, EAST,
SOUTHEAST, SOUTH,
SOUTHWEST, WEST,
NORTHWEST, CENTER)
UTM zones for which the
emissions modeling grid exists
(1-60)
southwest x comer origin
(measured from the origin of the
projection) of the emissions
modeling domain (meters)
southwest y comer origin
(measured from the origin of the
projection) of the emissions
modeling domain (meters)
x grid cell size (meters)
y grid cell size (meters)
number of grid cells in the x
direction
number of grid cells in the y
direction
number of grid cells in the z
direction (currently not used by
ARC/INFO)
projection name (LAMBERT,
UTM, ALBERS, STATE,
GEOGRAPHIC)
4-21
-------
EPA/600/R-99/030
Table 4-2. Data Items That Affect the Gridding Processor
Variable Name
projunit
xshift
yshift
projzone
fipszone
datum
p_lsp_dd
p_lsp_mm
p_lsp_ss
Menn
Identifier
Units: (S,E)
X-shift: (E)
Y-shift: (E)
UTM zone:
(S,E)
PIPS zone for
state plane
projection: (E)
Datum
conversion
name for
projection:
(S3)
1st standard
parallel DD:
(S3)
1 st standard
parallel MM:
(S3)
1st standard
parallel SS:
(S3)
Type
C
N
N
N
N
C
N
N
N
Example
METERS
0.0
0.0
16
3701
MM5
(Perfect
sphere)
30
0
0.00
Description
units used to make measurements
in the selected projection
(METERS, FEET, DD [decimal
degrees])
constant value to add to x input
coordinates (value in projunit —
typically 0.0)
constant value to add to y input
coordinates (value in projunit -
typically 0.0)
State Plane or UTM zone (used
only when projectn is STATE or
UTM -- see ARC™ Command
References, Commands J-Z;
Table PROJECT-3 and Table
PROJECT-4 for valid values)
FIPS code for State Plane zone
(used only when projectn is
STATE -- see ARC™ Command
References, Commands J-Z;
Table PROJECT-5 for valid
values)
datum upon which input
coordinates are based (used only
when projectn is STATE -
NAD27, NAD83, UNKNOWN)
degree of first standard parallel
(only for projectn ALBERS or
LAMBERT)
minutes of first standard parallel
(only for projectn ALBERS or
LAMBERT)
seconds of first standard parallel
(only for projectn ALBERS or
LAMBERT)
4-22
-------
Table 4-2. Data Items That Affect the Gridding Processor
EPA/600/R-99/G30
Variable Name
p_2sp_dd
p_2sp_mm
p_2sp_ss
p_cen_dd
p_cen_mm
p_cen_ss
p_lpo_dd
p_Ipo_mm
p_lpo_ss
p_f_east
p_fhorth
Menu
Identifier
2nd standard
parallel DD:
(S3)
2nd standard
parallel MM:
(S3)
2nd standard
parallel SS:
(S3)
Central
meridian DD:
(S3)
Central
meridian MM:
(S3)
Central
meridian SS:
(S3)
Latitude of
projection
origin DD:
(S3)
Latitude of
projection
origin MM:
(S3)
Latitude of
projection
origin SS:
(S3)
False easting:
(E)
False northing:
(E)
Type
N
N
N
N
N
N
N
N
N
N
N
Example
60
0
0.00
-90
0
0.00
40
0
0.00
0.0
0.0
Description
degrees of second standard
parallel (only for projectn
ALBERS or LAMBERT)
minutes of second standard
parallel (only forprojectn
ALBERS or LAMBERT)
seconds of second standard
parallel (only forprojectn
ALBERS or LAMBERT)
degrees of central meridian of
projection (only forprojectn
ALBERS or LAMBERT)
minutes of central meridian of
projection (only forprojectn
ALBERS or LAMBERT)
seconds of central meridian of
projection (only forprojectn
ALBERS or LAMBERT)
degrees of latitude from
projection origin (only for
projectn ALBERS or LAMBERT)
minutes of latitude from
projection origin (only for
projectn ALBERS or LAMBERT)
seconds of latitude from
projection origin (only for
projectn ALBERS or LAMBERT)
false easting (only forprojectn
ALBERS or LAMBERT -
meters)
false northing (only forprojectn
ALBERS or LAMBERT --
meters)
4-23
-------
EPA/600/R-99/030
Variable Name refers to the name as it is found in the SAS® data set run_desc.ssd01 which can be
found in the directory referenced by the UNIX environment variable EMS_RUN (see Models-3
Volume 9B: User Manual, Chapter 6). Menu Identifier refers to the descriptive text found on
the grid definition window in Science Manager (S) or the EMPRO Grid Processor window (E)
within MEPPS. Type refers to the variable type ~ N is a numeric variable, and C is a character
variable. Example gives an example of what might be entered for each data item. Description
provides descriptive text about the data item, supported units, and a list of valid values that each
data item recognizes — valid values and units are listed in parentheses.
4.2.3.2 Temporal Allocation of Emission Data
Emission data that are based on annual, seasonal, weekly, or daily values must be temporally
allocated (usually disaggregated) to hourly data for compatibility with the time scale of episodic
air quality modeling. Generally, this procedure applies to regional inventories of point- and area-
source emission data. Modeled emission data, such as biogenic- and mobile-source emissions are
generated as hourly data for the time period of interest. Allocation of emission data from time
periods greater than hourly (e.g., annual total, monthly total, weekly total), to hourly data is
accomplished by use of seasonal, monthly, weekly, and daily diurnal temporal allocation factors to
translate the data to daily total emission values. The daily values are then transformed into
emission values for each hour of a typical day by using user-supplied or default temporal
allocation profiles. The profiles assign proportions of the total daily emissions to each of the 24
hours of a typical day. Profiles are ideally source or source-category specific, but often are used
for a range of similar source categories because of limited data on source category temporal
variability. The majority of diurnal temporal source profiles have been developed for point
sources, since more detailed reporting and monitoring exist for point sources, particularly large
point sources, than for area sources which are spatially diverse and variable. Consequently are
area source temporal allocation profiles tend to be less detailed and specific that point source
temporal allocation profiles. Most of the temporal profiles used as defaults in MEPPS, were
developed for the National Acid Rain Precipitation Assessment Program (NAPAP) (Fratt et al.,
1990) and used in FREDS, with some more recent supplements (Moody et al., 1995).
In the EMPRO module of MEPPS, temporal allocation of point and area source category
emission data is performed after spatial allocation of data in each main processor (point, area,
biogenic, and mobile) and prior to speciation (see Chapter 6 of the Models--3 User Manual), This
is transparent when running from Study Planner. If running in MEPPS under Tools Manager, the
user may enter source category or source-specific daily or diurnal hourly temporal allocation data,
or elect to use profiles computed by EMPRO (procedure described in the following sections) or
use temporal allocation profiles based on those developed for use in the FREDS processors as
supplemented. If a combination of emission inventory data at different temporal resolutions is
used, EMPRO uses the most time-resolved (specific) data first, filling in with less time-specific
data and applying temporal profiles as needed. If hourly, daily, and annual emission data are
selected, EMPRO will provide hourly emission values taken from available data in the following
order:
4-24
-------
EPA/600/R-99/030
• Loading of hourly diurnal emission inventory (including Continuous Emission Monitoring
(CEM) data);
• Loading and gridding of day-specific (daily) emission inventory data;
• Computing hourly emission data from annual or other longer-term emission inventory
data.
Further details concerning temporal allocation are discussed as part of the following descriptions
of the basis for processing each basic type of emission source.
4.2.3.3 Point Source Emission Data Processing
The point-source emission processor prepares gridded, temporally allocated point-source
emission estimates suitable for speciation and reformatting for input to Models-3 framework for
air quality modeling (e.g., CMAQ) (see Chapter 6 of the Models-3 User Manual). As previously
indicated, the point-source emission processor does not compute basic emission estimates using
emission factors and source-specific information. Instead, it reduces annual, point-source
emission estimates to hourly, point source emissions estimates, unless the user is able to provide
hourly emission data.
The point-source emission processor begins by establishing study-specific foundation emission
estimates taken by INPRO from an user-specified emission inventory database. Foundation
estimates (referred to as Foundation Files in MEPPS) are the basic annual emission inventory files
imported to a processor (point, area, or mobile source) after quality control corrections have been
applied and the data have been reduced to cover the study area only. After a point-source
foundation file has been created, the processor spatially allocates the emission data into the spatial
allocation grid structure, temporally allocates the emission estimates, and updates the hourly
emission estimates derived from the foundation annual inventory with day-specific hourly
emission estimates or hourly CEM data (if available). Usually, the point source-specific emission
data are processed to files corresponding to the major hierarchical point-source data elements
(i.e., Facility, Stack, Device, and Process), and translated to SAS® data sets.
Hourly emission estimates are computed using data from one of the following:
• Gridded emission estimates derived from annual emission inventory data by applying
factors from source category-specific temporal allocation profiles
• Day and source-specific hourly emission data provided by the user
• Source-specific hourly CEM data.
Typically, the primary inputs to the point-source processor are the foundation emission annual
estimates which are supplemented by available day-specific emission estimates. The primary
output of the processor is the spatially and temporally allocated emission estimates.
Computation of Point Source Emission Estimates
4-25
-------
EPA/600/R-99/030
The point-source emission processor treats emission estimates that have been prepared on one of
the following bases: annual average; average day; or day-specific.
Day-Specific Emissions Estimates
Day-specific emission estimates are not computed by the MEPPS EMPRO point source
processor. The user must supply any available day-specific emission estimates to the processor.
Because day-specific emission estimates are more representative of actual conditions, they replace
the hourly emission estimates derived from disaggregated annual emission data, when available.
Day-specific emission data formats are made consistent with the internal formats of the SAS® data
sets derived from other (e.g., annual) point-source emission data when imported through IDA.
CEM Hourly Emission Data
The CEM data are a subset of hourly emission data derived from continuous air pollutant
concentration monitors attached to components of specific facilities, usually boilers or stacks of
large point sources such as electric utilities. The data elements included are hourly emission rates
for Carbon Monoxide (CO), nitrogen oxide, (NOx), and sulfur dioxide (SO2). The CEM emission
data are treated as hourly data imported to MEPPS via IDA. Electric utility CEM data are
identified by specific source using their ORIS (Office of Regulatory Information Systems)
identification number and are mapped to the corresponding point source identification number
from an emission inventory. The CEM data are read in for a specified grid domain and day from a
CEM dataset and substituted for hourly emission data derived from annual data. The CEM
emission data are used in the format of data available for electric utility emissions provided by the
U.S. EPA Acid Rain Division. The Acid Rain Division has electric utility CEM data for the
United States for 1995 and 1996.
Spatial Allocation
Point sources are spatially allocated to an emission modeling grid by the geographic (latitude and
longitude, UTM position) coordinates of a stack or by the geographic coordinates of the facility.
The point-source emission data processor prepares an ASCII file of point-source identifiers and
point-source geographic coordinate locations. The processor reads title ASCII file that was
generated by the location processor, generates the appropriate ARC/INFO® coverages, and
prepares two ASCII files:
* A file which contains point source identifiers and grid cell location; and
» A file which contains point-source identifiers and latitude/longitude coordinates. The
process of assigning grid coordinates to point sources is an ARC/INFO® function;
therefore, the underlying procedure follows the user documentation and proprietary code
for the ARC/INFO® software.
Temporal Allocation
4-26
-------
EPA/600/R-99/030
The EMPRO temporally allocates (produces hour-by-hour) foundation file emission estimates
within each main emission processor (point, area, biogenic, and mobile emissions) based on
operating schedule data that are provided in one of two ways. The operating schedule data may
be passed into the MEPPS EMPRO module via an ASCII foundation file, which corresponds with
the segment (process) level of point source inventory information. For point sources, the
temporal factors may be assigned by the user at the segment (process) level of the EMPRO point
source hierarchy. The hierarchy is closely aligned with that of regional emission inventories.
Temporal allocation factors may be also be applied at the device level in the facility, stack, device
hierarchy. Alternatively, default source-category-specific temporal allocation factors may be
selected. These factors were used in a MEPPS predecessor, FREDS, and are referred to as
FREDS temporal allocation factors. The point source hierarchy for temporal allocation factors hi
FREDS is: plant, point, temporal factors. Source-specific temporal allocation factors are
typically not available for large regional modeling domains. Consequently, the default temporal
allocation factors are commonly used in regional modeling, perhaps more than 90 percent of the
time. Note that EMPRO temporally allocates foundation file emission estimates, but not day-
specific emissions estimates since the latter are entered on an hour-by-hour basis.
EMPRO provides a variety of methods for identifying operating schedule data:
• Seasonal throughput fractions (winter [Dec, Jan, Feb], spring [Mar, Apr, May], summer
[Jun, Jul, Aug], and fall [Sep, Oct, Nov])
• Hours per year hi operation
• Days per year in operation;
• Weeks per year in operation;
• Days per week in operation; and
• Hours per day in operation.
Any, none, or all of the operating schedule data can be supplied. If no operating schedule data are
supplied, EMPRO uses a default continuous operating schedule of 24 hours per day, 7 days per
week, and 52 weeks per year; otherwise, EMPRO filters through a hierarchy of the operating
schedule data to determine how to compute the temporal factors:
• Weekly temporal factor; .
• Daily temporal factor; and
• 24 hourly temporal factors.
To determine the weekly temporal factor, EMPRO determines the number of days in the year of
interest and divides that value by 7 days per week. For example, in a leap year, there are 29 days
in February; therefore, the weekly temporal factor is 366 days/year •*• 1 days/week).
EMPRO computes a daily temporal factor through a lookup table based on the value of average
8-hour work days (days per week in operation) which is passed into MEPPS through the ASCII
foundation files. To determine the daily temporal factor, EMPRO examines the values for days,
4-27
-------
EPA/600/R-99/030
hours per year (houryear), days per year (dayyear), and weeks respectively. If days has a valid
value, EMPRO takes no further action to determine the days per week in o peration for the
source. If days does not have a valid value, EMPRO attempts to assign an operation code value
to days by examining houryear, dayyear, and weeks respectively:
• houryear (hours per year in operation)
if houryear > 0 and houryear <= 850 then days = 2,
if houryear > 850 and houryear <= 1250 then days = 3,
if houryear > 1250 and houryear <= 1670 then days = 4,
if houryear > 1670 and houryear <= 2100 then days = 5,
if houryear > 2100 and houryear <= 2500 then days = 6,
if houryear > 2500 then days = 7;
• dayyear (days per year in operation)
if dayyear > 0 and dayyear <= 110 then days = 2,
if dayyear > 110 and dayyear <= 160 then days = 3,
if dayyear > 160 and dayyear <= 210 then days = 4,
if dayyear > 210 and dayyear <= 260 then days = 5,
if dayyear > 260 and dayyear <= 315 then days = 6,
if dayyear > 315 then days = 7;
• weeks (weeks per year in operation)
if weeks > 0 and weeks <= 7 then days = 1,
if weeks > 7 and weeks <= 13 then days = 2,
if weeks > 13 and weeks <= 19 then days = 3,
if weeks > 19 and weeks <= 26 then days = 4,
if weeks > 26 and weeks <= 33 then days = 5,
if weeks > 33 and weeks <= 39 then days = 6,
if weeks > 39 then days = 7.
EMPRO computes an hourly temporal factor through a lookup table based on the value of hours
(hours per day in operation) which is passed into EMPRO through the ASCII foundation files. If
hours has a valid value, the value is used and no further action is taken to determine the hours per
day in operation for the source. If hours does not have a valid value, EMPRO attempts to assign
a value to hours by examining houryear, dayyear, and weeks respectively, using the above tables.
If houryear, dayyear, or weeks has a valid value, then hours is assigned a value of 8, again based
on an average work day of 8 hours. If hours or days cannot be assigned through the method
described above, operation is assumed to be continuous and hours is assigned a value of 24, and
days is assigned a value of 7. The following calculations are performed.
4-28
-------
EPA/600/R-99/030
week_fac = daysinyear / 7 (4-2)
day_sumd = £ dy_actf, i = 1,... 7 (4-3)
day_faCj = dy_actf / day_sumd, i = 15... 7 (4-4)
hour_sumh = £ hr_aet], i = 1, ...24 (4-5)
hrprofj = hr_acr? / hour_sumh, i = 1,... 24 (4-6)
where week_fac is the weekly temporal factor
daysinyear is the number of days in the year (365 or 366)
day_sum is the total relative activity for the week
dy_act is an array that contains relative daily activities
day_fac is the day of week temporal factor
hour_sum is the total relative activity for the day
hrprof is an array of hourly temporal factors
hr_act is an array that contains relative hourly activities
d is an index indicating which days per week in operation code (days) to use
h is an index indicating which hours per day in operation code (days) to use
i is the integer count that applies (e.g., 7 days or 24 hours)
EMPRO uses the temporal factors to allocate emission estimates to hourly values. Application of
the temporal factors depends on the temporal basis (annual average, day-specific, average day)
that emissions estimates were input to EMPRO.
« if the temporal basis code estt = "AA" (annual average) then
hremiSj = aceekg/( week_fac* day_fac * hrprofj), i — 1,... 24 (4-7)
where hremis is the array of gridded, hourly emissions estimates (NOX, TOG, CO,
etc) (mass)
aceekg is the emission estimates (mass)
weekjfac, dayjfae, and hrprof are the weekly, daily and hourly temporal
factors, respectively, computed by EMPRO
• if the temporal basis code estt = "AD" (average day) and (week_fac - 0 or day_fac
= 0)then
hremiSj = 0.0, i = 1,... 24 (4-8)
• if the temporal basis code estt = "AD" (average day) and not (week_fac = 0 or
day_fac = 0) then
hremiSj = aceekg * hrprofj, i = 1,... 24 . (4-9)
• if the temporal basis code estt = "DS" (day-specific) then
4-29
-------
EPA/600/R-99/030
hremis; = aceekg * profj, i = 1,... 24 (4-10)
where prof is an array of day-specific, hourly temporal factors
4.2.3.4 Area Source Emission Data Processing
The area-source emission data processor prepares area-source emission estimates for speciation
(lumped-model or discrete speciation) and reformatting for input to CMAQ or other Models-3
framework components. The area-source processor does not compute emission estimates from
fundamental emission data. It reduces annual, county-wide emission estimates to emissions on an
hourly, grid cell-by-grid cell basis. Most of the area source emission data processor is written in
the SAS® programming language. The area-source spatial surrogates allocation component of the
processor is written in the ARC® Macro Language (AML®). The primary inputs to the processor
are the area-source emission estimates for the spatial domain of interest, arid day-specific emission
estimates (if available). The primary outputs are the spatially allocated, temporally resolved,
emission estimates.
The processor begins with an area-source emission data base, extracted by INPRO from an
emission inventory as an ASCII file and automatically converted to a SAS<5> file. The area-source
processor updates the estimates with any available day-specific estimates, spatially allocates the
emissions into the emission modeling grid, and temporally allocates the criteria emissions.
Emission Data Processing
This section discusses how area-source emission estimates are read into, manipulated by, and
quality-assured by the area-source emission data processor. References are made to MEPPS
source code where necessary to provide further details concerning area source emission
allocation.
The processor uses county-wide area-source emission estimates, that have been prepared
externally as part of an emission inventory, to produce spatially allocated, temporally resolved,
pollutant emission estimates. To the extent that the inventory used includes small point sources in
area source estimates, they are treated as area sources. All data provided as point sources,
regardless of magnitude, are processed as point sources in EMPRO. The area-source emission
data processor accommodates area-source emission estimates that have been prepared on the
following bases: annual average; average day; or day-specific.
Day-specific emission data enter the area-source emission data processor during the temporal
allocation step. Day-specific emission data are not computed by EMPRO. Like the foundation
file emissions estimates, day-specific emission estimates are supplied from an external source to
the emissions modeler. The day-specific emissions estimates replace the area-source emission
estimates derived from annual inventory data because day-specific emission data are more
representative of actual conditions.
4-30
-------
EPA/600/R-99/030
Day-specific area-source emission estimates are summed to a daily total so that they are
consistent with the internal formats of the SAS® data sets that maintain the other (e.g., derived
from annual emission inventories) area-source emissions estimates, which are passed to EMS-95
on an hourly basis.
Spatial Allocation
Area-source emission estimates are allocated to a modeling domain through the application of
spatial surrogates. In general, surrogates approximate the value of unknown quantities. For
example, population can be used to estimate the number of gasoline service stations. A spatial
surrogate not only helps estimate the value of an unknown quantity, but a spatial surrogate also
helps locate the unknown quantity.
For each spatial surrogate, it is necessary to specify what data (categories of land use/land cover,
population counts, housing counts, etc.) contribute to the surrogate. In EMPRO, county area,
land cover, population (census), Federal Highway Administration major roadway, and
TIGER/LINE roadway data for the United States are provided as ARC® coverages for use as
surrogates. Others may be added by providing ARC® coverages generated external to MEPPS.
County area, and less detailed population and land cover coverages are provided for Canada and
Mexico. Each area source category must be assigned a unique spatial surrogate value by grid cell.
After the appropriate files have been populated, EMPRO grids the necessary data sets according
to the requirements of the user-defined spatial surrogates. Each area source category (asct)
which has been assigned to a spatial surrogate (k) can be allocated to grid cells (1, m) through the
application of Equation 4-11.
aceeij>k, l>m = aceeij>k, * ratioyw>m (4-11)
where i is the index on states
j is the index on counties within the states
k is the spatial surrogate index
k' is the area source category index (directly related to k)
1 is the x cell index
m is the y cell index
acee is the county-wide area source emissions estimate (mass)
ratio is the gridded surrogate ratio by state/county/surrogate/cell
Each surrogate is computed through the application of Equations 4-12 and 4-13.
surtotij>k = S1Smattribute,w>m (4-12)
ratiOij,kj.m = attribute jikAm/surtotjj>k (4-13)
where: i is the index on states
j is the index on counties within the states
4-31
-------
EPA/6QO/R-99/030
k is the spatial surrogate index
1 is the x cell index
m is the y cell index
surtot is the total value of a surrogate within specified (indexed) states and
counties
attribute is the gridded value of the surrogate attribute by
state/county/surrogate/cell
ratio is the gridded surrogate ratio for specified (indexed) states/counties/
surrogates/cells
In the case of census data or other area-based or length-based surrogate information, it is
necessary to area-apportion or length-apportion the surrogate (such as population or housing)
information prior to aggregation to the cell level. This is because the locale of interest crosses cell
boundaries. The assumption is that the surrogate information (such as population or housing) has
a constant density across the locale (in the case of population or housing data, a constant density
across the census tract). Therefore, in some cases, it is necessary to apply Equations 4-14
through 4-16 prior to the application of Equations 4-12 and 4-13.
arlgj<%k = S,Sm arealengjJiW>m (4-14)
ap_ratiOj jifc>lim = arealengy>kil>m / arlgjDt, jjc (4-15)
attributes J|1(in = attribute, jjk * ap_ratiojjjk>ljm (4-16)
where i is the index on states
j is the index on counties within the states
k is the locale of interest index (such as census tract or roadway)
1 is the x cell index
m is the y cell index
arlg__tot is the total area or length of a locale of interest within specified (indexed)
states and counties
arealeng is the gridded value of the specified (indexed) locale of interest by state/
county/surrogate/cell
apjratio is the gridded ratio of the locale of interest specified (by indices) states/
counties/locales/cell
attribute is the surrogate data (population, road length, etc.) for the area specified
by indices.
As a check for Equations 4-14 through 4-16, the following must be true:
SjSjSkratio^'M.O . (4-17)
Temporal Allocation
4-32
-------
EPA/600/R-99/030
Temporal allocation of area source emission data is performed in an identical manner as described
for point source processing in Section 4.2.3.3, although it is almost always necessary to use
default temporal allocation profiles. Please refer to that section for a discussion of temporal
allocation of area sources.
4.2.4 Modeled Emission Data
Emission data from available emission inventories are spatially and temporally allocated as
described in the preceding paragraphs. Many anthropogenic area and point sources of emissions
vary little or predictably with time, and can reasonably be disaggregated for regional modeling.
However, some emission data are highly variable, diurnally and .seasonally, because they are
dependent on environmental variables, such as temperature, humidity, and solar insolation. It is
more accurate to model these kinds of emissions on an hourly basis for direct use in episodic air
quality modeling. The two principal kinds of sources for which hourly emission data are normally
modeled are mobile sources and biogenic sources. In both cases, MEPPS takes hourly
meteorological data from MCIP output derived from MM5.
4.2.4.1 Biogenic Emissions
Hourly biogenic emission rates for biogenic VOC compounds (including isoprenes,
monoterpenes, and soil NO) for each grid cell are estimated using the Biogenic Emission
Inventory System, Version 2 (BEIS-2) within MEPPS. In order to estimate biogenic emissions
for modeling, it is necessary to apply biogenic emission and biomass factors to a geographic
distribution of land cover. The BEIS-2 was developed to fill this need separately from Models-3
as an improvement on BEIS-1 (Pierce et al., 1990; Geron et al., 1994), and has been used in
conjunction with different air quality modeling systems, including the Regional Acid Deposition
Model (RADM) (Pierce et al., in press), the Regional Oxidant Model (ROM), and now Models-3
CMAQ. Additional information concerning BEIS-2 can be found in the Emission Inventory
Improvement Program report on Biogenic Sources Preferred Methods (EIIP, 1996) and Pierce et
al. (1998). The BEIS-2 model applies emission flux factors specific to tree genera and
agricultural crop types by geographic area for biogenic emission species in accordance with
equation 4-18.
ER^LjtVEF'F^T)] (4-18)
where: ER is the emission rate (in grams/sec/model cell)
i is the chemical species (e.g., isoprene, monoterpene)
j is the vegetation type
A is the vegetation area (meter2) in a grid cell
EF is the emission factor (micrograms/gram of leaf biomass/hour), and
Fy(S,T) is an environment factor to account for solar radiation S and leaf temperature T
4-33
-------
EPA/600/R-99/030
Vegetation emission flux factors were adapted from those compiled by Geron et al. (1994) for 77
tree genera. Emission flux factors for 16 agricultural crops were taken primarily from Lamb et al
(1993), and the emission factors applied to 34 land cover types are from the work of Guenther et
al. (1994), Emission flux factors by land cover type are used principally for Canada and the
western United States where genus-level forest cover and agricultural crop data were not
available. Biogenic emission flux factors for summer and winter conditions are given in Table 4-3
and Table 4-4, respectively. The vegetation classes are taken from the Biogenic Emission Land
Use Database (BELD) (Kinnee et al., 1997), which in turn is drawn from other land cover data
sets, with an emphasis on forest and agricultural land cover. Spatial resolution of the raw
(original) land cover data is at the county-level for the United States and sub-province level for
Canada. Emission flux factors are based on foil leaf summer conditions normalized to leaf or soil
temperatures of 30°C and photosynthetically active radiation (PAR) of 1000 micromoles/nrVsec.
For use in regional modeling, the mass of biogenic emissions are converted to moles by dividing
the mass (grams-compound) by the molecular weight of the compound. The emission factors for
tree genus and agricultural vegetation, and used in the Models-3 application of BEIS-2 are
compiled by Pierce et al, (1998).
Table 4-3. Summer Biogenic Emission Flux Factors (^g-compound m"2hr"!) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Abie
Acac
Acer
Aeso
Aila
Aleu
Alfa
Ainu
Amel
Asim
Avic
Barl
Isoprene
170.0
79.3
42.5
42.5
42.5
42.5
19.0
42.5
42.5
42.5
42.5
7.6
Terpene
5100.0
2380.0
680.0
42.5
42.5
42.5
7.6
42.5
42.5
42.5
42.5
19.0 .
Other
.. VOCs'
2775.0
1295.0
693.7
693.7
693.7
693.7
11.4
693.7
693.7
693.7
693.7
11.4
NO
4.5
4.5
4.5
4.5
4.5
4.5
12.8
4.5
4.5
4.5
4.5
256.7
Leaf Area
Index (m2
m-2)
7
5
5
5
5
5
0
5
5
5
5
0
Description
Abies (fir)
Acacia
Acer (maple)
Aesculus (buckeye)
Ailanthus
Aleurites (tung-oil tree)
Alfalfa
Alnus (European alder)
Amelanchier (serviceberry)
Asimina (pawpaw)
Avicennia (black mangrove)
Barley
4-34
-------
EPA/600/R-99/03Q
Table 4-3. Summer Biogenic Emission Flux Factors (Xg-eompound nrtir"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
land Cover
Code
Barr
Beta
Borf
Bume
Carp
Gary
Casp
Cast
Casu
Cata
Cedr
Celt
Cere
Cham
Citr
Cnif
Conf
Cora
Com
Coti
Cott
Crat
Cswt
Desh
Isoprene
0.0
42.5
910.0
42.5
42.5
42.5
42.5
42.5
29750.0
42.5
79.3
42.5
42.5
170.0
42.5
745.4
1550.0
0.5
42.5
42.5
7.6
42.5
1050.0
65.0
Terpene
0.0
85.0
713.0
42.5
680.0
680.0
42.5
42.5
42.5
42.5
1269.3
85.0
42.5
340.0
680.0
1366.6
1564.0
0.0
680.0
42.5
19.0
42.5
660.0
94.5
Other
VOCs
0.0
693.7
755.0
693.7
693.7
693.7
693.7
693.7
693.7
693.7
1295.0
693.7
693.7
2775.0
693.7
993.9
1036.0
0.0
693.7
693.7
11.4
693.7
770.0
56.7
NO
0.0
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
577.6
4.5
4.5
256.7
4.5
0.2
57.8
Leaf-Area
Index (m*
. m*)
0
5
5
5
5
5
5
5
7
5
7
5
5
'7
5
9
6
.0
5
5
0
5
2
0
Description
Barren
Betula (bkch)
Boreal forest (Guenther*)
Bumelia (gum bumelia)
Carpinus (hombean)
Carya (hickory)
Castanopsis (chinkapin)
Castanea (chestnut)
Casuarina (Austl pine)
Catalpa
Cedrus (Deodar cedar)
Celtis (hackberry)
Cercis (redbud)
Chamaecyparis (prt-orford
cedar)
Citrus (orange)
BEIS conifer forest
Conifer forest (Guenther)
Corn
Cornus (dogwood)
Cotinus (smoke tree)
Cotton
Crataegus (hawthorn)
Herbaceous Wetlands
(Guenther)
Desert shrub (Guenther)
4-35
-------
EPA/600/R-99/Q30
Table 4-3. Summer Biogenic Emission Flux Factors (//g-compound m"2hr"') for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Dios
Euca
Fagu
Frax
Gled
Gord
Gras
Gymn
Hale
'Harf
Hay
Ilex
Jugl
Juni
Lagu
Lari
Liqu
Liri
Macl
Magn
Malu
Meli
Mixf
Moru
Isoprene
42.5
29750.0
42.5
42.5
42.5
42.5
56.2
42.5
42.5
8730.0
37.8
42.5
42.5
79.3
42.5
42.5
29750.0
42.5
42.5
42.5
42.5
42.5
11450.0
42.5
Terpene
42.5
1275.0
255.0
42.5
42.5
42.5
140.5
42.5
42.5
436.0
94.5
85.0
1275.0
476.0
42.5
42.5
1275.0
85.0
42.5
1275.0
42.5
42.5
1134.0
85.0
Other
VOCs
693.7
693.7
693.7
693.7
693.7
693.7
84.3
693.7
693.7
882.0
56.7
693.7
693.7
1295.0
693.7
693.7
693.7
693.7
693.7
693.7
693.7
693.7
1 140.0
693.7
NO
4.5
4.5
• 4.5
4.5
4.5
4.5
57.8
4.5
4.5
4.5
12.8
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
Leaf Area ••
Index (m2
itr2)
5
5
5
5
5
5
0
5
5
5
0
5
5
7
5
5
5
5
5
5
5
5
5
5
Description
Diospyros (persimmon)
Eucalyptus
Fagus (american beech)
Fraxinus (ash)
Gleditsia (honey locust)
Gordonia (loblolly-bay)
Grass
Gymnocladus (KY
coffeetree)
Halesia (silverbell)
Hardwood forest (Guenther)
Hay
Ilex (holly)
Juglans (black walnut)
Juniperus (east, red cedar)
Laguncularia (white
mangrove)
Larix (larch)
Liquidambar (sweetgum)
Liriodendron (yellow poplar)
Maclura (osage-orange)
Magnolia
Malus (apple)
Melia (chinaberry)
Mixed forest (Guenther)
Morus (mulberry)
4-36
-------
EPA/600/R-99/030
Table 4-3. Summer Biogenic Emission Flux Factors (//g-compound m"2hr"') for Principal
Biogenic Compounds by Vegetation Category ,
Vegetation or
Land Cover
Code
Mscp
Nmxf
Nyss
Oak
Oats
Odcd
Ofor
Oksv
Ostr
Othe
Oxyd
Pacp
Past
Paul
Pean
Pers
Pice
Pinu
Plan
Plat
Popu
Pota
Pros
Prun
Isoprene
7.6
10150.0
5950.0
3108.3
7.6
2112.4
56.2
7350.0
42.5
56.2
42.5
55.0
56.2
42.5
102.0
42.5
23800.0
79.3
42.5
14875.0
29750.0
9.6
42.5
42.5
Terpene
19.0
1100.0
255.0
255.5
19.0
368.8
140.5
100.0
42.5
140,5
255.0
79.8
140.5
42.5
255.0
255.0
5100.0
2380.0
42.5
42.5
42.5
24.0
42.5
42.5
Other
VOCs
11.4 :.
850.0
693.7
894.2
11.4
871.8
84.3
600.0
693.7
84.3
693.7
' 47.9
84.3
693.7
153.0
693.7
2775.0
1295.0
693.7
693.7
693.7
14.4
693.7
693.7
NO
12.8
4.5
4.5
4.5
256.7
4.5
4.5
4.5
4.5
'57.8
4.5
35.3
57.8
4.5
12.8
4.5
4.5
4,5
4.5
4.S
4.5
192.5
4,5
4.5
Leaf Area
Index (m2
m-2)
0
5
5
6
0
6
0
2
5
0
5
0
0
5
0
5
7
3'
5
5
5
0
5
5
^Description
Misc crops
Northern Mixed Forest
(Guenther)
Nyssa (blackgum)
BEIS oak forest
Oats
BEIS other deciduous forest
Open forest
Oak Savannah (Guenther)
Ostrya (hophornbeara)
Other (unknown, assume
grass)
Oxydendrum (sourwood)
Pasture crop land (Guenther)
Pasture
Paulownia
Peanuts
Persea (redbay)
Picea (spruce)
Pinus (pine)
Planera (water elm)
Platanus (sycamore)
Populus (aspen)
Potato
Prosopis (mesquite)
Prutius (cherry)
4-37
-------
EPA/600/R-99/030
Table 4-3. Summer Biogem'c Emission Flux Factors (/^g-compound rr^hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Pseu
Quer
Rang
Rhiz
Rice
Robi
Rye
Sabl
Sali
Sapi
Sass
Scru
Scwd
Sere
Shrf
Smxf
Snow
Sor
Sorg
Soyb
Spin
Swie
Taxo
Thuj
Isoprene
170.0
29750.0
37.8
42.5
102.0
5950.0
7.6
5950.0
14875.0
42,5
42.5
37.8
2700.0
14875.0
10750.0
17000.0
0.0
42.5
7.8
22.0
1460.0
42.5
42.5
170.0
Terpene
2720.0
85.0
94.5
42.5
255.0
85.0
19.0
42.5
42.5
42.5
42.5
94.5
349.0
42.5
530.0
1500.0
0.0
42.5
19.5
0.0
1983.0
42.5
1275.0
1020.0
Other
VOCs
2775.0
693.7
56.7
693.7
153.0
693.7
11.4
693.7
693.7
693.7
693.7
56.7
651.0
693.7
910.0
1250.0
0.0
693.7
11.7
0.0
1252.0
693.7
693.7
2775.0
NO
4.5
4.5
57.8
4.5
0.2
4.5
12.8
4.5
4.5
4.5
4.5
57.8
31.2
4.5
4.5
4.5
0.0
4.5
577.6
12.8
4.5
4.5
4.5
4.5
Leaf Area
Index (m2
nr2)
7
5
0
5
0
5
0
5
5
5
5
0
2
5
5
4
0
5
0
0
3
5
5
7
Description
Pseudotsuga (douglas fir)
Quercus (oak)
Range
Rhizophora (red mangrove)
Rice
Robinia (black locust)
Rye
Sabal (cabbage palmetto)
Salix (willow)
Sapium (chinese tallow tree)
Sassafras
Scrub
Scrub woodland (Guenther)
Serenoa (saw palmetto)
Southeast/Western
Deciduous Forest
Southeast Mixed Forest
Snow
Sorbus (mountain ash)
Sorghum
Soybean
Southern pine (Guenther)
Swietenia (W. Indies
mahogany)
Taxodium (cypress)
Thuja (W. red cedar)
4-38
-------
EPAy600/R-99/030
Table 4-3. Summer Biogenic Emission Flux Factors (jug-compound m^hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Till
Toba
Tsug
Tund
Ufor
Ugra
Ulmu
Uoth
Urba
Utre
Vacc
Wash
Wate
Wcnf
Wdcp
Wetf
Whea
Wmxf
Wwdl
Isoprene
42.5
0.0
79.3
2411.7
1988.7
56.2
42.5
0.0
408.6
5140.0
42.5
5950.0
0.0
4270.0
2550.0
3820.0
15.0
5720.0
525.0
Terpene
42.5
58,8
158.7
120.6
663.7
140.5
42,5
0.0
161.9
1000.0
42.5
42.5
0,0
1120.0
663.0
923,0
6.0
620.0
250.0
Other
VOCs
693.7
235.2
1295.0
150.7
920.0
84.3
693.7
0.0
200.5
959.0
693.7
693.7
0.0
1320.0
2053.0
1232.0
9.0
530.0
360.0
NO
4,5
256,7
4,5
0.2
4.5
57.8
4.5
0.0
12.5
4.5
4.5
4.5
0.0
4,5
8.7
0.2
192.5
4.5
4.5
Leaf Area
Index (m4
m-2)
5 •
0
7
0
0
0
5
0
0
5
5
5
0
5
3
5
0
4
3
Description
Tilia (basswood)
Tobacco
Tsuga (Eastern hemlock)
Tundra
BEIS urban forest
BEIS urban grass
Ulmus (American elm)
BEIS other urban (barren)
BEIS urban (.2 grass/,2
forest)
Urban trees (.5 Harf/.S Conf)
Vaccinium (blueberry)
Washingtonia (fan palm)
Water
W Coniferous Forest
(Ouenther)
Woodland/crop land
(Guenther)
Wetland forest (Guenther)
Wheat
Western Mixed Forest
(Guenther)
Western Woodlands
(Guenther)
*Guenther references biogenic emission factors taken from Guenther et al. (1994) using AVHRR (Advanced Very
High Resolution Radiometer) satellite imagery.
4-39
-------
EPA/60Q/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors (^g-eompound m~2hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Abie
Acac
Acer
Aesc
Aila
Aleu
Alfa
Ainu
Amel
Asim
Avic
Bar!
Barr
Betu
Borf
Bume
Carp
Gary
Casp
Cast
Isoprene
170.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
42.5
0.0
0.0
0.0
640.0
42.5
0.0
0.0
0.0
0.0
Terpene
5100.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
42.5
0.0
0.0
0.0
706.0
42.5
0.0
0.0
0.0
0.0
Other VOCs
2775.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
693.7
0.0
0.0
0.0
634.0
693.7
0,0
0.0
0.0
0.0
NO
4.5
4,5
4.5
4.5
4.5
4.5
12,8
4.5
4.5
4.5
4,5
256.7
0.0
4.5
4.5
4.5
4.5
4.5
4.5
4.5
Leaf Area
Index (nt2
m-2)
7
5
5
5
5
5
0
5
5
5
5
0
0
5
6
5
5
5
5
5
Description
Abies (fir)
Acacia
Acer (maple)
Aesculus (buckeye)
A i Ian thus
Aleurites (tung-oil
tree)
Alfalfa
Alnus (European
alder)
Amelanchier
(serviceberry)
Asiminia (pawpaw)
Avicennia (black
mangrove)
Barley
Barren
Betula (birch)
Boreal forest
(AVHRR/G.*)
Bumelia (gum
bumelia)
Carpinus (hornbean)
Carya (hickory)
Castanopsis
(chinkapin)
Castanea (chestnut)
4-40
-------
EPA/600/R-99/030
Table 4-4. Winter Bidgenic Emission Flux Factors Oug-compound m^hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Casu
Cata
Cedr
Celt
Cere
Cham
Citr
Cnif
Conf
Corn
Coru
Coti
Cott
Crat
Cswt
Desh
Dios
Euca
Fagu
Frax
Isoprene
29750.0
0.0
79.3
0.0
0.0
170.0
42.5
0.0
1400.0
0.0
0.0
0.0
0.0
0.0
1050.0
0.0
0.0
29750.0
0.0
0.0
Terpene
42.5
0.0
1269.3
0.0
0.0
340.0
680.0
1353.0
1548.0
0.0
0.0
0.0
0.0
0.0
660.0
0.0
0.0
1275.0
0.0
0.0
Other VOCs
693.7
0.0
1295.0
0.0
0.0
2775.0 '
693.7
835.0
870.0
0.0
0.0
0.0
0.0
0.0
770.0
0,0
0.0
693.7
0.0
0.0
NO
4.5
4.5
4.5
4.5
4.5
4.5 '
4.5
- '4.5
4.5
' ' 577.6
4.5
4.5
256.7
4.5
0.2 ' •
57.8
4.5
4.5
4.5
' '4.5 ' '
Leaf Area
Index (m2
m-2)
7
5
7
5
5
7
5
9
6
0
5
5
0
5
1
0
5
5
5
5 "
Description
Casuarina (Austl
pine)
Catalpa
Cedras (Deodar
cedar)
Celtis (hackberry)
Cercis (redbud)
Chamaecyparis
(prt-orford cedar)
Citrus (orange)
BEIS conifer forest
Conifer forest
(AVHRR, G. )
Com
Comus (dogwood)
Cotinus (smoke tree)
Cotton
Crataegus (hawthorn)
Herbaceous Wetlands
(AVHRR, G.)
Desert shrub
(AVHRR, G.)
Diospyros
(persimmon)
Eucalyptus
Fagus (american
beech)
Fraxinus (ash)
4-41
-------
EPA/600/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors (tig-compound m^hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Gled
Gord
Gras
Gymn
Hale
Harf
Hay
Ilex
Jugl
Juni
Lagu
Lari
Liqu
Liri
Macl
Magn
Malu
Meli
Isoprene
0.0
0.0
0.0
0.0
0.0
0.0
0.0
42.5
0.0
79.3
42.5
0.0
0.0
0.0
0.0
42.5
0.0
0.0
Terpene
0.0
0.0
0.0
0.0
0.0
371.0
0.0
85.0
0.0
476.0
42.5
0.0
0.0
0.0
0.0
1275.0
0.0
0.0
CJther VOCs
0.0
0.0
0.0
0.0
0.0
185.0
0.0
693.7
0.0
1295.0
693.7
0.0
0.0
0.0
0.0
693.7
0.0
0.0
NO
4.5
4.5
57.8
4.5
4.5
4.5
12.8
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
4.5
Leaf Area
Index (m*
i m-2)
5
5
0
5
5
3
0
5
5
7
5
5
5
5
5
5
5
5
Description
Gleditsia
(honeylocust)
Gordonia
(loblolly-bay)
Grass
Gymnocladus (KY
coffeetree)
Halesia (silverbell)
Hardwood forest
(AVHRR, G.)
Hay
Ilex (holly)
Juglans (black
walnut)
Juniperus (east, red
cedar)
Laguncularia (white
mangrove)
Larix (larch)
Liquidambar
(sweetgum)
Liriodendron (yellow
poplar)
Maclura
(osage-orange)
Magnolia
Malus (apple)
Melia (chinaberry)
4-42
-------
EPA/600/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors dug-compound m"2hr'') for Principal
Biogenic Compounds by Vegetation Category
• Vegetation or
Land. Cover
Code
Mixf
Moru
Mscp
Nmxf
Nyss
Oak
Oats
Ocdf
Ofor
Oksv
Ostr
Othe
Oxyd
Pacp
Past
Paul
Pean
Pers
Pice
Pinu
Isoprene
0.0
0.0
0.0 .
175.0
0.0
0.0
0.0
0,0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
42.5
23800.0
79.3
Terpene
1077.0
0.0
0.0 .
1 100,0
0.0
217.0
0.0
313.0
0.0
100.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
255.0
5100.0
2380.0
Other, VOCs
581.0
0.0
0.0
850.0
0.0
188.0
0.0
183.0
0.0
200.0
0.0
0.0
0.0
0.0
0.0
0.0
0.0
693.7
2775.0
1295.0
NO
4.5
4.5
. 12.8
4.5
. 4.5
. 4.5
236.7
4.5
4.5
4.5
4.5
57.8
4.5
35.3 . .
57.8
4.5
12.8
4.5
4.5
4.5
Leaf Area
Index (m2
m"2) ;
4
- 5
0
1
5
6
0
6
0
1
5
0
5
. 0
0
5
0
5
7
3
Description.
Mixed forest
(AVHKR, G.)
Morus (mulberry)
Misc crops
Northern Mixed
Forest (AVHKR, G.)
Nyssa (blackgum)
BEIS oak forest
Oats
BEIS other deciduous
forest
Open forest
Oak Savannah
(AVHRR, Guen)
Ostrya
(hophornbeam)
Other (unknown,
assume grass)
Oxydendrum
(sourwood)
Pasture cropland
(AVHRR, G.)
Pasture
Paulownia
Peanuts
Persea (redbay)
Picea (spruce)
Pinus (pine)
4-43
-------
EPA/600/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors (^g-eompound m~2hr~l) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Plan
Plat
Popu
Pota
Pros
Prun
Pseu
Quer
Rang
Rhiz
Rice
Robi
Rye
Sabl
Sali
Sapi
Sass
Scru
Scwd
Sere
Shrf
Isoprene
0.0
0.0
0.0
0.0
0.0
0.0
170.0
0.0
0.0
42.5
0.0
0.0
0.0
5950.0
0.0
0.0
0.0
0.0
0.0
14875.0
0.0
Terpene
0.0
0.0
0.0
0.0
0.0
0.0
2720.0
0.0
0.0
42.5
0.0
0.0
0.0
42.5
0.0
0.0
0,0
0.0
332.0
42.5
0.0
Other VOCs
0.0
0.0
0.0
0.0
0.0
0.0
2775.0
0.0
0.0
693.7
0.0
0.0
0.0
693.7
0,0
0.0
0.0
0.0
332.0
693.7
0.0
NO
4.5
4.5
4.5 '
92.15
4.5
4.5
4.5
4.5
57.8
4.5
0.2
4.5
12.8
4.5
4.5
4.5
4.5
57.8
31,2
4.5
4.5
Leaf Area
Index (m*
tn-2)
5
5
5
0
5
5
7
5
0
5
0
5
0
5.
5'
5
5
0
2
5
0
Description
Planera (water elm)
Platanus (sycamore)
Populus (aspen)
Potato
Prosopis (mesquite)
Prunus (cherry)
Pseudotsuga (douglas
fir)
Quercus (oak)
Range
Rhizophora (red
mangrove)
Rice
Robinia (black locust)
Rye
Sabal (cabbage
palmetto)
Salix (willow)
Sapium (chinese
tallow tree)
Sassafras
Scrub
Scrub woodland
(AVHRR, Q.)
Serenoa (saw
palmetto)
SE/W Deciduous
Forest (AVHRR, G.)
4-44
-------
EPA/600/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors (^g-compound m^hr"1) for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Smxf
Snow
Sorb
Sorg
Soyb
Spin
Swie
Taxo
Thuj
Tili
Toba
Tsug
Tund
Ufor
Ugra
Ulmu
Uoth
Urba
Utre
Isoprene
0.0
0.0
0.0
0.0
0.0
0.0
42.5
42.5
170.0
0.0
0.0
79.3
0.0
0.0
0.0
0.0
0.0
0.0
700.0
Terpene
1500.0
0.0
0.0
0.0
0.0
1963.0
42.5
1275.0
1020.0
0.0
0.0
158.7
0.0
631.0
0.0
0.0
0,0
' 154.0
960.0
Other VOCs
500.0
0.0
0.0
0.0
0.0
1052.0
693.7
693.7
2775.0
0.0
0.0
1295.0
0.0
469.0
0.0
0.0
0.0
102.0
528.0
NO
..'•••!
4.5
0.0
4.5
577.6
12.8
4,5
4.5
4.5
4.5
4.5
256.7
4,5
0.2
4.5
57.8
4.5
0.0
12.5
4.5
Leaf Area
Index (m2
«*>
2
0
5
0
0
3
5
5
7
5
0
7
0
0
0
5
0
0
6
Description
SE Mixed Forest
(AVHRR, G.)
Snow
Sorbus (mountain
ash)
Sorghum
Soybean
Southern pine
(AVHRR, G.)
Swietenia (W. Indies
mahogany)
Taxodium (cypress)
Thuja (W. red cedar)
Tilia (basswood)
Tobacco
Tsuga (Eastern
hemlock)
Tundra
BEIS urban forest
BEIS urban grass
Ulmus (American
elm)
BEIS other urban
(barren)
BEIS urban (.2
grass/.2 forest)
Urban tree (.5 Harf/.5
Conf)
4-45
-------
EPA/600/R-99/030
Table 4-4. Winter Biogenic Emission Flux Factors (/ig-compound m"2hr"') for Principal
Biogenic Compounds by Vegetation Category
Vegetation or
Land Cover
Code
Vacc
Wash
Wate
Wcnf
Wdcp
Wetf
Whea
Wmxf
Wwdl
Isoprene
0.0
5950.0
0.0
3500.0
0.0
0.0
0.0
0.0
0.0
Terpene
0.0
42.5
0.0
1120.0
630.0
877.0
0.0
620.0
250.0
Other VOCs
0.0
693.7
0.0
1200.0
1047.0
628.0
0.0
330.0
360.0
NO
4.5
4.5
0.0
4.5
8.7
0.2
192.5
4.5
4.5
Leaf Area
Index (m?
rn-*)
5
5
0
5
2
3
0
3
3
Description
Vaccinium
(blueberry)
Washingtonia (fan
palm)
Water
Western Coniferous
Forest (A VHRR, G. )
Woodland/cropland
(A VHRR, G.)
Wetland forest
(A VHRR, G.)
Wheat
Western Mixed Forest
(A VHRR, G.)
Western Woodlands
(A VHRR, G.)
* AVHRR/G. references biogenic emission factors from Guenther et al (1994) which used land use classifications
from Advanced Very High Resolution Radiometer (AVHRR) satellite imagery.
The BEIS-2 applies environmental correction factors to account for the effect of leaf temperature
and visible solar radiation on isoprene (Pierce et at, 1998). Specifically:
= 1,* CL* CT
(4-19)
where: I is the adjusted isoprene emission flux,
Is is the isoprene emission flux standardized to leaf temperature 30°C and PAR of 1000
micromoles/mVsec.
The light adjustment factor CL is estimated by:
CL = (ac^PAR) / V(l+a2PAR2)
(4-20)
where: a = 0.0027, and CL1= 1.066 are empirically derived coefficients. The leaf temperature
adjustment factor CT is derived from laboratory data and is computed using the following formula:
4-46
-------
EPA/600/R-99/030
CT = (exp[cTI(T-Ts)/RTsT])/ (l+exp[cT2(T-TJ/RTsT]) (4-21)
where: CT, = 95,000 J
Ts is the standardized temperature (303°K),
R is the ideal gas constant (8.3 14 °K mol'1),
CT2 = 23 0,000 J mol' ', and
Tm = 314°K.
PAR is computed as a function of height by:
PAR, = PARo ( exp[-0.42LAIz]) (4-22)
During March 1998, the factor used in BEIS-2 within Models-3 to convert solar radiation values
from watts per square meter (W/m2) to micromoles per square meter-second (u.m/m2-sec) was
changed from 2.2982 to 2, based on Alados et al. (1996). This had the effect of reducing PAR
by ~ 1 5 percent and isoprene emissions by ~ 5 percent based on limited simulation tests for the
July 1 995 ozone maximum period.
Emissions for VOCs other that isoprene are assumed to vary only as a function of leaf
temperature in accordance with the following:
E = Es * exp[0.09 (T-TS)] (4-23)
where: Es is the standardized emission flux for monoterpenes and other VOCs,
T is the leaf temperature in degrees Kelvin, and
Ts is the standardized temperature (303°K).
The soil NO emission flux factors used in BEIS-2 were adapted from Williams et al. (1992). For
temperatures greater than 0°C, soil NO temperature corrections follow the formulation of
Williams et al. As follows:
NO = N00 * exp[0.07 1 (T-TS)] (4-24)
where NO is the adjusted soil NO emission flux,
NO0 is the emission flux standardized to a soil temperature Tsof 30° C, and
T is the soil temperature.
NO emissions at temperatures less than 0° C are set to zero. During March 1998, it was decided
to cap the exponential increase of soil NO emissions with temperatures above 30°C, based on the
findings of Yienger and Levy (1995).
4-47
-------
EPA/600/R-99/Q30
The inclusion of BEIS-2 into MEPPS biogenic emission modeling required creating an efficient
connection such that BEIS-2 is easy to use from within MEPPS, and also so that the input and
output data related to BEIS-2 can take advantage of MEPPS and Models-3 data handling and
analysis features. The changes included:
• BEIS-2 is now automatically invoked from MEPPS or the Study Planner when the user
requests that biogenic emission factors be calculated. It is not necessary to work directly
with the BEIS-2 software.
* Solar radiation and temperature input files are not directly supplied by the user when
BEIS-2 is run. These meteorological data are supplied automatically from MCIP files for
the case (time period) specified, presuming that MM5 and MCIP have been run for that
period.
Within MEPPS, BEIS-2 can now use ARC-INFO® generated grids and use gridded
surrogate data to disaggregate county-level data to grid cells.
• BEIS-2 results are in terms of hourly emissions per grid cell within Models-3.
• Biogenic emission data are subject to the extended quality control and reporting features
in MEPPS, and visualization using the ARC-INFO® based GIS-View feature.
The MEPPS uses BEIS-2 by allowing the user to assume summer, winter seasons, or access frost
data in order to allow for the seasonal change in vegetative biomass. For summer and winter
conditions, vegetation genus and emission factors for isoprene, monoterpene, other biogenic VOC
emissions, emission factors for NO, and the leaf area index (LAI) are provided based on the work
of Geron et al. (1994). Frost data include the federal identification protocol (FIPS) codes by state
and county along with the biomass of the first and last day of summer. The MEPPS draws
hourly temperature and solar radiation from MCIP for use in the biogenic processor. Land cover
and vegetation are provided in gridded ARC/INFO coverage. The biogenic modeling input data
are then provided to BEIS-2 and output as hourly gridded emission values of the VOCs isoprene,
monoterpene, and "other" (unspecified); along with NO to the EMPRO speciation processor,
where they are grouped with chemical species from other source types depending upon the
conventions of either the CB4 or RADM2 speciation split factors (Section 4.2.5).
4.2.4.2 Mobile Source Emissions
Emissions from mobile sources to the air are established as one of the primary contributors to
pollution problems in many localities. Unlike many anthropogenic emissions, mobile source
emissions are strongly affected by the rapid variations of atmospheric temperature and
anthropogenically-influenced, geographically-varying factors. Consequently, it is necessary to
model hourly mobile emissions, rather than to temporally disaggregate annual totals. The
capability to do this is in the MEPPS mobile-source emission processor. For gaseous emissions
4-48
-------
EPA/600/R-99/030
the mobile source processor estimates hourly emissions of VOC total organic gases (TOG),
carbon monoxide (CO), and oxides of nitrogen (NOx) from on-road mobile sources (vehicles).
The processor, which is located in the EMPRO module of MEPPS, uses a combination of air
temperature data at 1.5 m above the surface (provided by MM5 through MCIP), mobile source
emission factors (computed by the EPA regulatory model Mobile 5a for gaseous emissions from
on-road mobile sources), fleet vehicle type composition, road type, and traffic data in the form of
vehicle miles traveled (VMT) to generate emission estimates. Vehicle emission controls, and
Inspection and Maintenance programs may be accounted for in the user-defined input settings for
Mobile 5a (U.S. EPA, 1991; U.S. EPA, 1996). The VMT data are usually available by county, or
occasionally by specific road segments. The resulting county or road segment-specific hourly
emissions are spatially allocated to the cells of a user-defined study grid. The gridded emission
data are combined with corresponding species output from the mobile source particulate emission
model, and allocated to different groups of lumped species for processing by the Speciation
Processor (Section 4.2.5).
Emissions of particulate species (particulate matter less that 10 micrometer and 2.5 micrometers
in diameter - PMIO and PM2 5, respectively) from mobile sources are modeled analogously to
gaseous emissions, using vehicle fleet composition, road type (urban or rural), and VMT data.
The mobile particulate emission factors are modeled with PARTS (U.S. EPA, 1995), the
particulate companion of Mobile 5a. Vehicle emission controls, Inspection and Maintenance
programs, and use of reformulated fuel may be accounted for with the input settings of PARTS.
In addition input information for road silt content and precipitation is required. The particulate
species modeled include lead, direct and indirect particulate sulfate (any remaining sulfur in fuel is
assumed to be exhausted as gaseous SO2), total exhaust particulate (sum of lead, direct sulfate,
and carbon including soluble organic material and other remaining carbon), and (separately)
soluble organic fraction (SOF) and a remainder carbon portion (RCP), which are added to the
mobile source emission output file.
4-49
-------
EPA/600/R-99/030
INPUT DATA
Vehicle Fleet Information
PARTICOLATE GASEOUS
EMISSIONS EMISSIONS
Silt Loading Controls
Read Type Fuel Type
METEOROLOGY
(MM5)
MCIP
Temperature
PARTS
Mobile
Emission
Factors
MobileSa
Mobile
Emission
Factors
Compute Gridded Hourly
MobEe Source Emissions
SPECIATION
OUTPUT
PROCESSOR
Gridded VMT Data
Figure 4-5 General Flow of Mobile Source Emission Modeling
The procedure by which mobile source emission data are estimated in MEPPS and flow diagrams
for the process are given in Section 6,5.3.4 of the Models-3 Volume 9B: User Manual. Briefly,
the sequence is as follows:
• VMT vehicle fleet data files are established for the study area (gridded spatial domain) of
interest by either extracting county-level annual aggregate VMT data from the annual
emission input files previously established by INPRO, or by loading "link-node" VMT data
that are available for specific road segments (links) between nodes with geographic
coordinates . The link-node data are generally available only for selected urban areas and
dates in conjunction with special studies. The process is similar for VMT data which are
occasionally available by quarter section (1/4 square mile) areas.
« ARC/INFO® spatial coverage files of surrogate data (Federal Highway Administration
major highways, or United States Census Tiger Line road coverages are provided) are
used to spatially allocate county-level aggregate VMT data to the cells of the gridded
domain being used with a particular study. County area may also be used as a coarse
surrogate coverage.
• Spatial allocation ratios are computed for use in assigning county-level VMT data to the
surrogate coverages. The EMPRO calculates the proportion of road links or county area
4-50
-------
EPA/600/R-99/030
in the spatial domain attributable to each grid cell overlain by the road link (FHA or
TIGER/LINE) or county.
• Mobile source emission factors are computed using Mobile 5a and/or PART 5- U.S. EPA
regulatory models. Mobile 5a requires user-defined information on driving operations,
vehicle fleet and fuel use, and hourly ambient temperature data. PART 5 requires user-
defined information on vehicle fleet and fuel use, emission controls, non-attainment, road
dust silt content, moisture, and percent of unpaved roads by geographic area (usually
county). Temperature and moisture data are from .the MM5 model as processed by
MCIP.
• Hourly mobile emissions are computed for each grid cell, with adjustments applied to
VMT to indirectly reflect the effect of temperature on gaseous emissions, and moisture on
re-entrained dust for particulate emissions.
• For mobile emissions of VOC, hourly gridded mobile emissions are speciated using either
RADM 2.0 or CB-4 emission profile splits. Speciation is not needed for particulate
emission data.
• The hourly, gridded, speciated mobile emission data are merged with gridded, speciated
hourly area-source emission data in OUTPRO, and converted to NetCDF I/O API format,
to provide a two dimensional emission data set acceptable to the chemical-transport model
and other Models-3 framework applications.
The Mobile 5a Model
Detailed descriptions of Mobile 5a are available from EPA publications (US EPA 1991, US EPA
1996). An optional Mobile 5b has also been adopted (US EPA 1996) to accommodate slight
modifications in Inspection and Modification credits for hybrid fuels. Both Mobile 5a and 5b are
improvements in the Mobile 4.1 model (US EPA, July 1991). Mobile 6 is in development and is
tentatively expected in late 1999. A new off-road mobile emission model is also due in late 1998.
Portions of the following description of Mobile are adapted from Wilkinson et al (1994). The
incorporation of Mobile 5a into MEPPS was extensively revised from its treatment in GEMAP,
including:
• The application of Mobile 5a factors was revised to substantially reduce redundancy and
computational space required. The application of Mobile 5a in GEMAP computed
specific emission factors for individual state/counties by area type/road type and hour.
The combinations of hour and road type were derived by mapping operating modes
(percent of hot and cold starts) to the hours and road types that used them, causing a
duplication of Mobile 5a runs and factors. It was necessary to recompute factors if there
was a change in how the emission factors were used by hours or road type. The computed
mobile emission factors were merged into permanent (for the processing run) emission
4-51
-------
EPA/600/R-99/030
factor tables without regard to calendar year. In MEPPS, Mobile 5a emission factors are
computed and stored by calendar year, state/county, and operating mode. Also, the
system now has separate user-specified tables that allocate emission factors by hour and
road type, using operating mode. The tables are used directly in the emission estimate
calculation. In addition start and end times may now be specified for mobile emission runs
rather than number of complete hours. This has also shortened the processing time for
county-level mobile emission processing.
• Calculations were separated into county-level, link-node level, and land-survey level
options, depending upon the spatial detail of VMT information available to the user.
« The ability to assign alias identification by state or county was added. This allows the user
to assign the values in a Mobile input file for a given area to a similar area which lacks
specific input data.
» Geographic coverages of major highways (from the Federal Highway Administration) and
all roads (TIGER-Line data) were made available as surrogate data to allow better spatial
allocation of mobile emissions.
• Mobile emission data are now subject to the additional quality control and ARC-INFO®
based visualization capabilities in MEPPS.
Mobile 5a is an ANSI FORTRAN 77 computer program designed to estimate hydrocarbon (HC),
carbon monoxide (CO), and oxides of nitrogen (NOx) emission factors for gasoline-fueled and
diesel-fueled highway motor vehicles. The computation methods that are embedded in Mobile 5a
are based on the procedures that are presented in the Compilation of Air Pollutant Emission
Factors — Volume XX: Highway Mobile Sources (US EPA, 1995). Mobile 5a computes
emissions factors for eight vehicle categories in two regions of the country (high-altitude and low-
altitude). The eight vehicle categories include the following:
« LDGV — light-duty gasoline vehicles;
« LDGT1 ~ light-duty gasoline trucks (up to 6000 pounds);
* LDGT2 ~ light-duty gasoline tracks (6001 to 8500 pounds);
• HDGV — heavy-duty gasoline vehicles (over 8500 pounds);
» LDDV — light-duty diesel vehicles;
• LDDT - light-duty diesel tracks (0 to 8500 pounds);
« HDDV — heavy-duty diesel vehicles (over 8500 pounds); and
• MC ~ motorcycles.
The Mobile 5a emission factors depend on various conditions including, but not limited to,
ambient temperature, vehicle speed, operating modes, and vehicle mileage accrual rates. Much of
the data required by Mobile 5a may be specified by the user through the Mobile input file. Mobile
computes emissions factors for any year from 1960 to 2020. This date range is important for
4-52
-------
EPA/600/R-99/030
mobile-source emission estimated projections, and for the-- application of regulatory control
factors. Mobile version 5a.01 (which was modified for use in GEMAP/EMS-95 and later for
MEPPS) is currently implemented in the EMPRO module of MEPPS. Consult the User's Guide
to MOBILE 5, EPA-AA-TEB-92-01 for further details on Mobile 5a. Mobile model
documentation may be obtained from the EPA Office of Mobile Sources World Wide Web home
page located at: http://www.epa.gov/OMSWWW/models.htm.
A comprehensive discussion of the technical formulation of Mobile 5a is found in the above
references. This section contains a description of modifications to Mobile 5a for use in GEMAP
and later in MEPPS. Mobile 5a was modified for use in EMS-95 to compute and report diurnal
evaporative emissions factors (grams/mile) separately. The unmodified Mobile 5a model
computes and reports diurnal evaporative emission factors as part of a composite evaporative
emission factor. Also, Mobile 5a was modified to report a total non-diurnal evaporative emission
factor. The total non-diurnal evaporative emission factor includes hot soak, crankcase blow-by,
running losses, and resting losses. In the MEPPS EMPRO module, the Mobile SAS® table
generators have been rewritten and segmented for greater computational efficiency, as have been
the VMT and surrogate coverage grid processors and mobile source emission calculation
procedures. The technical procedure and rationale are the same, but processing time is shortened
by at least a factor of two for regional modeling of spatial domains.
The mobile-source emission processor relies on a SAS® lookup table to find appropriate emissions
factors to compute the motor vehicle emission estimates. The SAS® lookup table is generated
through iterative runs of Mobile 5a. The necessary number of runs of Mobile 5a is. performed
automatically, and is based on the standard Mobile ASCII input file supplied by the user. For
urban scale modeling, an input file for one state may suffice. However, for regional modeling,
multiple Mobile 5a ASCII input files can be concatenated to one file and run through the
processor. This capability is necessary to handle study domains where motor vehicle activity
differs spatially and temporally. For example, in a multiple-state study domain, it is unlikely that
adjoining states have the same inspection and maintenance (I/M) program or the same vehicle
fleet distribution (to name just two inputs to Mobile 5a). Differences in regional inputs to Mobile
5a such as I/M programs and vehicle fleet distributions result in different motor vehicle emissions
factors. A concatenated input file template (based on example values used for the OTAG project)
is provided which may be copied and edited. In MEPPS, the Mobile input file (m5a,mv) is
located in the directory structure at:
The only difference in the concatenated input file from the original input file is that the state title
lines contain state (and where needed, county) FIPS identifier codes to identify the area for which
each section of the Mobile 5a input file is applicable:
• FIPS state code — Mobile input file applies to a state; or
» FIPS state code and FIPS county code — Mobile input file applies to a particular county.
4-53
-------
EPA/600/R-99/030
The mobile-source emission processor varies the vehicle speed and ambient temperature records
of the user-supplied input file and runs Mobile for each variation. The Mobile source processor
varies the vehicle speeds from 4 miles per hour to 64 miles per hour in increments of 2 miles per
hour, although the user may adjust the increment (Mobile supports compulation of emission
factors for speeds between 2.5 MPH and 65 MPH). The mobile-source processor varies the
minimum ambient temperature from 50° F to 110° F in increments of 2° F, and the mobile source
processor varies the maximum ambient temperature from 50°FtollO°Fin increments of 2° F
(Mobile 5a supports computation of exhaust emissions factors for temperatures between 0° F and
110° F and evaporative emissions factors for temperatures between 40° F and 110° F). The
vehicle speed and ambient temperature values can be set interactively in EMPRO under the mobile
source model, using the "Generate Mobile Emission Factors" screen.
In MEPPS, the mobile-source processor generates two emission factor SAS® lookup tables.
These are compact tables that contain the information held in seven generated SAS lookup tables
inEMS-95. They are:
• State/area type/facility type diurnal emissions factors (I/M and non-l/M HC diurnal
evaporative emissions factors for eight vehicle categories by state, minimum ambient
temperature, maximum ambient temperature, and calendar year). The user specifies
whether the operating mode varies by the hour of the day in the Temporal Allocation
screen of the mobile emission model of EMPRO);
» State/area-type/facility-type non-diurnal emissions factors (I/M and non-I/M HC
[evaporative and exhaust], CO [exhaust only], and NOX [exhaust only] emissions factors
for eight vehicle categories by state, area type [urban or rural], facility type [eg. interstate,
collector, arterial, local], vehicle speed, ambient temperature, and calendar year. The user
can interactively specify whether the operating mode varies by hour of the day when
establishing temporal profiles in the Temporal Allocation screen of the mobile emission
model of EMPRO.
Operating mode is an important variable in the computation of motor vehicle emissions factors.
Attention is specifically called to operating mode because the mobile source processor uses up to
four different operating mode mixes to compute emissions factors for the Mobile 5a lookup tables
(refer to the Mobile 5a user's guide for further discussion of operating modes).
If area type/facility type records are to be used in the Mobile-generated lookup tables, the user
must specify the mix of operating modes. The following example of operating mode mixes was
used for the Lake Michigan Ozone Study: 5% PCCN, 12.1% PCHC, and 10.9% PCCC, where
PCCN is percent VMT generated by non-catalyst vehicles in cold-start mode, PCHC is percent
VMT generated by catalyst vehicles in hot-start mode, and PCCC is percent VMT generated by
catalyst vehicles in cold-start mode. If hourly records are used in the Mobile-generated lookup
tables, the user must define the operating mode mixes by time of day, consistent with Mobile 5a
guidance. The following is an example of the use of four operating mode mixes:
4-54
-------
EPA/600/R-99/030
to 0700 operating mode is 12,9% PCCN, 7% PCHC, and 17% PCCC;
0700 to 0900 operating mode is 15% PCCN, 10% PCHC, and 15.4% PCCC;
0900 to 1600 operating mode is 5% PCCN, 12.1 % PCHC, and 10.9% PCCC;
1600 to 1800 operating mode is 10,1% PCCN, 9.1% PCHC, and 13.9% PCCC;
1800 to 2200 operating mode is 5% PCCN, 12.1% PCHC, and 10.9% PCCC;
2200 to 0000 operating mode is 12.9% PCCN, 7% PCHC, and 17% PCCC.
As with the vehicle speeds and ambient temperatures, the operating mode mixes can be changed
interactively on the Operating mode screen in THE MEPPS EMPRO module under Models,
Mobile, Mobile Source.
The mobile source processor generates emission factors for the following area type (urban or
rural) and facility type (interstate, local, etc.) combinations:
• rural/principal arterial - interstate
• urban/principal arterial - interstate
• rural/principal arterial - other
• urban/principal arterial - free/express-ways
• rural/minor arterial
• urban/principal arterial - other
• rural/major collector
• urban/collector
• rural/minor collector
• urban/minor arterial
• rural/local
• urban/local
As with vehicle speeds, ambient temperatures, and operating modes; area type/facility type
combinations can be changed interactively using the "Generate Mobile Emission Factors" screen
under EMPRO, Models, Mobile, Mobile Source to edit the ASCII input file. When changes are
made, the same changes are automatically made to the S AS® data sets.
User Input Data
The mobile source emission processor requires user-defined specifications using interactive
screens and an input file. While some of the specifications are required, some are not required. In
each case, there is provision to import the data file through IDA or in the mobile source processor
of MEPPS.
The principal input data ASCII files to the mobile source model include the following
(environment variables are defined in Chapter 6 of the Models 3 Volume 9B: User Manual):
• On-network link-specific area type/facility type ($EMS/onnet.mv)
4-55
-------
EPA/600/R-99/030
• On-network link-specific percentage of vehicles that fall under an I/M program
($EMS/onnetim. mv)
• On-network link-specific daily or hourly VMT (SEMS/onnetvmt.mv);
» Hourly or daily, link-specific or area type/facility type-specific vehicle mix
• Off-network, area type/ facility type average speed ($EMS_LOC/offnetspd.mv)
* Off-network, area type/facility type VMT ($EMS/offnvmt.mv);
* Off-network, area type/facility type-specific vehicle mix profile ($EMSJLOC/offvmix.mv)
« Public land survey quarter sections area type/ facility type hourly oir daily VMT, hourly
speeds, and daily vehicle mix profile ($EMS/ofnvmtp.mv)\
» Area type/facility type seasonal and daily adjustment factors ($EMS/adjstvmt.mv)
• Area type/facility type hourly average speed profile ($EMS_CAT/spdadju,mv)
• On-network, link-specific hourly or daily average speed ($EMS_GRD/onnetspd.mv)
• Area type/facility type hourly VMT fractional profile ($EMS/fracvmtu.mv)
If off-network VMT is to be used to compute emissions estimates (typical when county VMT
data are taken from regional emission inventories), the mobile emission processor can assign
VMT to grid cells using ARC/INFO® coverage files for either county area, Federal Highway
Administration (FHA) major highways, or TIGER/LINE road data. These coverages for the
contiguous United States are provided with MEPPS. Subsets of the coverages for the spatial
domain of interest are extracted either when a grid is established or during mobile source
processing.
If Public Land Survey Quarter Section-based VMT is to be used to compute emissions estimates,
the mobile source emission processor requires an ARC/Info® export data set of the polygon-
network system.
Gridding VMT
After the ASCII user input (foundation) data files have been read and checked, the user directs
the mobile source emission processor to prepare the necessary network coverages (unless they
were prepared when the grid was established). The on-network system is prepared from the user-
provided ASCII input files. The off-network system is prepared via county boundary, FHA, or
TIGER/LINE data file extracts. The polygon system is prepared from an ARC/INFO® export
coverage. The networks are gridded by overlaying the emission modeling grid on the network
coverages in much the same manner that the area source spatial surrogates are gridded. Each link
in the networks (or area type/facility type polygon) are length (or area) apportioned to a grid cell
in the emissions modeling domain. Once the networks have been gridded, the corresponding
VMT can also be gridded. Equations 4-25 through 4-30 show how the on--network, off-network,
and polygon VMT are gridded respectively.
lenratio,jiUm = lengridjj k, m / lenparijik (4-25)
VMTjjMm = vmtorigy>k * lenratioiJMm (4-26)
, = cellsumijAam / ctysumjjAf (4-27)
4-56
-------
EPA/600/R-99/030
(4-28)
ateapctij.pAf.ijn = areagrdijip>aif>l)m / areaparijip (4-29)
VMTij,P,a ,f,i.m = vmtorigjj pa f * areapctijip>aifil>m (4-30)
where lenratio is the percentage of a link in a given grid cell
lengrid is the length of a link in a given grid cell
lenpar is the total length of the link
VMT is the gridded VMT value
vmtorig is the original VMT
cellpct is the percent of the FHA or TIGER/Line-based area type/facility type
roadway in a given cell
cellsum is the total length of FHA or TIGER/Line-based area type/facility type
roadway in a given cell
ctysum is the total length of the FHA or TIGER/Line-based area type/facility type
roadway in a county
areapct is the percent of the polygon-based area type/facility type roadway in a
given cell
areagrd is the total length of the polygon-based area type/facility type roadway in a
given cell
areapar is the total length of the polygon-based area type/facility type roadway in
a county
i is the state index
j is the county index
k is the link index for on-network systems
1 is the east-west grid cell index
m is the north-south grid cell index
a is the area type index for off-network systems
f is the facility type index for off-network systems
p is the polygon index for Public Land Survey Quarter Sections network systems.
For on-network and polygon VMT, the indices may also include an identifier for hour since
polygon and on-network VMT can also be supplied on an hourly basis.
Apply Default Information .
If the user supplies limited data to the mobile-source emission processor, it can apply a variety of
default values to compute motor vehicle emissions estimates. These defaults include vehicle mix,
speeds, VMT fractional profiles, and I/M vehicle percentages. If the user supplies limited data,
the mobile-source emission processor applies the following defaults:
Default Vehicle Mix Profile LDGT1 0. 177
Vehicle Class Percentage LDGT2 0.077
LDGV 0.618 HDGV 0.035
4-57
-------
1
4
6
7
8
9
1
2
4
5
6
9
60
56
56
54
40
35
60
59
55
30
55
30
EPA/600/R-99/030
LDDV 0.008 Default Speed Profile
LDDT 0.002 Area Type Facility Type Speed
HDDV 0,077 0
MC 0.008 0
0
• - 0
0
0
1
1
1
1
1
1
Default Hourly VMT Fractional Profile
Emission Hour 1 23 4 5 67 8 9 10 11 12
Diurnal 0.000 0,0000.000 0.000 0.000 0.1290.021 0.100 0.095 0.0950.1660.199
Other 0.016 0.0100.003 0.006 0.010 0.0260.053 0.064 0.055 0.0480.0500.052
13 14 15 16 17 18 19 20 21 22 23 24
Diurnal 0.079 0.116 0.000 0.000 0.000 0.000 0.000 0.000 0,000 0.000 0.000 0.000
Other 0.054 0.055 0.059 0.070 0.074 0.070 0.058 0.046 0.037 0.033 0.028 0.022
If the user does not specify what the percentage of vehicles are under an I/M program, the mobile
source emission processor assumes that no vehicles are under an I/M program. The result is
higher emission estimates.
The mobile source emission processor does not override user-supplied data. Default information
are applied only when data are missing from the ASCII user input files.
Add Temperatures and Adjust VMT
Prior to computing the motor vehicle emission estimates, the mobile-source emission processor
adds the gridded, hourly temperatures and adjusts the VMT to the specific modeling day. The
temperature data are used to obtain the correct emission factor from the Mobile emission factors
lookup tables. Through the application of Equations 4-3 1 through 4-34, the day-specific, diurnal
and nondiurnal (other than full days), hourly VMT are computed, respectively.
dvn%lihim>n = ddayvmty>lim>n * adjday * adjmonth * dvmt_profij>l)h (4-31)
ovmt^,^,, = odayvmtjj,,^,, * adjday * adjmonth * ovmt_prof; jil-h (4-32)
dvmtjj^h^,, = ddayvmty^,, * adjday * adjmonth * dvmt_profia,fjm>n * adjday * adjmonth * ovmt_profjjia_f h (4-34)
4-58
-------
EPA/600/R-99/030
where , ,
dvmt is the diurnal VMT; . • -
dayvmt is the total day diurnal VMT;
adjday is the day-specific VMT adjustment fector supplied through an ASCII input
file;
adjmonth is the month-specific VMT adjustment factor supplied through an ASCII
input file;
dvmt_prof is the hour-specific, diurnal VMT fractional profile factor;
ovmt is the nondiurnal VMT;
odayvmt is the total day nondiurnal VMT;
ovmtjsrof is the hour-specific, nondiurnal VMT fractional profile factor;
i is the state index;
j is the county index;
h is the hour index;
1 is the link identifier index;
a is the area type index; , .
f is the facility type index;
m is the east-west grid cell index; and . •• ,
n is the north-south grid cell index.
Compute Motor Vehicle Emissions Estimates
The mobile source emission processor computes day-specific, gridded, hourly motor vehicle
emissions estimates of VOC total organic gases (TOG), carbon monoxide (CO), and oxides of
nitrogen (NOX). The actual emission estimate computations performed and maintained in the
MEPPS data base are as follows: ,
• TOG for gasoline (noncatalyst/catalyst composite) evaporative diurnal processes;
• TOG for gasoline (noncatalyst/catalyst composite) evaporative, nondiurnal processes; .
• TOG for gasoline (noncatalyst/catalyst composite) exhaust operations;
• NOX for gasoline (noncatalyst/catalyst composite) exhaust operations;
• CO for gasoline (noncatalyst/catalyst composite) exhaust operations;
• TOG for diesel (noncatalyst/catalyst composite) exhaust operations;
* NOX for diesel (noncatalyst/catalyst composite) exhaust operations; and
• CO for diesel (noncatalyst/catalyst composite) exhaust operations.
The mobile emission processor computes motor vehicle emissions estimates based on three types
of VMT files:
• On-network VMT;
• Off-network VMT; and
• Public land survey quarter sections VMT.
4-59
-------
EPA/600/R-99/030
The differences in the VMT types are:
• On-network VMT are network link-specific;
• Off-network VMT are county/area type/facility type-specific; and
• Public land survey quarter sections VMT are county/polygon/area type/facility type-
specific.
The mobile-emission processor estimates motor vehicle emissions for all combinations of VMT.
Regardless of the VMT type that is used to generate motor vehicle emission estimates, all motor-
vehicle emission estimates have some degree of uncertainty. However, the most desirable motor-
vehicle emissions estimates usually are generated from on-network, link-specific data, followed by
public land survey quarter sections data, and finally off-network data. Motor vehicle emission
estimates that are generated from on-network data are not more certain, but that they are more
spatially representative.
Before the mobile-source emission processor computes the motor vehicle emission estimates, it
computes a fleet composite emission factor. Equations 4-44 through 4-67 show how the
processor computes the fleet composite emission factors. In all cases, the emission factors are
extracted from the Mobile lookup tables based on the appropriate indices. Please refer to the
section on generating the Mobile emission factors lookup tables earlier in this section for
additional information.
If the user has provided non-I/M-specific VMT, the mobile source emission processor computes
the fleet composite emission factors by vehicle classification for temperature and speeds
combinations through the application of Equations 4-35 through 4-42.
hcdifSjt = SjhcdiJAt * vmix, (4-35)
hcexfs t = Sj heexj s t * vmiXj (4-36)
noxfs, = SjnoxXj s, * vmiXj (4-37)
S,t J JjS,t J \ •/
hcevfs>t = Sj hcevjiS, * ymfac, . (4-39)
rlfir'p'YT "5 7, r1rif*f*v * \7tniv ( &,—XlTY\
UULlv/wAJ.. * ***\ VillWwA.: «. * VI. Hi A: \" ^\/ /
5,1 J J»»*1 J \. S
dnoxfj, = Sj dnoxXjAt * vmiXj (4-41)
dcoexfj t = Sj dcoeXj s, * vmiXj (4-42)
where: j is the vehicle classification index
s is the speed index
t is the temperature index
hcdif is the hydrocarbon diurnal fleet composite emission factor for all vehicle types
(grams/mile)
hcdi is the non-I/M hydrocarbon diurnal emission factor (grams/mile)
vmix is the vehicle mix profile
4-60
-------
EPA/600/R-99/030
hcexf is the hydrocarbon exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
hcex is the non-I/M hydrocarbon exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
noxf is the oxides of nitrogen exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
noxx is the non-I/M oxides of nitrogen exhaust nondiurnal emission factor for
gasoline vehicles (grams/mile)
coexf is the carbon monoxide exhaust nondiurnal fleet composite emission factor for
gasoline vehicles (grams/mile)
coex is the non-I/M carbon monoxide exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
hcevf is the hydrocarbon evaporative nondiumal fleet composite emission factor for
gasoline vehicles (grams/mile)
hcev is the non-I/M hydrocarbon evaporative nondiurnal emission factor for gasoline
vehicles (grams/mile)
dhcexf is the hydrocarbon exhaust nondiurnal fleet composite emission factor for
diesel vehicles (grams/mile)
dhcex is the non-I/M hydrocarbon exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
dnoxf is the oxides of nitrogen exhaust nondiurnal fleet composite emission factor for
diesel vehicles (grams/mile)
dnoxx is the non-I/M oxides of nitrogen exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
dcoexf is the carbon monoxide exhaust nondiurnal fleet composite emission factor for
diesel vehicles (grams/mile)
dcoex is the non-I/M carbon monoxide exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
If the user has provided I/M-specific VMT data, the processor computes the fleet composite
emission factors by vehicle category for temperature and speed combinations through the
application of Equations 4-43 through 4-50.
hcdifSit = 2j ihcdijAt * vmiXj . (4-43)
hcexfs>t = 2,. ihceXjAt * vmiXj (4-44)
noxfs>t =SjinoxxjAt*vmixj (4-45)
^rw^vf" = yi if*n^v * vmiv (A-A.{\\
COeXls, ^ ICUCXj st VIIllXj • (* ^°)
hcevfS)t = 2j ihcevjAt * vmiXj (4-47)
dhcexfs t = 2j idhcexjA, * vmiXj (4-48)
dnoxfs t = 2j idnoxXj s t * vmiXj (4-49)
dcoexfs, = 2j idcoeXj s t * vmiXj (4-50)
where: ihcdi is the I/M hydrocarbon diurnal emission factor (grams/mile)
4-61
-------
EPA/600/R-99/030
ihcex is the I/M hydrocarbon exhaust nondiurnal emission factor for gasoline vehicles
(grams/mile)
inoxx is the I/M oxides of nitrogen exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
icoex is the I/M carbon monoxide exhaust nondiurnal emission factor for gasoline
vehicles (grams/mile)
ihcev is the I/M hydrocarbon evaporative nondiurnal emission factor for gasoline
vehicles (grams/mile)
• idhcex is the I/M hydrocarbon exhaust nondiurnal emission factor for diesel vehicles
(grams/mile)
idnoxx is the I/M oxides of nitrogen exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
idcoex is the I/M carbon monoxide exhaust nondiurnal emission factor for diesel
vehicles (grams/mile)
If the user has provided both I/M and non-I/M VMT data, the mobile source emission processor
computes the fleet composite emission factors by vehicle category for temperature and speed
combinations through the application of Equations 4-51 through 4-58 This set of equations can be
applied only to on-network VMT data because the I/M percentages are input to the mobile source
emission processor on a link-specific basis.
hcdifw - Sj ((ihcdijs, * imvmt,) + (hcdijAt * (1 - imvmt,))) * vmiXj (4-51)
hcexfs>, = Sj ((ihcexJAt * irnvmt,) + (hceXjAt * (1 - imvmtj))) * vmiXj (4-52)
noxf,t = Sj ((inoxxjA, * imvmt,) + (noxXj s, * (1 - irnvmt,))) * vmiXj (4-53)
coexJfSit = Sj ((icoexJA, * imvmt,) + (coexjAt * (1 - imvmt,))) * vmiXj (4-54)
hcevf., = Sj ((ihceVjst * imvmt,) + (hceVjA, * (1 - irnvmt,))) * vrniXj (4-55)
dhcexfj, = Sj ((ielhceXjAt * imvmt,) + (dhceXjSt * (1 - imvmt,))) * vmiXj (4-56)
dnoxf,, = Sj ((idnoxXj st * imvmt,) + (dnoxxjA, * (1 - imvmt,))) * vmiXj (4-57)
dcoexfst = Sj ((idcoexjst * irnvmt,) + (dcoexjst * (1 - imvmtj))) * vrniXj (4-
58)
where: imvmt is the fraction of vehicles that are under an I/M program; and
1 is the link identifier index.
Many indices have been left out of Equations 4-35 through 4-58. The composite emission factors
that are generated are gridded, hourly values. Because temperature is a gridded, hourly value, the
temperature index t in Equations 4-31 through 4-54 implies that the composite emissions factor
are gridded, hourly values.
Once the fleet composite emission factors have been computed, the mobile source emission
processor computes the motor vehicle emission estimates through the application of Equations 4-
59 through 4-66,
4-62
-------
EPA/6DO/R-99S030
dhc_vee = hcdif * divmt * 10**-3 .. (4-59)
hcjcee = hcexf * ovmt * 10**-3 (4-60)
noxxee = noxf * ovmt * 10**-3 (4-61)
co_xee = coexf * ovmt * 10**-3 (4-62)
hc_vee = hcevf * ovmt *' 10**-3 .v (4-63)
dhc_xee = dhcexf * ovmt * 10**-3 (4-64)
dnoxxee = dnoxf * ovmt * 10**-3 (4-65)
dco_xee = dcof * ovmt * 10**-3 (4-66)
where: dhc_vee is the gridded, hourly diurnal hydrocarbon emission estimate (kilograms)
divmt is the gridded, hourly diurnal VMT (miles)
hc_xee is the gridded, hourly exhaust hydrocarbon nondiurnal emission estimate for
gasoline vehicles (kilograms)
ovmt is the gridded, hourly nondiurnal VMT (miles)
noxxee is the gridded, hourly exhaust oxides of nitrogen nondiurnal emission estimate
for gasoline vehicles (kilograms)
co_xee is the gridded, hourly exhaust carbon monoxide nondiurnal emission estimate
for gasoline vehicles (kilograms)
hc_vee is the gridded, hourly evaporative hydrocarbon nondiurnal emission estimate
for gasoline vehicles (kilograms)
dhc_xee is the gridded, hourly exhaust hydrocarbon nondiurnal emission estimate for
diesel vehicles (kilograms)
dnoxxee is the gridded, hourly exhaust oxides of nitrogen nondiurnal emission
estimate for diesel vehicles (kilograms)
dco_xee is (the gridded, hourly exhaust carbon monoxide nondiurnal emission
estimate for diesel vehicles (kilograms)
The mobile-source emission processor computes the final motor vehicle emission estimates file by
summing the emission estimates computed by Equations 4-59 through 4-66 over each state
identifier, county identifier, east-west grid cell identifier, north-south grid cell identifier, process
type (EV for evaporative, EX for exhaust), technology type (1 for gasoline, 2 for diesel), and
pollutant identifier (HC for hydrocarbon, CO for carbon monoxide, NOX for oxides of nitrogen).
The PARTS Model
The current version of PARTS was released in 1995 by the U.S. EPA Office of Mobile Sources.
The description given here is taken principally from the user's guide, A Draft User's Guide to
PARTS: A Program for Calculating Particle Emissions from Motor Vehicles (U.S. EPA,
1995b), which provides information in addition to that given in this section. The guide is available
from the Office of Mobile Sources Internet web site at http:www.epa.gov/omsw/models.htm
The PARTS is a FORTRAN program used to model emission factors needed to estimate
emissions from gasoline and diesel powered on-road vehicles. It calculates particle emission
factors in grams per mile for particle sizes 1 to 10 micrometers. The emission factors include
4-63
-------
EPA/600/R-99/030
exhaust particulate matter and components, brake wear, tire wear, and re-entrained road dust.
The program includes default data for most inputs, but allows user-supplied data for most items.
The Interactive aspects of running PARTS, as it is provided by the Office of Mobile Sources,
have been subsumed into MEPPS to allow the user to enter and edit data, specify program control
flags, and run the model via a series of windows within the EMPRO mobile source processor.
Controls specified within PARTS may vary spatially and temporally. PARTS output report flags
are not germane within MEPPS because the outputs are automatically processed with emissions
from other emission sources in MEPPS internal format. They flags used include the following
items:
• Vehicle fleet mix. The user may specify of whether default or user-supplied VMT vehicle
fleet mixes are used. The VMT mix is the fraction of total VMT for all vehicles
contributed to by each vehicle class. The default VMT is based on national averages and
trends over the years. The trends reflect the sales shift from automobiles to light-duty
trucks, and the increasing use of diesel engines in both light and heavy-duty trucks. VMT
mix may vary spatially. The following are the descriptions of the vehicle classes and
corresponding Federal Highway Administration (FHA) and gross vehicle weight (GVW)
used in PARTS.
Vehicle Class FHA Class GVW (Ibs)
1 = LDGV (light-duty gasoline vehicle)
2 = LDGT1 (light-duty gasoline truck, I) 1 <6000
3 = LDGT2 (light-duty gasoline truck II) 2A 6001 -8500
4 = HDGV (heavy-duty gasoline truck) 2B-8B >8500
5 = MC (motor cycle)
6 = LDDV (light-duty diesel vehicle) 1 <6000
7 = LDDT (light-duty diesel truck) 2A 6001 -8500
8 = 2BHDDV (class 2B heavy duty diesel vehicle) 2B 8501 -10000
9 = LHDDV (light heavy-duty diesel vehicle) 3,4,5 10001-
19500
10 = MHDDV (medium heavy-duty diesel vehicle) 6,7,8A 19501-
33000
11 = HHDDV (heavy heavy-duty diesel vehicle) 8B 33000+
12 = BUSES (buses)
• Mileage accumulation rates. The user may supply or use default mileage accumulation
rates and vehicle registration distributions. The mileage accumulation rate is the expected
number of miles that a typical vehicle (for each class) is expected to travel in a year,
divided by 100,000. The rates must be provided for each vehicle type for each of 25 years
(12 years for motor cycles).
4-64
-------
EPA/600/R-99/030
• Inspection and maintenance. The user specifies whether an inspection and maintenance
program is assumed for gasoline vehicle vehicles only.
• Reformulated gasoline. The specifies whether the use and effects of reformulated gasoline
on particulate emissions .are assumed. The effects are partially.based on the sulfur content
of gasoline used. An average sulfur weight percent of 0.034 is used for any gasoline used
in all years prior to 2000. If reformulated gasoline is used for the year 2000 or later, the
sulfur weight percent is assumed to be 0.0138.
The purpose of MEPPS is to estimate temporally and spatially-varying emission data on a gridded
basis. This is consistent with some additional input data which PARTS requires be entered on
spatial basis, which are defined as PARTS scenarios. In EMPRO mobile source processing the
user is prompted to edit existing or new scenario data. The information includes:
• Region, calendar year, speed cycle (type of driving, ie. cruise), and average speed.
• Percent of unpaved road silt, paved road silt loading in gm /m2, and optionally, the average
number of wheels per vehicle. Modeled dust emission factor are highly sensitive to the
unpaved road silt percentage which is extremely variable. Consequently, measured data
are advised when possible.
• The number of days each year with greater than 0.01 inches of precipitation for use in
modeling re-entrained dust. This is climatological average data. In the future, the
emission processing system may be modified to use more spatially and temporally
accurate modeled precipitation data by grid cell taken from MM5 and MCIP for the
PARTS calculations.
• A scenario name may be applied. In MEPPS a FIPS code a geographic identifier is
entered which is used to apply PARTS emission factors computed for the county level to
the appropriate modeling grid cells.
• PARTS expects the user to define a maximum particle size cutoff (maximum allowed is 10
micrometers). In MEPPS the maximum size is defaulted to 10 micrometers to ensure that
the full range of particle sizes are represented in the regional emission estimates.
• Average vehicle weight (Ibs)
PARTS Emission Factors
The emission factors calculated by PARTS include the particulate pollutant compounds of lead,
sulfate, soluble organic fraction particulate matter, remaining carbon portion particulate matter,
and total exhausted particulate. The lead and sulfate are formed from the lead and sulfur
contained in the fuel. The soluble organic fraction consists primarily of hydrocarbons coming
4-65
-------
EPA/600/R-99/030
from unburned or partially burned fuel and lubricating oil. The remaining carbon portion consists
of soot-like carbon (elemental carbon) and trace amounts of other components from the fuel and
lubricating oil. The total exhaust particulate is the sum of these four categories. In addition to
these categories of exhaust emissions, idle exhaust emissions (for heavy diesel vehicles only),
brake wear, tire wear, fugitive dust, indirect sulfate, and gaseous sulfur dioxide are calculated by
PARTS.
The model calculates the emission factors for the 12 vehicle classes previously described and a
fleet-wide average (estimated by VMT weighting of the emission factors for each of the 12
vehicle classes. The factors are composites of the emission factors for the 25+ years prior to the
year of interest, in order to allow for the effects of older vehicles. The composite emission factor
for each vehicle class is calculated by weighting the emission factor calculated for each model year
by the travel fraction for that model year, and then summing the 25 weighted factors. The travel
fraction of a model year is the fraction of VMT by a vehicle of that model year out of the total
number of miles traveled by all model years' vehicles in that vehicle class.
25
EFCOMPV = EF m>v * TF m>v (4-67)
m=I
where: EFCOMPV is the composite emission factor for vehicle class v,
EF m,v is the emission factor for vehicle class v, model year m,
TF m,v is the travel fraction for vehicle class v, model year m.
The overall travel fraction of a vehicle class represents the fraction of the total number of VMT of
that class of the total highway VMT by all 12 classes. The VMT fractions for each vehicle class
are multiplied by the corresponding composite emission factors (EFCOMPV), and the sum of the
adjusted emission factors is reported as the emission factor for all vehicles.
EFALL = £ EFCOMPV * TFCLASSV (4-68)
where: EFALL is the weighted emission factor for all vehicles,
TFCLASSV is the VMT of vehicle class v.
The emission factor for all vehicles, EFALL, represents the grams/mile of emissions
Lead Emission Factors for gasoline-fueled vehicles
Lead particulate emission factors for gasoline-fueled vehicles assume that almost all lead in the
fuel is exhausted. Therefore, the emission factors (grams/mile) depend on the lead content of fuel
and fuel economy of the vehicle (miles/gallon). The factors allow for the fact that the lead content
of leaded fuel is much greater than for unleaded fuel. The assumption is made that because diesel
fuel has negligible lead content, the lead emissions from diesel vehicles will be negligible. The
4-66
-------
1PA/60WR-W/030
following formulae are used to compute lead emission factors for all gasoline-fueled automobiles
and motorcycles, respectively.
LEADrav = PLNOCTmv*VLNOCTmv + PUNOCTmv*VUNOCTmv+ .
PLYSCTm_> VLYSCTm,v + PUYSCT^. * VUYSCT^ (4-69)
where: m denotes a specified model year
v denotes a specified vehicle class
LEADro v is the lead particulate emissions for any given vehicle (grams/mile)
PLNOCTm v is the emissions for a non-catalyst, leaded fuel vehicle (grams/mile)
PUNOCTm v is the emissions for a non-catalyst, unleaded fuel vehicle (grams/mile)
PLYSCTm v is the emissions for a catalyst, leaded fuel vehicle (grams/mile)
PUYSCTm v is the emissions for a catalyst, unleaded fuel vehicle (grams/mile)
VLNOCTm v is the emissions for a non-catalyst, leaded fuel vehicle (grams/mile)
VUNOCTm v is the emissions for a non-catalyst, unleaded fuel vehicle (grams/mile)
VLYSCTm v is the fraction of catalyst leaded fuel vehicles
VUYSCTm v is the fraction of catalyst, unleaded fuel vehicles
The emission rate is adjusted for speed by the factor FEC such that:
CLEADm>v = FEC * LEADm>¥ (4-70)
where: CLEADm v is the lead emissions for a vehicle of model year m and vehicle class v,
which has been adjusted for the effect of speed (grams/mile)
FEC is 1/SCFC (4-71)
SCFC is the speed correction factor, based either transient driving cycle (c=l) or
steady cruise driving cycle (c=2)
SCF, = 0.17930 + (0.038561 * SPEEDV ) - (0.00041067 * SPEEDV2) (4-72)
SCF2 = 0.26929 + (0.054607 * SPEEDV) - (0.00069818 * SPEEDV2) (4-73)
SPEEDV is the average speed for vehicle class v (mph) - user input
Motorcycles do not have catalytic emission controls, therefore organic emission factors are not
calculated and sulfate emission factors are deemed negligible. PARTS emission factors for
motorcycles are almost entirely for lead particulate matter. The lead particulate emission factor
for 2-stroke engines is 0.33 g/mi and for 4-stoke engines 0.046 grams/mile. For model years
before 1978 there were 51 percent 4-stroke engines and 49 percent 2-stroke engines, computed as
follows:
4-67
-------
EPA/600/R-99/030
LEADm,motoreycIe = [(0.49 * 0.33) + (0.51 * 0.046)] * PSL (4-74)
For model years 1978 and later motorcycles are assumed to consist entirely of 4-stroke engines:
LEADm,motorcycle - 0.046 * PSL (4-75)
where: PSL is the fraction of all particles that are emitted based on a specified upper particle
size cutoff (10 micrometers running in MEPPS)
The specific derivations of the lead particulate emission formulae are given in Appendix 1 of the
PART 5 user's guide (U.S. EPA, 1995b).
Sulfate Emission Factors for Gasoline-Fueled Vehicles
Particulate sulfate emission factors consist of direct and indirect sulfate material. The direct
sulfate is exhausted as sulfuric acid, and the indirect sulfate is formed later in the atmosphere from
exhausted SO2. The indirect sulfate in the model is calculated based on the assumption that it
consists entirely of ammonium sulfate and ammonium bisulfate. The direct sulfate, indirect
sulfate, and gaseous sulfate emission factors are computed in PARTS, and the emission factors
reported as grams/mile traveled.
The direct sulfate from non-catalyst vehicles using leaded feel (includes catalyst-equipped vehicles
which are misfueled, making the catalyst ineffective) is calculated as:
DSULFN = .002, for speeds equal to or less than 19.6 mph (4-76)
DSULFN = .001 , for speeds equal to or greater than 34.8 mph (4-77)
The direct sulfate from catalyst vehicles is calculated as :
DSULFCm>v = [FRAC^* (.005) + FRACcat/air (.016)] (4-78)
for speeds equal to or less than 19.6 mph
DSULFCm>v = [FRACox/noair (.005) + FRAC3w/noair (.001) + FRACox/air (.020)
+FRAC3w/air(.025)] (4-79)
for speeds equal to or greater than 34.8 mph
For speeds between 19.6 and 34.8 mph, DSULFN and DSULFC are interpolated between
Equations 4-76 and 4-77 and 4-78 and 4-79, respectively.
where: m denotes a specified model year
v denotes a specified vehicle class
FRACcat/,noair is the fraction of vehicles which are catalyst equipped with no air pump
4-68
-------
EPA/6007R-99/030
FRACcat/air is the fraction of vehicles which are catalyst equipped with an air pump
FRACox/noajr is the fraction of vehicles which are oxygen catalyst equipped with no air
pump .
FRAC3w/noair is the fraction of vehicles which are 3-way catalyst equipped with no air
pump
FRACox/aif is the fraction of vehicles which are oxidation catalyst equipped with an air
pump
FRAC3w/air is the fraction of vehicles which are 3-way catalyst equipped with air pump
The direct sulfate from all gasoline-fueled vehicles is computed as:
DSULFm v = CTLFRCm v * DSULFCm v + (1.0 -CTLFRCm v) * DSULFN (4-80)
where: CTLFRCmiV = CATFCTm>v(l-RMISmiV) (4-81)
the fraction of the vehicle class that has an effective catalyst
The PARTS assumes that all sulfur in fuel is exhausted as either sulfate or gaseous sulfur dioxide
(SO2). Therefore when the direct sulfate emission factor is calculated, the remaining sulfur in the
fuel is considered to be exhausted as SO2. The amount of sulfur remaining in the fuel after the
direct sulfate emission factor has been determined must be calculated to find the amount of sulfur
exhausted as SO2 (grams/mile).
The following equation is used to determine the fraction of sulfur in the fuel that has been directly
converted to sulfate (DSULFm v calculated in Equation 4-80 above). The equation calculates
direct sulfate as a function of the fuel sulfur content, DCNVRT (the fraction of sulfur in the fuel
that is converted to direct sulfate), and the fuel economy.
DSULFmv - 13.6078 * (1.0 + WATER) * FDNSTY * SWGHT * DCNVRT
/FEm,v (4-82)
where: m denotes a given model year
v denotes a given vehicle class
DSULFm v is the direct sulfate emission factor (grams/mile)
WATER is the weight ratio of seven water molecules to sulfate, 1.2857, based the
estimate that at 50 percent humidity, seven water molecules bond with each
sulfuric acid molecule
FDNSTY is the fuel density in lb/gal (6.09 pounds/gallon)
FEm v is the fuel economy
SWGHT is the weight percent of sulfur content in fuel (.034, except for reformulated
fuel phase II, for year 2000 and later .0138)
DNVRT is the percent of sulfur in the fuel that is directly converted into sulfate (2
percent)
4-69
-------
EPA/6QO/R-99/030
13.6078 is a unit conversion factor equal to (453.592 * 3.0)/100, where 453. 592 is
equal to the number of gram in a pound, 3.0 is the weight ratio of sulfate to
sulfur, and division by 100 corrects for the weight percent of sulfur, SWGHT
If DSULFCm-v and DSULFN (from Equations 4-76 through 4-79) are substituted in Equation 4-
82, one can solve for the fractions of sulfur in the fuel that are converted to sulfate separately for
catalyst and noncatalyst vehicles:
FCr4¥RCm v = DSULFCm¥ * FErav / (13.6078 * (1.0 + WATER) *
FDNSTY* "SWGHT) (4-83)
FCNVRNmv - DSULFN * FEmv / (13.6078 * (1.0 + WATER) *
FDNSTY * SWGHT) (4-84)
where: FCNVRCm v is the fraction of the percent of fuel that is directly converted into sulfate
for catalyst equipped vehicles
FCNVRNmv is the fraction of the percent of fuel that is directly converted into sulfate
for non-catalyst vehicles
The gaseous sulfur emission factors, which are dependent on the above fractions, are calculated
from the following equation:
SO2m,v = 9.072 * FDNSTY * SWGHT * (1.0 -DCNVRT) / FEm-v (4-85)
where: SO2m v is the gaseous sulfur emission factor of a vehicle
9.072'is a unit conversion factor equal to (453.592 *2)/100, where 453.592 is the
number of grams in a pound, 2 is the weight ratio of SO., to sulfur, and the
division by 100 corrects for the weight percent of sulfur, SWGHT
Additional details concerning the calculation of gaseous sulfur emission factors are given in the
PARTS user's guide (U.S. EPA, 1995b). In addition to direct sulfate and gaseous sulfate
emission factors, PARTS estimates an indirect sulfate emission factor by assuming that a fraction
of the gaseous sulfur dioxide emissions is later converted in the atmosphere to sulfate material.
Based on ambient sulfur and sulfate measurements in 11 cities, it is estimated that 12 percent of
all gaseous sulfur is converted to sulfate. Additional information on the calculation of indirect
sulfate is given in the PARTS user's guide.
Sulfate Emission Factors for Diesel-Fueled Vehicles
For diesel-fueled vehicles, PARTS calculates sulfate emission factors again assuming that all
sulfur in the fuel is exhausted as either sulfuric acid or gaseous sulfur dioxide. The direct sulfate
emission factor (grams/mile) is calculated using the following equation:
4-7.0
-------
EPA/600/R-99/030
DSULFm v = 13.6078 * (1.0 + WATER) * FDNSTY * SWGHTD *
DCNVRT/FEm.v (4-86)
where: m is a specified model year
y is a specified vehicle class
DSULFm v is the direct sulfate emission factor for a vehicle (grams/mile)
DCNVRT is the fraction of sulfur in the fuel that is converted directly to sulfate (2.0
percent
FDNSTY is the density of diesel fuel (7.11 pounds/gallon)
FEm v is the fuel economy for a vehicle (miles/gallon)
SWGHTD is the weight percent of sulfur in diesel fuel (0.25 for high sulfur fuel, 0.05
for low sulfur fuel used in 1993 and later)
WATER is the weight ratio of seven water molecules to sulfate (1.2857)
13.6078 is a unit conversion factor equal to 493.592 * 3/100, where 493.592 is the
number of grams in a pound, 3 is the weight ratio of sulfate to sulfur, and the
division by 100 corrects for the weight percent of sulfur SWGHTD
The gaseous sulfur emission factor is calculated by the following equation:
SO2m v = 9.072 * FDNSTY * SWGHTD * (1.0 - DCNVRT) / FEm ¥ (Eq.4-
87)
where:
SO2mv is the sulfur emission factor (grams/mile) of a vehicle
9.072 is a unit conversion factor equal to (453.592 * 2/100) where 453.592 is the
number of grams in a pound, 2 is the weight ratio of SO2 to sulfur, and the
division by 100 corrects for the weight percent of sulfur, SWGHTD
The indirect sulfate emission factor, consisting mainly of ammonium sulfate and ammonium
bisulfate is calculated using the following equation:
ISULFm v = ICNVRT * SO2m>v * (3/2) * AMNWGT (4-88)
where:
ISULFm y is the indirect sulfate emission factor of a vehicle (grams/mile)
ICNVRT is the fraction of SO2 converted to sulfate in the atmosphere (12 percent)
3/2 is the weight ratio of SO2 to sulfate
AMNWGT is the estimated weight ratio of the combination of ammonium bisulfate
and ammonium sulfate in the atmosphere to sulfate (1.6)
Total Exhaust Particulate Emission Factors for Gasoline-Fueled Vehicles
4-71
-------
EPA/600/R-99/030
The total exhaust participate emission factors for light-duty gasoline fueled vehicles are calculated
by summing lead, direct sulfate, and a carbon emission factor which includes soluble organic
material and other remaining carbon. Table 4-5 presents a summary of the carbon emission
factors by vehicle model year and type of technology.
Table 4-5. Carbon Emission Factors for Gasoline-Fueled Vehicles (grams/mile)
Vehicle Type/
Model Year
Light-Duty
Gasoline Vehicles
pre-1970
1970-1974
1975-1980
1981 +
Light-Duty
Gasoline Trucks I
pre-1970
1970-1974
1975-1986
1987 +
Light-Duty
Gasoline Trucks II
pre-1979
1979-1986
1987 +
Heavy-Duty
Gasoline Vehicles
pre-1987
1987 +
Leaded Fuel
.193
.068
.030
.017
.193
.068
.030
.017
.370
.068
.030
.370
.163
*' UrdeaMj/ jGaialyst ; ; ;
.. ^tppjed^^iPf
.:y^;^pUmp3;;^.:;2v
.;- -- -. ...v_ .-: -.:•:::•:•:•:•:-- :•:•;•:::-:-.•:•:•:•: '••"••';'•<•'••'• - >--'--'-;-;-;-'
.v;!,*.!;:,:.,,;.:.:.:.; :;;.;:.* ; •?;;;;?,, ;;,¥;;:;;::; :
NA
.0060
.0060
.0043
: '. •. . . :'.:::.- :- ":- ''•,-,' ':-.-. •-".-•" " '-...'.'.'..".' '.'.;.
NA
.0060
.0060
.0043
.:•!'"•:::-• ••:•:•::•••:•-•:'::--' :':'::: '"!•'".'• • ' •;• •:••-*-:-
- .••.:•:'•'••• '-;:• .•::::::•::••••:••':•:•:':•:'::• ': :-:W::::
NA
.0060
.0043
.054
.054
; '. • .l|ii |e^|el|:|SSa^it:;;; >
i:i|ili||!i|||p|li:|f;:-:;;
•:- ; :f • ::;:: :-;i:purnp) :; - W< "; r ' :
NA
.0250
,0250
.0043
: : :_:-: :; • : ::::;::_:::::;':::'-'_::::'::::-:::-:-:-:-x- -:-•-:•:•:-"-:-:
• , --•;•::.::.• :-":.-"-"-., :".--: -" "- " '•: . .-,-,
NA
.0250
.0250
.0043
ff^^^^^mmiiM
NA
.0250
.0043
.054
.054
s:;W:;.tJ&audeji?.MonS:::;:*:s
i;:ca|alyst eqwljiped: :
•.-...' - -•-.-•"-• .'--'- '--:' -'--'-'
.030
.030
.030
.017
.030
.030
.030
,017
i ;]:;:}! ;:-:;.;: V. :xll|l|&;:;;;:;'C ;•' '•;. .
.054
.030
.017
.054
.054
Total Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles
4-72
-------
EPA/600/R-99/030
The total exhaust particulate emission factors are for diesel-fueled vehicle categories and model
years (EFDPMm>v), The emission factors for heavy-duty vehicles are in units of grams/brake-
horsepower hour (g/BHP-hr), which are converted to grams/mile by PARTS. The conversion
factors and emission factors vary by model year. The emission factors for light-duty diesel
vehicles are in units of grams/mile. The total exhaust emission factors given in Table 4-6 are
based on high-sulfur fuel. The sulfur content in diesel fuel was reduced in 1993 by EPA
regulatory requirements. Consequently, when a the specified calendar year is 1993 or later,
PARTS will adjust the exhaust emission factors for lower sulfur foel. Particulate emission factors
for diesel-fueled vehicles are not adjusted for speed.
Table 4-6, Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles
Vehicle Type/ Model Year Group
Light-Duty Diesel Vehicles (grams/mile)
pre-1981
1981
1982-1984
1985-1986
1987
1988-1990
1991-1993
1994-1995
1996 +
Light-Duty Diesel Trucks (grams/mile)
pre-1981 '
. 1981
1982-1984
1985-1986
1987
1988-1990
1991-1993
1994-1996
1997 +
Exhaust Particulate Emission Factor
.700
,259
.256
.255
,134
.132
.131
.128
.100
-;";^::^:'-v-%:^:l:r^
.700
.309
.354
,358
.334
.291
.294
.130
.109
4-73
-------
EPA/600/R-99/030
Table 4-6. Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles
Vehicle Type/ Modal Yew Group
Class 2B Heavy-Duty Diesel Vehicles (grams/brake
horsepower-hour)
pre-1988
1988-1990
1991-1993
1994 +
Light Heavy-Duty Diesel Vehicles (grams/brake
horsepower-hour)
pre-1988
1988-1990
1991-1993
1994 +
Medium Heavy-Duty Diesel Vehicles (grams/brake
horsepower-hour)
pre-1987
1988-1990
1991-1993
1994 +
Heavy Heavy-Duty Diesel Vehicles (grams/brake
horsepower-hour)
pre-1987
1988-1990
1991-1993
1994 +
Buses (grams/brake horsepower-hour)
pre-1987
1988-1990
1991
Exhaust Partieuiate Emission Factor
,5156
.5140
.2873
.1011
.
.5156
.5140
.2873
.1011
.6946
.4790
.2747
.0948
^';*^^s^^ss^;r';v;'^^sp^
^^y*'fi&^f^^Vl:^?^^^Mt- ":'^^^:K::&dr::^m:
.6444
.4360
.2709
.0836
.6931
.4790
.2772
4-74
-------
EPA/600/R-99/030
Table 4-6. Exhaust Particulate Emission Factors for Diesel-Fueled Vehicles
Vehicle Type/ Model Year Group
1992 without traps
1992 with traps
1993 without traps
1998 with traps
1994 +
Bxhaust Partjcufafe Emission Factor
.1716
.0257
.1457
.0240
.0591
Soluble Organic Fraction and Remaining Carbon Portion Emission Factors for Diesel-Fueled
Vehicles
The PARTS model calculates the Soluble Organic Fraction (SOF) emission factor as a fraction of
the remaining mass, using the following equation:
SOFm>v, = [EFm,v- DSULFm>v ] * (fraction^ (4-89)
The Remaining Carbon Portion (RCP) is defined the remainder (everything else):
RCPm,v = Efm>v - DSULFm;v - SOFm>v (4-
90)
where: m is the model year of a selected vehicle
v is the class of a selected vehicle
SOFm v is the Soluble Organic Fraction of the exhaust particulate emission factor
(grams/mile) , .',"'.
RCPmv is the remaining carbon portion (elemental carbon) of the exhaust particulate
emission factor (grams/mile)
EFm v is EFDPMm v * CFm v , the exhaust particulate emission factor for a vehicle
(grams/mile)
EFDPMm v is the exhaust particulate emission factor for a vehicle (grams/Brake
Horsepower-hour)
CFm v is the conversion factor from grams/Brake Horsepower-hour to grams/mile
(Brake Horsepower-hour/mile)
fractionSOFv is the fraction of the non-sulfate portion (ie. the carbon portion) of the
diesel exhaust particulate emission factor which is organic carbon for a vehicle
(Brake Horsepower-hours/mile)
The Soluble Organic Fractions (fractionSOFv) for different vehicle classes are as follows (U.S.
EPA, 1990):
4-75
-------
EPA/600/R-99/030
• 0.18 for LDDVs (Light-Duty Diesel Vehicles)
0.50 for LDDTls (Class 1 Light-Duty Diesel Tracks)
0.48 for LDDT2s (Class 2 Light-Duty Diesel Tracks)
• 0.51 for LHDDVs (Light Heavy-Duty Diesel Vehicles) and 2BHDDVs (Class 2B Heavy-
Duty Diesel Vehicles)
0.44 for MHDDVs (Medium Heavy-Duty Diesel Vehicles) and BUSES
• 0.24 for HHDDVs (Heavy Heavy-Duty Diesel Vehicles)
Idle Emission Factors for Heavy Diesel-Fueled Vehicles
Idle emission factor data in grams/hour were collected from manufacturers! for heavy-duty diesel
vehicle classes only. Consequently, the idle emission factors are not included into the "all
vehicles" emission category. The vehicle class emission factors are calculated by averaging
together model-year-specific emission data, where the model-year-specific: emission data are
weighted by the estimated travel fraction of that model year within the vehicle class.
The idle emission factors are model-year-specific but the model year emission rates do not vary by
vehicle class. Consequently, the same model year emission factors are used for all the heavy-duty
diesel classes, and the differences between idle emission factors between classes reflects only the
differing travel fractions between model years for a class. As a result, the emission factors
reported for the smaller of the heavy-duty vehicle classes, such as 2BHDDV and LHDDV, may
be over estimated. The base idle emission factors in PARTS for all heavy-duty diesel vehicles are
as follows:
• 5.370 grams/hour for models prior to 1988
» 3.174 grams/hour for model years 1988-1990
• 1.860 grams/hour for model years 1991 -1993
• 1.004 grams/hour for model year 1994 +
Re-entrained Dust from Unpaved Roads
Re-entrained road dust emission factors for PM10 (particulate matter less man 10 micrometers in
size) in PARTS are estimated using an equation (4-91) based on ambient measurements. Because
these measurements include particulate matter from brake wear, tailpipe exhaust, tire wear, and
ambient background particulate matter concentrations, these factors must be subtracted from the
unpaved road particulate emission factors before the latter factors can be applied. It is necessary
to obtain a data base of unpaved and paved road types by county for regional modeling purposes.
The MEPPS contains procedures and estimates for 1995 developed for EPA by calculating re-
entrained emissions by the month at the state road-type level for the average vehicle fleet, and
then allocated to the county road type by population for unpaved roads and by total VMT for
paved roads. These procedures should be consulted for more details, such as the means of
estimating unpaved and paved VMT data. The equation used to calculate PMIO from unpaved
roads is as follows:
4-76
-------
EPA/600/R-99/030
UNPVD = PSDUNP * 5.9 * (UNSILT/12.0) * (SPD/30.0) * (WEIGHT/3.0)OJ *
(VWHEEL/4.0)0'5 * (365-IPDAYS)/365 * 453.592 (4-91)
where: UNPVD is the fleet average unpaved road dust emission factor (grams/mile)
PSDUNP is the fraction of particles less than or equal to the particle size cutoff (the
cutoff is 10 micrometers in MEPPS)
UNSILT is the percent silt content of the surface material (user input)
SPD is the average vehicle speed in miles/hour (user input)
WEIGHT is the fleet average vehicle weight (user input in pounds)
V WHEEL is the fleet average number of wheels (user input, default is 4)
IPDAYS is the average number of precipitation days per year with greater than 0,01
inches of rain (user input - MEPPS contains climatological default data for
1995)
453.592 is the number of grams in a pound
Emission factors for brake and tire wear (in addition to exhaust emissions) must be calculated so
that they may be subtracted from the unpaved road emission factors. The brake wear emission
factor is assumed to be the same for all vehicle classes. It is set equal to:
BRAKE = 0.0128 * PSBRK (4-92)
where: PSBRK is the fraction of particles less than or equal to the particle size cutoff. The
emission factor 0.0128 grams/mile is taken from U.S. EPA, 1985c.
The tire wear emission factor is calculated using the following equation:
EFTIREV = 0.002 * PSTIRE * IVEHWLV ' (4-93)
where: v is a selected vehicle class
EFTIREV is the tire wear emission factor
0.002 is the emission rate of airborne particulate matter from tire wear for light-duty
vehicles (U.S. EPA, 1985)
PSTIRE is the fraction of particles less than or equal to the particle size cutoff
IVEHWLV is the average number of wheels on a vehicle of a given class, where
LDGV=4, LDGT1,2=4, HDGV=6, MC=2, LDDV=4, LDDT=4, 2BHDDV=4,
LHDDV=6, MHDDV=6,HHDDV=18, BUSES=4
Re-entrained Dust from Paved Roads
The PM10 emission factor for paved roads is estimated similarly to that for unpaved roads. The
VMT data for paved roads may be developed following the procedures described in U. S. EPA,
1998. The equation used to estimated paved road re-entrained dust emission factors again is
based on ambient measurements, and tailpipe, brake wear, and tire wear emission factors must be
4-77
-------
EPA/600/R-99/030
subtracted prior to use of the paved road dust emission factor. The paved road emission factors
are calculated by:
PAVED = PSDPVD * (PVSILT/2.0)065 * (WEIGHT/3.0)15 (4-94)
where: PAVED is the fleet average paved road dust emission factor (grams/mile)
PSDPVD is the base emission factor for the particle size cutoff (10 micrometers in
MEPPS)
PVSILT is the road surface silt loading (grams/square meter) (user input)
WEIGHT is the fleet average vehicle weight (input by the user in pounds)
Application in MEPPS
The PARTS model is a companion of MobileSa in MEPPS. When running MEPPS interactively
through the Tools Manager, the user is prompted to specify input data and options under
EMPRO, Mobile Source Model. Standard lookup data files can be found in the Models-3
directory structure at /horae/models3/datasets/nostudies/part5/. Default/example input data sets
are included for the 1995 calendar year. When a study is created, the study input data are located
atSEMS_HOME/project/$EMS__PROJECT/ravf_data/$EMS_DOMAIN/coirimon/.
If the user is processing mobile emission data using the Study Planner, it is necessary to ensure
that the appropriate PARTS input data sets are in place first. This may be accomplished when
using MEPPS interactively to establish MEPPS directory structure prior to running through Study
Planner. If the data are in place, Study Planner will automatically process them.
When processing mobile particulate emission data, the input and output data of PARTS are
assigned to and summed within each grid cell in the same fashion given in detail under the
Gridding VMT discussion of the implementation of the Mobile 5a model (above),
4.2.5 Chemical Speciation of Emission Data
Chemical transport models, such as CMAQ, require that emission data be provided for either
individual species or specific species groups or "lumped" species. This is necessary so that the
atmospheric chemistry of pollutants may be more accurately modeled. However, an initial
processing step is required because emission data are often reported for pollutants that are
aggregates of many species, such as VOC. These aggregate pollutants must be split into their
component species, or "speciated".
The speciation takes two forms, discrete and lumped-rnodel.
In discrete speciation, a pollutant is split into the individual components which comprise the
pollutant. For the organic pollutant TOG (total organic gas), the individual components which
comprise the pollutant are dependent on a variety of factors including the process, fuel type, and
device from which the emissions occurred. For example, TOG from the exhaust of automobile
4-78
-------
EPA/6QO/R-99/03G
may contain approximately fifty discrete organic compounds (benzene, methane, toluene, hexane,
etc.), while TOG from degreasers may contain approximately seventy discrete organic
compounds. NOX (nitrogen oxides), which is related to combustion processes, is speciated
(discrete) into NO and NO2 (and sometimes HONO).
The discrete components in an emission stream are determined by a number of methods including
source testing, surrogate application, and engineering knowledge of the process. Many sources
have been inventoried and a compendium of species profiles has been assembled by the US EPA
in the Air Emissions Species Manual (US EPA, 1988). The Air Emissions Species Manual
contains a list of TOG and particulate matter (PM) species profiles to which a substantial number
of emission sources have been assigned. A compendium of species profiles currently available is
compiled in the US EPA Speciate database. The database is available on the Air Chief CD ROM,
which is updated annually (US EPA, 1997). Each of the species profiles identifies the mass
percent of the discrete compounds that comprise TOG and PM. Note that NOX, SOX (sulfur
oxides), and CO (carbon monoxide) do not have source-specific speciation profiles. For all
sources, NOX is discretely speciated into NO and NO2 (and sometimes HONO), and CO is
treated explicitly. The speciation processor currently assumes that NOx is speciated to 95 percent
NO and 5 percent NO2, based on average observed values.
The MEPPS EMPRO speciation processor provides chemical speciation of hydrocarbons, oxides
of nitrogen, and sulfur oxides. For example, aggregations of hydrocarbon species such as TOG
are disaggregated to their component individual chemical species. The processor speciates the
spatially and temporally allocated emission estimates that are prepared by the emission estimation
processors and models (e.g., point source processor, mobile emission model). The speciation is
accomplished using look-up tables of profiles containing source-category specific and specie-
specific chemical split factors (percent of mass of source emissions attributable to each specie for
a given source category). The general components of emission speciation processing are
illustrated in Figure 4-6.
It is computationally prohibitive to model the chemistry of all discrete VOC compounds in the
emissions stream in photochemical grid models. Therefore, individual organic species comprising
Total Organic Gas are assigned to one or more model species (groups of species) according to the
chemical mechanism that is being used. Thus, instead of modeling with a larger number of
discrete compounds, the discrete compounds are lumped into a much smaller number of
mechanism species. The rules for assigning the discrete compounds to the mechanism species are
mechanism dependent and typically involve lumping in one of two ways: 1) lumping compounds
with similar reactivity characteristics into a single mechanism species (the lumped molecule
approach) or 2) assigning molecular fragments of an individual compound to one or more
mechanism species on the basis of molecular structure (lumped structure approach). In Models-3,
the lumping of discrete compounds to form mechanism species is carried out using tables of split
factors that assign the discrete compounds to mechanism species. Currently, two mechanisms, the
Regional Acid Deposition Model 2.0 (RADM2) and Carbon Bond 4 (CB-4) are available in
Models-3. The user may also define lumping procedures for an alternative mechanism if the
4-79
-------
EPA/600/R-99/030
assignment methods for that mechanism are known, or the user may create a modified version of
the lumping procedure for one of the two mechanisms in Models-3.
In the CB-4 mechanism, discrete compounds are assigned to mechanism species on the basis of
the compound's carbon bond structure (Gety et al.,1989). For instance, single carbon-carbon
hydrocarbon bonds are assigned to a paraffin group (PAR), and carbon-carbon double bonds are
assigned to an olefkuc group (OLE). Thus, an individual discrete VOC could be dis-aggregated
into more than one mechanism species depending on its structure. Descriptions of the CB-4
mechanism species are given in Table 4-8.
The RADM2 mechanism lumps discrete compounds on the basis of their prevalence in the
atmosphere, common reactivity, and/or molecular weight (Stockwell et al.> 1990). For RADM2,
individual VOC compounds are first assigned to one of 32 lumped groups. The 32 lumped
groups are then further reduced to 15 groups for increased computational efficiency. The
relationship between the 32 and 15 lumped groups is shown in Table 4-9. The EMPRO does not
include chemical or size fractionation of particulate matter, although size fractionation of
particulate matter is planned in the future.
Temporally Allocated Gridded Emission Data
Point
Source
Data
Area
Source
Data
Mobile
Source
Data
Biogenic
Source
Data
Application of Speciation Split Factors
o Chemical Mechanism Dependent
(e.g., RADM2, CB4)
o ROG to TOG Adjustment
o Non-VOC Split Factors
(e..g,NQx,SQx)
Output Processor
Figure 4-6 General Components of Emission Speciation Processing
4-80
-------
EPA/600/R-99/030
4-81
-------
EPA/600/R.-99/030
Table 4-8. Carbon Bond-4 Lumped Species
CB4 Lumped Species
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
CB4 Lumped Species
Code
OLE
PAR
TOL
XYL
FORM
ALD2
ETH
MEOH
ETOH
ISOP
NR
NO
NO2
CO
SO2
AERO
Lumped Species Description
Olefinic carbon bond (C=C)
Paraffin carbon bond (C-C)
Toluene (C6HS-CH3)
Xylene (C6H4-(CH3)2)
Formaldehyde (CH2=O)
High MW aldehydes (RCHO, R>H)
Ethene (CH2=CH2)
Methanol (CH3OH)
Ethanol (C2H5OH)
Isoprene
Non Reactives as methane
Nitric Oxide
Nitrogen Dioxide
Carbon Monoxide
Sulfiir Dioxide
Aerosols (PM)
4-82
-------
Table 4-9. RADM 2.0 Lumped Species Descriptions
BPA/600/R-99/030
RADM 2,0 32
. Lamped
Speejes
1
2
3
4
5
6
7 .
8
8
9 •
10
11
12
13
13
14
15
16
17
18
18
19
20
21
22
RADM 2.0 .16
Lumped Species
CH4
ETH
HC3
HC3
HC5
HC8
HC8
HC8
XYL
OL2
OLT
OLT
OLI
OLT
OLI
TOL
TOL
XYL
CSL
OLT
TOL
HCHO
ALD
KET
KET
Allocation
Factor
(fraction of
one)
1
1
1
1
1
1
. i
0.91
0.09
1
1
1
1
0.5
0,5
1
1
1
1
0.5
0.5
1
1
1
I
Reactivity Factor
(ratio)
1
1
0.519
0.964
0.956
0.945
1.141
1.101
1
1
1
1
1
1
1
0.293
1
1
1
1
1
I
1
a.253
1
Description of the 32 RADM
Lumped Species Before Grouping to
1 6 Lumped Species.
' * % •*'
Methane
Ethane
Propane
Alkanes (0.25-0.50)
Alkanes (0.50-1. 00)
Alkanes (1.00-2.00)
Alkanes (>2.00)
Alkane/Aromatic Mix
Alkane/Aromatic Mix
Ethene
Propene
Alkenes (Primary)
Alkenes (Internal)
Alkenes (Prim/Int Mix)
Alkenes (Prim/Int Mix)
Benzene/Halobenzenes
Aromatics (<2 react)
Aromatics (>2 react)
Phenols and Cresols
Styrenes
Styrenes
Formaldehyde
Higher Aldehydes
Acetone
Ketones
4-83
-------
EPA/600/R-99/030
Table 4-9. RADM 2.0 Lumped Species Descriptions
RADM 2.0 32
Lumped
Species
23 •
24
25
26
27
28
29
30
31
32
RADM 2.0 16
Lumped Species
ORA2
HC3
HC3
.
HC3
HC3
HC5
HC8
.
Allocation
Factor
(fraction of
one)
1
1
1
1
1
1
1
1
1
1
Reactivity Factor
(ratio)
1
0.343
0.078
1
0.404
1.215
1.075
1.011
1
1
Description of the 32 RADM
Lumped Species Before Grouping to
1 6 Lumped Species
Organic Acids
Acetylene
Haloalkenes
Unreactive
Others (<0.25 react)
Others (0.25-0.5 react)
Others (0.5- 1.0 react)
Others (> 1.00 react)
Unidentified
Unassigned
In Models-3, the user initiates speciation by selecting the chemical mechanism to be used for a
study in the Study Manager under the Models-3 framework, and within the MEPPS main
window under Tools Manager. The procedure is explained in detail in Chapter 6 of Models-3
Volume 9B: User Manual. The speciation processor is written primarily in SAS®; however, it
includes some FORTRAN programs. ARC/INFO® is not used in the speciation processor.
Speciation Processing in MEPPS
In MEPPS, speciated gaseous emissions can be calculated in either of two units — moles/hour or
Kg/hour. (Particle emissions are always calculated as Kg/hr.) Emissions in moles are required by
the Models-3 CMAQ, but some analyses require that emissions be expressed in mass per unit
time. Emissions in mole and mass units are calculated using the following two generic equations:
chemestjj = £p Em hrjmp * 1000 * rogtotogmp * factorjmp / divisorj;
jmp
(4-95)
where: chemestjj is the gridded emissions estimate for hour i and mechanism species j
(moles/hour)
hrimp is the gridded hourly emission estimate of pollutant p (NOx, SOx, CO, PM,
NH3, or ROG) for hour i and source category m (Kg /hr). The factor of 1000
is used to convert kilograms to grams.
4-84
-------
EPA/600/R-99/030
rogtotogmp is-a ROG to TOG conversion factor (discussed below)
factorjmp is a mole-based split factor to allocate total emissions of pollutant p from
source category m to mechanism species j
divisorjmp is a second conversion factor to allocate total emissions of pollutant p from
source category m to mechanism species j
and
estKgjj = Ep Em hrimp * rogtotogmp * xmassjmp (4-96)
where: estKgy is the gridded emissions estimate for hour i and mechanism species j (Kg/hour)
hrjmp is the gridded hourly emission estimate of pollutant p (NOx, Sox, CO, PM,
NH3, or ROG) for hour i and source category m (Kg /hr)
rogtotogmp is a ROG to TOG conversion factor (discussed below)
xmasSjmp is a mass-based split factor to allocate total emissions of pollutant p from
source category m to mechanism species j
These equations show that emissions for mechanism species are computed by summing the
contributions of pollutants from the different source categories that emit that pollutant. As
indicated previously, the assignment of a pollutant to a specific mechanism species usually
mechanism dependent, although some assignments may be handled the same way in different
mechanisms. In Equations 4-95 and 4-96, the terms factor^,,, divisorjmp, and xmassjmp are
mechanism specific apportioning factors that will be defined further below. Note that, in some
cases a pollutant will contribute to only one mechanism species, but in other cases a pollutant will
contribute to more than on species. For example, emissions of ammonia are always assigned to a
model species named NH3, but NOx is allocated between two mechanism species NO and NO2.
The details of the allocation procedure for both the RADM2 and CB4 mechanisms will be
described further below. First, however, the derivation and use of the rogtotogmp conversion
factor is described.
ROG-to-TOG Adjustment
The ROG (Reactive organic gas) to TOG conversion portion of the speciation process is
selected when the emission inventory is loaded by INPRO (see section 4.2.2). VOC substances
deemed non-reactive (e.g., methane and ethane) are often excluded from inventories, and some
emission measurement techniques do not capture all discrete compounds in an emission stream
(e.g, formaldehyde). The ROG-to-TOG adjustment factor is used to account for those missing
components. It is calculated using Equations 4-97 and 4-98:
mistogm - Ek ntsrdk>m (4.97)
4-85
-------
EPA/600/R-99/030
rogtotogm = 1.0 / mistogm (4_98)
where: mistogj is the sum of mass fractions of the discrete VOC compounds deemed non-
reactive or not included in an emission inventory for source category m (grams
of missing compound per gram of TOG)
ntsrdfcra is the mass fraction of the missing or non-reactive compound k in the
emission stream for source category m
rogtotogm is the ROG to TOG conversion factor
Note that the rogtotogmp adjustment factor applies only to anthropogenic emissions of VOC, and
thus the p subscript used in Equation 4-95 and 4-96 has been dropped here. In equations 4-95
and 4-96, rogtotogm p is set to 1 .0 for all pollutants other than ROG and also set to 1 .0 when
VOC emissions of biogenic origin are being computed.
Carbon Bond 4 (CB4> Speciation Factors
The CB-4 chemical mechanism is widely used in air quality model simulations of ozone
concentrations at urban and regional spatial scales. As described in the introductory part of this
section, the basis of the CB-4 mechanism is that reactivity of organic compounds in the
atmosphere can reasonably be simulated by mechanism species that represent different carbon
bond types. A detailed description of the CB-4 mechanism is contained in Section 8.2.1. The
focus of the discussion here is on how the CB-4 apportioning factors used in Equations 4-95 and
4-96 are calculated. Since speciation of anthropogenic VOC emissions is done differently than for
other pollutants, the discussion is divided into two subsections.
Anthropogenic VOC speciation. For the CB4 mechanism, the apportioning factor divisor^ in
equation 4-95 is set to 1 ,0 for all anthropogenic VOC emission calculations(i.e., it is essentially
not used). The factorjmp term for anthropogenic VOC is set equal to a split; factor term Sj, that is
computed for a specified VOC emission profile 1, which in turn is assigned to a source category m
(i.e., an SCC or ASC). The split factor for each profile and mechanism species is computed as
follows:
sfj, = Ek xmfy / mwk * xnumjk (4-99)
where: sfj j is a molar split factor for CB4 species j and VOC profile 1 (moles of CB4
species/gram of TOG)
xmffc, is a mass fraction of discrete VOC compound k in VOC profile 1 (grams of
' discrete VOC/grams of TOG)
mwk is the molecular weight of discrete VOC compound k (grams of discrete
VOC/mole of discrete VOC)
xnurrijk is the moles of discrete VOC compound k assigned to CB4 species j (moles of
CB4 species/mole of discrete VOC)
4-86
-------
EPA/600/R-99/030
In Equation 4-96, the xmassjmp term apportions mass emissions to CB4 species for a given
pollutant and source category. Analogous to the CB4 molar split factor defined above, the
xmasSjmp term is set equal to a mass split factor that appoitions the mass of a discrete VOC
compound according to the fraction of carbon atoms that is assigned to a mechanism species:
xmassj, = Ek (xmfk>| * enp / cnk (4-100)
where: xmasSj, is the mass split factor for CB4 species j and VOC profile 1 (moles of CB4
species/gram of TOG)
xmfk, is the mass fraction of discrete VOC compound k in VOC profile 1 (grams of
' discrete VOC/grams of TOG)
cnj is the number of carbon atoms in CB4 species j
cnk is the number of carbon atoms in discrete VOC compound k
Again, one VOC profile is assigned to each source category, and thus split factors are generated
for each source category/mechanism species combination.
Other speciation. For pollutants and source categories other than anthropogenic VOC, the terms
factorjmp, divisorjmp, and xmasSjmp are simply assigned numeric values that determine the allocation
of pollutant emissions. These are summarized in Table 4-10.
For all pollutants (except particles) in Table 4-10, the divisor corresponds to the molecular
weight of the pollutant and is used in equation 4-76 to convert mass emissions of the pollutant to
moles of pollutant. The two split factors factor^, and xmasSjmp are used to apportion the
pollutant emissions to the mechanism species. A factor of 1.0 indicates a 1:1 correspondence.
Factors different from one indicate a disaggregation or lumping of pollutant emissions into
individual species.
CO and NH3 are treated explicitly in many air quality models, and therefore emissions of these
compounds are not split into other components. Similarly, particle emissions are treated explicitly
so no lumping or dis-aggregation is necessary for them either.
When SOx emissions are speciated, 97% of the SOx mass is treated as SO2. The remaining 3%
of the SOx mass corresponds to SO4 (or SULF), but is dropped from further consideration. SO2
and SO4 emissions are treated as explicit species.
4-87
-------
EPA/600/R-99/030
Table 4-10. CB4 Split Factors for Pollutants Other Than Anthropogenic VOC
All
CO
CO
1.0
28.0
1.0
All
NH3
NH3
1.0
17.0
1.0
All
AERO
AERO
1.0
1.0
1.0
All
PM10
PM10
1.0
1.0
1.0
All
PM25
PM2_5
1.0
1.0
1.0
All except BIO
NOX
NO
0.62
30.0
0.62
AH except BIO
NOX
NO2
0.05
46.0
0.05
All
SO2
SO2
1.0
64.0
1.0
All
SOX
SO2
0.97
64.0
0.97
AH
SO4
SULF
1.0
96.0
1.0
BIO
NO
NO
1.0
30.0
1.0
BIO
Isoprene
ISOP
1.0
68.12
1.0
BIO
OVOC
NR
0.5
148.0
0.05
BIO
OVOC
OLE
0.5
148.0
0.10
BIO
OVOC
PAR
8.0
148.0
0.85
BIO
TERP
ALD2
1.5
136.23
0.3
BIO
TERP
OLE
0.5
136.23
0.1
BIO
TERP
PAR
6.0
136.23
0.6
BIO
TERP
TERPB
1.0
136.23
1.0
In MEPPS, it is assumed that NOx is composed of 95% NO and 5% NO2 (by mass). However,
actual NOX composition can vary from 89%/l 1% to 95%/5%. In a few cases, a small percentage
(<2%) of NOX emissions is assumed to be nitrous acid (HONO). Since NOx emissions are
typically reported as NO2 mass, it is necessary to normalize the NOx emissions by the molecular
weight of NO2. Hence, the molar split factors (factor^) for NO and NO2 are calculated as
follows:
4-88
-------
EPA/600/R-99/030
sfNO = xmfNO * mwNO / mwN02 (4-101)
sfN02 = xmfN02 * mwN02 / rnwN02
With a NOx composition of 95% NO (as NO2) and 5% NO2 by mass, the split factors for NO
and NO2 become 0.62 and 0.05, respectively. The mass split factors are calculated in the same
manner, and thus have the same values as for the molar split factors.
The speciation of biogenic emissions in CB4 is the same as used in the Regional Oxidant Model
(EPA, 1989), Isoprene is treated as an explicit species and terpene emissions are speciated into
1.0 mole of OLE, 6.0 moles of PAR, and 3.0 moles of ALD2. Terpenes are also assigned to the
model species TERPB for special processing in the aerosol module of the CMAQ. The apparent
double counting of terpene emissions is accounted for in the CMAQ processing however. Finally,
the category OVOC (other VOCs from biogenic sources) is apportioned to 1.0 mole of OLE, 8.5
moles of PAR, and 0.5 mole of NR. Here it assumed that number of carbon atoms in a OVOC
molecule is 10.
RADM2 Speciation Factors
The RADM2 chemical mechanism was developed for and has been used principally in regional air
quality simulations of sulfur dioxide (SO2) and oxides of nitrogen for acid rain assessment
(Walters and Saegar, 1990). A detailed description of the RADM2 mechanism is given hi section
8.2.2. The speciation procedure used for RADM2 is fully described in Walters and Saeger
(1990), and will only be summarized here. This section focuses on how the RADM2 apportioning
factors used in Equations 4-95 and 4-96 are calculated. Again, the discussion is divided into two
subsections, one dealing with the speciation of anthropogenic VOC emissions and one with all
other speciation.
Anthropogenic VOC speciation. The RADM2 mechanism requires that discrete organic VOC
compounds be lumped into 15 mechanism species based on common reactivity and reaction
products. As with the CB4 mechanism, the apportioning factor divisor^ in Equation 4-95 is set
to 1 .0 for all anthropogenic VOC emission calculations, and thus is not used. Also as with the
CB4 mechanism, The factor^ term for anthropogenic VOC is set equal to a split factor term Sj,
that is computed for a specified VOC emission profile 1, which in turn is assigned to a source
category m (i.e., an SCC or ASC). However, the RADM2 split factors are calculated using the
following equation:
sfj,! = Ek xm^j * afacj,k * rfacj,k / mwk (4-103)
where: sfj, is the molar split factor for RADM2 species j and VOC profile 1 (moles of
RADM2 species/gram of TOG)
4-89
-------
EPA/600/R-99/030
xmfk, is the mass fraction of discrete VOC compound k in VOC profile 1 (grams of
' discrete VOC/grams of TOG)
afaCj k is the allocation factor listed iri Table 4-6 for RADM2 species j and discrete
VOC compound k
rfacji)c is the reactivity factor listed in Table 4-6 for RADM2 species j and discrete
VOC compound k
mwk is the molecular weight of discrete VOC compound k (grams of discrete
VOC/mole of discrete VOC)
In the RADM2 speciation process, each discrete VOC compound is assigned to one of the 32
lumped species categories, and that assignment is used to further lumped to one of the 15 species
in the condensed group. See Middleton et al. (1990) for information on how the allocation and
reactivity factors are derived.
In Equation 4-96, the term xmassjmp apportions mass emissions to RADM2 species for a given
pollutant and source category. Analogous to the RADM2 molar split factor defined above, the
term xmasSjmp is set equal to a mass split factor that apportions the mass of a discrete VOC
compound according to the mass fraction of the discrete VOC compound in the emission stream
and the mechanism specific allocation factor:
xmasS;
= £k xmfkl * afack, (4-104)
where: xmasSj, is the mass split factor for RADM2 species j and and VOC profile 1 (moles of
CB4 species/gram of TOG)
xmfy is the mass fraction of discrete VOC compound k in VOC profile / (grams of
' discrete VOC/grams of TOG)
afaCj k is the allocation factor listed in Table 4-6 for RADM2 species j and discrete
VOC compound k
Other speciation. For pollutants and source categories other than anthropogenic VOC, the terms
factor^, divisorjmp, and xmasSjmp are simply assigned numeric variables that control the allocation
of emissions, similar to what is done for the CB4 mechanism. The factors for RADM2 are
summarized in Table 4-11.
Except for the biogenic categories, the contents of Table 4-11 are identical to those iri Table 4-10
for the CB4 mechanism. Thus, the reader is referred to the corresponding CB4 section for
information on those source category/pollutant combinations. The biogenic portions of the two
tables are similar in that isoprene is treated as an explicit species in RADM2, and terpenes and
other VOCs (OVOC) of biogenic origin are apportioned to mechanism species. In RADM2,
terpenes are apportioned entirely to OLI (Middleton et al,, 1990). As with the CB4 mechanism
,terpenes are also assigned to the model species TERPB for special processing in the aerosol
4-90
-------
EPA/600/R-99/030
module of the CMAQ, but the apparent double counting of terpene emissions is accounted for in
the CMAQ processing however. Finally, the assignment of other VOC species to RADM2
species is carried out in a manner analogous to that for CB4: 85% is assumed to correspond to
slow reacting alkanes, 15% to fast reacting olefins, and 5% is assumed to be non-reactive. Thus,
in RADM2 85% of the OVOC mass is assigned to the category containing the slowest reacting
alkanes (category 27 in Table 4-6). The split factor is obtained by multiplying the mass fraction
(.85) by the reactivity factor (0.404) to give a molar split factor of 0.343. The 10% of the mass of
OVOC assumed to be reactive olefins is assigned to the OLI category. Both the mass and molar
split factors for this compound are simply 0.1.
Table 4-11. RADM2 Split Factors for Pollutants Other Than Anthropogenic VOC
All
All
AH
All
All
All except BIO
All except BIO
All
All
AH
BIO
BIO
BIO
BIO
BIO
BIO
CO
NH3
AERO
PM10
PM2_5
NOX
NOX
SO2
SOX
SO4
NO
Isoprene
OVOC
OVOC
TERP
TERP
CO
NH3
AERO
PM10
PM2_5
NO
NO2
SO2
SO2
SULF
NO
ISO
HC3
OLI
OLI
TERPB
1.0
1.0
1.0
1.0
1.0
0.62
0.05
1.0
0.97
1.0
1.0
1.0
0.343
0.1
1.0
1.0
28.0
17.0
1.0
1.0
1.0
30.0
46.0
64.0
64.0
96.0
30.0
68.12
148.0
148.0
136.23
136.23
1.0
1.0
1.0
1.0
1.0
0.62
0.05
1.0
0.97
1.0
1.0
1.0
0.85
0.1
1.0
1.0
4.2.6 Output Processor (OUTPRO)
The results of MEPPS processing must be in a form that is useful to the Models-3 framework and
readily evaluated for the substantive content and quality of the data. The OUTPRO processes
spatially and temporally allocated, speciated emission data files to prepare them for use by the
4-91
-------
EPA/600/R-99/030
Models-3 framework and its components, including CMAQ. In addition, OUTPRO prepares
many standard and user-defined emission summary reports. The processing includes the following
items:
• Merging of temporally allocated point, area, biogenic, and mobile source emission data
files.
• Merging of speciated point, area, biogenic, and mobile source emission data files .
« Merging of point, area, biogenic, and mobile source files into a consolidated two-
dimensional emission file. Merging is optional; and data type files may be maintained and
output separately.
• Preparation of three-dimensional point source emission files for use by the plume rise and
Plume-in-Grid models available in CMAQ. The user may define groups of similar stacks
by specifying percentage tolerance differences in different physical stack properties. In
addition, the user may categorize minor, major, and major elevated point source emission
(MEPSE) stacks by their emission rates and/or stack properties . The classification
criteria are not pre-specified, but are set by the user. Specifically, any combination of
emissions of specified pollutants in tons/day and/or physical stack parameters including
height, diameter, flow rate, or exhaust velocity may be used to define a major or MEPSE
point source. Generally, MEPSE stacks are the largest of point sources, such as electric
utility stacks. Major point sources are usually somewhat smaller, but significant point
sources. Source-specific information is necessary to classify MEPS1E or major point
sources. Smaller sources not classified as MEPSE or major point sources or without
source-specific information are referred to as minor point sources, and are usually
included with the surface area sources by OUTPRO, unless specified otherwise by the
user.
• Conversion of output file format to NetCDF I/O API format. This allows other
components of the Models-3 framework, including the CMAQ and visualization tools, to
use the emission files. The CMAQ accepts the emission files and processes them for input
to CMAQ using the Emission Chemical Input Processor (ECIP). The ECIP is described in
Section 7.
• Preparation of summary reports of the processed emission files, including summaries by
primary emission type (point, area, biogenic, mobile), by geographic area (grid area, grid
cell, state, county, etc.), by source category code and groups of codes (tiers, or
combinations of these items. In addition, the reports rank emission values by amount, type,
geographic area, etc.
4-92
-------
EPA/600/R-99/030
4.3 ModeIs-3 Emission Projection Processor (MEPRO)
It is often necessary to iteratively project emissions and emission control combinations to future
years in order to model future year ambient pollutant concentrations relative to regulatory
standards. Consequently the Models-3 system includes a projection and control system, the
Models-3 Emission Projection (MEPRO) processor to perform the necessary work. The MEPRO
was developed from the Multiple Projection System (MPS). The MPS was originally designed as
a stand-alone software tool to assist the EPA in projecting and tracking the "reasonable further
progress" of regulatory emission reduction programs for criteria pollutants (Monroe et al., 1994).
It may be invoked from Strategy Manager from the Models-3 framework or run separately.
Because MPS was designed as a PC application using the Superbase® programming language,
MEPRO must be run under an emulator (SoftWindows®) if it used in an UNIX operating system.
Currently, MEPRO is more efficiently run on a PC workstation with a Windows NT® operating
system. Figure 4-7 illustrates the relationship of MEPRO with the rest of MEPPS.
Using MEPRO, the user may edit regulatory control factors, control efficiency, rule penetration,
rule effectiveness, as well as "across-the-board" emission adjustments and source category code-
specific emission projections . Projected emission inventory data, and/or inventory data with
revised emission controls and efficiencies are passed to EMPRO (in the
/raw_data/S£MS_JDOM4/Mcommon/ directory) for processing to the NetCDF I/O API format
required for air quality modeling by CMAQ. Mobile source emissions are not projected in
MEPRO. Instead, VMT is projected, and the projected mobile sources emissions are computed in
the mobile source model of EMPRO.
The MEPRO will project emissions of NOX, VOC, and CO for each year from 1991 through
2010, using the 1990 EPA inventory for the base year. The capability to project SO2 and PM
emission data may be provided in the future. Projections are by source category code of
emissions for any area in the United States. The base year and the annual projection factors will
be updated as new data become available. The annual growth factors for VOC, NOX and CO are
taken from look-up tables containing economic growth factors by county for the United States.
The growth factors are based on economic forecasts applied to specific source category codes
by the Economic Growth Analysis System (EGAS) (US EPA, 1995a).
The EGAS system is a PC-based tool which uses a hierarchical three-tiered approach to generate
growth factors. Tier 1 is the National Economic Tier. It includes an economic model by
Regional Economic Models, Inc (REMI) (Treyz et al., 1994) which is primarily based on the
Bureau of Labor Statistics (BLS) American Workforce 1.992-2005 projections. After 2005, the
BLS moderate growth labor force participation rates and the Census Bureau's middle population
projections for the United States are used to forecast the labor force. The second tier is the
Regional Economic Tier, the results of which are overlain on the Tier 1 results. The Regional
Economic Tier contains separate economic models developed by REMI for each non-attainment
area and attainment area of each state. The largest area addressed by one model is a state. The
third tier is the Growth Factor Tier, which contains commercial, residential, industrial, and electric
4-93
-------
EPA/600/R-99/030
utility models, and a VMT growth module. The commercial, residential, and industrial energy
models were developed by the Argonne National Laboratory and were used in the National Acid
Precipitation Assessment Program (Boyd et al., 1990). Electrical Utilities projection was
accomplished using the Neural Network Electric Utility Model (EUMOD). The default economic
projection tables from EGAS are based on U.S. Bureau of Labor Statistics forecasts. Projection
factor tables based on forecasts by the Wharton School of Economics are provided as an option.
[ Unprocessed emission
I inventory data
[VOC,CO.andNOx
growth factor files
MEPRO Output Files
EMPRO
(iaw_data/ffiMS_DOMAlN/caminon)
Giidded,temparally
allocated, spetiated
emission data files
Figure 4-7 MEPRO Relationship to Other MEPPS Components
The method of applying regulatory and growth projection factors to point and area source
emission data to obtain future year daily controlled emission data is summarized by Equation 4-
105, Because MEPRO projects only the VMT data for mobile source emissions, the projection
method is slightly different (Equation 4-106). The projected VMT data are passed to the
EMPRO mobile-source emission model to be converted into projection mobile source emission
data.
PCONE = DCONE * PGF * (1 + AF/100) *
[1-(PCE * PRE * PRP)]/[1-(CE * RE * RP)] (4-105)
where: PCONE is the point or area source future-year daily controlled emissions
4-94
-------
EPA/600/R-99/030
DCONE is the base year daily-controlled emissions
PGF is the projected year growth factor (percent)
AF is the projected year emission adjustment factor (percent)
PCE is the projected year control efficiency
PRE is the projected year rule effectiveness
PRP is the projected year rule penetration
CE is the (adjusted) base year control efficiency
RE is the (adjusted) base year rule effectiveness
RP is the (adjusted) base year rale penetration
For mobile source data:
FPE - B VMT * EF * PGF * (1 + AF/100) (4-106)
where: FPE is the future projected emissions (on-road mobile)
BVMT is the base VMT by vehicle class by facility class by year
EF is the emission factor
PGF is the projected year growth factor
AF is the projected year emission adjustment factor
Additional details for EGAS and MPS are given the references cited above.
4.4 Emission Processing Interface
An accurate characterization of the spatial and temporal variability of emissions at the surface and
aloft is vital for realistic air quality grid modeling. The MEPPS creates separate emission files for
surface area and elevated point sources for a particular domain and time period to be modeled.
Consequently, an interface processor program was needed to efficiently consolidate these various
emissions types into a single, hourly gridded data file for use in grid model simulations.
4.4.1 Overview of Key Features of ECIP
The Emission-Chemistry Interface Processor (ECIP) serves as the key link between the MEPPS
system and the CMAQ Chemistry Transport Model (CCTM), The primary function of ECIP is to
generate an hourly 3-dimensional (3-D) emission data file for the CCTM from the individual
emission file types produced by the MEPPS.
The schematic diagram in Figure 4-8 shows the principal input files used to drive ECIP.
The notable elevated point sources were likely separated into major and MEPSE (Major Elevated
Point Source Emissions) source groups, based on a user-specified emission rate criterion during
MEPPS processing. The MEPSE group contains the largest point source emissions and is
intended to be specially simulated by the CCTM Plume-in-Grid (PING) treatment. However, the
PING treatment is an optional capability of the CCTM and when it is not exercised, the MEPSE
4-95
-------
EPA/600/R-99/030
emission file must be processed by ECIP for the modeling scenario so that the significant
emissions from these point sources are included in the 3-D emission data file with the other major
point source emissions. Thus, this capability to include or omit the MEPSE emissions in ECIP
allows for CCTM runs to be performed without PING or with the PING treatment, respectively,
while not requiring separate runs of the MEPPS emissions system. A companion stack parameter
file for each point source emission file is needed by ECIP for plume rise calculations. In addition,
meteorological data files generated by MCIP are also required to run ECIP in order to simulate
point source plume processes. The ECIP 3-D emission file displayed in Figure 4.6 contains the
surface area anthropogenic and biogenic emissions, and the elevated emissions from major point
sources and the MEPSEs, if appropriate. The 3-D emissions output file from ECIP is ready for
direct input into the CCTM.
AREA
EMIS
MEPSE
PS EMIS
(opt)
STACK
PARMS
MCIP
MET
DATA
ECIP
3-D EMIS
CCTM
Figure 4-8 Schematic Flow Diagram of the Input Files and 3-D Emission Data File for
CCTM generated by ECIP
A specific set of tasks is performed by ECIP for preparation of the 3-D emission file. The 2-D
gridded area emissions are incorporated into the first model layer since they represent near-
surface releases. In contrast, each elevated point source plume must be subjected to plume rise
and initial vertical spread'processes prior to the allocation of the plume emissions into the proper
grid cells.aloft. Another capability of ECIP can be applied in case the CCTM domain is smaller
4-96
-------
1PA/6QO/R-W030
than the emission domain. ECIP can perform "spatial windowing" of the emissions needed for a
particular CCTM domain from a larger MEPPS emissions domain.
The methods used in ECIP to treat the processes impacting point source plumes are described in
subsequent sections. While these approaches employ existing scientific techniques, it is
recognized that other formulations exist. However, the modular design of the Models-3 coding
structure allows for the implementation and application of alternative algorithms to treat a
particular process.
4.4.2 Plume Rise of Point Source Emissions
The rise of a buoyant plume above stack height is strongly dependent upon the initial stack
parameters and atmospheric vertical structure at the time of release. A realistic determination of
the height of final plume rise is important to incorporating the plume emissions into the proper
vertical layer(s) of the model. The initial buoyancy flux (Fb), which is a key parameter in plume
rise formulas, is given by
,, g(Ts-Ta)(Vsd2)
s s (4-107)
where Ts and Ta are the stack exit temperature and ambient temperature at stack top,
respectively. Other notable stack parameters in Equation 4-107 include the plume exit velocity
(Vs) and stack diameter (d), while g is gravity. Clearly, the magnitude of Fb is greatly influenced
by these stack exit parameters. For buoyant plumes, which exist for the vast majority of point
sources, Fb is greater than zero since Ts > Ta. However, if Ts < Ta, then Fb is set to zero.
The vertical profile of wind and temperature also greatly impact plume rise. In particular,
advances in the accuracy of plume rise estimates have resulted from taking into consideration the
vertical variations in the thermal stability and wind structure which frequently display a strong
height dependency in the atmosphere. A key atmospheric stability parameter (s) used to
distinguish the different stability regimes aloft is defined by
s = (g/Ta)(d6/dz) (4-108)
where d8/dz is the vertical potential temperature gradient. The value of s is employed as a
criterion to apply the appropriate stability-dependent plume rise formula. However, for the initial
calculation of plume rise, the convective velocity scale (H.) is used as an indicator variable to
identify the particular stability regime and it is defined by
4-97
-------
EPA/600/R-99/030
H = -
(4-109)
where T, is the 1.5m air temperature, and wT is the surface heat flux covariance. Unstable
conditions are defined for H. > 0.03H»min, while stable conditions exit when H, < -0.03H.min.
Neutral conditions occur for the range of values between these criteria. The default value of
H,m|n has been set to 10"4 m2/s3.
The layer-by-layer approach described by Turner (1985), as originally suggested by Briggs
(1975), has been applied in ECIP for the determination of final plume rise (Ah) based on the
stability of each vertical layer. This practical scheme takes advantage of the vertical resolution of
the hourly, temperature and wind profiles and other 2-D meteorological parameters provided by
the MM5 dynamic mesoscale meteorological model outputs as postprocessed through MCIP.
The method, as outlined by Turner (1985), is an iterative approach which computes plume rise
through each layer. An initial plume rise calculation is performed using meteorological variables
derived at the stack top height with a stability-dependent plume rise formula at this level. Two
methods, linear interpolation or a surface similarity scheme by Byun (1990), are available for
deriving temperature and wind at stack height from the modeled profile values. For tall stacks,
negligible differences were found in the derived values between these methods. If the projected
effective plume rise height (he = h,. + Ah) exceeds the top of the layer containing the stack top, the
amount of rise is limited to the height of the current layer top. Then residual buoyancy flux (FR)
is determined with an inverted form of the plume rise equation just applied. Using FR, the
procedure is repeated to determine the plume rise using the profile parameters for the next higher
layer. This method is applied over successive layers until the buoyancy flux is completely
exhausted. The plume rise ceases at the level where Fb = 0.
A set of analytical plume rise equations presented by Briggs (1984) for different atmospheric
stabilities have been utilized in ECIP for all point sources. The various plume rise equations are
provided below. The final effective plume centerline height (he) is found by adding the computed
plume rise (Ah) to h^.
4.4.2.1 Plume Rise Treatment for Stable Conditions
For stable atmospheric conditions, plume rise is taken from Briggs (1984) equation for bent-over
plumes.
Ah = 2.6— (4-110)
us
4-98
-------
EPA/6QO/R-9W03Q
where u is the wind speed for the layer. Equation 4-88 is applied when H. < -0.03H. for the
initial plume rise computation at the stack, and when s > 10~s for subsequent layers in the current
approach.
4.4.2.2 Plume Rise Treatment for Unstable Conditions
The plume rise formula proposed by Briggs (1984) during unstable conditions is given by
Ah '= 3[Fb/u]3/5H;2/s (4-111)
However, Briggs (1983) suggested a reasonable approximation for H, which permits Equation 4-
111 to be applied in the following form.
Ah = 30[Fb/u]3/s (4-112)
The rationale for the simplification is due to the lack of data for evaluation to justify a more
complicated form. Equation 4-112 is applied with H. > 0.03Hmin for the first plume rise
computation at the stack and for higher layers for s < 10"5.
.x , • '
4.4.2.3 Plume Rise Treatment for Neutral Conditions
The neutral formula developed by Briggs (1984) plume rise equation has been modified into the
following expression.
Ah = l.2[Fb/(uu2)]3/5[hs + 1.3Fb/(uu2)f5 (4-113)
where u. is the surface friction velocity. Equation 4-113 introduces minor differences from the
other form. Equation 4-113 is a more computationally efficient form suggested by Briggs (1983,
communication) since his original neutral formula requires iteration to solve. In the plume rise
algorithm of ECIP, the neutral plume rise equation is also solved during other stability conditions
in the layers aloft. A comparison between the two plume rise estimates is made and the lower
value is selected before proceeding.
4.4.2.4 Special Conditions
For the cases when Fb = 0, plume rise can occur due to momentum provided by the exit velocity
of the plume out of the stack. Therefore, a momentum rise formula (Turner, 1985) has also been
implemented to consider these situations and is given by
4-99
-------
EPA/600/R-99/030
Ahm = 3d v/u (4_i 14)
If Equation 4-114 is selected, no further plume rise computations are performed.
Another situation occurs when the stack is below the PBL height (zf) under unstable conditions
and the condition (hs - z;) < 200 m applies. A treatment for limited plume penetration above the
PBL is determined when this situation is triggered. This condition most often occurs during the
morning period. Since z; is generally growing rapidly, this period is generally brief. To consider
plume penetration of the overlying stable layer, a practical algorithm employed previously by
Byun and Binkowski (1991) has been implemented based on Briggs (1984). If hs is less than 200
m below Zj, then the following equation is solved.
zb = 3.9[Fb/us]1/3 (4-115)
If Zj is greater than z,,,, then the plume top height (z,) is set to z{ and he is defined to be 2/3 z,.
However, plume penetration is permitted when z{ < zj, in Equation 4-115. For this case, the
plume top height is defined to be the height of the top of the next higher layer and the effective
plume height is again computed as 2/3 h,
4.4.3 Method for the Treatment of Initial Vertical Plume Spread
Buoyancy-induced turbulence promotes plume expansion during the rise phase. A widely-used
method from Briggs (1975) designates the vertical thickness of a plume to be equivalent to the
amount of plume rise. With this method, the heights of the top and bottom of the plume are
determined by
h = h + 1.5 Ah
where h, and hb are the heights of the plume top and bottom, respectively. Since the plume
thickness is directly related to the amount of plume rise, this approach leads to rather thick plumes
during the nocturnal period. Experimental plume dimension data suggest more limited vertical
thickness for plumes during the nighttime hours. As an alternative, an empirical form has also
been included. It is based on analyses of observed plume dimensions and vertical temperature
gradients (Gillani, 1996 communication). He found the best-fit empirical result is given by
= Ae(-BdT/dz) 4_U7
4-100
-------
EPA/600/R-99/03O
where the standard deviation of plume depth (az) is a function of the vertical temperature
gradient (dT/dz) at hr Values for A and B are given by 10 and 117, respectively. A minimum
value specified for oz is 3 m. With this method, h, and hb are determined to be ±2.150Z above
and below hs, respectively. This approach provides for smaller plume thicknesses during the
nocturnal period.
4.4.4 Vertical Allocation of Plume Emissions
Rather than dumping the entire emissions of a plume into a single layer, an approach has been
developed to allow for the allocation of plume emissions into multiple layers since a plume can
often span more than one layer. This situation occurs often as more vertical layers are used in the
model since model layers are thinner.
Once h, and hb have been computed for each plume, these values, along with the heights of the
model layer interfaces (Zj.) are employed to determine the fractional amount of plume overlap
across each layer. The method uses the fractional amount of the plume depth residing within a
layer in order to weight the amount of plume emissions Incorporated into a particular layer. If
both ht and hb are contained within a particular layer, all the plume's emissions are allocated into
one layer. As noted above, the number of layers receiving plume emissions is also dependent on
the number of vertical layers in the model. A model configuration with fewer vertical layers
generally implies greater layer thicknesses.
4.4.5 Generation of 3-D Emissions
Once the plume rise and plume partitioning functions have been performed, the emissions from
each point source plume are transferred to the 3-D emission array which also contains the surface
area emissions in the first layer. The 3-D emission array is written at an hourly interval for the
entire simulation period to a data file in a format compatible for use in CCTM simulations.
4.5 Data Requirements
The following items are the data input requirements for operation of MEPPS. Those items that
must be supplied by the user are marked (U). Those items that are fixed internal lookup tables,
provided with Models-3, or that may be generated in MEPPS are marked (I).
• Complete annual (point, area, and mobile-source data by source category code) Regional,
National, or international emission inventories are necessary for regional modeling. The
inventories preferably should be in the ASCII format of the EPA National Emission
Trends (NET) inventories. However, IDA allows import and conversion of any inventory
with known fields formats in ASCII, SAS, or NetCDF format. Currently, the 1985
NAPAP, 1988 National Inventory, 1990 National Interim Inventory, and 1990 National
Emission Trends (NET) inventories for criteria pollutants are included. Limited data for
4-101
-------
EPA/600/R-99/030
southern Canada are available as a part of these inventories (I). Emission inventories for
other areas, years, or pollutants must be supplied by the user (U).
Existing hourly emission data, such as CEM data, may be used directly or substituted for
hourly emission data derived by temporal disaggregation from annual emission inventories.
The hourly data may be imported through the File Converter and IDA. The Models-3
system is includes 1995 CEM data in the SAS data set format provided by the US EPA
Office of Acid Rain (I). Other hourly data must be supplied by the user if desired (U).
Hourly meteorological data for surface temperature and solar radiation, in NetCDF I/O
API format, (from MM5 processed via MCIP) must be available for modeling biogenic or
mobile source emission data. If data files for the appropriate case area available, the user
may select them in MEPPS or when running Study Planner. Otherwise, it is necessary to
first run MM5 and output the meteorology files through MCIP (I).
Chemical speciation profiles matched to source category codes are necessary for the
speciation processor (I). These may be updated with new information.
Temporal allocation profiles, by source category code, must be available to accomplish
temporal allocation of emission data to hourly emission data (I). These may be updated
with new information
Geographic coverages for surrogate spatial allocation (gridding) of emission data. The
user may supply additional coverages: Those supplied with MEPPS include (I):
Political boundaries at the county level for North America
Land-water boundaries and features for North America
County-level population information for North
America (with a gridded surrogate for Canada)
Federal Highway Administration major highway
coverage for the United States
TIGER-LINE detail road coverage for the United
States
Land cover for North America. Currently is at
county-scale for the United States and gridded
at coarser resolution for Canada and Mexico.
Land cover for the United States at one
kilometer resolution is anticipated in 1999
Species emission factors for biogenic emission modeling (I)
Standard Mobile 5a model input information by political area (usually state and/or
county), including vehicle fleet composition data, fuel type use by geographic area,
4-102
-------
EPA/60Q/R-99J030
inspection and maintenance program information,, etc, (U). A template with examples is
provided for the user to edit.
Road silt loading, geographic distribution of paved and unpaved roads, vehicle fleet
composition data, fuel use data, and inspection and maintenance data are necessary as
input to the PARTS mobile particulate model.
Geographic, source category code-specific growth projection factors. These are provided
for use with MEPRO (I).
Source category and/or geographically specific emission control data, and regulatory
factors including control efficiency, rule effectiveness, and rule penetration for use in
MEPRO. These must be supplied by the user because controls are not standard, and in
fact are a key variable in examining different emission scenarios (U).
4.6 Plans for Improvement
Plans for improvement of MEPPS may be divided into long and short-term improvements. The
short-term improvements are those that are anticipated to be in the Models-3 release scheduled
for the summer of 1999.
Short-term Improvements
• The IDA will be improved to consolidate and further automate much of the quality
control processing, format and unit conversion, and data file manipulation. In particular,
format templates of internal formats will be added to assist importing of emission data
files into the system, and quality control for the CEM data will be enhanced.
• Although SAS® has proved to be a useful tool to date, increasing data handling
requirements will likely overwhelm the data handling capabilities of SAS®. Therefore, it is
necessary to convert MEPPS to a fully Models-3 framework compliant system in order to
take full advantage of the or object-oriented data base architecture of the system (eg.,
Orbix®) to more efficiently manage very large amounts of data. The use of the Sparse
Matrix Operator Kernel Emission (SMOKE) system, in conjunction with the Models-3
object-oriented architecture and expansion of existing functionality should substantially
improve performance because relatively inefficient processing sequences will not be
necessary, and it will not be necessary to manipulate all elements of large files for each
operation. More information about SMOKE can be found in Coats et al. (1995). An
initial (but not complete) version of SMOKE is planned for installation by summer 1999.
Long-term Improvements
4-103
-------
EPA/600/R-99/030
* New mobile source emission estimation models are being developed by the U.S. EPA
Office of Mobile Sources. They are scheduled for completion late summer of 1999.
Mobile 6 will replace Mobile 5a and Mobile 5b, and an Off-road Mobile Source Model
will be introduced. These models will be installed in MEPPS when they are available and
resources allow.
« The MEPPS contains the split factor assignments for two common chemical speciation
mechanisms, CB-4 and RADM 2, that may be selected by users. These speciation
mechanisms will also be in SMOKE. There are plans to also install the split factor
information and computational mechanisms necessary to use for the Statewide Air
Pollution Research Center (SAPRC) (Carter, 1988), and eventually, code necessary to
support the more complex Morphecule mechanism being developed at the University of
North Carolina.
• Additional quality control and reporting capabilities will be added to SMOKE, equaling or
surpassing those capabilities in currently in MEPPS.
• The emission data processing system could be enhanced to support nested grid structures.
Currently, there is only limited support for nested grid structures arid no support for
multiscale grid structures. Under the current formulation, EMPRO (including the gridding
processor) must be run consecutively for each grid structure that exists within the nested
grid structure. Repetitive runs are inefficient, and computer resources are poorly utilized
because certain areas in the modeling domain will be processed more than once. In
addition, the gridding processor cannot generate rotated grids, which may be a limitation
for some applications. If an air quality modeling study requires a rotated emission
modeling grid, a knowledgeable ARC/INFO® user must prepare the grid independently
from menu options provided in Model-3 and MEPPS.
• Tools such as NetCDF and the I/O API will evolve to directly accommodate geographic
data references in their structure. This will allow manipulation of geographic data
(gridding, for example) to be accomplished without the use of non-conforming
commercial software tools now in Models-3 system.
4.7 References
Alados, I., I. Foyo-Moreno, and L. Alados-Arboledas, 1996: Photosynthetically Active
Radiation: Measurements and Modeling, AgrL and Forest Meteor,, 78, pp. 121-131.
Beck, L., L.A.. Bravo, B.L. Peer, M.L. Saeger, and Y. Yan, 1994: A Data Attributes Rating
System, Presented at the International Conference on the Emission Inventory: Applications and
Improvement, Raleigh, NC., November 1-3,1994,12 pp.
4-104
-------
EPA/600/R-99/030
Boyd, G.A., E.G. Kokkelenberg, and M.H. Ross, 1990: Sectoral Electricity and Fossil Fuel
Demand in U.S. Manufacturing: Development of the Industrial Regional Activity and Energy
Demand (INRAD) Model,' Argonne National Laboratory, Argonne, IL.
Briggs, G.A., 1975: Plume Rise Predictions. In: Lectures on Air Pollution and Environmental
Impact Analyses, Workshop, Proceedings, Boston, MA, 1975, pp 59-111.
Briggs, G.A., 1983 communication: Plume rise equations used in EPA models. Meteorology
Division, Environmental Sciences Research Lab., Research Triangle Park, NC.
Briggs, G.A., 1984:. Plume Rise and Buoyancy Effects. In: Atmospheric Science and Power
Production, D. R. Anderson, Ed., DOE/TIC-27601 (DE84005177), Technical Information
Center, U.S. DOE, Oak Ridge, TN, 850 pp.
Byun, D.W., 1990: On the. Analytical Solutions of Flux-profile Relationships for the
Atmospheric Surface Layer. J, ofAppl. Meteorol., 29, No. 7, 652-657
Byun, D.W. and F.S. Binkowski, 1991: Sensitivity of RADM to Point Source Emissions
Processing. Seventh AMS/AWMA Joint Conf. on Applications of Air Poll. Meteorol., New
Orleans, LA, Jan. 14-18,1991, Preprints, Amer. Meteorol. Soc., Boston, MA., 1991, pp. 70-73.
Carter, W.P.L., 1988: Documentation for the SAPRC Atmospheric Photochemical Mechanism
Preparation and Emissions Processing Programs for Implementation in Airshed Models, Prepared
for the California Air Resources Board, Report AS-122-32.
Coats, C.J., Jr., 1995: High Performance Algorithms in the Sparse Matrix Operator Kernel
Emissions (SMOKE) Modeling System, Microelectronics Center of North Carolina,
Environmental Systems Division, Research Triangle Park, NC, 6 pp.
E.H. Pechan and Associates, 1994: Industrial SO2 and NOX Tracking System, EPA Contract No.
68-Dl-0146,38pp. • . -. •
Emission Inventory Improvement Program (EIIP), 1996: Biogenic Sources Preferred Methods,
Volume V, prepared for the EIIP Area Sources Committee by the Radian Corporation, Research
Triangle Park, NC, 104 pp.
Fratt, D.B., D.F. Mudgett, and R.A. Walters, 1990: The 1985 NAPAP Emissions Inventory:
Development of Temporal Allocation Factors, EPA~60Q/7-89-01Qd, U.S. Environmental
Protection Agency, Office of Research and Development, Washington, D.C., 209 pp.
Geron, C.D., A.B. Guenther, and T.E. Pierce, 1994: An Improved Model for Estimating
Emissions of Volatile Organic Compounds from Forests in the Eastern United States, J. Geophys.
Res., 99, D6, pp. 12773-12791.
4-105
-------
EPA/6QO/R-99/030
Gery, M.W., G.Z. Whitten, J.P. Killus, and M.C. Dodge, M.C., 1989: A Photochemical Kinetics
Mechanism for Urban and Regional Scale Computer Models, J, Geophys. Res,, 94, D10, 12295-
12956.
Gillani, N.V., 1996: personal communication: Analyses results of plume data.
Guenther, A.B., P. Zimmermann, and M. Wildemuth, 1994: Natural Volatile Organic Compound
Emission Rate Estimates for U.S. Woodland Landscapes, Atmos. Environ., 28,1197-1210.
Kinnee, E., C. Geron, and T. Pierce, 1997: United States Land Use Inventory for Estimating
Ozone Precursor Emissions, Ecol AppL, 1,46-58,
Middleton, P., W.R. Stockwell, and W.P.L. Carter, 1990: Aggregation and Analysis of Volatile
Organic Compound Emissions for Regional Modeling, Atmos. Environ., 24A, pp. 1107-1133.
Modica, L.G., D.R. Dulleba, R.A. Walters, and J.E. Langstaff, 1989: Flexible Regional Emissions
Data System (FREDS) Documentation for the 1985 NAPAP Emissions Inventory, EPA-600/9-
89-047, U.S. Environmental Protection Agency, Research Triangle Park, NC, 574 pp.
Monroe, C.C., T.A. Dean, and W.R. Barnard, 1994: Multiple Projections System: Version 1.0
User's Manual, EPA-60Q/R-94-085, 70 pp.
Moody, T., J.D. Winkler, T. Wilson, and S. Kersteter, 1995: The Development and Improvement
of Temporal Allocation Factor Files, EPA-600/R-95-004,457 pp.
Moran, M.D., M.T. Scholtz, C.F. Slama, A. Dorkalam, A. Taylor, N.S. Ting, D. Davies, C.
Sobkowicz, P.A. Makar, and S. Venkatesh, 1998: An Overview of CEPS1.0: Version 1.0 of the
Canadian Emissions Processing System for Regional-Scale Air Quality Models, in the Proceedings
of a Specialty Conference on Emission Inventory: Planning for the Future, Air and Waste
Management Association, Pittsburgh, VIP-77, Volume 1,95-106.
Morris R.E., M.A. Yocke, T.C. Myers and V. Mirabella, 1992: Overview of the Variable-Grid
Urban Airshed Model (UAM-V), Paper presented to the Air and Waste Mangement Association,
85th Annual Meeting, Kansas City, MO, June 21-26,1992,13 pp.
Ozone Transport Assessment Group, 1997: Emission Inventory Development Report (Draft), 3
Volumes.
Pierce, T.E., B.K. Lamb, and A.R. VanMeter, 1990: Development of a Biogenic Emissions
Inventory System for Regional Scale Air Pollution Models, Paper presented to the Air and
Waste Management Association, 83rd Annual Meeting, Pittsburgh, PA, June 24-29,1990,16 pp.
4-106
-------
EPA/SOO/R-99/030
Pierce, T., C. Heron, L. Bender, R. Dennis, G. Tennyson, and A. Guenther, 1998: The Influence
of Increased Isoprene Emissions on Regional Ozone Modeling, J. Geophys. Res., in press.
Saeger, M., J. Langstaff, R. Walters, L. Modica, D. Zimmerman, D. Fratt, D. Dulleba, R. Ryan, J.
Demmy, W. Tax., D. Sprague, D. Mudgett, and A.S. Werner, 1989: The 1985 NAPAP
Emissions Inventory (Version 2): Development of the Annual Data and Modelers' Tapes, U.S.
Environmental Protection Agency, Office of Research and Development, EPA~600/7-89-012a,
692pp.
Stockwell, W.R., P. Middleton, and J.S. Chang, 1990: The Second Generation Regional Acid
Deposition Model Chemical Mechanism for Regional Air Quality Modeling, J. Geophys. Res., 95
(DID), pp. 16,343-16367.
Treyz, G. et al.: 1994: Model Documentation for the REMI EDFS-14 Forecasting and Simulation
Model, Regional Economic Models, Inc., Amherst, MA.
Turner, D.B., 1985: Proposed Pragmatic Methods for Estimating Plume Rise and Plume
Penetration through Atmospheric Layers. Atmos. Environ., 19, 1215-1218.
U.S. EPA, Office of Mobile Sources, 1985: Size Specific Total Particulate Emission Factors for
Mobile Sources, EPA-46Q/3-85-Q05, 294 pp.
U.S. EPA, Office of Air Quality Planning and Standards, 1988: Air Species Manual: Volume 1,
Volatile Organic Compound Species Profiles, EPA-450/2-88-003a, 492 pp.
U.S. EPA, Office of Research and Development, 1989: Development of the Regional Oxidant
Model Version 2.1, U.S. EPA Technical Report.
U.S. EPA, Office of Mobile Sources, 1991: User's Guide to MOBILE 4.1 (Mobile Source
Emission Factor Model), EPA-AA-TEB-91-Q1, 354 pp.
U.S. EPA, Office of Air Quality Planning and Standards, 1.992: User's Guide for the Urban
Airshed Model, Volume IV: User's Manual for the Emissions Preprocessor System 2.0, EPA-
450/4-90-007D(R),497pp.
U.S. EPA, Office of Air Quality Planning and Standards, 1993: Regional Interim Emission
Inventories 1987-1991), EPA-454/R-93-0212, 54 pp.
U.S. EPA, Office of Air Qaulity Planning and Standards, 1994: Emissions Inventory for the
National Particulate Matter Study, Final Draft, Prepared by E.H. Pechan, Inc., EPA Contract No.
68-D3005, WA No. 0-10, 83 pp.
4-107
-------
EPA/600/R-99/030
U.S. EPA, Office of Air Quality Planning and Standards, 1995: Compilation of Air Pollutant
Emission Factors: Volume II: Mobile Sources, AP-42, Fifth Edition, 293 pp.
U.S. EPA, Office of Research and Development, 1995a: Economic Growth Analysis System:
User's Guide Version 3.0, EPA-600/R-95-132b, 89 pp.
U.S. EPA, Office of Mobile Sources, 1995b: Draft User's Guide to PARTS: A Program for
Calculating Particle Emissions from Motor Vehicles, EPA-AA-AQAB-94-2, 67 pp.
U.S. EPA, Office of Mobile Sources, 1996: User's Guide for MOBILE 5a (Mobile Source
Emission Factor Model), EPA-AA-TEB-92-01,176 pp.
U.S. EPA, Office of Air Quality Planning and Standards, 1997: Air CHIEF, Version 5 (Compact
Disk containing air quality speciation profiles, emission factors, and other emission-related data),
Research Triangle Park, NC.
U.S. EPA, Office of Air Quality Planning and Standards, 1998: National Air Pollutant Emission
Trends Procedures Document, 1900-1996 Projections 1999-2010, EPA-454/R-98-QQ8,712 pp.
U.S. EPA, Office of Research and Development, 1995: Models-3: Volume 7, System
Requirements (Draft) ,
U.S. EPA, Office of Research and Development, 1995: Models-3: System Design (Draft).
Walters, R.A. and M.L. Saeger, 1990: The 1985 NAPAP Emissions Inventory: Development of
Species Allocation Factors, EPA-600/7'-89-01Of, 470 pp.
Wilkinson, J.G., C.F. Loomis, D.E. McNally, R.A. Emigh, and T.W. Tesche, 1994: Technical
Formulation Document: SARMAP/LMOS Emissions Modeling System (EMS-95), Final Report,
Lake Michigan Air Directors Consortium and the California Air Resources Board, AG-90/TS26
andAG-90/TS27,\2Qpp.
Williams, E., A. Guenther, and F. Fehsenfeld, 1992: An Inventory of Nitric Oxide Emissions from
Soils in the United States, J. Gaffes. Res., 97, 7511-7519.
Yienger, J.J. and H. Levy II, 1995: Empirical Model of Global Soil-Biogenic Nox Emissions, J.
Gaffes. Res., 100, No. D6, 11447-11464.
4-108
-------
EPA/600/R-99/030
This chapter is taken from Science Algorithms of (he EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
4-109
-------
EPA/600/R-99/030
Chapter 5
FUNDAMENTALS OF ONE-ATMOSPHERE DYNAMICS
FOR MULTISCALE AIR QUALITY MODELING
Daewon W. Byim
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
_r
This chapter provides essential information needed for the proper use of meteorological data in
air quality modeling systems. Sources of meteorological data are diverse and many difficulties
can arise while linking these with air quality models. To provide an integral view of atmospheric
modeling, a robust and fully compressible governing set of equations for the atmosphere is
introduced. Limitations of several simplifying assumptions on atmospheric dynamics are
presented. Also, concepts of on-line and off-line coupling of meteorological and air quality
models are discussed.
When the input meteorological data are recast with the proposed set of governing equations,
chemical transport models can follow the dynamic and thermodynamic descriptions of the
meteorological data closely. In addition, this chapter introduces a procedure to conserve mixing
ratio of trace species even in the case meteorological data, are not mass consistent. In summary,
it attempts to bridge the information gap between dynamic meteorologists and air quality
modelers by highlighting the implication of using different meteorological coordinates and
dynamic assumptions for air quality simulations.
* On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Daewon W. Byun, MD-80, Research Triangle Park, NC 27711. E-mail:
bdx@hpcc.epa.gov
-------
EPA/600/R-99/030
5.0 FUNDAMENTALS FOR ONE-ATMOSPHERE MODELING FOR MULTISCALE
AIR QUALITY MODELING
To simulate weather and air quality phenomena realistically, adaptation of a one-atmosphere
perspective based mainly on "first principles" description of the atmospheric system (Dennis,
1998) is necessary. The perspective emphasizes that the influence of interactions at different
dynamic scales and among multi-pollutants cannot be ignored. For example, descriptions of
processes critical to producing oxidants, acid and nutrient depositions, and fine particles are too
closely related to treat separately. Proper modeling of these air pollutants requires that the broad
range of temporal and spatial scales of multi-pollutant interactions be considered simultaneously.
Several chapters (Chapters 4, 8, 9, 11 and 16) of this document present the one-atmosphere
modeling perspective related with the multi-pollutant chemical interactions. Another key aspect
of the one-atmosphere perspective is the dynamic description of the atmosphere. This is the
focus of the present chapter.
r»
Air quality modeling should be viewed as an integral part of atmospheric modeling and the
governing equations and computational algorithms should be consistent and compatible.
Previously, many atmospheric models have been built with limited atmospheric dynamics
assumptions. To simplify the model development process, the governing equations were first
simplified to match with the target problems, then computer codes were implemented. This
approach enabled rapid development of models. However, we believe that dynamic assumptions
and choice of coordinates should not precede the computational structure of the modeling
system. To provide the scalability in describing dynamics, a fully compressible governing set of
equations in a generalized coordinate system is preferable. Once the system is based on the fully
compressible governing equations, simpler models can be built readily. The characteristics of
the vertical coordinates and other simplifying assumptions need to be considered as well. For
successful one-atmosphere simulations, it is imperative to have consistent algorithmic linkage
between meteorological and chemical transport models (CTMs).
The present chapter addresses the issue of consistent description of physical processes across
scales in meteorological and air quality modeling systems. It intends to provide appropriate
background information to properly link air quality and meteorological models at a fundamental
level. It deals with dynamic scalability issues, such as hydrostatic and nonhydrostatic modeling
covering wide range of both temporal and spatial scales. Some of the contents are extracts from
Byun (1999a and b) and others are complementary information to them. It includes mass
correction methods, mass conservative temporal interpolation method, and the coupling
paradigm for meteorology and chemical transport models.
5.1 Governing Equations and Approximations for the Atmosphere
In most weather prediction models, temperature and pressure, as well as moisture variables, are
used to represent thermodynamics of the system. Often these thermodynarnic parameters are
represented with the advective form equations hi meteorological models. Most of time, the
5-2
-------
EPA/600/R-99/030
density is diagnosed as a byproduct of the simulation, usually through the ideal gas law. For
multiscale air quality applications where the strict mass conservation is required, prognostic
equations for the thermodynamic variables are preferably expressed in a conservative form
similar to the continuity equation. Recently, Ooyama (1990) has proposed the use of prognostic
equations for entropy and air density in atmospheric simulations by highlighting the
thermodynamic nature of pressure. Entropy is a well-defined state function of the
thermodynamic variables such as pressure, temperature, and density. Therefore, entropy is a
field variable that depends only on the state of the fluid. The principle he uses is the separation
of dynamic and thermodynamic parameters into their primary roles. An inevitable interaction
between dynamics and thermodynamics occurs in the form of the pressure gradient force.
In this section, a set of governing equations for fully compressible atmosphere is presented.
Here, density and entropy are used as the primary thermodynamic variables. For simplicity, a
dry adiabatic atmosphere is considered. Most of the discussions in this section should be
extensible for moist atmosphere if Ooyama' s approach is followed.
5.1.1 Governing Equations in a Generalized Curvilinear Coordinate System
Using tensor notation, the governing set of equations for the dry atmosphere in a generalized
curvilinear coordinate system can be written as:
(5-2)
, = =0 (5.4)
" V7t! ( '
where vj and v4 are contravariant and covariant wind components, respectively, v?k represents
the covariant derivative of contravariant vector, e'u is the Levi-Cevita symbol, Ok is the angular
A . r*j^"
velocity of earth's rotation, FrJ represents frictional forcing terms, ^7 is the Jacobian of
coordinate transformation,
-------
EPA/600/R-99/030
In - pRd ln(--) (5-5)
oo r^oo
where r is temperature, T00 and p^ are temperature and density of the reference atmosphere,
respectively, at pressure pot> - 1000 mb = 105 Pascal, CytS is the specific heat capacity at
constant volume, and Rj is the gas constant for dry air. The g-terms represent sources and sinks
of each conservative property. Although the source term for air density (Qp) should be zero in
an ideal case, it is retained here to capture the possible density error originating from numerical
procedures in meteorological models. It is important to understand how this error term
influences computations of other parameters such as vertical velocity component. Effects of the
error term on trace gas simulation are discussed later.
To close the system we need to utilize the ideal gas law and the thermodynamie relations for
temperature, entropy, pressure gradients, and density. Here, atmospheric pressure is treated as a
thermodynamie variable that is fully defined by the density and entropy of the atmosphere.
Then, pressure gradient terms can be computed using the thermodynamie relations with the
density and entropy (e.g., Batchelor, 1967; Ooyama, 1990; DeMaria, 1995) in terms of the
general vertical coordinate s = x3, as:
(5-6a)
(5-6d)
(5-6e)
where C^ is the specific heat capacity at constant pressure for dry air, and
V. = i
-------
EPA/600/R-99/030
xl = mx
"2
x =my
t =t
-
x = m x
-I *2
y = m x
(5-7a,b)
where »i is the map scale factor, zsrc is the topographic height, and h is the geometric height, and
^AOL represents height above the ground (AGL). In the derivation of Equations 5-7a,b, we
neglected the first-order variations of the map scale factor in x- and y-directions. The
approximation establishes a quasi-orthogonality of the vertical coordinate to the horizontal plane
on the confomal map. The covariant metric tensor, for example, and its determinant are given as
dh dh
dh. .dh.
* *
dh 3k
(5-7c)
(5-7d)
With above relations, one can rewrite the governing momentum equation, Equation 5-1, into the
horizontal and vertical components of the curvilinear coordinates as (Byun, 1999a):
dt
ds
p ds ds
(5-8)
(5-9)
where V, = v'i + v2j, <&(x{,x2,x3,t) = gz represents the geopotential height, Fs is the horizontal
•** •"»
forcing vector, T3 is the vertical tangential basis vector and 1^ is the forcing term in the
momentum equation for x3 direction.
An alternative equation for the Cartesian vertical velocity component is given as:
5-5
-------
EPA/600/R-99/030
S l d(pjswv3)
dt * V m ds
\ p )
where Vf = Ul + Vj = (vl /m)i + (vz /m)j is the horizontal wind vector represented in the
Cartesian coordinate system, w is the vertical velocity component, Js is the Jacobian for vertical
coordinate transformation (Js =
ds
ds
= m2-^), and F3 is forcing term for the w-
component. Note that the contravariant vertical velocity component is related to the Cartesian
vertical velocity as:
^3 ds ds _, „ ( ds\ ds . If, „ /i^s'>l /c m\
v3= — = — + Vz.Vz5 + w — = — + ( — V,«V,4> + w) — L (5-10)
o(? OT \ozy «f ^ V°fe/
where 7=id/dx\ +}dldy\
I lz=con.« J •'h-const
The conservation equations for air density, entropy density, and tracer concentrations are found
to be:
(5-12)
,
m ds '^
-
5.1.2 Assumptions of Atmospheric Dynamics
In this subsection, several popular assumptions used in meteorological models are reviewed.
Here, the dynamic and thermodynamic assumptions are discussed separately because they have
been applied as independent approximations in many atmospheric models. However., readers
should be aware of the inseparable nature of the dynamics and thermodynamics of the
atmosphere. This study focuses on the impact of basic assumptions of the mass conservation
issues and limits of applications in air quality applications,
5.1.2.1 Boussinesq Approximation
The crux of the Boussinesq approximation is that variation in density is important only when it is
combined as a factor with the acceleration of gravity. Originally, it was applied for studying
5-6
-------
EPA/600/R-9ft/030
shallow convection or boundary layer dynamics. Descriptions of the Boussinesq approximation
can be found in the literature (e.g., Arya, 1988; Pielke, 1984; Stall, 1988; and Thunis and
Bornstein, 1996). Although the Boussinesq approximation was originally developed for
incompressible fluid, Dutton and Fichtl (1969) expanded the concept for anelastic deep
convection applications. The results of the approximation lead to the following simplifications
of the equations of motions in the planetary boundary layer (PBL):
(1) Flows can be treated essentially as solenoidal either in velocity field (incompressible) or
in momentum field (anelastic).
(2) The equation of state for the fluctuating component is simplified because the ratio of
fluctuating density to total density can be approximated by the ratio of temperature
fluctuation to the reference temperature.
(3) Molecular properties including diffusivity are constant. These approximations are often
used in air quality modeling to simplify the equations of motions and trace gas
conservation equations. The effect of the Boussinesq approximation on mass continuity
is in the limitation of the flow characteristics, such as incompressible or anelastic. For
multiscale atmospheric studies, this approximation may be used only in the
parameterization of the surface fluxes where the density can be treated essentially
independent of height.
5.1.2.2 Nondivergent Flow Field Assumption
Essentially, this is an assumption about flow characteristics. The basis of this assumption is
purely dynamic although an incompressible assumption leads to the nondivergent flow
approximation. A priori, there is no connection with atmospheric thermodynamics. Therefore
this assumption does not provide any information about the state variables, such as density,
temperature, and pressure fields. For atmospheric applications, this approximation should be
viewed as a result of the incompressible atmosphere assumption linked through the continuity
equation of air. Because of the characteristics that the nondivergent velocity field can be
expressed as the curl of a vector stream function, the field is also called solenoidal. Implications
of this assumption on mass conservation of trace species are presented below in the description
of the incompressible atmosphere assumption. In the generalized coordinate system, the
nondivergent flow field is represented with following equation
-°../- 1.2.3. (5-14),
This is somewhat different from the meteorological nondivergent flow field assumption in the
Cartesian coordinate system, V • V = 0. In the generalized meteorological coordinate system,
Equation 14 can be rewritten as
5-7
-------
EPA/600/R-99/030
(5-14')
The two additional terms represent essentially the effects of the map projection and the gradient
of the vertical Jacobian on the divergence of wind. For a small domain and for a coordinate
whose vertical Jacobian is constant with respect to height (e.g.,
-------
EPA/600/R-99/030
in an Eulerian expression one can show that in order for the atmosphere to be treated as
incompressible, the following conditions must be satisfied:
U2 U2 KL
U (5-19)
c c c
sound sound sound
where U is the phase speed of dominant atmospheric waves. The first condition states that the
movement of air should have a Mach number much smaller than one, say 10%; the second
condition states that energy-carrying waves should not propagate as fast as 10% of the speed of
sound; and the last condition limits the vertical extent of motion to less than about one kilometer.
Similarly, Button and Fichtl (1969) showed that the nondivergent wind relation is generally
applicable up to half a kilometer above ground level through a scale analysis of the continuity
equation. Because of these limitations, a meteorological model with incompressible flow
approximation may not be suitable for multiscale air quality simulations that require descriptions
of atmospheric motions over a wide range of temporal and spatial scales. The second condition:
:
-------
EPA/600/R-99/030
pollutant species as long as there is inconsistency in air density and wind fields. It is not a
surprising statement, but in general this fact has not been actively addressed in air quality
modeling studies. Because the nondivergent relation simply disassociates density and wind
fields, it cannot be used to estimate the mass consistency error in the meteorological data set. On
the other hand, the diagnostic relations applicable for the family of hydrostatic pressure
coordinates based on total air density maintain the consistency in wind and air density fields.
5.1.2.4 Anelastic Atmosphere Assumption
Another popular limiting approximation applied in meteorological modeling is the anelastic
assumption. It simplifies the continuity equation as a diagnostic relation for the momentum
(P0V» where pa is density of reference atmosphere) components as follows:
= 0 (
Ogura and Phillips (1962) and Dutton and Fitchl (1969) found that for deep atmospheric
convection, if the characteristic vertical scale of motions is smaller than the atmospheric scale
height, the anelastic assumption is satisfied. For shallow convection, the Boussinesq
approximation allows us to treat the fluid as incompressible; for deep convection, the
approximate continuity equation requires the momentum field to be solenoidal, and the
expansion or contraction of parcels moving in the vertical is taken into account. Lipps and
Hemler (1982) also performed a scale analysis to propose a set of approximate equations of
motion which are anelastic when the time scale is larger than the inverse of Brunt- VSisala
frequency. The anelastic approximation leads to a divergent wind field, i.e.:
• V. top. + v3 top.) (5-22)
Usually, the right hand side of Equation 5-22 does not vanish. Like the nondivergent wind field
approximation, this assumption provides a diagnostic relation among wind components although
it cannot be used to estimate the inconsistency in the total air density, p, and wind field data
provided by a meteorological model. However, unlike the incompressible atmosphere
assumption, the pressure, temperature and wind fields are not completely independent with the
anelastic assumption. The distribution of pressure must be such that the wind fields predicted by
the momentum equations continue to satisfy the anelastic relation (Gal-Chen and Somerville,
1975). For this reason, most anelastic meteorological models solve for the elliptic equation for
the pressure that is derived from Equation 5-22. Refer to Nance and Durran (1994) for a recent
review on the accuracy of anelastic meteorological modeling systems.
For air quality application, the anelastic approximation still requires use of a full continuity
equation for the perturbation density component. However, most anelastic meteorological
models do not solve for the perturbation air density directly. Therefore, one needs to infer it
from other thermodynamic fields. Also, because the trace gas concentration depends on the total
5-10
-------
EPA/600/R-99/030
density of air, not on just the reference density, it does not simplify the pollutant continuity
equation and the concentration distribution represented in density units cannot be interchanged
with trace species mixing ratio.
5.1.2.5 Hydrostatic Atmosphere Approximation
Perhaps one of the most popular assumptions of atmospheric dynamics used in meteorological
models is the hydrostatic approximation. In the case of a hydrostatic atmosphere, the
acceleration and the fiictional force terms in the z-direction of the earth-tangential Cartesian
coordinates are considered negligible. In earlier days of atmospheric modeling, the hydrostatic
approximation was usually applied with the pressure coordinate. It is well known that the
hydrostatic pressure coordinate applied to a hydrostatic atmosphere has a special property that
simplifies the continuity equation into a solenoidal form and provides a diagnostic equation for
the vertical velocity component. On the other hand, the geometric height coordinate was not
used extensively for studying a hydrostatic atmosphere. Recently, Ooyama (1990) and DeMaria
(1995) have presented a diagnostic vertical velocity equation. Extending this, a general
diagnostic equation for the vertical velocity component can be obtained with the hydrostatic
approximation for a coordinate whose Jacobian is independent of time (Byun, 1999a):
(5-23)
The diagnostic Equation 5-23 can be used to maintain mass consistency in meteorological data
for air quality simulations.
It is worthwhile to note that the hydrostatic or nonhydrostatic atmospheric description, which is a
characterization of the vertical motion, is rather independent from either the
compressible/incompressible atmosphere or the anelastic atmosphere assumption, which is an
approximation of the mass continuity equation. Choices of the assumptions from the two distinct
groups have been used to simplify atmospheric motions, although some of the combinations,
such as compressible but hydrostatic atmosphere, are rarely used in atmospheric studies.
5.2 Choice of Vertical Coordinate System for Air Quality Modeling
Figure 5-1 provides a pedigree of vertical coordinates used in many atmospheric models.
Definitions of the coordinates are provided in Tables 5-1, 5-2 and 5-3. The hierarchy of
classification is: (1) temporal dependency of coordinates, (2) base physical characteristic of
coordinate variables, and (3) method of topography treatments. Application assumptions, such
as hydrostatic or nonhydrostatic atmosphere approximations, are not part of the classification
criteria. Isentropic coordinates are not included here because they are not suitable for the
5-11
-------
EPA/600/R-99/030
regional and urban scale air quality simulation due to their inherent difficulties representing
planetary boundary layer (PBL) structure. For larger-scale simulations, an isentropic coordinate
system can serve as an interesting alternative (Arakawa et al., 1992). Also, there are new
developments of hybrid coordinates that combine isentropic coordinates with other coordinates
to mitigate the problem.
Many different types of vertical coordinates have been used for various meteorological
simulations. For example, the geometric height is used to study boundary layer phenomenon
because of its obvious advantage of relating near surface measurements with modeled results.
Pressure coordinates are natural choices for atmospheric studies because many upper
atmospheric measurements are made on pressure surfaces. Because most radiosonde
measurements are based on hydrostatic pressure, one may prefer use of the pressure coordinate
to study cloud dynamics. This idea of using the most appropriate vertical coordinate for
describing a physical process is referred to as a generic coordinate concept (Byun et al., 1995).
Several different generic coordinates can be used in a CTM for describing different atmospheric
processes while the underlying model structure should be based on a specific coordinate
consistent with the preprocessor meteorological model. The Models-3 Community Multiscale
Air Quality (CMAQ) modeling system allows users to choose a specific coordinate without
having to exchange science process modules (i.e., subroutines with physical parameterizations
for describing atmospheric processes) which are written in their generic coordinates. The
coordinate transformation is performed implicitly through the use of Jacobian within CMAQ.
Byun (1999a) discusses key science issues related to using a particular vertical coordinate for air
quality simulations. They include a governing set of equations for atmospheric dynamics and
thermodynamics, the vertical component of the Jacobian, the form of continuity equation for air,
the height of a model layer (expressed in terms of geopotential height), and other special
characteristics of a vertical coordinate for either hydrostatic or nonhydrostatic atmosphere
applications. Tables 5-1,5-2 and 5-3 summarize properties of the popular time-independent
vertical coordinates (e.g., terrain-influenced height and the reference hydrostatic pressure
coordinate systems) and the time-dependent terrain-influenced coordinate systems, respectively.
Not only the assumptions on atmospheric dynamics, but also the choice of coordinate can affect
the characteristics of atmospheric simulations. For the time-independent vertical coordinates (z,
pm sigma-z, sigma-p^), the vertical Jacobians are also time-independent. Especially with the
hydrostatic assumption, one can obtain a diagnostic equation for the vertical velocity component
, which includes soundwaves together with meteorological signals. Further assumptions on flow
characteristics, such as anelastic approximation, provide a simpler diagnostic equation for the
nonsolenoidal air flow. For such cases, with or without the anelastic approximation, one can
maintain trace species mass conservation in a CTM by using the vertical velocity field estimated
from the diagnostic relation. The scheme works whether the horizontal wind components,
temperature, and density field data are directly provided from a meteorological model or
interpolated from hourly data at the transport time step. This suggests that the mass error can be
estimated with the diagnostic relations that originate from one of the governing equations of the
5-12
-------
EPA/600/R-99/030
preprocessor meteorological models. For a nonhydrostatic atmosphere, which does not have a
special diagnostic relation for time independent coordinates, one should rely on the methods
described below to account for the mass consistency errors.
For time dependent coordinates, the vertical Jacobians sue also time dependent. In general, this
makes it more difficult to derive a diagnostic relation from the continuity equation. However,
for a coordinate with the Jacobian-weighted air density independent of height, a diagnostic
equation for the vertical velocity is available when appropriate top and bottom boundary
conditions are used. Vertical layers defined with this type of vertical coordinate are considered
as material surfaces because mass continuity can be satisfied in a diagnostic fashion. Ak
particles are not expected to cross material surfaces during the advection process. An
atmospheric model based on this type of coordinate may not have a mass consistency problem
except for numerical reasons. The dynamic pressure coordinates based on true air density belong
to this category, which includes such coordinates as n -coordinate, crff-coordinate, and the rj -
coordinate defined in conjunction with o^ (See Table 5-3), A meteorological model using one
of these coordinates will conserve mass within the limits of numerical errors expected from finite
differencing and computer precision. For these coordinates, one can apply the same mass
conservation procedure for both hydrostatic and nonhydrostatic cases. Note that the diagnostic
relations obtained by appropriate choices of coordinates and assumptions on atmospheric
dynamics allow estimation of the density error term in the continuity equation. This information
can be used to reconstruct mass-consistent air density and wind fields that ensure mass
conservation of pollutant species in air quality models.
5-13
-------
EPA/600/R-99/030
Figure 5-1. Pedipee of meteorological vertical coordinates. The encircled T symbol represents
that the associated coordinates are identical when temporal dependency is ignored. Dashed-
circles show that all the coordinates can be used for hydrostatic (HYD) and nonhydrostatic
(NHY) application, regardless of the dynamic characteristics of the variables used to define
vertical coordinates.
5-14
-------
EPA/600/R-99/030
Table 5-1. Summary of Characteristics of the Geometric Height and Pressure Coordinate
Systems. [Note: HYD and NHY stand for hydrostatic and nonhydrostatic applications,
respectively. D() and P( ) symbols are assigned for diagnostic and prognostic formulas with
equation numbers. pg and p are the reference and dynamic (time-dependent) hydrostatic
pressure, respectively.]
Coordinate
geometric height
(z)
reference
hydrostatic
pressure (pg}
%=- P.«te
at
dynamic
hydrostatic
pressure (7t),
Ste , .
— =-p(jr,y,z,Og
DZ
large-scale
hydrostatic
pressure (JJ),
!~ = -p(.x,y,z,f)s
vertical velocity
NHY: w with P(5-9) or P(5-
9')
HYD: D(5-23)
NHY: P(5-9) or P(5-9')
HYD: D(5-23)
v'=* = -|[m%.(j]-|]^
for both NHY & HYD
NHY: P(5-9) or P(5-9') for
perturbation component
HYD:
tf.f.-fat.K\-$»
n v 1 r
Vertical Jacobian
Jfl
constant in
(Jt1,*2,*3,?)
JP. = (P<,STl constant
in (x1,.?,!)
Jn = (pg)~l but,
pJR — 1 / g constant
> x- A I «5 ^3 x
in (x ,x ,x ,t)
J-P = (pgYl
Geopotential height
® = gz
<* * f- %'
*** — w , i
J"** P0
— .-Cf
— .-Cf
5-15
-------
EPA/600/R-99/G30
Table 5-2. Summary of Time Independent Terrain-influenced Height and Reference Hydrostatic
Pressure Coordinate Systems. [Note: D() and P( ) symbols are assigned for diagnostic and
prognostic formulas with equation numbers, respectively, and V f(y represents that the
parameter is not dependent on the argument.]
Coordinate
normalized
geometric
height (
o- . p~~pr
'" />.<**> -ft
Application
hydrostatic
generalized
hydrostatic
non-
hydrostatic
hydrostatic
non-
hydrostatic
Vertical Momentum
D(5-23)
P(5-9)or,
P(5-9')withEq.(5-10)
D(5-23)
P(5-9) or, P(5-9') with Eq.(5-
10) for perturbation
component
Vertical
Jacobian
•* et ~ H- Z#.
*f(Z3,t)
, Pa
"* P,(^g
*:/(0>
Pl-Pa(Z#)-Pr
Geopotential height
5-16
-------
EPA/600/R-99/030
Table 5-3. Summary of Characteristics of the Time Dependent Terrain-influenced Coordinate
Systems.
Coordinate
terrain-
influenced
hydrostatic
pressure
„ K-KT
n -n
terrain-
influenced
large-scale
hydrostatic
pressure (cT.)
step-mountain
eta (T]) with
P0(.0)-PT
step-mountain
ETA ( 7L. )
P
.with a.p,
Applicatio
n
hydrostatic
non-
hydrostatic
non-
hydrostatic
hydrostatic
non-
hydrostatic
non-
hydrostatic
Vertical Momentum
. . VJm
fc a,\
**>*
"a P
-<«.-«*^
P(5-9) or, P(5-9') with Eq.(S-lO) for
perturbation component when p
and p given
<5te* 1 "fr 'Qf 2 **%
»^.Hf^.+
P(5-9) or, P(5-9') with Eq.(5-10)for
perturbation component when f]a ,
p provided
Vertical
Jacobian
^=7
#f(x3)
n
— #
j p
/aV
#
/i-
* ~ PS%<
p*
psn*
Geopotential height
_*
f&f, £?
£p ^ y3 ™- 1 Smmmm^fff *
<*&%$• 0
(fi = iA m. 1 -Jj '
Sfc \a r- X
-*-£**-•
rl p
* ^pntff n
& = & _f ?t' .jrf
^ = 0f-l" P dij
5-17
-------
EPA/600/R-99/030
5.3 Coupling of Meteorology and Air Quality
Characteristics of air quality model simulations are heavily dependent on the quality of the
meteorological data. Meteorological data for air quality can be provided either by diagnostic
models, which analyze observations at surface sites and upper air soundings, or by dynamic
models with or without four-dimensional data assimilation (FDDA). Readers are referred to
Seaman (1999) for a state-of-science review on this topic. In the next section a dynamic
modeling with FDDA approach, which is used in the Models-3 CMAQ system, is described.
5.3.1 Meteorological Data for Air Quality Modeling
Meteorological simulations are applied to drive a CTM for solving atmospheric diffusion
equations for trace species. For regional scale simulations, whose problem size is continental
scale or somewhat smaller, hydrostatic meteorological models have been used, usually with
FDDA. For small scale simulations where topographic effects are important, nonhydrostatic or
compressible atmospheric models are used. These differences in the assumptions used for
atmospheric characterization affect air quality simulations greatly.
Meteorological data can be supplied by running dynamic models prognostically, or with the
archived reanalysis data routinely available as a part of numerical weather forecasting for air
quality simulations (Schulze and Turner, 1998). Currently, GCIP (GEWAX Continental-scale
International Project) provides an archive of the Eta model reanalysis of surface and upper air
fields at 48 km resolution (Leese, 1993; Kalany et al, 1996). Based on the success of GCIP, the
National Center for Environmental Prediction (NCEP), NOAA, is planning to archive regional
reanalysis at a higher resolution. Similarly, the Mesoscale Analysis and Prediction
System/Rapid Update Cycle (MAPS/RUC) of the Forecast Systems Laboratory (FSL), NOAA,
produces accurate and timely analyses and short-term forecasts at 40-60 km resolutions
(Benjamin et al., 1995,1998). The output data are archived at 1-3 hour intervals on 25-34 levels.
These alternative data sources are promising because of the wealth of observation data used for
the reanalysis and the availability of long-term meteorological characterization data suitable for
seasonal or annual assessment studies.
5.3.2 Off-line and On-line Modeling Paradigms
Air quality models are run many times to understand the effects of emissions control strategies
on the pollutant concentrations using the same meteorological data. A non-coupled prognostic
model with FDDA can provide adequate meteorological data needed for such operational use.
This is the so-called off-line mode air quality simulation. However, a successful air quality
simulation requires that the key parameters in meteorological data be consistent. For example, to
ensure the mass conservation of trace species, the density and velocity component should satisfy
the continuity equation accurately. Details of this issue will be discussed below.
If air quality is solved as a part of the meteorology modeling, this data consistency problem
would be much less apparent. Dynamic and thermodynamic descriptions of operational
5-18
-------
EPA/600/R-99/030
meteorological models should be self-consistent, and necessary meteorological parameters are
readily available at the finite time steps needed for the air quality process modules during the
numerical integration. The ultimate goal within atmospheric community is the development of a
fully integrated meteorological-chemical model (Seaman, 1995). This is the so-called on-line
mode air quality simulation. There have been a few successful examples of integrating
meteorology and atmospheric chemistry algorithms into a single computer program (e.g., Vogel
et al., 1995). For certain research purposes, such as studying two-way interactions of radiation
processes, the on-line modeling approach is needed. However, the conventional on-line
modeling approach, where chemistry-transport code is imbedded in one system, exhibits many
operational difficulties. For example, in addition to tremendously increasing the computer
resource requirements, differences in model dynamics and code structures hinder development
and maintenance of a fully coupled meteorological/chemical/emissions modeling system for use
in routine air quality management.
Figure 5-2 shows structures of the on-line and off-line air quality modeling systems,
respectively, commonly used at present time. Table 5-4 compares a few characteristics of on-
line and off-line modeling paradigms. Each method has associated pros and cons. Therefore, in
the future versions of the Models-3 CMAQ system, we intend to realize both on-line and off-line
modes of operations through the use of an advanced input/output (I/O) applications programming
interface (API) (Coats, 1996). Figure 5-3 provides a schematic diagram of the implementation
idea. A proof-of-concept research effort using MM5 and a prototype version of CMAQ is
underway (Xiu et al., 1998). However, to accomplish the goals of multiscale on-line/off-line
modeling with one system, a full adaptation of the one-atmosphere concept is needed.
Development of the fully coupled chemistry-transport model to a meteorological modeling
system requires a fundamental rethinking of the atmospfneric modeling approach in general.
Some of the suggested requirements for a next generation mesoscale meteorological model that
can be used as a host of the on-line/off-line modeling paradigms are;
• Scaleable dynamics and thermodynamics: Use fully compressible form of governing
set of equations and a flexible coordinate system, that can deal with multiscale dynamics.
• Unified governing set of equations: Not only the weather forecasting, dynamics and
thermodynamics research but also the air quality studies should rely on the same general
governing set of equations describing the atmosphere.
• Cell-based mass conservation: As opposed to the simple conservation of domain total
mass, cell-based conservation of the scalar (conserving) quantities is needed. Use of
proper state variables, such as density and entropy, instead of pressure and temperature,
and representation of governing equations in the conservation form rather than in the
advective form are recommended.
5-19
-------
1PA/60Q/R-99/030
State-of-the-art data assimilation method: Not only the surface measurements and
upper air soundings, but also other observation data obtained through the remote sensing
and other in situ means must be included for the data assimilation'.
• Multiscale physics descriptions: It has been known that certain parameterizations of
physical processes, including clouds, used hi present weather forecasting models are
scale dependent. General parameterization schemes capable of dealing with a wide
spectrum of spatial and temporal scales are needed.
The Weather Research & Forecasting (WRF) Modeling System (Dudhia et al, 1998), which is
under development by scientists at NCAR and NOAA, could meet most of the above
requirements. Therefore, the WRF modeling system has the potential to be the future
meteorological model of the Models-3 CMAQ system to provide the multiscale on-line/off-line
air quality modeling capability simultaneously.
Table 5-4. Comparison of On-line and Off-line Modeling Paradigms
Off-line Modeling
On-line Modeling
Dynamic
Consistency
• Need sophisticated interface
processors
* Need careful treatment of
meteorology data in AQM
• Easier to accomplish, but must have
proper governing equations.
«Meteorology data available as
computed
Process Interactions
• No two-way interactions between
meteorology and air quality
Two-way interaction
• Small error in meteorological data
will cause large problem in air
quality simulation (positive feedback
problem).
System
Characteristics
• Systems maintained at different
institutions
• Modular at system level. Different
algorithms can be mixed and tested
• Large and diverse user base
• Community Involvement
* Proprietary ownership
• Expensive in terms of computer
resource need (memory and CPU)
«Unnecessary repeat of
computations for control strategy
study
• Low flexibility
• Limited user base
* Legacy complex code, which
hinders new development
Application
Characteristics
• Easy to test new science concept
• Efficient for emissions control
study
• Good for independent air quality
process study
« Difficult to isolate individual
effects
* Excellent for studying feedback of
met. and air quality
5-20
-------
EPA/6WR-99/03G
Off-line Modeling
Met, Landuse,
& Assimilation
Data
Surrogate &
Emisssions
Data
On-line Modeling
Environmenta
Data
Atmospheric
Modeling System
MET.
Figure 5-2. Current On-line and Off-line Air Quality Modeling Paradigms
5-21
-------
EPA/600/R-99/030
Met., Landuse,
& Assimilation
Data
Meteorology
Modeling System
I/O API
A
inlilll
Emisssions
Data
Chemistry
Data
t
CPU/Network
Boundary
I/O API
CPU/Network
Boundary
Unified Governing
Set of Equations
+
Consistent Numerics
real-time I/O API
feedback on : On-line model
file-based I/O API : Off-line model
feedback off
Figure 5-3. Proposed One-atmosphere Air Quality Modeling Paradigms, Double arrowhead
lines represent possibility of two-way coupling. The coupling of independent modeling
components is accomplished through the I/O API linking the cooperating executables.
5.4 Mass Conservation
For air quality simulations, mass conservation is the most important physical constraint. This is
because it is unrealistic to have injection of primary pollutant mass through any other means than
a real source emission process, and also because the little perturbations in the mass of both
primary and secondary pollutants will jeopardize the correct simulation of reactions among trace
species. Therefore, conserving mass of a passive primary trace species is a necessary property of
an air quality model.
5.4.1 Mass Consistency in Meteorological Data
The mam objective of many meteorological models has been to predict synoptic or mesoscale
weather phenomena. Therefore, major design considerations are focused on such issues
important for energy conservation, resolving a spectrum of different wavelengths, and energy
cascade under nonlinear wave-wave interactions. Conservation of mass is not usually
emphasized as the other constraints listed. Also, the predictive quantities are generally
thermodynamic parameters, such as temperature and pressure. The conservation equation for air
density is rarely solved directly in meteorological models because of little operational use of air
density for weather forecasting and no direct measurements to compare. Usually it is estimated
from the equation for the state of ideal gas or from a hydrostatic relation when hydrostatic
assumptions are made. Even the predictive equations for the moisture variables are often written
5-22
-------
in an advective form rather than a continuity equation form. On the other hand, air quality
simulation relies mostly on the continuity equation. The success of a simulation is heavily
dependent on the consistency of density and wind data (i.e., how well they satisfy the continuity
equation).
The mass inconsistency in density and wind fields from a meteorological model is most likely
caused by one or more of the following reasons:
1. Many meteorological models do not use the proposed ideal set of governing equations. A
continuity equation for air is not used as one of the prognostic equations and air density is
usually a diagnostic parameter in meteorological models.
2. The prognostic equation for temperature is often used to represent thermodynamics of the
atmosphere. It is well known that temperature is not a good conserving parameter.
3. Removal of hydrometeors due to condensation or sublimation may subtract and add mass
and heat to the moist atmosphere making the system nonadiabatic (thermodynamically
irreversible) and not mass-conserving.
4. Numerical schemes used in meteorological models are designed to conserve energy,
entropy, rather than the mass of air.
5. The FDD A and overall assimilation process, including the effects of Newtonian forcing
terms in the momentum and temperature equations, may cause inadvertent modification
of the energy balance and subsequent perturbation of air density resulting in mass
conservation problems.
6. Heat, moisture and momentum flux exchanges at the surface-atmosphere interface may
affect the air density distribution. Usually this effect is not significant as it is often
neglected with the Boussinesq approximation.
7. Flux exchanges at the nesting boundaries for nested runs affect mass balance.
8. Energy and mass balance characteristics of cloud modules used influence air and
moisture density fields.
9. Data output time steps are too large to capture the dynamic variations in the
meteorological models. If temporally averaged data are provided from the meteorology
model this problem can be minimized (Scamarock, 1998).
5.4.2 Techniques for Mass Conservation in Air Quality Models
As presented in Byun (1999b), species mixing ratios (c(. /p) is a useful conserved quantity for
photochemical Eulerian air quality modeling, in particular. In limited area atmospheric modeling
like an urban or a regional scale simulation, the total air mass in the simulation domain is subject
5-23
-------
EPA/600/R-99/030
to the inflow conditions determined by large synoptic scale weather systems. In this situation,
the conservation of pollutant mass in the modeling domain can be difficult unless the density and
wind fields are perfectly mass consistent. When the mass inconsistency in the meteorological
fields is expected, the conservation equation for mixing ratio must be used as a necessary
condition to ensure exact conservation of pollutant mass. This is accomplished by replacing the
right-hand-side term of Equation 5-11 with Qc = c/—^-. Then, the conservation equation for
P
pollutant species is rewritten as:
This adjustment alone is not sufficient to conserve pollutant mass when the density error term is
not small. Equation 5-24 shows that the correction term has the same form as a first-order
chemical reaction whose reaction rate is determined by the normalized air density error term.
Table 7-5 in Chapter 7 in this document summarizes correction methods discussed in Byun
(1999b). Among these, the method based on the two-step procedure (i.e., solving the Ihs of
Equation 5-24 first followed by the mass correction step solving for rhs) is expected to be the
most accurate:
T
(5-25)
where superscripts cor, int, and T represent corrected, transported (adveeted), and interpolated
quantities, respectively. It should be noted that Js in Equation 5-25 must not be canceled out
even for a coordinate with time independent Jf because the spatial variation of the Jacobian must
be taken into account for the numerical advection. In the event the total air mass in the
computational domain fluctuates, this correction procedure would affect air quality predictions.
In general, the air quality prediction can be as good as the density prediction of the
meteorological model. However, considering the nonlinear interactions of trace species in the
chemical production/loss calculations, one could expect serious effects on air quality simulations
when the quality of meteorology data is in doubt.
Byun (1999b) also provides an alternative method to deal with the mass inconsistency in
meteorological data through the modification of wind field, while keeping the density field
intact, before solving the species conservation equation. Assuming a modified wind field exists
that eliminates the source term hi the continuity equation for air, the relationship between the
original and modified wind components is given as:
(5-26a)
5-24
-------
EPA/600/R-99/030
(5_26b)
where A- is the Lagrangian multiplier to be determined and a/ and % are the weights for the
horizontal and vertical wind components. /L must satisfy the Poisson equation
•V.A
with the associated boundary conditions:
A.= 0 for flow-through boundaries; and
dkl ds = Ofor impenetrable boundaries (i.e., at the topographic surface).
The modified wind components are subject to the same top and bottom boundary conditions
imposed by the given coordinate system and dynamic assumptions.
The main difference in the two proposed correction methods, correction after advection versus
correction of wind fields before advection, is practically philosophical. Should we process a
CTM using meteorological data as supplied, then correct possible errors in the species
concentrations, or should we modify the velocity field to be mass consistent before the
computation of trace gas concentrations in the CTM? The answer to this question lies in whether
the air quality modeling need is satisfied with simple mixing ratio conservation with the
adjustment process or not. In case the source-receptor relation is important, it is preferable to
maintain the linearity of transport process using the mass-consistent wind components, which
have been modified at the expense of truthfulness of meteorological fields. In practice, a
combination of both methods is needed. The mass consistency error in the meteorological data
must be corrected before air quality simulations with the wind-field adjustment method and the
mixing ratio correction method Equation 5-25 should be applied to compensate the numerical
differences in advection processes between meteorological and air quality models.
5.4.3 Temporal Interpolation of Meteorological Data
Byun (1999b) discusses a mass-conservative temporal interpolation method to complement the
mass inconsistency correction. Temporal interpolations of density and velocity data are often
required in a CTM because the meteorological model output has a coarser temporal resolution
than the transport time step (which is usually the synchronization time step for a CTM using a
fractional time-step method).
The Jacobian and density at a time ta=(l — a)tn + (Xtn+l between the two consecutive output time
steps, tn and fn+1, are interpolated with linearity assumed:
(/,)„=(!-«)(/,)„ +a(/,)a+1 (5-28a)
5-25
-------
EPA/600/R-99/030
(pjs)a = (1 - a)(A/x),, + aCMU (5-28b)
*
where 0 ^ a < 1. It is obvious that the functional form of the Jacobian (which depends on a
vertical coordinate) changes the characteristic of density interpolation. The premise used here is
that the Jacobian is a fundamental quantity that determines the coordinate system. When the
Jacobian is interpolated to define the vertical layers through linear interpolation, all other
components involved in the mass conservation equation need to be interpolated accordingly.
Wind components multiplied with the Jacobian- weighted density are interpolated linearly:
). = 0 - «)(ATOn + alpj )fl+1 (5-29a)
3)« = (\-a)(pJsv\ +a(pJ/)M (5-29b)
and interpolated wind components are derived with:
(V ) = * (5-30a)
' "
However, the proposed scheme, Equation 5-28b, has a problem in such cases where the finite
difference value of (pJs) cannot approximate the linear interpolation of the time rate change of
the quantity, ^. , adequately. Usually, this tendency term is not available with the
at
meteorological data. However, when the tendency is available or can be estimated with the
diagnostic relations for certain meteorological coordinate systems, a different interpolation rale
must be sought. Because the tendency term, not (pJs) itself, is a component of the continuity
equation, linear interpolation of the tendency may be more appropriate. Then, (p/,) at the
interpolation time step must be estimated in such a way that satisfies the continuity as well as the
tendency term (Byun 1999b).
5.5 Conclusion
In this chapter I attempted to bridge the information gap between dynamic meteorologists and air
quality modelers and to promote the proper use of meteorological information in air quality
modeling studies. It highlights the importance of dynamic consistency in meteorological and air
quality modeling systems. The effects of the common assumptions used for the atmospheric
study on the mass conservation for trace species have been reviewed. Although meteorological
data provided by operational meteorological models are usually self-consistent, air quality
modelers need to evaluate the data for exact consistency before they can be used in air quality
simulation. Minor adjustment of the meteorological data may be needed to assure mass
conservation of trace gas species in CTMs, Also, characteristics of vertical coordinates have
5-26
-------
EPAJ600JR-9&030
been discussed. Certain coordinates provide diagnostic relations that can be used to maintain
mass consistency in meteorological fields. When meteorological data are needed at sub-output
time steps within CTM, the interpolation of the data should be done in such a way that the mass
conservation and consistency in the thermodynamic variables are not compromised.
In addition, the on-line and off-line modeling concepts are discussed to provide design guidance
for fully integrated meteorological-chemical models. To realize the noble goal of implementing
the one-atmosphere modeling system, both the multi-pollutant chemistry and multiscale physics
capability in meteorology are needed. The following are the features that make the CMAQ air
quality model a suitable key component of an one-atmosphere modeling system:
• Flexible chemistry representations through a mechanism reader;
• Comprehensive list of atmospheric processes that are implemented;
* Modular coding structure and versatile data handling method;
• Capability to handle multiscale dynamics and thermodynamics;
• Fully compressible governing set of equations in generalized coordinates; and
• Robust mixing ratio conservation scheme, even with mass inconsistent meteorology data.
At present, we are encouraged by the efforts of the WRF meteorology model development
groups that focus on issues such as choice of coordinates, grid staggering method, state variables
in the governing equations (e.g., fully compressible), conservation properties (mass and energy)
both in the model equations and numerics, modularity of code, data communication methods,
and coding language. This entails continuous exchange of ideas between the Models-3 CMAQ
and WRF modeling groups.
To achieve the true one-atmosphere modeling system, we must address multi-pollutant and
multiscale processes that are typically broader than any one group (or institution) has expertise to
address. The need is well summarized in Dennis (1998):
Considering additional needs for emerging environmental problems such as
coastal eutropbication and ecological damage issues related with cross-media
purview, encompassing the one-atmosphere scope is needed. This means we have
to work with a more complete one-atmosphere description to facilitate
interactions within it as efficiently as broadly as possible. One potential answer is
to foster a community modeling perspective and model system framework that is
supported and used by a critical fraction of the scientific community.
5-27
-------
EPA/600/R-99/030
5.6 References
Arakawa, A», C. R. Mechso, and C. S» Konor, 1992: An isentropic vertical coordinate model:
Design and application to atmospheric frontogenesis studies. Meteor. Atmos. Phys. 50, 31-45.
Arya, S, Pal, 1988: Introduction to Micrometeorology, Academic Press, Inc., 307 pp.
Batchelor, G, K., 1967: An Introduction to Fluid Mechanics, Cambridge University Press, 615
PP.
Benjamin, S.G., D. Kim, and T.W. Schlatter, 1995: The Rapid Update Cycle: A new mesoscale
assimilation system in hybrid-theta-sigma coordinates at the National Meteorological Center.
Second International Symposium on Assimilation of Observations in Meteorology and
Oceanography, Tokyo, Japan, 13-17 March, 337-342.
Benjamin, S.G., J.M. Brown, KJ. Brunge, B.E. Schwarts, T.G. Smirnova, and T.L. Smith, 1998:
The operational RUC-2. 16th Conference on Weather Analysis and Forecasting, Phoenix, AZ,
Amer. Meteor. Soc., 249-252.
Bishop, R. L. and S. I. Goldberg, 1968: Tensor Analysis on Manifolds. Dover Publications, Inc.
New York.
Byun, D, W., 1999a: Dynamically consistent formulations in meteorological and air quality
models for multiscale atmospheric applications: I. Governing equations in a generalized
coordinate system. J. Atmos. Sci., (in print)
Byun, D. W., 1999b: Dynamically consistent formulations in meteorological and air quality
models for multiscale atmospheric applications: II. Mass conservation issues. J. Atmos. Sci., (in
print)
Byun D. W., A.. Hanna, C. J. Coats, and D. Hwang, 1995a: Models-3 Air Quality Model
Prototype Science and Computational Concept Development. Trans. TR-24 Regional
Photochemical Measurement and Modeling Studies, San Diego, CA, of Air & Waste
Management Association, 197-212.
Coats, C. J., cited 1996: The EDSS/Models-3 I/O Applications Programming Interface. MCNC
Environmental Programs, Research Triangle Park, NC. [Available on-line from
http://www.iceis.mcnc.Org/EDSS/ioapi/H.AA.html.]
Defiise, P., 1964: Tensor Calculus in Atmospheric Mechanics. Advances in Geophysics 10, 261-
315.
DeMaria, M., 1995: Evaluation of a hydrostatic, height-coordinate formulation of the primitive
equations for atmospheric modeling. Man. Wea. Rev., 123, 3576-3589,
5-28
-------
EPA/600/R-99/030
Dennis, R.L., 1998: The environmental protection agency's third generation air quality modeling
system: an overall perspective. Proceedings of the American Meteorological Society 78th
Annual Meeting, Phoenix, AZ, Jan. 11-16, 1998. 255-258.
Dudhia, J., D. Gill, J. Klemp, and W. Skamarock, 1998: WRF: Cuurrent status of model
development and plans for the future. Preprints of the Eighth PSU/NCAR Mesoscale Model
User's Workshop. Boulder, Colorado, 15-16 June, 1998.
Dutton, J. A., 1976: The Ceaseless Wind, an Introduction to the Theory of Atmospheric Motion.
McGraw-Hill, 579 pp.
Dutton, J. A., and G. H. Fichtl, 1969: Approximate equations of motion for gases and liquids. J.
Atmos. Set., 26, 241-254.
Gal-Chen, T., and R. C. J. Somerville, 1975: On the use of coordinate transformations for the
solution of the Navier-Stokes equations. J. Comput. Phys., 17, 209-228.
Kalany, E., and Co-authors, 1996: The NCEP/NCAR 40-year Reanalysis Project. Bull Amer.
Meteor. Soc., 77, 437-471.
Leese, J. A., 1993: Implementation Plan for the GEWEX Continental-Scale International Project
(GCIP). Int. GEWEX Project Office #6, 148 pp. [Available from IGPO, 1100 Wayne Ave., Suite
1225, Silver Springs, MD 20910].
Lipps, F. B., and R. S. Hemler, 1982: A scale analysis of deep moist convection and some related
numerical calculations. J. Atmos. ScL, 39,2192-2210.
Nance, L. B., and D. R. Durran, 1994: A comparison of the accuracy of three anelastic systems
and the pseudo-incompressible system. J. Atmos. Sci., 51,3549-3565.
Ogura, Y., and N. W. Phillips, 1962: Scale analysis of deep and shallow convection in the
atmosphere. J. Atmos. Sci., 19,173-179.
Ooyama, K. V., 1990: A thermodynamie foundation for modeling the moist atmosphere. J.
Atmos. Sci., 47,2580-2593.
Pielke, R. A., 1984: Mesoscale Meteorological Modeling. Academic Press, 612 pp.
Scamarock, W. 1998: Personal communication.
Schulze, R. H., and D. B. Turner, 1998: Potential use of NOAA-archived meteorological
observations to improve air dispersion model performance. EM, March 1998,12-21.
Seaman, N. L, 1995: Status of meteorological pre-processors for air-quality modeling.
International Conf on Paniculate Matter, Pittsburgh, PA, Air & Waste Management
Association, 639-650.
5-29
-------
EPA/600/R-99/030
Seaman, N.L., 1999: Meteorological modeling for air-quality assessments. (Submitted to Atmos.
Environ., 32, 87pp.)
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic
Publishers. 666 pp.
Thunis, P. and R. Bornstein, 1996: Hierarchy of mesoscale flow assumptions and equations. J.
Atmos. Set., 53,380-397.
Vogel, B., F. Fiedler, and H. Vogel, 1995: Influence of topography and biogenic volatile organic
compounds emission hi the state of Baden-Wurttemberg on ozone concentrations during
episodes of high air temperatures. /. Geophys. Res., 100, 22,907-22,928.
Xiu, A., R. Mathur, C. Coats, and K. Alapaty, 1998: On the development of an air quality
modeling system with integrated meteorology, chemistry, and emissions. Proceedings of the
International Symposium on Measurement of Toxic and Related Air Pollutants, Research
Triangle Park, North Carolina, 1-3 September, 1998. 144-152.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
MultiscaleAir Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K, S.
Ching, 1999.
5-30
-------
EPA/600/R-99/030
Appendix 5 A. Tensor Primer and Derivation of the Continuity Equation in a Generalized
Curvilinear Coordinate System
The Appendix 5 A summarizes essential information needed for understanding the governing
equations represented in tensor form. It includes tensor primer and derivation of the continuity
equation in a generalized coordinate system. Readers are referred to classic references such as
Dutton (1976), Defrise (1964), and Pielke (1984) for the details.
5A.1 Tensor Analysis in a Curvilinear Coordinate System
Cartesian coordinates are those curvilinear systems in which the positions of fluid elements are
determined by their distance from intersecting planes. Although the Cartesian coordinates with
orthogonal intersecting planes are specifically called rectangular, the adjective rectangular is
often dropped. To represent formulations governing atmospheric phenomena in a coordinate
system other than a rectangular Cartesian one, a tensor representation is often used. This
generally involves determination of the unit vectors in the new system, determination of the
components of a tensor with respect to theses unit vectors, and determination of the differential
derivatives (e.g., divergence, curl, and gradient) of a tensor. All these quantities depend
explicitly on the form of the new coordinate system and it is always convenient to express these
quantities in a rectangular Cartesian coordinate system for comparison purposes.
In atmospheric modeling one is frequently led to adopt a curvilinear coordinate system other than
the Cartesian coordinates depending on the problem under consideration. A general curvilinear
coordinate system can be defined relative to a Cartesian system x = (x1 »x2, x3) represented by
three families of curved surfaces
x? = y,(jcl,jc2,jc3,/), /- 1,2,3 (5A-1).
Here, the symbols with carat (A) are used to denote a transformed curvilinear system. In vector
form, it is given as:
x=tfx,0. (5A-2)
When the curvilinear system i^is at rest relative to the rectangular Cartesian system, i.e.,
independent of time, x1 = yf,(xl,x2,x3), then the system is called a Euclidean system. Here we
assume that components of vector x, (Jc'.jc2,*3), are three independent, single-valued, and
A
differentiable scalar point functions such that to every pobit of some region 9t of three-
dimensional Euclidean space, there is a corresponding unique triple of values (x1, x2, x3) in the
Cartesian space 91. In other words, the function iff prescribes one and only one value of A: and is
such that the three coordinates are independent of each other. Also, we assume continuity of the
function y. Then the new coordinates x are called curvilinear and the surfaces x1 = ifft =
constant, x2= y/2 = constant, *3= \f/3 = constant are called coordinate surfaces. The curvilinear
coordinates (Jc1, Je2, jc3) should be independent, single-valued, and differentiable. As shown in
5-31
-------
EPA/600/R-99/030
Figure 5A-1, the vector OF pointing a parcel of air enclosed by the boundary dQ can be
represented either in Cartesian or Euclidean curvilinear coordinate systems.
Figure 5A-1. Coordinates of the Cartesian and Curvilinear Coordinates. 9t and
Cartesian and Euclidean spaces, respectively.
represent
Note that the transformation involves with not only the spatial variables but also time as an
independent variable. We need a tensor calculus in the four variables of space-time with regard
to the coordinate transformations. Defrise (1964) used the term 'world tensor' to distinguish it
from the time independent Euclidean tensor.
5A.2 Basis Vectors
In a rectangular coordinate system, directions of the basis vectors are constant in space.
However, in a general curvilinear coordinates, directions of the basis vectors will vary from point
to point and no one set of directions can be regarded as more natural than any other for the
directions of base vectors to define the local base vectors. Usually, an upper index denotes
contravariant, and a lower index denotes covariant tensors, respectively.
With the coordinates defined by Equation 5A-2, the chain rule provides the two expansions:
ax1
(5A-3)
5-32
-------
EPA/600/R-9W030
dxl = ~dxi = (Vi') * dx (5A-4)
dx]
Here, the Einstein summation convention (i.e., the repeated indices on two quantities that are
multiplied by each other are summed over) has been implied. The symbol (•) represents the
inner product. An inner product of two vectors yields a scalar that is invariant of the coordinate
system. An inner product of two tensors results in contraction of the rank in the resulting tensor.
From Equations 5A-3 and 5A-4, we can form two distinct: sets of basis vectors. One is the
tangential vectors:
A. CsA. s ft , _»,
*- (5A-5)
that reveals the variation of the position vector as it traces out a curve in which jc-' varies and the
other two coordinates are constant. Hence ij is tangent to the curve along which only x' varies.
The other set of basis vectors is the normal vectors of the surfaces where jc' = constant:
ri = Vjc' (5A-6)
While there could be many choices, the tangential ( ^.) and normal ( fj) vectors are considered as
a natural choice for the local basis vectors for the curvilinear coordinate system. Using
Equations 5A-5 and 5A-6, one can show that:
(5A-7)
where Sf is the Kronecker delta and the symbol ® represents the outer product. Outer product
of two tensors with rank rj and r^ yields a tensor with rank >
A curvilinear system is not orthogonal when not all the off-diagonal components of 77' ® r\j and
f,- ® $j vanish. The orthogonal curvilinear coordinate system is often used for interesting
engineering problems that can be described with simple geometric orthogonal coordinates, such
as spherical, cylindrical coordinate systems. Usually, meteorological coordinates are not
orthogonal and therefore, the vector calculus specific for the orthogonal curvilinear coordinates
must not be used.
5A.3 Distance and Metric Tensor in a Curvilinear Coordinate System
The differential element of distance ds can be expressed in terms of the curvilinear coordinates
as:
*2* (5A-8)
5-33
-------
EPA/600/R-99/030
~ OX OX
» „, A ,_ . _.
*>®T> (5A-9)
Because of its obvious role in the measurement of distance, the quantity y'* is called the metric
tensor. It is a symmetric tensor. As such, it has an inverse matrix y '* , which will satisfy
following condition:
C/JC
e oar
y *• i * i»» \.
(5 A- 10)
The Levi-Cevita symbol e used in Equation 5-1 is an antisymmetric tensor defined as
{0 if i— }•, j = k, or i = k
I if i,j,k are an even permutation of 1,2,3 (5 A- 12)
—1 if i,j,k are an odd permutation of 1,2,3
Using the Levi-Cevita symbol, the cross vector product A = B x C can be written as
One of the uses of the metric tensor and its inverse is for converting a co variant tensor to a
contravariant tensor, and vice versa. Another important usage of the metric tensor is the
estimation of the Jacobian determinant of the transformation, which is defined as:
where J =
(5A-13)
. Note that the Jacobian matrix and the metric tensor are related
as:
ru={J*}T{J*} (5A-14)
A necessary and sufficient condition that (x',x2,z3) be orthogonal at every point in 9t is that the
components of the metric tensor vanish for i &j.
5A.4 Covariant Tensor and Contravariant Tensor
In this section, the covariant and contravariant tensor concepts are presented using a vector,
which is a simple form of a tensor (i.e., a tensor of rank one). A distance in a Euclidean space
can be represented in two corresponding sets of tangential basis vectors:
5-34
-------
EPA/600/R-99/030
d$ = dxJtj=dxlTl (5 A- 15)
where
*y=*, (5A-16)
Any vector that transforms similarly to the tangential basis vector %"j is called as a covariant
vector. On the other hand, when a vector transforms like the local normal basis vectors fjf, we
call it a contravariant vector:
v*=t< (5A-17)
where v; is the components of V with respect to the normal base vectors.
Since a vector A is invariant between coordinate systems, we can express it using either
contravariant components (i.e., with the tangential basis vectors) or covariant components (i.e.,
with the normal basis vectors):
A = Aji]j=Attl (5 A- 18)
Using Equation 5 A-7, one can readily find the covariant and contravariant components with:
Aj=A.»fj (5A-19a)
A''=A»f" (5A-19b)
5A.5 Derivatives, Total Derivative, and Divergence in Euclidean Coordinate
Covariant derivative of a contravariant vector is defined as:
' ' d2l
. - • - x
v> = + r1 VJ T' = —
* * * ' *
Similarly, covariant derivative of a covariant vector is defined as:
A. .
The Christoffel symbol /^J is not a tensor but it is an important quantity relating the derivatives
in the curvilinear coordinate system with those in the original Cartesian coordinate system. Its
relation with the metric tensor is:
5-35
-------
EPA/600/R-99/030
=
o / T*t "V»l "W '
2 GO: oa etc
A
Divergence of a eontravariant vector W , wind for example, can be expressed as:
dx
(5A.23)
The total derivative of a covariant vector^ is represented in a Cartesian coordinate as
dA. <9
... _..
(5A-24)
dt dt dt '
d dx1 d /9
where v « V = vl — -T = & -**• -. > = & "^r was used. This expression is correct in any holonomic
abc ax1 dx dx1
coordinate system where the covariant component^, metric tensor, velocity, and x* all refer to
the same system whether or not the coordinate system is time dependent.
5A.6 Continuity Equation in Generalized Curvilinear Coordinate System
Many practical coordinate systems used for atmospheric studies are time dependent. Consider
the case when a volume element that confines the fluid moves with the fluid. Then, this is also
the velocity of the fluid in the respective coordinate system. A direct conversion from the
continuity equation expressed in a Cartesian coordinate system does not work because the
divergence term should take into account for the time rate change of volume element as well the
same for the tune dependent curvilinear coordinates. In this situation, the Lie derivative concept
(e.g., Bishop and Goldberg, 1968) becomes appropriate. A Lie derivative is obtained by
differentiating a function with respect to the parameters along the moving frame of reference.
Following Defrise (1964), one can show that a Lie derivative of a mass volume integral along the
moving frame vanishes:
Q (5A-25)
where <$lf = pSV = p-JfSV, 8V = SxlSx2Sx3 , and6V = Sx*Sx2®c* . Therefore:
(5A-26)
5-36
-------
EPA/600^R-99/030
where, index /i = 1 ,4; Jc4 = t and v4 = 1. Using the same notation convention, the contravariant
velocity is defined as for the coordinates that moves with fluid:
v"=-va;a,/i=l,4 (5A-27)
Note that one cannot derive the same result by directly replacing the divergence term in the
continuity equation for a Cartesian coordinate system because the volume element is dependent
on time as well.
Alternatively, one can obtain Equation 5A-26 by a method based on the finite derivative of a
volume integral with the application of the Leibnitz rule. This method of derivation helps to
visualize the meaning of terms in the equation more clearly than the procedure based on the Lie
derivative. Volume integral in a Cartesian coordinates is defined as:
(5A-28)
an
where/is a conservative quantity, such as density or total kinetic energy. Equation 5A-28 can
be rewritten in the curvilinear coordinate as:
(5A-29)
XI
For example, if /=! :
F = HI SV = V* = JJJ ^SV = VH, (5A-30)
an sa
The meaning of the metric becomes very clear-it is a measure of volume correction for the
transformed coordinates.
h(Xl,x\X3)8Xl$x2&3 (5A-31)
an 3fi
where fc(Je',Jc2, x3) = /(Jc1,*2,*3) was used.
Consider a time derivative following the control volume. Applying a three-dimensional version
of Leibnitz rule for the tune differential of the integral, we get:
SF
following
volume
5-37
-------
EPA/600/R-99/030
dt
2 c^l
at
-J
Then, using the following relation with the aid of Figure 5A-2:
we obtain an integral equation:
I\jt I i *•*"
+ ^\di
dx-
dt
bd
(5A-32)
(5A-33)
(5A-34)
Figure 5A-2. Volume Element in a Curvilinear Coordinate System
When / = P (density), thenF = jJJpCje'.je2,*3)^:^2^*:3 = M = mass of the volume element.
Therefore, the conservation law states:
5-38
-------
EPA/600/R-99/030
St
8M
following
volume
Here, for example,
St
j~2
ax
following
volume
= 0
(5A-35)
dt
is the velocity of the boundary of the volume element in the curvilinear
bd
coordinate x2. Because the volume element confines the fluid and moves with the fluid, this is
velocity component of the fluid in the curvilinear coordinate Jc2. Then, we have
\\\
30.
dt
(5A-36)
Since above integral should be satisfied for an arbitrarily infinitesimal volume element, we
obtain the continuity equation in differential equation form for the time-dependent curvilinear
coordinate as follows:
(5A-37)
5-39
-------
EPA/600/R-99/030
Chapter 6
GOVERNING EQUATIONS AND COMPUTATIONAL STRUCTURE OF THE
COMMUNITY MULTISCALE AIR QUALITY (CMAQ) CHEMICAL TRANSPORT
MODEL
Daewon W. Byun* and Jeffrey Young**
Atmospheric Modeling Division,
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
M. Talat OdmaiT*
MCNC-Environmental Programs
P.O. Box 12889,3021 Cornwallis Road
Research Triangle Park, NC 27709-2889, USA
ABSTRACT
The chemical transport model (CTM) of the Models-3/CMAQ (Community Multiscale Air
Quality) modeling system can be configured to follow the dynamics of the preprocessor
meteorological model. A science process module in the CMAQ CTM is not specific to a
coordinate system. The generality is accomplished through the use of the coordinate
transformation Jacobian within the CMAQ CTM. In this chapter, we derive the governing
diffusion equation in a generalized coordinate system, which is suitable for multiscale
atmospheric applications. We describe the CMAQ system's modularity concepts, fractional
time-step formulation, and key science processes implemented in the current version of the
CMAQ CTM. We examine dynamic formulations of several popular Eulerian air quality models
as emulated by the governing diffusion equations in the generalized coordinate system. Also, a
nesting technique for the CMAQ CTM is introduced. Finally, because the amount of a
substance in the atmosphere can be expressed in many different ways, we summarize the most
popular expressions for concentration and their transformation relations.
"On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Daewon W. Byun, MD-80, Research Triangle Park, NC 27711.
E-mail: bdx@hpcc.epa.gov
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Present Affiliation: Georgia Institute of Technology, Atlanta, GA.
-------
EPA/600/R-99/030
6.0 GOVERNING EQUATIONS AND COMPUTATIONAL STRUCTURE OF THE
COMMUNITY MULTISCALE AIR QUALITY (CMAQ) CHEMICAL TRANSPORT
MODEL
In Chapter 5, "Fundamentals of Atmospheric Modeling ..." we discussed the fimdamental set of
equations for atmospheric dynamics and thermodynamics in a generalized coordinate system. In
this Chapter, we investigate the diffusion equation for the trace species in the atmosphere in the
generalized coordinate system and the computational structure of the Community Multiscale Air
Quality chemical transport model (CMAQ CTM or, hereafter, CCTM).
One requirement of the CMAQ modeling system is to maintain a consistent description of the
atmosphere for different meteorological and chemical transport models. This is a feature that is
essential for spatial scalability. Various coordinate systems are used in atmospheric models.
Selection of a suitable coordinate system is an important step of model formulation. There are
numerous criteria to be considered in selecting a coordinate system, such as the dynamic
characteristics it can handle and how well it can deal with curvature of the earth's surface and
features of the terrain. Formulation of the models may vary substantially for different coordinate
systems. If a CTM can be formulated and coded using a generalized coordinate system, it would
be easy to switch from one coordinate to another depending on the application. The generalized
coordinate concept is useful because a single CTM formulation can adapt to any of the
coordinates commonly used in meteorological models. It is also desirable to compare the benefits
of various coordinate systems and to be able to link the CTMs to meteorological models and
databases in different coordinates.
Conformity of the coordinates to the physics of the problem is very important. Unlike a model
with a fixed coordinate system, a generalized coordinate system allows use of generic coordinates
for the specific science processes within a model. Although the model's overall structure is
determined by the choice of a coordinate system, the individual science modules can still use their
own generic coordinates that best suit the physical processes they model. This means that each
science process can utilize the parameterizations based on the best coordinate to represent the
problem. For example, the planetary boundary layer (PEL) parameterizations can be expressed
in terms of geometric height, or dimensionless height scaled with PBL height, while for cloud
physics, they can be represented in terms of pressure. The linkages between the generic
coordinate parameterizations in the science processes and the governing conservation equation in
the generalized coordinates are established through the application of appropriate coordinate
transformation rules.
Here, we intend to provide a comprehensive and rational development of the governing
conservation equation in generalized coordinates, which can be readily implemented in an Eulerian
model. The operating assumptions used for the derivations are listed below (see Srivastava et al.,
1995).
« Assumption 1: Pollutant concentrations are sufficiently small, such that their presence
would not affect the meteorology to any detectable extent. Hence, the species
6-2
-------
EPA/600/R-99/030
conservation equations can be solved independently of the Navier-Stokes and energy
equations. The conditions which could invalidate this assumption are for cases where
sufficient heat is generated by chemical reactions to influence the temperature of the
medium or where an atmospheric layer become so concentrated with pollutants that
absorption, reflection, and scattering of radiation alter the air flow (Seinfeld, 1986).
• Assumption 2: The velocities and concentrations of the various species in atmospheric
flow are turbulent quantities and undergo turbulent diffusion. Because turbulent diffusion
is much greater than molecular diffusion for most trace species, the latter can be ignored.
• Assumption 3: The metric tensor that defines the coordinate transformation rules is not a
turbulent variable. This means that we can define the coordinates based on the Reynolds
averaged quantities. The vertical grids will be defined incrementally between time steps
when a time-dependent vertical coordinate is used.
* Assumption 4: The ergodic hypothesis holds for the ensemble averaging process. This
means that the ensemble average of a property can be substituted with the time average of
that property.
• Assumption 5: The turbulence is assumed stationary for the averaging time period of
interest (say 30 minutes to one hour for atmospheic applications).
• Assumption 6: The source function (i.e., emissions of pollutants) is deterministic for all
practical purposes and there is no turbulent component.
• Assumption 7: The effect of concentration fluctuation on the rate of chemical reaction is
negligible, i.e., contributions of covariance effects among tracer species are neglected.
• Assumption 8: Because the large-scale motions of the atmosphere are quasi-horizontal
with respect to the earth's surface, science processes can be separately represented in
horizontal and vertical directions (i.e., quasi-orthogonal in transformed coordinates).
6.1 Derivation of the Atmospheric Diffusion Equation
In Chapter 5, we derived the species continuity equation in generalized coordinates. It is given
as:
t m as '
where ^ is the trace species concentration in density units (e.g., kg m"3), J, is the vertical
*.
Jacobian of the terrain-influenced coordinate s, m is the map scale factor, VY and s are horizontal
and vertical wind components in the generalized coordinates, and **
-------
EPA/60Q/R-99/03Q
To make the instantaneous species continuity equation useful for air quality simulation, we need
to derive the governing diffusion equation. This is done by decomposing the variables in
Equation 6-1, except for the Jacobian and map scale factor, in terms of mean and turbulent
components. The Reynolds decompositions of species concentration and mixing ratio are
expressed as:
-------
EPA/600/R-99/030
m
(6-5)
where we used J^ = J^ and ^ = Q^t based on Assumption 3 and Assumption 6, respectively.
The Reynolds flux terms in Equation 6-5 can be approximated in terms of the mixing ratio as:
V^" (6-6a)
*" • (6-6b)
(6-6c)
(6-6d)
in which we have neglected the second order perturbation terms based on the scale analysis of the
equations. Equation 6-5 can be rewritten using Equations 6-3a-c and 6-6a-c to give:
m
(6-7)
The turbulence flux terms can be parameterized using a simple closure scheme such as the eddy
diffusion concept (K-theory):
(6-8)
where ^;' denotes the eddy difrusivity tensor in the transformed coordinate. The eddy
diffusivity tensor for the generalized meteorological coordinates is related to the diffusivity
tensor in Cartesian coordinates as: .
1* 3S.I
(6-9)
6-5
-------
EPA/60Q/R-99/03Q
If we postulate that the diffusivity tensor in Cartesian coordinates is diagonal (i.e., all the off-
diagonal components vanish), then the eddy diffusivity tensor in the generalized meteorological
coordinates becomes:
K
0
m-
0
dJP
dx
dy
»
dy
-.
dx
(6-10)
where K^ — Kn, Kff = K22, and Ku = KJ3 are the diagonal components of eddy
diffusivity tensor in the Cartesian coordinate. To match with the computational grid, the gradient
terms in Equation 6-10 must be rewritten in terms of the generalized coordinates Jc3 (defined
based on height above ground hAGL = h-ztfc, where ztfc represents the height of topography)
•A •* u- i f i l(dA\ (dA\ dA(dA\(dh\ ___ ,
using the appropnate chain rules, for example, — — - = •— r - -r- -r-r -=-r . When A
m\dxjz VoK )# dz. \dx~ )\dx J^
= x3, we get
K =
0
0
-m
dz
„
•K.,
(6-11).
Then the non-zero diffusion terms in Equation 6-7 can be parameterized with the eddy diffusion
theory as follows:
m
\ J
= tn
m
''<• * "at J_ V'-' "' "^
2^ ^ + R __.)
(6-12)
and
(6-13)
6-6
-------
EPA/600/R-99/030
Rewriting Equation 6-7 with Equations 6-12 and 6-13, and separating the diagonal and off-
diagonal diffusion terms with an explicit description of the source terms, one can obtain the
governing atmospheric diffusion equation in the generalized coordinates where the turbulent flux
terms are expressed with the eddy diffusion theory:
• + i
m
-vj3
oa
(a)
(b)
(c)
a?;)
m
n/
P
(d)
(e)
T£l
ofir
(f)
s(*"i^
(g)
%1
"1*2 ^
^ J
dt
0)
+
«
M
+ •
(k)
dt
0)
(6-14)
ping
The terms in Equation 6-14 are summarized as follows:
(a) time rate of change of pollutant concentration;
(b) horizontal advection;
(c) vertical advection;
(d) horizontal eddy diffusion (diagonal term);
(e) vertical eddy diffusion (diagonal term);
(f) off-diagonal horizontal diffusion;
(g) off-diagonal vertical diffusion;
(h) production or loss from chemical reactions;
(i) emissions;
Q) cloud mixing and aqueous-phase chemical production or loss;
(k) aerosol process; and
6-7
-------
EPA/600/R-99/030
(1) piume-in-grid process.
Note that the dry deposition process can be included in the vertical diffusion process as a flux
boundary condition at the bottom of the model layer.
Alternatively, we can express the turbulent flux terms in Equation 6-7 using the Reynolds flux
terms defined as:
(6-15)
•
and the turbulent flux terms are related with the Cartesian counterpart using f* = -^-^K '•
V, j 1,
(6-16)
In comparison with Equation 6-14, the Reynolds flux terms shown in Equation 6-15 include the
off-diagonal components. One can now rewrite the governing conservation equation for trace
species equivalently to Equation 6-14 hi terms of the Reynolds flux terms:
m
+m
£2
m
2 "
cltl
dt
dt
(6-17)
ping
The governing equation can be simplified for a domain with gentle topography for which one may
ignore all the terms involved with the horizontal gradients of the surface normal to the vertical
-jr i ii '•"•', ' ii
coordinate. This forces the vertical diffusion terms in the curvilinear coordinate system to be
identical to those of the orthogonal Cartesian coordinate system. Then the trace species
conservation equation can be written in a simpler form:
-------
EPA/600/R-99/030
where
-------
BPA/600/R-99/030
subroutines) in a CTM. While the emissions processing and the meteorology model are modular
at the second level, the CCTM achieves the third-level of modularity by employing the operator
splitting, or fractional time step, concept in the science processes. The next level of modularity
is based on the computational functionality in a processor module, e.g., science parameterization,
numerical solver, processor analysis, and input/output routines. The lowest meaningful
modularization level is the isolation of sections of code that can benefit from machine dependent
optimization.
6.2.2 Modularity Concept of CMAQ
To allow for both the continuous improvement of science and for the addition of new capabilities
in a unified fashion, it is critical to have efficient modular schemes in the CMAQ design.
Currently, the modularity within CMAQ is based mostly on the fractional time-step
implementation of the science processes. This level of modularity involves the distinction of a
driver, processor modules, data provider modules, and utility subroutines in CMAQ. We have
chosen this method because it provides a natural disciplinary distinction for different science
processes through which developments in specific research areas can readily be incorporated
(Refer to Figure 6-2).
In some of the process modules, such as the aqueous-phase chemistry module, the science
algorithms and numerical solvers are tightly linked. For other types of modules, the science
parameterization components and numerical solvers have a looser association. In such cases the
modularity can be defined either at the parameterization level, the numerical algorithm level, or
both. For example, the module definition for the advection process is based on the numerical
advection algorithm used. For the gas-phase chemistry process, the modularity is based on the
ordinary differential equation numerical solvers. The chemistry mechanism description is
generalized and the ModeIs-3/CMAQ framework provides a straightforward method to link
model species surrogate names with the species names in the data set. See Chapter 15 for details.
The use of different chemical mechanisms is accommodated through the mechanism reader and
generalized codes for setting up the production and loss terms of the chemistry reactions.
Therefore, the CCTM does not require different gas-phase chemistry modules for different
mechanisms.
The vertical diffusion process can be formulated using either local- or non-local-mixing
parameterization schemes. The current classification of vertical diffusion modules is based on the
process parameterization methods. The modularity of this process can be enhanced if we
distinguish the method used for computing the vertical diffusivities for local-mixing. In this case,
the modularity is defined at the level of data provider modules. The modularity level can be
deepened further if we identify different numerical solution methods for the diffusion.
With the current version of CMAQ, the level of science modularity is subordinated by the way
the science process codes are archived in the system. Here, we define a class as a collection of
different modules, for a given science process. The science classes are identified with the grouping
6-10
-------
EPA/600/R-99/030
of the terms in the governing conservation equation, Equation 6-18. Currently, nine science
process classes are defined in CCTM:
• DRIVER controls model data flows and synchronizes fractional tune steps;
• HAD V computes the effects of horizontal advection;
• VADV computes the effects of vertical advection;
• ADJCON adjusts mixing ratio conservation property of advection processes;
• HDIFF computes the effects of horizontal diffusion;
• VDIFF computes the effects of vertical diffusion and deposition;
• CHEM computes the effects of gas-phase chemical reactions;
• CLOUD computes the effects of aqueous-phase reactions and cloud mixing;
• AERO computes aerosol dynamics and size distributions; and
• PING computes the effects of plume chemistry.
CCTM does not have emissions as a separate science process because it can be either a part of
the vertical diffusion or the gas-phase chemical reaction process. It is worthwhile to mention
here that the current modular paradigm does not prevent establishment of combination of
processes in a larger single module. For example, one can develop a module describing the vertical
transport, chemistry, and emissions simultaneously when time scales of those processes become
comparable. Users could experiment with the combination of modules to best fit to their
problems at hand.
In addition to nine science process modules, CCTM includes two science process, classes. The
PHOT computes photolysis rates, and AERO_DEPV computes particle size-dependent dry
deposition velocities. These are typical "data-provider" science process classes, which do not
involve updating concentrations directly. There are some other classes that do not fall in any of
the above definitions. We have grouped these auxiliary routines as the UTIL class, which is a
collection of utility subroutines. As one can see, the current modularity of the CCTM is
implemented more on a practical basis rather than by strictly following a design paradigm. One
can also see that the present modularity definition of CMAQ is somewhat subjective. In the
future we intend to allow definition of the modularity at the user-defined granularity level.
Figure 6-1 describes the key science process modules in CCTM and their data linkage with
CMAQ's preprocessors, whose descriptions are available in other chapters. The only data
dependencies among the CCTM science modules are the trace species concentration field as seen
in the diagram and the model integration time step. Figure 6-2 shows the distinct data
6-11
-------
EPA/600/R-99/030
dependencies within the CCTM, To facilitate modularity and to minimize data dependency in
CCTM, we store concentrations in global memory while the environmental input data are
obtained from random-access files and interpolated to the appropriate computational
(synchronization) time step. This realizes the recommended "thin-interface" structure of the
model:
• Common timing data are managed through the science process main subroutine's call
arguments;
• Concentrations are the object of all process operations;
• Environmental data are provided through a standard I/O interface;
• Model structure data are provided through shared include files; and
• Standard physical constants are obtained from shared include files.
See Chapter 18 for further details on how the science codes are integrated in the Models-3
CMAQ system.
6-12
-------
EPA/6QG/R-99/03G
CMAQ Science Processors
— Emissions
.>---»-%- \ / \ / nun***
/HorizA/ Udvect'ionj M Diffusion
Boundary _J Advection | V / \\
Meteorology /Gas-Phase
1
_1 —JPhotolitic
I I rates
• - Plume Dynamics —
Figure 6-1. Science Process Modules in CMAQ. Interface processes are shown with rectangular
boxes. Typical science process modules are updating the concentration field directly and the data-
provider modules include routines to feed appropriate environmental input data to the science
process modules. Driver module orchestrates the synchronization of numerical integration across
the science processes. Concentrations are linked with solid lines and other environmental data with
broken lines. (From Byun et al., 1998.)
6-13
-------
EPA/600/R-99/030
Table 6-1. Interface Processors for the CMAQ Modeling System
Interface
Processor
Description
Reference
ICON Provides initial three-dimensional fields of trace species Chapter 13
concentrations for modeling domain
BCON Provides concentrations of trace species for the boundary Chapter 13
cells
ECIP Incorporates emissions from separate area and major Chapter 3
point sources to generate hourly 3-D emissions input file
MCIP Processes the output of a meteorological model to Chapter 12
provide the necessary meteorological data for CMAQ
models
JPROC Computes photolysis rates for various altitudes, Chapter 14
latitudes, and sun zenith angle
PDM Generates plume information needed to apply plume-in- Chapter 9
grid (PinG) processing in CCTM
6-14
-------
EPA/600/R-99/030
6.2.3 Description of Science Processes
In this section we describe individual science processes, shown in Figure 6-1, associated with the
groups of individual terms in the governing diffusion equation. Note that different concentration
units are used for different science processes in CMAQ CTMs. Appendix 6A provides the
relationships among the concentration units and their conversion factors from one unit to another.
CMAQ CTM 's Data Flow
Random-Access Disk F lie
NTTTTH LI I L LI II 11,11 Mil III ML
nvironmental Data (Meteorology, Emissions, B
Figure 6-2. Data Dependencies Among Modules in CCTJVI. P and Sk represent a science process
module and the related subroutines for the module, respectively. (From Byun et al., 1995.)
6.2.3.1 Driver Class for CCTM
The key function of the driver class module is hosting the science processors. It is responsible
for coordinating model integration time (synchronizing fractional-time steps of science process
call) and some input/output sequences. The driver struclure of the current CCTM is given in
Figure 6-3. A synchronization time step is used to ensure the global stability of the CCTM's
numerical integration at the advection time step, which is based on a Courant number limit.
Nesting requires finer synchronization time steps for the fine grid domain. The CCTM's process
synchronization time steps are represented as integer seconds because the Models-3 I/O API can
only handle integer seconds for I/O data. All the needed data are appropriately interpolated
based on the synchronization time step. For maintaining numerical stability and for other
reasons, an individual process module may have its own internal time steps. In general, each
science process module uses the synchronization time step ( A^n(.) as the input time step of
required environmental data. The global output time steps can be set differently from the
6-15
-------
EPA/600/R-99/030
l: '
synchronization time step. Usually, the output time step (AfnH,) is set as one hour, but sub-
hourly output down to the synchronization time step is possible.
Table 6-2. List of Science Process Subroutines Called by the CMAQ Driver
Subroutine
Science Class Description
CGRID MAP UTIL
INITSCEN
ADVSTEP
ADJADV
HDIFF
VDIFF
CHEM
PING
AERO
CLDPRC
INIT*
DRIVER
COUPLE/ COUPLE*
DECOUPLE
SCIPROC DRIVER
XADV, HADV
YADV
ZADV VADV
ADJCON
HDIFF
VDIFF
CHEM
PING
AERO
CLOUD
Sets up pointers for different concentration species: gas
chemistry, aerosol, non-reactive, and tracer species
Initializes simulation time period, time stepping constants, and
concentration arrays for the driver
Computes the model synchronization time step and number of
repetitions for the output time step
Converts units and couples or de-couples concentration values
with the density and Jacobian for transport
Controls all of the physical and chemical processes for a grid
(currently, two versions are available: symmetric and
asymmetric around the chemistry processes)
Computes advection in horizontal plane (x- and y-directions)
Computes advection in the vertical direction in the generalized
coordinate system
Adjusts concentration fields to ensure mixing ratio
conservation given mass consistency error in meteorology data
Computes horizontal diffusion
Computes vertical diffusion and deposition
Solves gas-phase chemistry
Computes effects of plume-in-grid process
Computes aerosol dynamics, particle formation, and
deposition
Computes cloud mixing and aqueous chemistry
'represents a process class that is part of DRIVER function.
6-16
-------
EPA/600/R-99/030
CMAQ-DRIVER
Asymmetric SCIPROC Symmetric SCIPROC
Output
Time Step
Synchronization t
Time Step Atsync /
XADV*
MX
YADV*
MX
ZADV
MX
ADJCON
MX
HDIFF
MX
DECOUPLE
E
C
E
E
Atsync
At
. '-"'sync
\ix
VDIFF
MX
PING
vj/
CHEM
M/
AERO
M/
PCLDPRC
MX
|COUPlE|
_|
I
1
1
I
1
* Alternating
XADV and YADV calls
for each asymmetric
SCIPROC call
XADV
•>
'
YADV
4^ /2
\lx
ZADV
\ /
ADJCON
\
HDl
>
i
FF
•
TDECOUPLE
MX
VDIFF
\
At
\ PING |
MX
I CHEM |
MX
I AERO |
MX
| CLDPRC |
vlx
VDIFF
T^
'
I'COUPLEp
MX
HDIFF
MX
ZADV
MX
YADV
MX
XADV
MX
ADJCON
Figure 6-3. Driver Module and Its Science Process Call Sequence.
Both asymmetric and symmetric call sequences in SCIPROC are presented. A^,. and A.tout are
model synchronization and output time steps, respectively. Refer to Table 6-2 for the description
of the subroutines.
The DRIVER program calls initialization routines to set up CCTM runs. It initializes the
concentration field and checks if the input files, run time, and grid/coordinate information are
consistent for a given scenario. Subroutines used for the initialization process are grouped into
the INIT class. Usually, initial concentrations for gaseous species are in molar mixing ratio units
(ppm) and aerosol species in density units (ug m"3), the same as the output units of CMAQ.
Also, DRIVER calls couple/decouple subroutines to convert concentration units for appropriate
data processing. The pair of couple/decouple calls, which are available in the class COUPLE,
limit the interchange of process modules between two different concentration units, such as
density versus mixing ratio. The classes INIT and COUPLE are introduced just for the
6-17
-------
EPA/600/R-99/030
convenience of code management from the point of view of science process modularity, and they
should be considered as part of the DRIVER class modules.
6.2.3.2 Advection Processes for CCTM: HADV, VADV and ADJCON
For convenience, the advection process is divided into horizontal and vertical components. This
distinction is possible because the mean atmospheric motion is mostly in horizontal planes.
Usually the vertical motion is related with the interaction of dynamics and thermodynamics. The
advection process relies on the mass conservation characteristics of the continuity equation:
(6-19,
Using the dynamically and thermodynamically consistent meteorology data from MCIP, we can
maintain data consistency for air quality simulations at the synchronization time step. In case
the meteorological data provided and the numerical advection algorithms are not exactly mass
consistent, we need to solve a modified advection equation:
(6-20)
where Qp is mass consistency error term (Byun, 1999). Equation 6-20 ensures conservation of
mixing ratio, which is a necessary (though not sufficient) condition for preserving total tracer
mass given significant fluctuations of density field in space and tune. The equation shows that
the correction term has the same form as a first-order chemical reaction whose reaction rate is
determined by the mass consistency error (normalized with air density) in the meteorology data.
Modules in HADV class solve for the horizontal advection:
and modules in VADV class solve for the vertical advection with boundary conditions v3 = 0 at
the bottom and top of the model.
In simulating air quality, one of fundamental characteristic of the model application should be
conservation of mass. Therefore, the modules in the ADJCON class solve for the mass
correction term:
6-18
-------
EPA/600/R-99/030
We wish to emphasize that the artificial distinction of advection modules between horizontal and
vertical processes is not adequate and that all three modules (HADV, VADV, and ADJCON)
should be considered as an integral unit for solving the physical advection process of trace
species. The advection and mass adjustment algorithms are described in detail in Chapter 7.
6.2.3.3 Diffusion Process Classes for CCTM: HDIFF and VDIFF
For convenience the atmospheric diffusion process is divided into horizontal and vertical
components. This distinction is needed because the vertical diffusion mostly represents the
thermodynamie influence on the atmospheric turbulence by the air-surface energy exchange
processes while the horizontal diffusion represents subgrid scale mixing due to the unresolved
wind fluctuations. To handle the atmospheric diffusion processes in the generalized coordinates,
we need to carefully examine the governing equation to properly set up the diffusion solver.
We start from the atmospheric diffusion equation in the same concentration units as used in
advection: . ....-•
1
, J -
(6-24)
diff
where '
=_v.
hdiff
\vdiff
vdiff
(6-25)
(6-26)
Emissions can be included either in vertical diffusion or gas-phase chemistry module. If we can
parameterize the turbulent fluxes directly in the curvilinear coordinates, we can implement
HDIFF and VDIFF modules following Equations 6-25 and 6-26. When the turbulent fluxes are
parameterized with eddy diffusion theory, the contributions of the off-diagonal (cross-
directional) diffusion terms show up explicitly as shown in Equation 6-14:
6-19
-------
EPA/600/R-99/030
. ' ' ' • • f
For a domain with a significant topographic feature, the module CDIFF must be implemented.
However, the current CMAQ version does not include CDIFF module as the off-diagonal terms
are often neglected in operational air quality models. In such a case, the HDIFF and VDIFF
modules solve for diagonal terms (with respect to the curvilinear coordinates) as follows:
Compared with above formulations, let's consider the case that we approximate the quantity
•\/^p, which defines the computational grid, to be constant for the duration of synchronization
time step for integrating the diffusion process with the fractional time-step method. Then, the
problem becomes equivalent to solving for the diffusion equations in terms of the mass mixing
ratio instead of density:
If we rely on Equation 6-28 for representing the atmospheric diffusion process, the concentration
must first be decoupled to obtain mass mixing ratio, #,. Once the new mixing ratio is computed, it
needs to be coupled with «J$p to give the updated concentration in terms of
-------
EPA/600XR-99/030
\
The effects of turbulence flux caused by the divergence of the grid boxes in the coordinate system
need to be included in order to describe the turbulence exchange processes precisely. One can
readily show that the coordinate-divergence term in Equation 6-30 vanishes for a mass conserving
vertical coordinate. Similarly, when topographical features vary significantly and horizontal
variations of the quantity -\/yp are large, one cannot neglect the last term in Equation 6-29.
Chapter 7 of this document describes physical parameterization schemes and numerical
algorithms for the horizontal and vertical diffusion processes in the CCTM.
One may wonder how deposition should be represented in the generalized coordinate system. In
Eulerian air quality models, the deposition process affects the concentration in the lowest layer
as a boundary flux condition. Considering the deposition process as the diffusion flux at the
bottom of the model, we can relate the boundary condition in the generalized coordinate system
to that of the Cartesian coordinate system as:
^ M
F?
dip
because F*
dep
dx
and Fl
1*31
*]>
dz
dep
(6-31)
dep
do not exist. Then, the effects of dry deposition on the species
concentration is accounted for by the following relationship:
d
-------
EPA/600/R-99/030
dt
layer I
(6-33')
<*!>
Equations 6-32 and 6-33 show that we do not need to estimate contravariant deposition
velocities if the deposition process is implemented as a bottom boundary condition in the
generalized coordinate formulation.
1 r : * . . ' - " " ' ' i **,'••.!••
In the current CCTM implementation, the concentration units for horizontal and vertical
•• • • ... ....... . . . • , ..... : ' • . • . ..,-;, .•..••: M
diffiision processes are density (coupled with Jacobian) and molar mixing ratio, m, — q,, "
respectively. We have chosen m. as the generic concentration unit for the vertical diffusion to
coordinate with the emissions units in the data. Subsection 6.2.3.6 provides a detailed
explanation for this. Therefore, HDIFF is placed outside and VblFF is placed in between the
pair of couple/decouple calls. Because the ratio of molecular weights are constant, equations for
the vertical diffusion in terms of molar mixing ratio are equivalent to those in terms of mass
mixing ratio, q, . Refer to Chapter 7 for details of the computational algorithms for HDIFF and
VDIFF.
6.2.3.4 Gas-phase Chemistry Process for CCTM
Instead of directly computing the time rate of change of
-------
EPA/600/R-99/030
(6-36)
chetfi
A A
where Rm and Qm represent chemistry reactions and source terms in molar mixing ratio units,
respectively.
CMAQ employs generalized chemistry solvers, such as QSSA (Young et al., 1993) and
SMVGEAR (Jacobson and Tureo, 1994), which are designed to solve the nonlinear set of stiff
ordinary equations presented in Equation 6-36. They can be applied independent of the
coordinate and grid descriptions. To accommodate the need for modified or new chemical
mechanisms, the CMAQ system is equipped with a generalized chemical mechanism processor.
Refer to Chapter 8 for detailed description of numerical solvers used for gas-phase chemistry.
The Models-3 framework provides a mapping table to link chemistry mechanism species with
surrogate species names in the initial and boundary condition files and emissions files. See
Chapter 15 for details. When a new mechanism is used, appropriate emissions data must be
supplied. It is possible to include emissions either in the gas-phase chemistry or In the vertical
diffusion process. It is preferable that the emissions are interpolated with the same temporal
interpolation schemes used in the transport processes.
6.2.3.5 Aerosol Process Class for CCTM
The fractional time step implementation solves for the effects of aerosol chemistry and dynamics
on trace gas and aerosol species concentrations with:
- v dp/' ffi 371
p, VS -,,- V0"-"/
dt
aero
where Rgerg represents processes such as new particle formation and growth and depletion of.
existing particles. Qaero. stands for all the external sink and source terms, and vg is the
contravariant sedimentation velocity. The generic concentration units for the aerosol process are
[fig m~3] (density) for aerosol mass and [number m~3] for aerosol particle number density.
Because the aerosol process is called within the pair of couple/decouple calls, the input
concentration is already decoupled and the following set of governing equations are solved in the
aerosol process module:
dt
n
afr0i
The present implementation of the aerosol module in CCTM is derived from the Regional
Paniculate Model (Binkowski and Shankar, 1995). Here, primary particles are divided into two
groups: fine particles and coarse particles. The fine particles result from combustion and
secondary production processes and the coarse group is composed of materials such as wind-
6-23
-------
EPA/600/R-99/030
blown dust and marine particles. The key scientific algorithms simulating aerosol processes for
the CCTM are: (1) aerosol removal by size-dependent dry deposition; (2) aerosol-cloud droplet
interaction and removal by precipitation; (3) new particle formation by binary homogeneous
nucleation in a sulruric acid/water vapor system; (4) the production of an organic aerosol
component from gas-phase precursors; and (5) particle coagulation and condensation growth.
Refer to Chapter 10 for details on aerosol process implemented in CCTM.
6.2.3.6 Emissions Process for CCTM
As mentioned earlier, the emissions process does not have its own science process class. Instead,
it is included Cither in vertical diffusion or in the gas-phase chemistry process. In the governing
conservation equations for the trace gases, the emissions process is represented simply as source
terms.
If emissions data are given in the unit of time rate of change of mass, for example for particulate
species, such as PM2.5 and PM10 in [g s~l], they are expressed as:
SV 8V
I d(pqiSV}
8V dt
where <5Fis volume of the cell andEf = —^-—
(6-39)
represents the emissions rate into the cell. If
the mass of air in the cell does not change for each time step (usually one hour), the concentration
expression, either as the time rate of change of density or as the mass mixing ratio can be used.
Otherwise, when the volume and density of a cell change substantially with time, the effect of
change in air mass must be accounted for in determining the emissions rates.
For gaseous species, the time rate of change of E; for each hour and each grid cell are provided in
the three-dimensional emissions data files in molar units, (i.e., mole s~l):
\5V
(6-40)
Emissions for gaseous species in molar units are preferred to those in mass units because molar
units are the natural units for chemistry and mass units must be transformed into [mole s~l]
eventually for the gas-phase chemistry process. For gaseous species the molar mixing ratio and
mass mixing ratio differ only by a simple multiplication factor, the ratio of molecular weights.
However, for lumped species, the molecular weights are variable depending on the fractional
compositions of the categorized hydrocarbon species in the emissions data. Therefore,
transformation of emissions in mass units into the molar units for the lumped species can
introduce misrepresentation of emissions amount. The data for the fractional compositions of
the categorized hydrocarbon species are available in the emissions processor, Models-3
Emissions Projection and Processing System (MEPPS) (See Chapter 4). Thus, when emissions
6-24
-------
EPA/600/R-99/030
data are processed in [mole s~l] units, we do not have this conversion problem. Then the
emissions process is represented as:
(6.41)
An additional benefit is that the same transformation rule can be applied when emissions are
included either in the vertical diffusion or in the chemistry.
6.2.3.7 Cloud Mixing and Aqueous-phase Chemistry (CLOUD) for CCTM
The rate of change in pollutant concentrations due to cloud processes is given by:
dm,\ din.
dt L dt
\cld
** (6-42)
subcld
dt
resold
where subscripts eld, subcld, and rescld represent cloud, subgrid scale cloud, and resolved cloud,
respectively. Although calls to the CLOUD module are made at every synchronization time
step, the subgrid cloud effects are accounted for once an hour while the resolved cloud effects are
impacted at each call. This is equivalent to assuming that the cloud life time of all sub-grid clouds
is one hour. The effects of subgrid cloud processes, such as mixing (mix), scavenging (scav),
aqueous-phase chemistry (aqchem), and wet deposition (wdep) on grid-average concentrations
are parameterized with a "representative cloud" within the grid cell:
——L ~ f(mix, scav, aqchem, wdep) (6-43)
^ \subcld
where/represents a function of its arguments. We chose this expression because of the implicit
nature of the algorithm representing the processes. For the resolved cloud, no additional cloud
dynamics are considered in CMAQ and only effects of the scavenging and aqueous-phase
chemistry are considered:
dm.
dt
dm.
resold
dt
(6-44)
aqchem
See Chapter 11 for details of the cloud process descriptions.
6.2.3.8 Plurae-in-Grid Process (PING) for CCTM
Anthropogenic precursors of the tropospheric loading of ozone, aerosols, and acidic species are
largely emitted from major point sources, mobile sources, and urban-industrial area sources. In
particular, inadequate spatial resolution of the major point source emissions can cause inaccurate
predictions of air quality in regional and urban Eulerian air quality models. A plume-in-grid
6-25
-------
EPA/600/R-99/030
(PinG) approach in CCTM provides a more realistic treatment of the subgrid scale physical and
chemical processes for major elevated point source emitters (MEPSEs).
The PING module solves for:
dmn dm.
dt dt
dap
dm
~ti
dm
emis
dm
chem
dt
(6-45)
dep
where mp is concentration of the subgrid plume (in molar mixing ratio) and the time-rate of
change terms with subscripts disp, emis, chem, and dep represent effects of plume dispersion,
point source emissions, plume chemistry, and dry deposition in the plume, respectively. The
location and shape of plumes are determined by the PDM and plume chemistry is computed in
the CCTM within plume subsections. When the subgrid scale phase of the plume simulation has
been completed, the PING module updates grid scale concentrations with:
= 5Vp_d(mr,-mibg) ,
5V ' dt ^ '
where mibg is the back ground concentration and SVp is the volume of plume in a grid cell with
volume 5V. Currently, only gaseous species are treated with the PING module. Readers are
referred to Chapter 9 for the details. The work for the inclusion of particulates in the PING
process has been started.
6.3 Equivalent Model Formulations for Different Vertical Coordinates
Because the CCTM is based on a generalized coordinate system, it is possible to emulate the
governing equations of other popular Eulerian air quality models. For most urban and regional
applications, the choice of horizontal map projection is handled with the map scale factors at the
individual grid points. Therefore, there are no real differences in formulations in horizontal
directions. One caveat is that the current CMAQ version is not tested with anholonomic
coordinates, such as spherical coordinates. A few implementation details must be taken into
account to accommodate the spherical coordinates. Most of the distinction of the dynamics is
attributed to the choice of the vertical coordinate of the system.
The generalized governing conservation equation for trace species, written in the Reynolds flux
form, is given in Equation 6-18. The same equation in eddy-diffusion form, in which the
components of the eddy difrusivity tensor are represented in terms of those in Cartesian
coordinates, is given below:
6-26
-------
EPA/600/R-99/030
. *
ctd
dt
ping
dt
(6-47)
In most popular air quality models, including the present implementation of the CCTM, the
cross-terms from the off-diagonal components of the diffusivity tensor are neglected. Note that
some models use wind components defined in Cartesian coordinates. The mass conservation
characteristic of Equation 6-47 is heavily dependent on the quality of the wind data provided. In
particular, depending on the dynamic assumptions used in atmospheric models, methods for
estimating the contravariant vertical velocity component vary considerably. Refer to Chapter 5
for the details.
Formulations of other air quality models with popular meteorological vertical coordinates, such
as z,crz, apg> and cr.; and the step-mountain eta coordinate, 77, can be obtained by substituting
the appropriate vertical Jacobian in Equation 6-47. Refer to Table 6-3 for the coordinate
definitions, associated Jacobians and contravariant vertical velocity components. Occasionally,
one may find discrepancies in the governing equation between the one represented by Equation 6-
47 and the one presented in the documentation of a specific model with the same vertical
coordinates. Some of these can be attributed to the explicit representation of the dynamic
characteristics and other idiosyncratic implementation practices used in those models. In the
CCTM, the vertical coordinate is defined to increase with geometric height as given in Equation
6-4. This restriction simplifies interpretation of terms in the governing equations and eventually
the computer coding of the algorithms. For example, the sign of the contravariant vertical
velocity component is kept the same (i.e., positive value represents upward motion) across the
different coordinate systems.
The terrain-influenced height coordinate crz has been used often for studying air quality
especially with some simplifying conditions such as the Boussinesq approximation and anelastic
assumption. For urban air quality simulations, there are a few examples of applying the <7Z
coordinate defined with time and space dependent H (thickness of model), which is often related
with the boundary layer height. The <7Z coordinate is used in air quality models, such as the
Urban Airshed Model (UAM) (Scheffe and Morris, 1993), STEM-II (Carmichael et al., 1991),
CIT (Harley et al., 1993), CALGRID (Yamartino et al., 1992), and others. The terrain-influenced
reference pressure coordinate is used in SAQM (Chang et al., 1997), which is designed to be
consistent with the nonhydrostatic MM5 meteorological model. The terrain-influenced time
dependent hydrostatic pressure coordinate is used in RADM (Chang et al., 1987). It is the same
coordinate used for MM4 or MM5 hydrostatic applications. Step-mountain eta coordinate is
6-27
-------
EPA/600/R-99/030
used for NCEP's Eta meteorological model (Black, 1994, and Mesinger et al., 1988), but no
operational air quality model using the eta coordinate is available.
Table 6-3. Vertical Coordinates and Associated Characteristics
CMAQ Definition Contravariant
Coordinate Vertical
jc3 = £ Velocity
v3
dz
7 w = —
^ dt
rjZ-Zsfc d<*z
°* H-Z*fc dt
Z-Z.rfc d°z
CTz H-Ztfc dt
i 0. _ Pa~Pr d(JPo
no P° „ „ j.
Pos ~~ PT &*
la a P~PT d(Jp
'" '" P.-PT dt
! fw-WrV -f^
""L-w r* rff
V7*-.? nT J
Vertical
Jacobian
1
H ~ z,fc
H
H-z#
Pos - PT
PoS
P* ~ PT
Pg
7ts -^ 7CT
im*
Remarks
Geometric height
H is the thickness of
model and a- is the
scaled height
Nondimensional height,
terrain-influenced
Nondimensional
reference pressure
Nondimensional
geostrophic pressure
Step-mountain ETA
^ (p0(W-PT)
6.4 Nesting Techniques
The nested grid CTM is needed to provide the required high resolution simulations. At present,
Models-3 CMAQ allows only static grid nesting. In static grid nesting, finer grids (FGs) are
placed (i.e., nested) inside coarser grids (CGs). The resolution and the extent of each grid are
determined a priori and remain fixed throughout the CTM simulation. Static grid nesting
conserves mass and preserves transport characteristics at the interfaces of grids with different
resolutions (Odman et al., 1995). It allows for effective interaction between different scales with
efficient use of computing resources. On the other hand, the nest domain is redefined during a
simulation with the dynamic nesting. Both static and dynamic nesting techniques allow one-way
or two-way exchange of information among FGs and the CG and periodically by independently
simulating CTMs at each grid (coarse or nested) with its own time step. The dynamic nesting
procedure is not implemented in the CMAQ system.
In one-way nesting, the primary concern is the mass conservation at the grid interface where
boundary conditions are input to the FG using the CG solution. The advective flux at the inflow
6-28
-------
EPA/600/R-99/03O
boundaries of the FG is the flux as determined by the CG solution that passes through this
interface. We also allow for time variation of the flux during the CG time step. This is
performed by computing the departure point of the last particle passing through the interface for
each FG time step. In the Lagrangian description, the mass crossing the interface is equal to the
spatial integral of the concentration distribution between the departure point and the interface.
Most flux-conserving advection schemes use the same Lagrangian concept to calculate the mass
transfer between grid cells (e.g., Bott, 1989). During each FG time step, we meter into the FG
the exact amount of mass that would have crossed the interface on the CG (Byun et al., 1996).
In two-way nesting, the concentrations are updated in each CG cell by averaging the
concentrations of all the FG cells overlapping with the CG cell. Special care is required to assure
strict mass conservation at the grid interface. The mass of some species (e.g., radicals) may no
longer be conserved because, when advancing the FG solution, we perform nonlinear chemical
transformations in addition to transport. However, the mass of the basic chemical elements such
as sulfur, nitrogen, and carbon must be conserved. The FG solution is used to compute the flux
of each element at the grid interface. When conservation principles are applied to the grid
interface, as described above, the CG concentrations near the interface must be corrected (when
the Courant stability limit is applied, only the first row of CG cells immediately outside the
interface need correction). This is done by renormalizing the concentration of each species based
on the assumption that the ratio of the species mass to element mass will remain the same before
and after the correction. This method is similar to the renormalization procedure used to make
slightly non-conservative chemical solvers strictly conservative, Alapaty et al. (1998) compares
different spatial interpolation schemes used for the two-way nesting.
Figure 6-5. Static Grid Nesting Used in CMAQ System
The CCTM provides a static nesting (see Figure 6-5) capability while maintaining a high level of
modularity by separately processing object codes for different grid domains and by enforcing the
protocol that each module reads its required input data independently from others. The scheme
6-29
-------
EPA/600/R-99/030
is also applicable for multiple and multi-level grid nesting. Multi-level nesting is a natural
extension of the single static grid nesting. Figure 6-6 presents a schema for multi-level nesting,
where three-levels are illustrated. The feedback processes update coarser grid concentrations at
each synchronization time steps of the nest grids. The information about the grid is
communicated to each process module through a set of FORTRAN include files specific to each
grid domain during compilation time. This allows use of the same process modules for different
grids. Customizing a nested model is as simple as preparing include files with grid dimensions for
each grid and a driver with the appropriate process calling sequence.
As seen in Figure 6-6, the only difference between the one-way and two-way nesting is whether
concentrations hi coarser grid simulations are updated with the finer grid simulations through the
feedback processors or not. Depending on the computer hardware and software configurations,
one could build a nested CTM model with one executable collectively simulating all the FGs and
CO, or with independent CTM executables for different grids that run simultaneously on
multiple CPUs accessing common data through appropriate I/O API. The latter approach relies
on the cooperating processors concept in a UNIX environment. As mentioned before, the
current CMAQ version lacks a feedback module, which is necessary for the two-way nesting.
(H_SynoJQiie)1
Update
IGricU
oQfid:
fa-id:
Emls:
Met:
J:
C:
EC:
Cone:
Sync
Proc:
Legend
coarse grid domain
fine grid domain
emissions file
meteorology files
photolysis rate
initial conditions
boundary conditions
concentration
synchronization
processor
Figure 6-6. Schematics for Multi-level Nesting (three levels illustrated)
6-30
-------
EPA/600/R-99/030
6.5 Summary
The CMAQ system achieves multi-pollutant and multiscale capabilities by combining several
distinct modeling techniques. The generalized governing conservation equations of the CCTM
allow transformation among various vertical coordinates (e.g., terrain-influenced geometric height,
or pressure) and transformation among various horizontal coordinates, especially map
projections (e.g., rectangular, Mercator, Lambert, and polar stereographic) by simple changes in a
few scaling parameters defining the boundary domain, map origin, and orientation. Therefore,
CMAQ can be configured to match the dynamic characteristics of the preprocessor
meteorological models. The CMAQ system uses a nesting technique and a plume-in-grid
approach to handle small scale air quality problems and subgrid scale plume dispersion and
chemical reactions, respectively. The multi-pollutant capability is provided by a generalized
mechanism reader and generalized chemistry solvers, linked cloud mixing and aqueous reaction
processes, and aerosol modules. The CCTM code uses a modular structure that allows for the
continuing improvement of the science and addition of new capabilities in a unified fashion. It
provides a natural disciplinary distinction among different science processes through which
developments in specific research areas can be readily incorporated.
As mentioned above, there remain a few implementation tasks, such as development of feed-back
modules for two-way nesting and adaptation of anholonomic coordinates (e.g., latitude-
longitude), which will provide additional functionalities in CCTM.
6.6 References
Alapaty, K., R. Mathur, and T. Odman, 1998: Intercomparison of spatial interpolation schemes
for use in nested grid models. Mon. Wea. Rev. 126,243-249.
Binkowski, F. and U. Shankar, 1995: The regional particulate model, Part I: Model description
and preliminary results. J. Geophys. Res., 100:26191-26209.
Black, T., 1994: The new NMC mesoscale Eta model: Description and forecast examples. Wea.
Forecasting, 9, 265-278.
Bott, A., 1989: A positive definite advection scheme obtained by nonlinear renormalization of
the advective fluxes. Mon. Wea. Rev. 117, 1006-1015.
Byun D. W., A. Hanna, C. J. Coats, and D. Hwang, 1995: Models-3 Air Quality Model
Prototype Science and Computational Concept Development. Trans. TR-24 Regional
Photochemical Measurement and Modeling Studies, 8-12 November 1993, San Diego, CA, Air &
Waste Management Association. Pittsburgh, PA., 197-212.
Byun, D.W., D. Dabdub, S. Fine, A. F. Hanna, R. Mathur, M. T. Odman, A. Russell, E. J.
Segall, J. H. Seinfeld, P. Steenkiste, and J. Young, 1996: Emerging Air Quality Modeling
Technologies for High Performance Computing and Communication Environments, Air Pollution
Modeling and Its Application XI, ed. S.E, Gryning and F. Schiermeier, 491-502.
6-31
-------
BPA/600/R-99/030
Byun, D.W., J. Young., G. Gipson., J. Godowitch., F. Binkowsk, S. Roselle, B. Benjey, J. Pleim5
J. Ching., J. Novak, C. Coats, T. Odman, A. Hanna, K. Alapaty, R. Mathur, J. McHenry, U.
Shankar, S. Fine, A. Xiu, and C. Jang, 1998: Description of the Models-3 Community Multiscale
Air Quality (CMAQ) model. Proceedings of the American Meteorological Society 78th Annual
Meeting, Phoenix, AZ, Jan. 11-16,1998. 264-268.
Byun, D. W., 1999: Dynamically consistent formulations in meteorological and air quality
models for multi-scale atmospheric applications: II. Mass conservation issues. J. Atmos. Sci., (in
print).
Carmichael, G.R., L.K. Peters, and R.D. Saylor, 1991: The STEM-II regional scale acid
deposition and photochemical oxidant model-L An overview of model development and
applications. Atmos. Environ. 25, 2077-2090.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madronich, P. Middleton, W.R. Stockwell, and C.J.
Walcek, 1987: A three-dimensional Eulerian acid deposition model: Physical concepts and
formulation, J. ofGeophys. Res., 92,14,681 -700.
Chang J. S., S. Jin, Y. Li, M. Beauhamois, C.-H. Lu, H.-C. Huang, S. Tanrikulu, and J. DaMassa,
1997: The SARMAP air quality model. Final Report, SJVAQS/AUSPEX Regional Modeling
Adaptation Project, 53 pp. [Available from California Air Resources Board, 2020 L Street,
Sacramento, California 95814.]
Coats, C.J., cited 1996: The EDSS/Models-3 I/O Applications Programming Interface. MCNC
Environmental Programs, Research Triangle Park, NC. [Available on-line from
http://www.iceis.mcnc.Org/EDSS/ioapi/H.AA.html.]
Harley, R.A., A.G. Russell, G.J. McRae, G.R. Cass, and J.H. Seinfeld. 1993: Photochemical
modeling of the Southern California air quality study. Envir. Sci. Technol. 27, 378-388.
Jacobson, M.Z. and R.P. Turco, 1994: SMVGEAR: A sparse-matrix vectorized Gear code for
atmospheric models. Atmos. Environ. 20, 3369-3385.
Mesinger, F., Z. I. Janjic, S. Nickovic, D. Gavrilov, and D. G. Deaven, 1988: The step-mountain
coordinate: model description and performance for cases of Alpine lee cyclogenesis and for a case
of Appalachian redevelopment. Man. Wea, Rev., 116,1493-1518.
Odman, M.T., R. Mathur, K. Alapaty, R.K. Srivastava, D.S. McRae, and R.J. Yamartino, 1995:
Nested and adaptive grids for multiscale air quality modeling. U.S. EPA Workshop on Next
Generation Models Computational Algorithms (NGEMCOM), Bay City, MI, June 1995.
Seaman, N. L., D. R. Stauffer, and A. M. Lario-Gibbs, 1995: A multiscale four-dimensional data
assimilation system applied in the San Joaquin Valley during SARMAP. Part I: Modeling design
and basic performance characteristics.J. Appl Meteor., 34,1739-1761.
6-32
-------
EPA/600/R-99/030
Seinfeld, J.H., 1986: Atmospheric Chemistry and Physics of Air Pollution, Wiley-Interscience,
New York.
Scheffe, R. D., and R, E. Morris, 1993: A review of the development and application of the
Urban Airshed Model. Atmos. Environ., 27B, 23-39.
Srivastava, R.K., D.S. McRae, M.T. Odman, 1995: Governing Equations of Atmospheric
Diffusion. Technical Report, MCNC [Available from MCNC-North Carolina Supercomputing
Center, P.O. Box 12889,3021 Comwallis Rd. Research Triangle Park, NC 27709-2889.]
Schwartz, E. S. and P. Warneck, 1995: Units for use in atmospheric chemistry. Pure & Appl.
Chem. 67, Nos 8/9, 1377-1406.
Toon, O.B., R.P. Turco, D. Westphal, R. Malone, and M.S. Liu, 1988: A multidimensional
model for aerosols: Description of computational analogs. J. Atmos. Set,, 45,2123-2143.
Venkatram, A., 1993: The parameterization of the vertical dispersion of a scalar in the
atmospheric boundary layer. Atmos, Environ. 27A, 1963-1966.
Yamartino, R.J., J.S. Scire, G.R. Carmichael, and Y.S. Chang, 1992: The CALGRID mesoscale
photochemical grid model -Part I. Model formulation, Atmos. Environ. 26A, 1493-1512.
Young, J.O., E. Sills, D. Jorge, 1993: Optimization of the Regional Oxidant Model for the Cray Y-
MP, EPA/600/R-94/065. [Available from U.S. Environmental Protection Agency, Research
Triangle Park, NC 27711.]
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
6-33
-------
EPA/6QO/R-99/030
Appendix 6A. Concentration Units Used for Air Quality Studies
As we have seen above, many different concentration units are used for air quality studies. In
this section we summarize the relations among the concentration units and their conversion
factors from one unit to another. For Models-3/CMAQ system we follow the International
System of Units (Systeme International, SI) as a framework for units in the formulations. The
fundamental assumption used here is that air and trace gases follow the ideal gas law, i.e.,
where Rg is the universal gas constant = 8.3 1 45 1 0 [ / l(mol • K)],
v is molar number,
Fis volume of the gas [m3 ],
p is pressure [Pa], and
Fis temperature
There are many different ways to express the amount of substance in the atmosphere. We
introduce most popular quantities and transformation relations among them are presented below.
Number Density, n
One way to express trace gas quantities is to count number of molecules in the unit volume. For
example, number of molecules of air in the unit volume, nair, is expressed as:
nair=~ (6A-2)
where valr is number of moles of air,
NA is the Avogadro's number = 6.0221367 x 1023, and
Similarly, number of molecules of trace gas per volume, i.e., number density of species «,, is
defined as:
n. = liE*. = VJN* — = JL r molecules m"3 1 (6A-3)
' V v^TIp, kBTl
where v, is number of moles of trace gas i in a given volume, and
kB = Rg I NA is Bolzmann's constant
Pi is the partial pressure of species i.
Molar Density, ct
The number of moles of air (cafr) and trace gas (c/) normalized for a unit volume of air are simply
defined as:
6-34
-------
EPA/600/R-99/030
c«*=% (6A-4a)
ci=^L (6A-4b)
with the unit [mole m~3]. Because the SI unit for the amount of substance is mol, this quantity
can be used conveniently for expressing chemical relationships.
Partial Pressure, pt
Although not used widely in atmospheric chemistry, the partial pressure has been playing an
important role specifying thermodynamic properties of the atmosphere, especially for water
vapor in the air. The Dalton's law states that the total pressure exerted by a mixture of gases is
equal to the sum of the partial pressure exerted by each constituent at the given temperature and
volume. Because we assume that each trace gas follows the ideal gas law, the partial pressure
can be used to express the trace gas quantity. The partial pressure of atmospheric constituent
gas is related to the number of moles per volume as:
p-Y&L = cRT .(6A-5)
and the standard SI unit for the partial pressure is Pascal [Pa].
Molar Mixing Ratio, m,
Often, the molar mixing ratio is used as a synonym for the volume mixing ratio, or the mole
fraction of a substance in air. Basically it is a unitless quantity. However, it is customary to
identify in terms of molar unit as [mole I mole]. Because the volume occupied by a mole of ideal
gas at given pressure is the same regardless of the constituent, the mole fraction is essentially
equal to the volume fraction. However, mole fraction is preferred because it does not require the
implicit assumption of the ideality of the gases, and more importantly because it is applicable
also to condensed-phase species (Schwartz and Warneck, 1995). For a given volume, the volume
mixing ratio of a trace species is expressed in terms of concentration units defined above as:
= -£-, (6A-6)
P
where p is the total pressure.
Because v,- « vair for trace gases, Equation 6A-6 can be approximated as:
Vi Ci Pi
V C 17
atr air ratr
6-35
-------
EPA/600/R-99/030
When dealing with trace gases in the real atmosphere, the contribution of moisture can be used in
the definition of mixing ratio. Therefore, pair represents total pressure of the atmosphere which
includes vapor pressure of water while the contribution of other trace gases are neglected.
Because the variation caused by the moisture can amount to several percent, some researchers
prefer to use dry air when expressing trace gas mixing ratios. However, in the Models-3/CMAQ
system, we use the trace gas mixing ratio with respect to the moist air because we rely on the
continuity equation for the total air density to represent atmospheric mass conservation.
Mass Mixing Ratio,
-------
Table 6A.1. Conversion formula among various trace gas concentration units used in the CMAQ system
Oi
number density,
[molecules/m3]
molar density, c,
[mol/m?]
partial pressure,
Pi [P Bi Pi =1" *l ~M~W
..^ .,.<£>.-.<«. .-fc), . ,.»..,,
«-io.^JW, ft-io.W), «-'o.y«". UJ. ,,=.O.(P.^
Note: Values for the Avogadro's constant (A^j) is 6.0221367xl()23 [number/mol], the universal gas constant (Rg) is 8.314510 [J/mol-K], and the Boltzmann's
constant (kg) is defined with R/N^. Molar mixing ratio is equivalent to volume mixing ratio and their units [ppmV], [ppb], and [ppt] are short hand notations for
[fi mol mot"1 ], [n mol mo/'1], and [p mol mol~l], respectively. The unit for mass mixing ratio, [ppm], is a short hand for [10"^ kg kg"*]. Both and may
represent values for dry air or moisture air depending on applications. However, in one processing system, their usage should be consistent. Mai-r and M,-
(molecular weight of species i) are in [g mo^ ] and pajr is in [kg m"^}.
M
o.
o
1
-------
EPA/600/R-99/030
Chapter 7
NUMERICAL TRANSPORT ALGORITHMS FOR THE COMMUNITY MULTISCALE
AIR QUALITY (CMAQ) CHEMICAL TRANSPORT MODEL IN GENERALIZED
COORDINATES
Daewon W. Byun,* Jeffrey Young,** and Jonathan Pleim*
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
M. Talat Odman* and Kir an Alapaty
MCNC-Environmental Programs
P.O. Box 12889,3021 Cornwallis Road
Research Triangle Park, NC 27709-2889
ABSTRACT
The transport processes in the atmosphere primarily consist of advection and diffusion, except
for the mixing of pollutants by the parameterized subgrid-scale clouds. In this chapter, numerical
algorithms for advection, vertical diffusion, and horizontal diffusion implemented in the
Community Multiscale Air Quality (CMAQ) chemical transport models are discussed. To
provide the CMAQ system with multiscale capability, we have formulated the transport
processes, both advection and diffusion, in conservation (i.e., flux) forms for the generalized
coordinate system. Therefore the numerical transport algorithms implemented in CMAQ will -
function under a wide variety of dynamical situations and concentration distribution
characteristics. Users can not only choose transport algorithms from optional modules available
in CMAQ, but also are encouraged to experiment with their own algorithms to test different
numerical schemes for air quality simulations.
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Daewon W. Byun, MD-80, Research Triangle Park, NC 27711.
E-mail: bdx@hpcc.epa.gov
"On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Resent Affiliation: Georgia Institute of Technology, Atlanta, GA.
-------
EPA/600/R-99/030
7.0 NUMERICAL TRANSPORT ALGORITHMS FOR THE COMMUNITY
MULTISCALE AIR QUALITY (CMAQ) CHEMICAL TRANSPORT MODEL IN
GENERALIZED COORDINATES
In this chapter, we study numerical algorithms for the transport processes in a turbulent
atmosphere. Many of the contents provided here are the results of a collaborative research
project, the EPA's Cooperative Agreement CR822053-01 with MCNC-Environmental Programs
(Exploratory Research on Air Quality Modeling Techniques: Research on Numerical Transport
Algorithms for Air Quality Simulation Models), and other related in-house projects at EPA.
Readers are referred to Alapaty et al. (1997), Byun (1999a, b), Byun and Lee (1999), and Odman
(1998) for additional information.
In principal, the transport process consists of advection and diffusion that cause the movement
and dispersion of pollutants in space and time. Transport of pollutants by the parameterized
subgrid-scale cloud modules is not considered here. We have assumed that the transport of
pollutants in the atmospheric turbulent flow field can be described by means of differential
equations with appropriate initial and boundary conditions. In Eulerian air quality models, the
transport process is modeled using numerical algorithms. These numerical algorithms for the
advection and diffusion processes must satisfy several properties that are essential for making
useful air quality simulations. As with all numerical methods, the numerical schemes for solving
the transport equations must meet convergence conditions and correctly model the conservation,
dissipation, and dispersion properties of the governing equations. A numerical scheme is said to
be convergent if the solution approaches the true solution of the corresponding partial differential
equation as the grid spacing and time-step size become infinitesimally small. Thus, a convergent
numerical scheme can provide a numerical solution of any desired accuracy within finite precision
bounds by reducing the grid spacing and the time-step size. For linear equations, consistency and
stability are both necessary and sufficient conditions for convergence (Lax's equivalence
theorem). In practice, machine precision and the computational resource availability limit the
reduction of grid spacing and time-step size. Therefore, numerical errors associated with using
limited grid spacing and time-step sizes must be of concern.
There have been many studies on the numerical advection algorithms used in air quality models.
The reason it attracted so much attention is that the equation is hyperbolic in nature and spatial
discretization of the solution generates a finite number of Fourier modes that travel at different
speeds and leads to constructive and destructive interference. If the high wave-number Fourier
modes are damped significantly, then numerical diffusion becomes prevalent. Solving the
diffusion equation, on the other hand, is a lot safer because the stiffness matrix is diagonally
dominant and the discretized solution is stable and sign preserving for a relatively wide range of
conditions (Chock, 1999).
7-2
-------
EPA/600/R-99/030
Transport processes are of central importance in turbulent flow studies, and in the literature there
are numerous transport algorithms that have different numerical characteristics and varying
degrees of accuracy and computational complexity. The skill needed here is to select appropriate
numerical schemes that provide solutions with the desired accuracy at reasonable computational
cost. This document does not intend to provide an extensive review of the transport algorithms
used in air quality modeling. Instead, we describe several popular numerical schemes
implemented in the Community Multiscale Air Quality Chemical Transport Model (CMAQ
CTM or, hereafter, CCTM), expecting users to choose the algorithms appropriate to their own
study objectives. We offer a few examples of good transport algorithms and present some key
numerical characteristics users should look for. With this information, users can find the best
algorithms through evaluation processes, and may even bring in their own algorithms to build a
transport model for their applications. To provide the CMAQ system with the multiscale and
multi-pollutant capabilities, we strive to incorporate schemes that can function under a wide
variety of dynamic situations and distribution characteristics (e.g., distributions of different
primary species and secondary species are quite distinct). Also, the schemes should be efficient
in the use of computer time and storage. Selected numerical transport algorithms for horizontal
and vertical advection and for vertical and horizontal diffusion are described below.
7.1 Numerical Advection Algorithms
Numerical advection algorithms for air quality models should satisfy several computational
requirements.
• They should be free of mass conservation errors to accurately account for pollutant
sources and sinks. .
• They should have small numerical diffusion to minimize the spread of a signal in every
direction and the smoothing of spatial gradients.
• They should also have small phase errors since disturbances that propagate at different
speeds produce spurious oscillations.
• Given initial positive concentrations, the schemes should be positive-definite (i.e., they
should not produce negative concentrations.
• They should be monotonic (i.e., they should not produce new extrema).
While it is essential that the schemes be positive-definite, this alone may not be sufficient
because the monotonic property, for example, is just as desirable for air quality modeling.
7-3
-------
EPA/600/R-99/030
Numerical algorithms have not been able to satisfy all the requirements listed above, and they are
imperfect, with varying degrees of accuracy. Advection schemes with different properties
introduce different errors, all of which are sources of uncertainty in air quality model predictions.
Before recommending its use, it is critical to identify which of the computational properties a
scheme possesses. Because an advection scheme with all the desired properties is not currently
available, a user needs to select a scheme with the most desirable properties and greatest
efficiency to meet the needs of the application.
•i. ' i miii :,•,"' to ' ' ' - '
7.1.1 Conservation Form Equation for Advection
The atmospheric advection process is expressed in conservation (flux) form as:
(
where *P< is the concentration of trace species / coupled with the coordinate Jacobian. Refer to
Chapters 5 and 6 for the definition of symbols used in Equation 7-1. For convenience, the
advection process is decomposed into horizontal and vertical advection processes, with the
fractional time-step implementation:
a/-. ,,*) ^^
where v1, v2 and v3 are contravariant components of wind velocity. Splitting of the three-
dimensional (3-D) advection into the horizontal and vertical components will lead some
difficulties, such as the representativeness of the mass continuity and setting up of proper
boundary conditions for non-orthogonal horizontal and vertical directions when simulating a
region with complex topography.
Many models further split the horizontal advection equation in two directions and solve for two
one-dimensional equations, one in each direction, using the solution of one as the initial condition
of the other. We refer to this scheme as a one-dimensional (1-D) algorithm. Others solve the
two-dimensional (2-D) form directly. Although using 1-D schemes is very common, it has been
found that problems can arise due to this additional splitting (Flatoy, 1993, and Odman and
Russell, 1993). Although 2-D schemes may be more desirable in this regard, fewer have been
tested and they are often more difficult to implement and less computationally efficient than 1-D
schemes. Also, there are general conditions in which the splitting scheme is actually more stable
7-4
-------
EPA/600/R-99/030
and accurate than the non-splitting case for higher-order approximations because the splitting
scheme intrinsically contains cross-spatial derivatives whereas the non-splitting scheme would
not (Leith, 1965). Yanenko (1971) has shown that time-splitting is second-order accurate if the
one-component advection operators commute. Alternating the sequence of operations would be
quasi-second-order accurate in the case of non-commutativity (Chock, 1999). Here, only 1-D
schemes will be discussed. When using appropriately interpolated contravariant wind
components, the 1-D advection in the generalized coordinate system is equivalent to the 1-D
equation in the Cartesian coordinate system. Therefore, it is sufficient here to discuss advection
algorithms in Cartesian coordinates.
The 1-D advection equation written in the Cartesian coordinate system is:
0 (7-4)
at ax
Equation 7-4 is the flux (or conservation) form and the quantity Fx = u
-------
EPA/600/R-99/030
advection alone, an implicit scheme may not have the Courant number restriction for stability
(Chock, 1999).
7.1.2 Classification of Advection Schemes
Numerical advection schemes in the literature were developed using several different approaches
(e.g., Chock and Dunker, 1983, and Chock, 1985,1991). Following Rood (1987), we classify
these schemes based on the methods used in their formulations. However, reviews in the
literature may not capture the most recent developments in advection research. Depending on
the methods used, the schemes may be classified as:
• Finite difference schemes;
• Finite volume schemes;
« Flux corrected schemes;
• Lagrangian Schemes;
• Finite element schemes; or
• Spectral schemes.
The distinction is somewhat arbitrary and only meant to convey the key intrinsic features of the
scheme. Current trends in advection scheme development show a merging of the methods to take
advantage of the most desirable properties of several schemes. For example, the Characteristic-
Galerkin method (Childs and Morton, 1990) combines the best of the finite element and
Lagrangian methods. Flux corrections are being used in the framework of finite element and
spectral schemes (LOhner et al., 1987). Also, the classical finite difference schemes are being
abandoned in favor of modern finite volume schemes. Refer to Odman (1998) for details of the
classification.
7.1.3 Description of Advection Schemes in CCTM
In this section we describe the schemes that are available with the first release of the CCTM
codes in the following order; the piecewise parabolic method (PPM), the Bott scheme (BOT),
and the Yamartino-Blackman Cubic scheme (YAM).
Odman (1998) provides additional descriptions of the Smolarkiewicz scheme (SMO), the
accurate space derivative scheme (ASD) (Chock, 1991, and Dabdub and Seinfeld, 1994), the flux-
corrected transport, the semi-Lagrangian method, and the chapeau function scheme with Forester
filter. These codes are not integrated into the CMAQ system yet, but along with other advection
modules, will be added to the system in the near future.
7-6
-------
EPA/600/R-99/030
To simplify the discussion, we will consider a uniform (i.e., constant &x = &Xj ) and staggered
grid ((pj represents the grid cell average of the concentration, while Uj+j/2 is the advection velocity
defined at grid cell interfaces). While discussing the finite-volume schemes (the piecewise
parabolic method, the Bott scheme and the Yamartino scheme) below, we use the explicit flux
formula presented in Equation 7-5. Further, a nondimensional coordinate r] is defined as 77 = (x -
xj-j/2)/A*> so that, in grid cell/, 0 £ r\ < 1. Now, suppose that the concentration has a certain
distribution q^rj) in each grid cell. Depending on the direction of the velocity, the flux FJ+ 1/2 can
be expressed as;
' A£ j-
At . a
l-fc.1/2 ' (?_7)
A Pi+\r ^ '
Ax r
where flj+1/2 is the Courant number at the right boundary of grid cell/.
The conditions of high-order accuracy and freedom from spurious oscillations are difficult to be
achieved simultaneously. The usual way to satisfy one of these conditions without significant
violation of the other is to introduce a correction mechanism. Typically, this mechanism is
provided by nonlinear flux-corrections, or by nonlinear filtering. In adveetion schemes, such
adjustments are either applied implicitly through the solution or explicitly as a subsequent step
to the linear solution. There is extensive literature on both solution algorithms (linear and
nonlinear) and explicit nonlinear mechanisms.
7.1.3.1 Piecewise Parabolic Method (PPM)
In the piecewise parabolic method (Colella and Woodward, 1984) the concentration distribution
is assumed to be parabolic in any given grid cell. In terms of the grid cell average concentration <
*
and the predicted values of the parabola at the left and right boundaries of the cell cp^- and (p^-,
this distribution can be written as:
(7-8)
Since the initial cell average is known, the construction of the parabola involves the determination
of the edge values. First, an approximation to (p at xj+i/2 is computed subject to the constraint
7-7
-------
EPA/600/R-99/030
that its value is within the range of the values at the neighboring cells. For the uniform AXJ, a
first guess for q>j+i/2 is estimated with:
In smooth parts of the solution away from extrema, (pij+i ~ X (7-10)
*=o
The polynomial can be made area-preserving by requiring:
7-8
-------
EPA/600/R-99/030
i-rt / ,
Vj+i = I IX*^» ' = 0,±l,±2,...,±^ (7-11)
/ 4=0 2 .
over a stencil of 7+1 grid cells by varying the value of z. The solution to this linear system yields
the coefficients a/^. The coefficients obtained this way for a quadratic (/=2) and quartic (1=4)
together with those of the donor cell (or upwind) scheme, and Tremback's scheme with second-
order polynomials (Tremback et al., 1987), are listed in Table 7-1,
Using Equation 7-7, integrating the polynomial of Equation 7-10 between appropriate limits, we
arrive at a first estimate of the fluxes. Finally, to make the scheme positive-definite, the total
outflux from cellj is limited by requiring that it should be positive and less than what the
available mass in the cell would allow:
AT • ' - •
0 < Foat < ——r (1-\2^
/ A, i ^ '
1 Ar
The outflux Fjmt is a combination of the boundary fluxes and its expression depends on the sign
of the velocities. In the CCTM implementation, we used, fourth-order polynomials as
recommended by Bott (1989) except for the boundary cells. The scheme is receiving increasing
attention in current air quality models because of its high accuracy and low computational cost.
Recently, a monotonic version of the scheme was also developed (Bott, 1992) and the time-
splitting errors associated with the use of one-dimensional operators in multidimensional
applications were reduced (Bott, 1993). Monotonicity is obtained by directly replacing the
positive-definite flux limiter of the original approach by new monotone flux limiters as:
>" ?; , if u > 0
i
+,,fl> , if u < 0
Although the new flux limited Bott scheme yields monotonic results, there is an inherent mass
conservation problem. This problem is directly related to the flux limiting that takes place. Near
the leading edge of a sharp wave the use of second or higher order polynomials causes an
underestimation of a certain advective flux, Fk-m- When this flux is not corrected it is less than
FfcH/2, and an undershoot occurs in cell k, as experienced with the original algorithm (Bott, 1989).
The motivation for the monotone flux limitation is to avoid such undershoots. However, there
are cases when the monotone flux limiter leaves the underestimated flux intact. Instead of
increasing the underestimated flux, the limiter reduces the advective flux downwind, F^i/2, in
order to avoid an undershoot hi cell k. This eventually reduces the net flux out of the domain
resulting in an accumulation of mass in the domain,
7-9
-------
EPA/600/R-99/030
Table 7-1, Coefficients of the Polynomials Used in Each Scheme
Donor Tremback-2 Bott-2 Bott-4
Cell
r(**i ~2
-------
EPA/600/R-99/030
The positivity of ^-(77) is ensured by various mechanisms. First, when 2~
-------
EPA/600/R-99/030
On the other hand, Odman (1998) introduced a "no condition at outflow boundary" [sic], for
Bolt's scheme where the scheme is modified to remove the need for the concentration j, cp 2 and (p 3 is used in
Cell 2 and a fourth-order polynomial is used in Cell 3. Mathematically, this condition is more
correct than the others. However, the order of the polynomial is reduced to one at the boundary
while the other conditions use a second-order polynomial to compute the flux out of the domain.
Because of the lack of generality of this approach, we have not implemented Odman's boundary
scheme in the^CCTM. The improved positive-definite zero-flux outflow boundary condition
scheme essentially reproduces his results without having to rely on the modified advection
algorithms near boundary.
•^r
%
X
0
-------
EPA/600/R-99/030
Table 7-2, Summary of Performance Measures Used to Test the Effects of Numerical
Advection.
-------
EPA/600/R-99/030
SMO produces ripples upwind from the pulse and leads to average and RMS errors larger than
other schemes.
Table 7-3. Gaussian signal test (Abridged from Odman, 1998)
Scheme
Peak Ratio
Background
Mass Ratio
Distribution
Average Error
RMS Error
ASD
0.99
0.05
1.00
0.99
0.08
0.01
EOT
0.87
0.01
' LOO
0.93
0.87
0.18
BOT-M
0.74
0.05
1.02
0.83
1.38
0.27
PPM
0.69
0.05
1.00
0.79
1.16
0.17
SMO
0.61
0.02
1.00
0.66
2.21
0.50
YAM
0.98
0.05
1. 00
0.92
0.51
0.12
7.1.5.2 Rotating Cone Test
• ^ . i
"... ' ';' • * j
In this test, a cone-shaped puff is introduced into a rotational flow field and followed for a certain
number of revolutions. The exact solution is a rigid-body rotation of the puff without any change
to its original shape. Various errors can be revealed in this test. For example, numerical diffusion
(or dissipation) manifests itself in the drop of the peak height during rotation. Also, by
observing the location of the peak, one can determine the leading or lagging phase-speed errors.
A 32x32 grid is used for this test (i.e., -!6Ax
-------
EPA/600/R-99/030
PPM predict similar peak heights (65% and 61%, respectively), but the shape distortions look
very different in each case. PPM has the worst peak clipping effect but the resulting shape has
the smallest base span among the three schemes. SMO is clearly the most diffusive scheme; it
also introduces a ripple upwind from the cone. Table 7-4 summarizes the performance measures
at the end of two rotations. Since there is no shear in the flow field, BOT-M and the two-
dimensional version of Bolt's scheme (BOT2D) (Bott, 1993) produce identical results. Again,
the mass conservation problem is revealed with BOT-M, BOT preserves 87% of the peak height
(third best after ASD and YAM), but it leads to ripples with an amplitude of 3% of the original
peak height. Performances of BOT-M and PPM are comparable in predicting the peak. But
PPM produces the lower distribution ratio, smaller absolute average and RMS errors. Notice
that the comparison results obtained from this test are very similar to those of the Gaussian
signal test. Additional test results such as skew advection of a point-source plume and advection
with shear flow are available in Odman (1998). Also, effects of density distribution on the
numerical advection are studied with a set of linear flows in Byun and Lee (1999). The solvers
integrated into the CCTM are BOT, PPM and YAM. Although ASD has very high accuracy
except for the 2 Ax wavelengths, ASD is neither strictly mass conservative nor monotonie. It is
also the most CPU-intensive scheme (taking about 4-5 times longer than BOT). In addition to
BOT's overall performance, results reported in Odman (1998) for a broad series of tests showed
that BOT had the best computational performance of all the schemes tested. However, because
of the concerns over the non-monotonicily about BOT and the mass conservation problem and
diffusive nature of BOT-M, we chose PPM for a number of demonstration executions (Byun et
al., 1998). Similar testing with BOT and YAM is underway. We intend to integrate other
methods into the CCTM at a later time.
7.1.6 Vertical Advection
Algorithms in the CCTM for vertical advection are essentially the same as those for the one-
dimensional horizontal algorithms. However, the vertical advection is performed in terms of the
generalized vertical coordinate in CMAQ. The contravariant vertical velocity component is used
as the transport wind for the irregular vertical grid spacing (usually expanding with altitude)
represented in the generalized coordinate. For the irregular grid, the computational time-step
should satisfy the CFL condition:
Vt-i
V »-i.
As in the case of Cartesian representation, we assume there is no mass exchange by advection
(i.e., v3 = 0) at the top and bottom boundaries of the model. Therefore, there is no need to apply
special algorithms for the boundary process. Because vertical grid spacing is usually irregular in
7-15
-------
EPA/600/R-99/030
most air quality models, a scheme that can accommodate irregular spacing must be used for
vertical advection. This means that numerical algorithms used to represent the vertical advection
may be different from that used for horizontal advection. For example, when the ASD algorithm,
which requires equal spacing hi the computational domain, is used for horizontal advection, a
numerical algorithm that allows irregular spacing (e.g., BOT and PPM) would need to be used for
vertical advection.
7.1.7 Adjustment of Mass Conservation Error
Recently, Byun (1999a, b) has highlighted the importance of dynamic consistency in
meteorological and air quality modeling for multiscale atmospheric applications. Mass
consistency quantifies how well the density and wind fields satisfy the continuity equation for
air. One of the fundamental requirements for the numerical transport algorithms used in air
quality models is the conservation of trace species in the domain. Ideally, the input
meteorological data for air quality simulations should be mass consistent. However, using
numerical models with highly parameterized physical and cloud algorithms, inappropriate set of
governing equations, misapplication of four-dimensional data assimilation (FDDA) schemes, or
..1 .»•'.• .... • • • ,i •„(„ • t
using incomplete objective analysis methods to characterize the atmosphere for a CTM could
result in meteorological conditions that are not mass consistent. In this situation, even precisely
mass conserving numerical algorithms may fail to conserve trace species mass in the domain.
Preferably, the mass inconsistency must be minimized before air quality simulation using a
suitable diagnostic relation or a variational wind field adjustment scheme as discussed in Byun
(1999b) and in Chapter 5, For certain vertical coordinates with appropriate dynamic
assumptions, the diagnostic methods can be used within the CTM to provide mass consistent
wind data. This is accomplished usually by adjusting the vertical wind component for the
advection process. However, the variational methods are applied during the meteorological data
preparation stage, instead of inside CTMs, mostly due to the computational efficiency reasons.
Whether the meteorological data are mass consistent or not, Byun (1999b) has shown that the
tracer mixing ratio must be conserved as a precondition for the tracer mass conservation. He has
reported several adjustment schemes used in current air quality models and proposed a two-step
time splitting numerical algorithm that satisfies the mixing ratio conservation equation. In the
event the meteorological data are not mass consistent, the mixing-ratio conservation scheme is
demonstrated to be useful for photochemical air quality models where chemical production and
loss terms are computed using molar mixing ratio. For this purpose, Equation 7-1 should be
modified as follows to conserve trace species:
ll +
-------
EPA/600/R-99/030
where Qp is the density error term for the meteorological data. This is a necessary condition for
the tracer mass conservation. The total tracer mass of the domain is conserved only when the
additional condition that total air mass of the domain is conserved, i.e.:
0 (7-21)
aa m
where BQ represents the boundary of a computational domain. However, even this condition
does not guarantee the cell base conservation of tracer mass except for the case of uniform mixing
ratio, because JJJV" — ^dV = JJJg,. — \QpdV * q{ \\\~\QpdV in general. Because of this, the
-------
EPA/600/R-99/030
Table 7-5. Trace Gas Mass Correction Schemes
Symbol Correction Method Mass correction algorithm
AO No correction
AI Advection of unity „,. _
-------
EPA/600/R-99/030
desirable because the interpolation of meteorological quantities in vertical direction would alter
the original turbulent flux exchange characteristics.
7.2.1 Closure Problem
Because of the stochastic nature of atmospheric motion, the primitive equations in the set
describing the atmosphere are averaged to form a set of deterministic equations before they can be
solved numerically. The decomposition of velocity components and concentrations into mean
and turbulent terms and the application of ensemble averaging produces Reynolds flux terms in
the species mass continuity equation. Introduction of the Reynolds flux terms generates a new
problem set in which the number of unknowns is larger than the number of equations. This
closure problem is caused by the attempt to represent nonlinear processes such as momentum
advection using a linear decomposition such as the Reynolds decomposition. We describe some
approaches on the closure of the Reynolds flux terms below.
7.2.1.1 Local Closure
Local closure assumes that turbulence is analogous to molecular diffusion, i.e., that an unknown
turbulence flux at any point in space is parameterized by values of known quantities at the same
point (Stull, 1988). First order closure retains the prognostic equations for only the mean
variables such as wind, temperature, humidity, and trace gas concentrations while the second-
order moments (Reynolds fluxes) are approximated. An example of a local closure scheme is the
approximation of Reynolds flux terms using a gradient transport theory, or a mixing length theory
resulting in an eddy diffusion method. One of the problems with the gradient transport theory is
finding a rational basis for parameterizing the eddy diffusivity. Also, the theory fails when
eddies larger than the grid size are present, like they are in a convective boundary layer.
The so-called one-and-a-half order closure retains the prognostic equations for the mean variables
and adds equations for the variances of those variables. The set of one-and-a-half order equations
is obtained by simplifying the full second-order turbulence equations. Instead of the velocity
component variance equations, the turbulent kinetic energy (TKE) equation is often used. By
including the variance equations, we have increased the number of unknowns that need to be
parameterized compared to the first-order closure approach. However, the benefit is that the
eddy diffusivity can be parameterized not only with the mean quantities but also with the TKE
and the temperature variance which characterize turbulence intensity. However, if an air quality
model is based on one-and-a-half order closure in a true sense, the prognostic equations of the
variances for the tracer concentrations should be included explicitly. In practical Eulerian air
quality models that deal with photochemical problems, additional prognostic variance equations
for the tracer species are very expensive computationally. Also, the additional closure problem
7-19.
-------
EPA/600/R-99/030
must be dealt with by parameterizing the Reynolds average terms involved with the variances for
tracer species. Therefore, the one-and-a-half order closure for an air quality model often actually
means that the diffusion equations for the tracer species are formulated with first order closure,
while eddy diffusivities (if gradient theory is applied for the closure) or turbulent fluxes (if non-
local flux-based closure is used) are estimated with the TKE information from a meteorological
model with one-and-a-half order schemes for wind, temperature, and humidity.
The set of second-border turbulence equations includes all the second moment terms. To derive
these terms, parameterizations on a full set of third-order moments are required. Similar to the
first-order case, the second-order closure approximates terms involving third moments. Several
basic closure assumptions such as down-gradient diffusion, return to isotropy, and turbulent
dissipation in the inertial subrange are used in the parameterization of the third moment terms.
These parameterizations must be valid, especially, for the scales of the energy-containing eddies
that are sensitive to atmospheric stability. Measurements of high-order moments in the real
atmosphere are difficult because of large scatter in the direct flux measurements and because a
long averaging time or a very large sample size of data is required because the events with a much
lower probability of occurrence must be gathered to estimate higher-order moments using eddy-
correlation methods. For air quality applications, especially for a complex chemical reaction
system, the technique requires too many ad-hoc assumptions that cannot be confirmed by
observations or other theoretical reasoning. In addition the second-order closure incurs a
prohibitively high computational cost. As before, we can solve the first-order tracer diffusion
equations with the variances and covariances for wind components, temperature, and humidity
from the second-order meteorological models. For air quality applications, the true second-order
closure formulation solves for the cross-species covariances explicitly. Some researchers have
attempted second-order closure for simple chemical mechanisms with a limited number of
photochemically reactive species. The introduction of additional parameterizations for the third-
order moment terms among the tracer species themselves and wind components (about which we
lack sufficient knowledge) and the added cost of solving for a large number of covariance terms
make this scheme impractical and prohibitively costly for operational Eulerian photochemical
models.
7.2.1.2 Non-local Closure
Non-local closure recognizes that larger-size eddies can transport fluid across finite distances
before the smaller eddies have a chance to cause mixing. This advection-like concept is supported
" " '"!!• ' " nl! ' ' ' ' ' , " ,;
by observations of thermals rising with undiluted cores, finite size swirls of leaves or snow, and
the organized circulation patterns sometimes visible from cloud photographs.
7-20
-------
EPA/600/R-99/030
Two main approaches of non-local closure methods are the transilient turbulence theory and the
spectral diffusion theory. Both allow a range of eddy sizes to contribute to the turbulent mixing
process. The spectral diffusion theory attempts to simulate mixing process by transforming
signals into a spectral space. For example, a spectral diffusion model of Otte and Wyngaard
(1996) represents the mean variables within the planetary boundary layer (PBL) by a truncated
series of Legendre polynomials. The first Legendre mode represents the layer average, and
additional modes add structure to the vertical profiles. Only a few modes are necessary to
resolve vertical profiles comparable to Mgh resolution diffusion models. However, the need to fit
a different number of spectral modes for each trace species makes the scheme less attractive for
air quality application and thus it have not been considered here. The transilient turbulence
theory (e.g., Stull, 1988) is a general representation of the turbulent flux exchange process. The
Latin word transilient, meaning to jump over, is used since turbulent eddies that exist in the PBL
can transport mass and momentum directly across several grid layers. A variety of mixing
processes can be modeled with the transilient scheme depending on the form of the transilient
matrix. Examples include complete mixing, top-down/bottom-up mixing, asymmetric convection
mixing, small-eddy mixing, cloud top entrainment, a detraining updraft core, patchy turbulence,
no turbulence, or eddies triggered by the surface layer. Non-local closure is most suitable for
describing vertical turbulence mixing process, which should represent turbulent diffusion and
atmospheric transport by eddies of different sizes simultaneously.
In the following, we describe only the first-order turbulent mixing schemes. The true one-and-a-
half and the second-order closure schemes are not discussed here. We organize the description of
vertical mixing algorithms based on the details of the turbulence parameterization; the eddy-
diffusion form or the Reynolds-flux form. Both the eddy-diffusion and the Reynolds-flux forms
are capable of accommodating information from higher order turbulence closure for momentum
and other meteorological parameters such as potential temperature and humidity. The vertical
diffusion modules in CMAQ will include two different ways of parameterizing the eddy
diffusivity (using PBL similarity theory and using TKE) and three flux form non-local algorithms
(Blackadar, ACM, and Transilient Turbulence).
7.2.2 Computing Vertical Mixing with the Eddy Diffusion Formulation; K-Theory
The eddy diffusion algorithm in the CCTM computes the following:
d
dt
vdiff
-x\ t>33
d?
(7-23)
7-21
-------
EPA/600/R-99/030
where K33 is the contravariant vertical component of eddy diffusivity in the generalized
coordinates. The contravariant eddy diffusivity is related to the diffusivities in Cartesian
coordinates as:
where
(7-24)
. Here we focus on the parameterization of the eddy diffusivity, K , in a
generic Cartesian vertical coordinate z (geometric height h). Parameterizations of the horizontal
eddy diffusivity are described later. However, in the current CMAQ implementation, the term
involving horizontal diffusivity KH in Equation 7-24 for the estimation of K33 is neglected.
• ,i ' ' .' • 't • . , • , • • • • ., . .. •.. •
Evaluation of the effects of this simplification is left for future work.
7.2.2.1 Parameterization of Vertical Eddy Diffusivity KK with PEL Similarity Theory
There are several eddy diffusivity parameterizations using different similarity theories. Since
these are. somewhat similar, we consider the formulations suggested by Businger et al. (1971) and
Hass et al. (1991) to represent the turbulent process in the surface layer and mixed layer.
Previous studies (Chang et al,, 1987, and Hass et al., 1991) indicated that this type of
formulation can represent turbulent mixing in air quality models adequately. With If-theory, we
assume that trace species have non-dimensional profile characteristics similar to potential
temperature,©, i.e., Ka = Kf,. We briefly describe the surface and boundary layer similarity
theory used for the parameterization of eddy diffusivity for different stability regimes of the
PBL below. The stability regime is defined with a nondimensional number z/L, where z is the
height above the ground and L is the Monin-Obukhov length.
For the surface layer, the non-dimensional profile functions of the vertical gradient of © are
expressed as:
0A = Pr0(l + ph —) for moderately stable conditions (1 > z/L > 0) (7-25a)
L*
0A=(l-yft.i)-"J for unstable conditions (z/L< 0} (7-25b)
LI
where Pr0 is the Prandtl number for neutral stability and fth and yh are coefficients of the profile
functions determined through field experiments. In addition, following Holtslag et al. (1990) we
7-22
-------
EPA/600/R-99/030
add a function for the very stable condition (z/L > 1) to extend the applicability of the surface
layer similarity:
Parameterizations for eddy diffusivity for the surface layer can be shown as:
. .. • _, ku,z • - ' • . .-••.'- ... . - • ..
K»=. • (7-26a)
Vh(ZfL) . . . . • . .
where u* is the surface friction velocity.
For the PEL (above the surface layer), eddy diffusivity is parameterized with:
K" = for .y> 0 (stable) (7-26b)
L
Kh=kw.z(l-z/K) for — < 0 (unstable) (7-26c)
JL* !
In the above expressions, h is the depth of the boundary layer, £the Von Karman constant, and
w* the convective velocity. Refer to Chapter 12 for the method used to estimate the PBL height
in the CMAQ modeling system.
These parameterizations for Kf, are sensitive to the boundary-layer height (h) and surface-layer
height. Therefore, when the vertical resolution is too coarse in the boundary layer, using a
"representative" eddy diffusivity together with the mean-concentration gradient at the interface
seems to be more appropriate for the estimation of the diffusive flux. In fact, the diffusive flux
across the interface can be estimated more accurately with the mean diffusivity and mean
concentration gradient than with local diffusivity and mean concentration gradient; the former has
an error of O[(A^f } (at best) while the latter has an error of 0[4£] • To estimate, the
"representative" eddy diffusivity at the layer interface, integrated eddy diffusivity formulas are
used as in RADM and CMAQ. They are summarized in the following equations (Byun and
Dennis, 1995)
• Surface Layer
(a) Stable conditions:
7-23
-------
EPA/600/R-99/030
-------
EPA/600/R-99/030
(7-29)
g
4
where the Richardson number is defined as^iB = - . ..^ .......... - - , Ko is the background value set at
fc)0& Az^
1 m2 s"1, and 5 is the vertical wind shear. Effects of the parameterization, Equation 7-29, on air
quality simulation will be evaluated in near future. Furthermore, not much is known of the eddy
diffusivity formulas for strong stable condition (i.e., z/L > 1) and therefore require further
research in this area.
7.2.2.2 Estimation of Vertical Eddy Diffusivity Using Turbulent Kinetic Energy
Various forms of higher order closure schemes are becoming common in mesoscale meteorology
models. Typically, such models are referred to as TKE models. The simplest form of a TKE
model, which Mellor and Yamada (1974) referred to as level 2.5, has only one second order
prognostic equation, for TKE itself. The next level up in complication is the true one-and-a-half
order closure model, level 3 in Mellor and Yamada (1974) nomenclature, which in addition to
TKE includes prognostic equations for the turbulent variances of other relevant quantities such as
temperature and humidity. The TKE scheme in the latest version of MM5 can be run as either
level 2.5 or level 3 where prognostic equations for temperature variance, moisture variance, and
temperature-moisture co variance are added to the TKE equation (Burk and Thompson, 1989).
Another similar form of a higher order closure model is known as TKE-e, which has prognostic
equations for TKE and the turbulent dissipation rate (e). An example of this type of model is
described by Alapaty et al (1996).
A common feature of TKE models, for level 2.5 or 3 or TKE-e, is that the turbulent fluxes of
momentum, heat, and moisture are represented as local gradient diffusion similar to Equation 7-
23. It is this characteristic that distinguishes 1.5 order closure from true 2nd order closure where
the flux covariances are explicitly represented by prognostic equations. Therefore, adaptation of
a CTM to a meteorology model that includes TKE closure is essentially simple if the TKE fields
are available to the CTM. Only a small part of the TKE model need be reproduced in the CTM,
namely the parameterization of eddy diffusion coefficients based on quantities already produced
in the meteorology model. For example, in the TKE-e model of Alapaty et al. (1996), the
governing TKE equations hi Cartesian coordinates are:
dE dE dE dE
u w __ -- VW -
at . ax ay dz dz dz
7-25
-------
EPA/600/R-99/030
\-e (7-30a)
0
ds de de ete c,ef -n— -du -z—r.dv g -
= _u__v___w__-h_!_ -u"w"- — v' w"— • + —
dt dx dy dz E { dz dz 0
(7-30b)
where u, v are horizontal wind components;
E is the turbulent kinetic energy per unit mass;
z is height;
Qy is the virtual potential temperature;
e is the rate of TKE dissipation;
8
u" w" , v" w" , and —0" w" are turbulence fluxes of momentum components and heat flux;
p is the density; and
p" is the fluctuating pressure.
The constants cj, 02, and ej can be estimated following Detering and Etling (1985). The first
three terms on the right-hand-side of the first equation represent advection of turbulent kinetic
energy, and the other terms, shear production, buoyancy production, turbulence transport and
the dissipation, respectively. Similarly, the first three terms on the right-hand-side of the second
equation represent advection of e. The rest of the terms represent the net rate change of e due to
shear and buoyancy production, the rate change of £ related to the time-scale of turbulence, and
the vertical transport, respectively. The coefficients of eddy viscosity for momentum and heat
can be written as:
Km=^- (7-31a)
where Km and Kh are the eddy exchange coefficients for momentum and heat, respectively. E
and £ need to be provided by meteorological models in their native coordinates, thus the
parameterization does not depend on the grid spacing explicitly. Obviously, we are assuming
here that the vertical layer structure in the CTM is compatible with that used in the preprocessor
meteorological model. Once Kh is computed, chemical fluxes can be modeled assuming similarity
7-26
-------
EPA/600/R-99/030
with heat flux using the same eddy diffusion numerical algorithm described below (section
7.2.3.1). The derivation of eddy coefficients in the Burk and Thompson (1989) model is
algebraically more complex (see Mellor and Yamada, 1974,1982) but based only on mean
quantities plus TKE. Therefore, given the TKE fields, which are output automatically from
MM5 when the Burk and Thompson (1989) option is invoked, Kh can be re-diagnosed within
the CTM.
7.2.2.3 Numerical Algorithm for Vertical Eddy Diffusion Modules
As described in Chapter 5, the vertical eddy diffusion module must solve the following equation
in terms of mixing ratio q:
d / £33 <%li % . £33 dln(V?p) dq.s _
— _^-3 \M\ ' ™^-i!J « *»• ^ "V-^ ' ^/".J-fa/
viiiff °^ «* ax ax
where the vertical mixing is represented with the pure diffusion term and the turbulent flux
exchange term, respectively.
Numerical algorithm for the diffusion kernel
In its generic form, the diffusion kernel solves for:
where £ is the generic vertical coordinate which increases with geometric height. To account for
the loss process due to deposition in the lowest model layer (J=l), dry deposition flux is
considered as the flux boundary condition at the surface, i.e.:
= _3Lft (7,34)
<* dep h
-------
EPA/600/R-99/030
At
K,.
(7-35a)
where # Is a time-step weighting factor and
"3/ti/z = (s/ + S/*i V 2; Af^.^j = Sjti/2 ~ b/-!/j» ^S/*i = >/*i ~~ S;
For the lowest model layer, we need to account for the loss due to deposition process:
*.4
(7-35b)
and for the top layer we have zero flux through the top boundary:
At Ku_,,
(7-35c)
Depending on t9, the finite difference scheme is explicit (z?=0), semi-implicit (t?=1/2), or fully
implicit (t?=l). In the current version of CMAQ, the semi-implicit (Crank-Nicholson) algorithm
is implemented. Equations 7-35a-c can be rearranged to yield a matrix equation:
Aq =
(7-36)
where A is tridiagonal whose coefficients (sub-diagonal component Op diagonal component
.'.'. ...H • • . , -;• ., ,
and super-diagonal component GJ) when/=l are given as:
c, =-
(7-37a)
7-28
-------
and forj=N:
EPA/600/R-99/030
(7-37b)
and for 2
-------
EPA/600/R-99/030
bj =
1-
(l—ff)At
Jy
(7-37e)
The numerical algorithm solving the tridiagonal system is based on the Thomas algorithm. Refer
to Appendix 7A.1 for details of the algorithm and stability characteristics.
Numerical algorithm for the coordinate divergence kernel
The coordinate divergence kernel solves for
=
at at
=
(7-38)
where Vmlx = K
. The differential equation is in advective from with the effective mass
transfer velocity Vmix. It can be solved with a vertical advection code. Because most operational
meteorological models rely on logarithmically spaced vertical layering based on sigma-p type
coordinates, Vmix is expected to be small. Currently, this component is not implemented in
CMAQ. However, a quantitative study is needed to assess the importance of this term.
Integration time-steps
According to Oran and Boris (1987), any numerical algorithm for the diffusion equation (with
equal grid spacing) should produce the following quantitative properties:
• the total integral of q(|, t) should be conserved;
» the amplitude jg(£,f)| should decay monotonically;
• there should be no phase errors introduced by the algorithm (for equal grid spacing); and
» positivity should be preserved.
Although the numerical solver algorithm for the semi-implicit scheme is stable in the sense that
the amplitude of the signal either decays or stays the same, the positivity condition may not be
satisfied if we choose a large time-step for the integration, especially for signals with small
Wavelengths. For example, with equal grid spacing, we can use the Von Neuman stability
analysis technique to demonstrate that some part of the short-wavelength spectrum shows a
7-30
-------
EPA/600/R-99/030
negative amplitude when the time-step is too long. The semi-implicit scheme is positive definite
given the rather stringent CFL condition, for a uniform vertical grid spacing:
:1.0 (7-39a)
For non-uniform grid spacing, we use the CFL condition:
1.0 (7-39b)
to ensure positive definiteness of the semi-implicit solution. The internal time-step for vertical
diffusion is thus determined in CMAQ with the following equation:
(7-40)
( K K V
where At, = AL\ -^- + -^~
1 \A^n A^J
7.2.3 Flux Form Representation of Vertical Mixing
Vertical mixing can be represented in flux form as:
dq_
dt~ dt; % ' ( }
where F^ = F* is the turbulent flux represented in the generalized vertical coordinate £ , whose
value increases monotonically with geometric height. Here, the flux should be parameterized in
coordinate instead of the generic height coordinate. The cross directional, (w.r.t. generalized
coordinate) diffusion terms, as well as the flux divergence due to grid spacing (second term in
Equation 7-41), are neglected.
Thus, the numerical solver kernel for the flux form vertical diffusion should solve for:
IHf . .
when the source term is zero. The flux form representation is extremely useful in describing the
algorithms based on non-local closure. Non-local closure recognizes that larger-size eddies can
7-31
-------
EPA/60Q/R-99/030
transport fluid across distances longer than the grid increment before the smaller eddies have a
chance to cause mixing (Stull, 1988).
To represent the turbulent mass exchange with the transilient parameterization, the boundary
layer height must coincide with the height of layer interfaces of the vertical grid. For most
situations, the index for the boundary layer top (Lp) is less than the total number of model layers
(i.e., Lp < N). With the transilient turbulence formulation, the new values of the trace species
mixing ratio q due to turbulent mixing for a layer/ at a future time (t + At) can be written as:
q}(t + At) = ^cjk(t,At)qk(t) (7-43)
4=1
where cjk are the components of a transilient matrix and subscripts j and k are indices of two
different grid boxes (vertical layers) below boundary layer top in a column of atmosphere. If we
consider turbulent mixing between grid boxes j and k, cjk represents fraction of air mass ending in
the grid boxy that came from grid box k. The grid box/ is considered as the "destination" box
while grid box k is considered as the "source" box. Thus, the change in the tracer concentration
due to turbulent mixing for grid box/ at a time interval At is a simple matrix multiplication with
concentration from the source cell. The transilient matrix representation is in fact applicable for
any physical process that involves mass exchange among grid boxes in a column. For example,
convective cloud mixing can be represented by a transilient matrix as well, similar to how mixing
in a convective boundary layer is handled.
The mass conservation requirements provide constraints for the coefficients of the transilient
matrix. The conservation of air mass requires that the sum over k of all mixing fractions be unity:
i
l (7-44)
*=i
and the conservation of trace gas amount requires that the sum over/ of all mass-ratio weighted
transilient coefficients be unity as well:
(7-45)
where A^l A%k represents the mass ratio (i.e., ratio of layer thicknesses for mixing ratio q)
between the source and destination boxes. In order that transilient turbulence theory be useful,
the coefficients should be determined using appropriate turbulent flux parameterizations.
7-32
-------
EPA/600/R-99/030
Consider how the mixing coefficients are related with the turbulent flux representations. Because
the transilient matrix describes the exchange of mass between grid boxes, the kinematic turbulent
fluxes across the/'-th level can be expressed, for a vertical layering with non-uniform grid spacing
(Stall, 1993), for 2
-------
EPA/600/R-99/030
resulting in a general matrix equation:
Pq"M=Rqw (7-50)
where the coefficients of P and R are given as:
m ; Pjk =
••. • • • ....•..
If the matrix P is nonsingular, we have a general expression for the transilient turbulence:
q"*1 =P"!Rq" = Cq" (7-51)
where C=P"!R.
The relationship between the coefficients of the transilient matrix and the mass exchange ratio
among the grid boxes are somewhat complicated for the semi-implicit scheme. However, the
semi-implicit scheme becomes attractive for the closure algorithms with sparse P matrix (i.e.,
when the matrix inversion is not so expensive) because it allows longer integration time-steps
than the explicit method. In the following, we describe non-local flux-form atmospheric
turbulence algorithms as a subset of the generalized transilient turbulence representations.
7,2.3.1 Blackadar Non-local Scheme
This scheme, first introduced by Blackadar (1978), has long been used as one of the PBL schemes
in the Mesoscale Meteorology model generation 4 (MM4) and generation 5 (MM5). The
Blackadar model is a simple non-local closure scheme designed to simulate vertical transport by
large convective eddies during conditions of free convection. Therefore, this scheme is used only
in the convective boundary layer and must be coupled with another scheme for non-eonveetive
conditions and above the boundary layer, such as K-theory. In general, the flux-form diffusion
algorithm can be written for the lowest layer as:
(7-52a)
k=l
arid for 2 <, j < L .as:
7-34
-------
EPA/600/R-99/030
(7-52b>
where mjk is the rate of mass exchange between two grid boxes in a column of the atmosphere
(below boundary layer top) per unit time. The convective mixing is assumed to be dominated by
eddies of varying sizes but all having roots in the surface layer, each eddy exchanging a certain
amount of its mass with the air around it as it ascends. The rate of change of mean potential
temperature caused by the mass exchange in the mixed layer can be expressed as:
(7-53)
t
where w(%) is a weight function that accounts for the variation of exchange rate with height.
The mass exchange rate, Mu, can be estimated from conservation of energy, which requires the
heat flux at any level to satisfy the equation:
(7-54)
,••.
where H. rfc is the sensible heat flux leaving the surface layer and Cpd is the specific heat at
constant pressure. When the integration limit is extended to the top of the boundary layer, where
H is assumed to be zero, we can estimate Mu with:
5*
Mu=Htfc/ICpJp(0^-e)w^)J^ . (7-55)
fin-
Usually the weight function w is approximated to be unity in the mixed layer. With the
Blackadar scheme, the mixing algorithm is represented for the lowest model layer by:
....- , . (7-56a)
and for 2 < j < Lp by:
(7-56b)
7-35
-------
EPA/600/R-99/030
A$- A$~
where we used m,t = Mu —^r, m,, = Mu—^-, and all other components of mik are zero.
J
Finite difference representations of the above equations are:
At h
lan<^ f°r 2 ^ y ^ Lp
k=\
(7-57b)
After rearrangement, we obtain the following matrix equation:
e2 d2 0 0 ••• 0
; o •-. o
ej : ': dj : :
: 0 0 0 '-. 0
e, -"000 d.
T
=
o,
*y
(7-58)
where the elements are defined with:
v,At
and for
7-36
-------
EPA/6QQ/R-99/Q3Q
*,=
The numerical algorithm to solve the sparse matrix system is similar to the Thomas algorithm for
the tridiagonal system. For the details of the numerical algorithm, refer to Appendix 7A.2,
Figure 7-2. Schematics of the Blackadar Scheme (a) and the Asymmetric Convective Model (b)
7.2.3.2 Asymmetric Convective Mixing
The Asymmetric Convective Model (ACM), developed by Pleim and Chang (1992), is based on
Blackadar's non-local closure scheme (Blackadar, 1978) but with a different scheme for
7-37
-------
EPA/600/R-99/030
downward mixing in the convective boundary layer (CBL). Blackadar's scheme is based on the
assumption that the turbulent mixing is isotropic (i.e., symmetric) in the CBL. However,
observational evidence and large-eddy simulation modeling studies indicate that mixing processes
in a connective boundary layer are essentially asymmetric (i.e., turbulence is anisotropic;
Schumann, 1989) with fast upward buoyant plumes and slow broad compensatory subsidence.
Therefore the direct, non-local downward transport of the Blackadar scheme is replaced with
layer by layer subsidence which increases to mass flux as it descends, like a cascading waterfall
(Figure 7-2). As with the Blackadar model, the ACM can only be used during convective
conditions in the PBL. For other stability regimes, one needs to rely on other schemes such as
K-theory.
Turbulent mixing in the PBL for any dynamic, or thermodynamic variables, or trace gas species
concentrations can be represented in essentially the same way as in the above transilient
parameterization. Also, the conceptual design of the ACM allows for considerable
simplification. Because the mass influx to the lowest model layer is from the secondlayer only
in ACM (refer to Figure 7-2), we can write the time rate change of mixing ratio as follows:
(7-59a)
For 2<>j<,Lp:
(7-59b)
where Mu represents upward mixing rate. Mjj represents downward mixing rate at layer/ and is
defined as:
(7-60)
____ . . "in, , • • . ., , , i
The scheme can be represented in terms of transilient mixing rates (/»/#), as shown in general form
in Equation 7-48:
da
Therefore, by equating Equations 7-59b and 7-61, we see thatw^, = Mu & m,^, which shows that
the transfer coefficients are asymmetric as expected, and m}\ + mjj+l = M^, which demonstrates
a recursive relationship for M^:
-7-38
-------
EPA/600/R-99/030
-^jr
Finally, we rewrite the prediction equation in terms of the upward mixing rate:
(7-62)
(7-63a)
and for
(7-63b)
Note that this scheme does include the effects of vertical wind shear in generating turbulent
mixing. The magnitude of the mixing rates of the transilient matrix is based on the conservation of
sensible heat flux in the vertical direction:
Mu =
Finite difference representations of the above equations are:
(7-64)
At
and for
(7-65a)
6s.
(7-65b)
This results in a sparse matrix of the form:
7-39
-------
EPA/600/R-99/030
0
0
0
0
0
0
0
0
0
0 N
0
0
;
/
/ fl+1 \
02*'
;
B+]
(
\"LP J
fM
b2
•
b,
j
*>L
*V Lf J
(7-66)
where;
MB(1 -
The numerical algorithm solving the sparse matrix system is presented in Appendix 7A.3.
7.2.33 Transilient Turbulence Parameterization
The general computational paradigm for the transilient turbulence parameterization has been
presented above. In order to use the transilient turbulence concept for mixing trace gases, one
• • : ' ' • • ' • . ' »
needs to know the mass exchange coefficient matrix. This is the closure problem with this
^/; u ' *, "^ '':,'' ' • - - r • ' ". • '• • r ' " ' ' • •
parameterization. A couple of methods have been presented in the literature. One method is
based on the TKE equation (Stull and Driedonks, 1987, and Raymond and Stull, 1990) and the
other is the one based on non-local Richardson number (Zhang and Stull, 1992). In the following
we describe briefly the TK£ based scheme and discuss its associated difficulties.
The horizontally homogeneous form of the TKE equation, Equation 7-3 Oa is given as:
7-40
-------
EPA/600/R-99/030
dt
&
(7-67)
Note that the pressure and turbulent transport terms have been ignored. After normalizing with
E, the non-local analogy of the finite difference form.of Equation 7-67 can be written as:
AtEJk
-7*
-f
8
(7-68)
where the symbol At represents temporal change while A represents spatial gradient. To close
the system, the unknown parameters are written in terms of known parameters by introducing
three scaling parameters T0, Ric, and D, which .are the time scale of turbulence, the critical
Richardson number, and dissipation factor, respectively. Thus, weighted kinematic fluxes can be
written as: .
(A0\
(7-69a)
(7-69b)
(7-69c)
Then Equation 7-68 can be rewritten as:
_ A,EJk _ T0A,t
E*
8
Ri&,
(7-70)
To use Equation 7-70 for the generalized coordinate £, the corresponding layer heights
should be computed for (A%) .k. .-.. ; .......
Since we are dealing with fraction of masses that are coming from and going to different layers
(i.e.,/ 96 K), Yj/c in the above equation is for off-diagonal elements only. Diagonal elements in
Equation 7-70, Yj,-, represent mass of air that remain in each layer without interaction with other
layers. Observations during convective conditions (Ebert et al., 1989) indicated that turbulent
eddies cause a well-mixed rather than convectively overturned boundary layer. This requires
741
-------
EPA/600/R-99/030
that the values of the mixing potential elements (1^) should increase monotonically from the
upper right-most element towards the main diagonal. Further, to account for the subgrid scale
I
mixing in each layer, an independent parameter Yref was introduced. Thus, the diagonal elements
i
can be written as:
Yji=MAX(YIJ_t,YIJJ + Yref (7-71)
given values for T0, Ric, and D. Usually, Yref is estimated based on observations (Stull, 1988).
Finally, the off-diagonal elements of the transilient matrix are estimated with:
'- (7-72)
where |Y||M is the infinite norm of matrix Y, max{|Y|}. The formulation presented in Raymond
and Stull (1990) and Alapaty et al. (1997) includes the additional weighting based on the mass in
the layer for irregularly spaced grids. However, we believe that Equation 7-72 should be valid
even for irregularly spaced grids when the constraint Equation 7-45 is satisfied. Also, Stull
(1993) states that the formulation Equation 7-72 causes too much mixing near the surface and
inclusion of the mass weighting in the formulation exacerbates the problem further. The diagonal
elements of the transilient matrix can be computed by rewriting Equation 7-44 as:
(7-73)
*«;
Once the transilient matrix is determined, the concentration due to turbulent processes in the
boundary layer can be estimated from Equation 7-43. The difficulties associated with this
: • i"i,,| . ''",!'
parameterization are:
• The scheme still depends on many free parameters (T0, Ric, D, and Yref) and they control
the behavior of the mixing algorithm; and
• The time-steps should be such that the trace species mixing ratio cannot be negative
I ' '-„ „ ... . ;,:, i '-;_, . | . . •
because Equation 7-43 is written in explicit form. Although explicit methods do not
require matrix inversion, the tune-step must be small enough to ensure positiviry and
numerical stability of the solution.
7.3 Horizontal Mixing Algorithms
Unfortunately, our understanding of horizontal turbulence is limited due to the lack of adequate
turbulence measurements as well as the scale dependency of the problem. In earlier days of
7-42
-------
EPA/600/R-99/03D
atmospheric modeling, the horizontal diffusion process was often ignored because the numerical
diffusion associated with the advection algorithms used was large. For problems with large scales,
such as regional to global transport studies, the coarse grid resolutions did not require an explicit
implementation of horizontal diffusion. However, with the advent of very accurate (i.e., less
diffusive) numerical advection schemes and the emerging need for high resolution grids for urban
scale problems, a good algorithm for horizontal diffusion is required. The skill needed is
balancing numerical diffusion associated with the advection schemes with the added explicit
diffusion to model horizontal diffusion in the atmosphere. A fundamental problem is that we do
not know much about the expected magnitude of the actual horizontal diffusion. In this section
we will describe the numerical algorithm for the horizontal diffusion implemented in the CMAQ
system.
The horizontal diffusion process in the curvilinear coordinate system (See Equation 6-25' in
Chapter 6) is given as:
(7-74)
i. "* J
Miff
Miff *
There are not many choices for the horizontal diffusion parameterizations. Frequently, the
horizontal turbulent fluxes are parameterized using eddy diffusion theory. The contributions of
the off-diagonal diffusion terms show up explicitly as in Equation 6^14 in Chapter 6. Often,
these off-diagonal terms are neglected in air quality simulations, and in the CMAQ
implementation, we solve for diagonal terms only:
d f r^"—/AH <^7>xl d f
3$H w>(^ ^fr) +3$H
etc L ax j ax i
3A2
ifa/if ""-i ~~ i "^ ' °*
-^/E>22
(7-75)
The contravariant eddy diffusivity components are related to the Cartesian counterparts as
K" = mK^ and Kn — mKn. In practice, for Eulerian air quality modeling, we do not distinguish
between eddy diffusivities in two different horizontal directions (i.e., K^. = K>y = KH), For a
Lagrangian simulation of atmospheric turbulence, the longitudinal (following the plume
movement) and lateral (perpendicular to the plume movement) dispersion are treated differently
according to characteristics from the isotropic turbulence analysis. Often the horizontal eddy
diffusivity in the Cartesian coordinates is parameterized with the magnitude of the deformation in
the gridded wind field. For that case, one must be careful whether the wind data are represented
in Cartesian coordinates or in the transformed coordinates.
743
-------
EPA/600/R-99/030
Unlike the vertical diffusion case, we do not separate the grid divergence term from the diffusion
equation. An explicit solution method for Equation 7-75 is:
(7-76)
*• j> vj
where K£ = -(K£rt + £,") and I5£ = -(C,m + O- At me boundary cells, a zero-gradient
*&f £*
boundary condition (Neumann) is applied. Because Equation 7-76 is an explicit scheme, the
™- ^
time-step should be chosen to prevent numerical instability and to maintain positivity. With an
i
appropriate Courant number for horizontal diffusion, /?My, the time-step can be determined
with:
- (7-77)
At present PMiff= 0.3 and a uniform eddy diffusivity J^//|Al._4fen = 2000 m2/s is used for the 4-km
grid resolution. To compensate for larger grid sizes for coarser grids, the eddy diffusivity is
modified to give:
(4000)%
where Ax is in meters.
i
Obviously, the above parameterization is too simple to be realistic in a variety of atmospheric
conditions. Also, depending on the numerical advection algorithms chosen, the artificial
diffusivity can be quite different. This calls for several in-depth studies on following two major
issues:
(1) Quantification of realistic horizontal sub-grid scale diffusion.
The simplest approach is to assume a space independent diffusivity (e.g., Kh — 50
Smagorinsky (1963) formulated a horizontal diffusivity that accounts for diffusion due to
distortion or stress in the horizontal wind field. For plumes which are several kilometers or more
across, the Briggs (1973) parameterizations of horizontal diffusion define the difrusivity as a
constant times the transport wind speed. The constant is usually based on the landuse (i.e.,
7-44
-------
EPA/600/R-99/030
urban or rural) and the stability class (i.e., stable through unstable). However, it is difficult to
quantify what the horizontal eddy diffusivity should be appropriate for a variety of atmospheric
conditions without more detailed wind field and turbulence information.
(2) Maintaining appropriate horizontal diffusion in the presence of numerical diffusion.
Most methods for simulating advective transport in current models yield an effective numerical
diffusivity much larger than physical horizontal diffusivities (Yamartino et al., 1992). Thus, the
physical process may be outweighed by the numerical errors in the model. A re-assessment of
this issue is required when the resolution of the model changes or when the method for simulating
advection is updated. For idealized concentrations and wind fields distributions, we may be able
to quantify the magnitude of the numerical diffusion in an advection scheme. However, for the
more general applications, estimating the magnitude of numerical diffusion with a specific
advection scheme is almost impossible. Refer to Odman (1997) for methodologies that quantify
numerical diffusion errors associated with advection algorithms.
7.4 Conclusions
In this chapter, we have described numerical advection and diffusion algorithms. It has two
purposes: to provide a description of the algorithms currently implemented in CCTM, and to
describe the fundamental formulations that would guide future implementation of advection and
diffusion modules. We encourage the development of algorithms in conservation (i.e., flux) forms
to ensure compatibility of new modules with existing ones.
Because of the concerns over the non-monotonicity of BOT and YAM schemes and the mass
conservation problem and diffusive nature of BOT-M, we have used PPM for a number of
demonstration executions (Byun et al., 1998). Similar testing with BOT and YAM is underway.
We intend to integrate other methods into the CCTM at a later time.
Also, we have identified several aspects in vertical and horizontal diffusion algorithms that
require additional quantitative studies:
• Effects of the parameterization for the free troposphere;
• Importance of coordinate divergence term for vertical diffusion, in particular for the
height-based constant coordinates;
• Characteristics among competing algorithms for the vertical diffusion, such as TKE and
transilient turbulence schemes; and
• Practical and theoretical concerns with the horizontal diffusion algorithms.
745
-------
EPA/600/R-99/030
7.5 References
Alapaty, K., R. Mathur, and D.W. Byun, 1996: A modeling study of vertical diffusion of passive
and reactive tracers using local- and nonlocal-closure boundary layer schemes. Air Pollution
Modeling and Its Application XL, ed. S.E. Gryning and F, Schiermeier. 433-442.
Alapaty, K.A., J.E. Pleim, S. Raman, D.S. Niyogi, and D.W. Byun. 1997: Simulation of
atmospheric boundary layer processes using local- and nonlocal-closure schemes. J. Applied
Meteor. Vol. 36.214-233.
Blackadar, A. K., 1978: Modeling pollutant transfer during daytime convection, Preprints,
Fourth Symposium on Atmospheric Turbulence, Diffusion, and Air Quality, Reno, Am. Meteor.
Soc., 443-447.
Bott, A., 1989: A positive definite advection scheme obtained by nonlinear renormalization of
the advective fluxes. Man. Wea. Rev. 117,1006-1015.
Bott, A., 1992: Monotone flux limitation in the area-preserving flux-form advection algorithm.
Mon. Wea, Rev. 120,2592-2602.
Bott, A., 1993: The monotone area-preserving flux-form advection algorithm: reducing the time-
splitting error in two-dimensional flow fields." Mon. Wea. Rev. 121.2637-2641.
Briggs, G. A., 1973: Diffusion estimates for small emissions. ATDL Contribution No. 79
[Available from Atmospheric Turbulence and Diffusion Laboratory, Oak Ridge, Term.]
Brooks, A. N., and T. J. R. Hughes, 1982: Streamline upwind/Petrov-Galerkin formulations for
convection dominated flows with particular emphasis on the incompressible Navier-Stokes
equations. Comp. Meth. Appl Meek Eng. 32,199-259.
. '• :\\ .":- ' . ' ' , ,'..'.• ,.\. . .'.;. . j
J" ' ' •"•" • ..',,, ,. . • I
Burk, S.D., and W.T. Thompson, 1989: A vertically nested regional numerical weather prediction
model with second-order physics. Mon. Wea. Rev. 117, 2305-2324.
1' • • ' i
Businger J. A., J. C. Wyngaard, Y. Izumi, and E. F. Bradley, 1971: Flux profile relationships in
the atmospheric surface layer, J. Attnos. Sci, 28,181-189.
Byun, D. W., 1999a: Dynamically consistent formulations in meteorological and air quality
models for multi-scale atmospheric applications: I. Governing equations in a generalized
coordinate system, J. Aimos. Sci., (in print).
7-46
-------
EPA/600/R-99/030
Byun, D. W., 1999b: Dynamically consistent formulations in meteorological and air quality
models for multi-scale atmospheric applications: II. Mass conservation issues.«/. Atmos. ScL, (in
print).
Byun, D. W., and R. L. Dennis, 1995: Design artifacts in Eulerian air quality models: Evaluation
of the effects of layer thickness and vertical profile correction on surface ozone concentrations.
Atmos. Environ., 29, 105-126.
Byun, D.W., and S.-M. Lee, 1999: Mass conservative numerical integration of trace species
conservation equation: I Experiment with idealized two-dimensional flows, (in preparation)
Byun, D.W., J. Young., G. Gipson., J. Godowitch., F. Binkowsk., S. Roselle, B. Benjey, J.
Pleim, J. Ching., J. Novak, C. Coats, T. Odman, A. Hanna, K. Alapaty, R. Mathur, J. McHenry,
U. Shankar, S. Fine, A. Xiu, and C. Jang, 1998: Description of the Models-3 Community
Multiscale Air Quality (CMAQ) model. Proceedings of the American Meteorological Society
78th Annual Meeting, Phoenix, AZ, Jan. 11-16, 1998.
Carpenter, R. L., K. K. Droegemeier, P. R. Woodward, and, C. E., Hane, 1990: Application of
the piecewise parabolic method (PPM) to meteorological modeling. Man. Wea. Rev, 118, 586-
612.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madronich, P. Middleton, W.R. Stockwell, and CJ.
Walcek, 1987: A three-dimensional Eulerian acid deposition model: Physical concepts and
formulation, J. ofGeophys. Res., 92,14,681 -700.
Childs, P. N., and K. W. Morton, 1990: Characteristic Galerkin methods for scalar conservation
laws in one dimension. SIAMJ. Numer. Anal, 27, 553-594.
Chock, D. P., and A. M. Dunker, 1983: A comparison of numerical methods for solving the
advection equation. Atmos. Environ. 17(1), 11-24.
Chock, D. P., 1985: A comparison of numerical methods for solving the advection equation - II.
Atmos. Environ. 19(4), 571-586.
Chock, D. P., 1991: A comparison of numerical methods for solving the advection equation-HI.
Atmos. Environ. 25A(5/6), 853-871.
Chock, D. P., 1999: Personal communication.
Colella, P., and, P. R. Woodward, 1984: The piecewise parabolic method (PPM) for gas-
dynamical simulations. J. Comp. Phys 54, 174-201.
7-47
-------
EPA/600/R-99/030
Dabdub, D., and Seinfeld, J. H., 1994: Numerical advective schemes used in air quality models -
sequential and parallel implementation. Atmos. Environ, 28(20), 3369-3385.
: • • ' !
Detering, H.W., and D. Etling, 1985: Application of E-e turbulence model to the atmospheric
boundary layer. Bound-Layer Meteor., 33,113-133.
Ebert, E.E., U. Schumann, and R. B. Stull, 1989: Nonlocal turbulence mixing in the convective
boundary layer evaluated from large-eddy simulation. J. Atmos. Sci., 46,2178-2207.
Flatoy, F., 1993: Balanced wind in advanced advection schemes when species with long lifetimes
are transported. Atmos. Environ, 27A( 12), 1809-1819.
Mass, H., H.J. Jacobs, M. Memmescheimer, A. Ebel, and J.S. Change, 1991: Simulation of a wet
deposition case in Europe using the European Acid Deposition Model (EURAD). Air Pollution
Modelling and Its Applications, Vol. VIII (edited by van Dop H. and D.G. Steyn), Plenum Press,
205-213.
»- . .1
Holtslag, A. A., E. I. F. de Bruin, and H.-L. Pan, 1990: A high resolution air mass transformation
model for short-range weather forecasting. Mon. Weather Rev., 118,1561-1575.
Leith, C, C., 1965: Numerical simulation of the earth's atmosphere. A chapter in Methods in
Computational Physics, Vol. 4, Applications in Hydrodynamics, p. 1. ed., B. Alder, Academic
Press, New York.
i
LOhner, R., K. Morgan, J. Peraire, and M. Vahdati, 1987: Finite element flux-corrected transport
(FEM-FCT) for the Euler and Navier-Stokes Equations. Finite Elements in Fluids. Vol. 7 . John
Wiley & Sons. 105-121.
.LI. ..i . ^ , I , ^
Mellor, G,L. andT. Yamada, 1974: A hierarchy of turbulence closure models for planetary
boundary layers, /. Atmos. Sci., 31,1791-1806.
Mellor, G.L. andT. Yamada, 1982: Development of a turbulence closure model for geophysical
fluid problems. Reviews ofGeophysica and Space Physics, 20, 851-875.
Odman, M. T., 1997: A quantitative analysis of numerical diffusion introduced by advection
algorithms in air quality models. Atmos. Environ. 31,1933-1940.
Odman, M. T., 1998: Research on Numerical Transport Algorithms for Air Quality Simulation
Models. EPA Report. EPA/660/R-97/142. [Available from National Exposure Research
Laboratory, U.S. Environmental Protection Agency, Research Triangle Park, NC 27711].
7-48
-------
EPA/600/R-99/030
Odman, M. T., and A. G. Russell, 1993: A nonlinear filtering algorithm for multi-dimensional
finite element pollutant advection schemes. Atmos. Environ. 27A, 793-799.
Odman, M. T., A. Xiu, and D. W. Byun, 1995: Evaluating advection schemes for use in the next
generation of air quality modeling systems, In Regional Photochemical Measurement and
Modeling Studies, eds A.J. Ranzieri and P.A. Solomon, Vol. Ill, Air & Waste Management
Association, Pittsburgh, PA. 13 86-1401.
Oran, E., and J. Boris, 1987: Numerical Solution of Reactive Flow, New York, Elsevier.
Otte, M.J. and J.C. Wyngaard, 1996: A general framework for an "unmixed layer" PBL model. J.
Atmos. Sci., 53, 2652-2670.
Pleim, I.E. and J. Chang, 1992: A non-local closure model for vertical mixing in the convective
boundary layer. Atmos. Envi., 26A, 965-981.
Raymond, W.H. and R.B. Stull, 1990: Application of transilient turbulence theory to mesoscale
numerical weather forecasting. Mon. Weather Rev., 118, 2471-2499.
Rood, R. B., 1987: Numerical advection algorithms and their role in atmospheric transport and
chemistry models. Rev. Geophysics 25(1), 71-100.
Schumann, U., 1989: Large-eddy simulation of turbulent diffusion with chemical reactions in the
convective boundary layer. Atmos. Envi., 26A, 965-981.
Smagorinsky, J., 1983: General circulation experiments with the primitive equations: 1. The basic
experiment. Mon. Wea. Rev. 91, 99-164.
Stull, R. B., 1988: An Introduction to Boundary Layer Meteorology. Kluwer Academic, 666 pp.
Stull, R. B., 1993: Review of non-local mixing in turbulent atmosphere: Transilient turbulence
theory. Bound-Layer Meteor., 62, 21-96.
Stull, R. B., and A.G.M. Driedonk, 1987: Applications of the transilient turbulence
parameterization to atmospheric boundary layer simulations. Bound-Layer Meteor., 40, 209-
239.
Tremback, G. J., Powell, J., Cotton, W. R. and Pielke, R. A., 1987: The forward-in-time
upstream advection scheme: extension to higher orders. Mon. Wea. Rev. 115, 540-555.
749
-------
EPA/600/R-99/030
Yamartino, R. J., 1993: Nonnegative, conserved scalar transport using grid-cell-centered,
spectrally constrained Blackman cubics for applications on a variable-thickness mesh, Man. Wea.
Rev. 121,753-763.
Yamartino, R.J., IS. Scire, G.R. Carmichael, and Y.S. Chang, 1992: The CALGRID mesoscale
photochemical grid model -Part I. Model formulation, Atmos. Environ. 26A, 1493-1512.
Yanenko, N. N., 1971: The Method of Fractional Steps. Spring-Verlag, New York, 160pp.
Zhang, Q., and Stull, R. B. Stall, 1992: Alternative nonlocal descriptions of boundary-layer
evolution. J. Atmos, Set., 49,2267-2281.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
7-50
-------
EPA/600/R-99>'030
Appendix 7A Numerical Solvers for Diffusion Equations
In this appendix we describe numerical procedures for eddy diffusion, the Blackadar mixing
scheme, and the asymmetric convective model.
7A.1 Stability of Tridiagonal Solver
The eddy diffusion formulation requires the solution of the linear equation
Aq = b (7A-1)
where:
' d\
a2
0
0
. 0
ci
d2
"3
0
c2
4
0
0
0
0
c3
a«-i
0
0 N
0
0
dn_}
b = (*i b2 b, - bn}T.
The system Equation 7A-1 can be solved by the Thomas algorithm (Gaussian elimination of a
tridiagonal matrix without pivoting) followed by back substitution. Assume that the following
stage of the elimination has been reached:
(7A-2a)
(7A-2b)
where a, = rf, and /?, =br
Fory-2,3,—n-1, eliminating q^ from Equations 7A-2a,b leads to:
where
(7A-3a)
(7A-3b)
7-51
-------
EPA/600/R-99/030
The last pair of simultaneous equations are:
n=A,-i . (7A-4a)
=bn (7A-4b)
Eliminating q , gives:
i • • " • « • . (• -
, _ . , -i _ .-.(.
: «A=A ; (7A-5a)
- j
and with Equation 7A-3a, we can obtain the solution by back substitution, i.e.,./ = n-1, «-2,'-%l :
,=± (7A.5b)
aJ
The algorithm described above is stable for the tridiagonal system if:
(i) dj > 0, a} < 0, andc; < 0;
(ii) d; > -(a;+, +CM) for 7=1,2, ,-n-l, defining ca = an = 0; and
(iii) dj > —(a.j +Cj) for /= 1,2, ,•••«-!, defining a, = cn_, = 0.
Th,e first two conditions ensure that the forward elimination is stable and the first and third
conditions ensure that the back substitution is stable.
. , • \
To prove that the forward elimination procedure is stable, it is necessary to show that the moduli
of the multipliers my = — a.} /a;-_, used to eliminate #/, q^ ... are positive and less than or equal to
one. From Equation 7 A-2, we get:
j^ (7A-6a)
and
_%±L = __lf/±J_ (7A.6b)
(Xj dj+mjej_i
Then, since d{ > —c^ > 0 = cc:
-------
EPA/600/R-99/030
Similarly, we have:
~ 1 (7A-7b)
since di > -(a^ + c, ) . In this way, we can show that 0 < m;. < 1 fory-1 , 2, — n- 1 ,
For back substitution, we can write:
~ ~~ ' ~ = ~
(7A"8)
with a, =cn_, =0 for y=l, 2, — n-\.
There will be no build-up of errors in the back substitution process if \pj+\ < 1, where:
«,
Now, 0 < p2 = -c, /rf, < 1, since a, = 0 and rf, > -c, by hypothesis. Then:
ft = . ~C2 (7A-9b)
^2 + «2ft
As -c2>0, 0 -«2 > 0, it follows that:
" (7A-10)
-(a2+c2)-a2
Similarly, we can show that 0 < p, < 1 for y'=l , 2,—«-l .
7A.2 Solver for Blackadar Scheme
The Blackadar scheme requires solving the sparse linear matrix equation Aq=b of the form
7-53
-------
EPA/600/R-99/030
AjL *""*
/a
*2
0 0
% ° 4 o
o
o
o o 4,-, o
0 0 0 d.
(7A-11)
The solver for this system uses a similar numerical procedure as for the tridiagonal system. The
elements of the linear set of equations are related as follows:
(7A-12a)
and for 2 <* j ^ n :
By substituting qj with q , for each/, we get;
€i="
7-2
andfor
(7A-12b)
(7A-13a)
(7A-13b)
Note that Equation 7A-13b involves neither a forward nor a backward substitution loop.
7A3 Solver for Asymmetric Convective Model (ACM)
The Asymmetric Convective Model requires solving the linear matrix equation Aq=b with a
sparse matrix of the form:
* -'V :" i
7-54
-------
EPA/600/R-99/030
A =
o ... o Q ^
0
0
0
0
o o 4,_, cn_,
0004
(7A-14)
n J
The solver for this system is based on a numerical procedure, similar to the tridiagonal solver.
The relation among the variables qf are given as:
(7A-15a)
(7A-15b)
(7A-1 5c)
and for 2< j „
(7A-17a)
d.a., - e.c.
and all other ^,s are computed with:
CJ
(7A-17b)
7-55
-------
BPA/600/R-99/030
for 1 £ j 5 n — 1, The final substitution stage can be implemented either in a forward or backward
sweep. >
7-56
-------
EPA/S00/R-99/030
Chapter 8
GAS-PHASE CHEMISTRY
Gerald L. Gipsoin*
Human Exposure and Atmospheric Sciences Division
National Exposure Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711, USA
Jeffrey O. Young"
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711, USA
ABSTRACT
This chapter describes the manner in which gas-phase chemistry is treated in the Models-3
Community Multiscale Air Quality (CMAQ) modeling system. The CMAQ system currently
includes two chemical mechanisms — RADM2 and CB4 — with plans to incorporate a third ~ the
SAPRC97 mechanism — in the near future. Each of these mechanisms is described, and the
manner in which the first two are linked to the aqueous chemistry and aerosol formation
processes is discussed. Enhanced isoprene chemistry that has been included in the RADM2
mechanism is also described, and procedures for entering new chemical mechanisms in the
CMAQ system are addressed. The representation of reaction kinetics in the CMAQ system and
the numerical modeling of gas-phase chemistry are also presented. The CMAQ system currently
employs two numerical solvers, SMVGEAR and a variant of the QSSA method. The numerical
procedures used in each are presented, and the relative computational efficiencies of each on
different computing platforms are noted.
'Corresponding author address: Gerald L. Gipson, MD-80, Research Triangle Park, NC 27711. E-mail:
ggb@hpcc.epa.gov
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
8.0 GAS-PHASE CHEMISTRY
Since atmospheric chemistry plays a major role in many air pollution problems, the representation
of chemical interactions among atmospheric constituents is often an essential element of an air
quality model. All important chemical transformations relevant to the problem being studied must
be included to make accurate predictions of ambient pollutant concentrations. Many atmospheric
pollutants or their precursors are emitted as gases and interact primarily in the gaseous phase.
However, some important atmospheric processes such as acid deposition and the formation of
aerosols involve the interaction of constituents in the gas, liquid, and solid phases, so
transformations taking place in all three phases often need to be represented. For computational
efficiency, these processes are usually modeled separately. This approach has been adopted in the
Chemical Transport Model (CTM) that is part of the Community Multiscale Air Quality (CMAQ)
modeling system (hereafter referred to as the CCTM). This section addresses the modeling of gas-
phase transformations alone in the CCTM. Descriptions of the linkages of gas-phase constituents
With aerosols and with aqueous chemistry are discussed below and in Chapters 10 and 11,
respectively. A potential future improvement to the CCTM would involve more closely coupling
the chemical interactions taking place in all three phases. Nevertheless, the current formulation
still enables the investigation and assessment of environmental problems using a multi-pollutant,
one-atmosphere modeling concept.
',',,", ' " ' ' '•! , i •
Interactions in the gas-phase are represented in air quality models by means of chemical
mechanisms. The CMAQ system currently includes two base chemical mechanisms that have
been developed primarily to address issues associated with urban and regional scale ozone
formation and acid deposition - the CB4 (Gery et al, 1989) and RADM2 (Stockwell et al., 1990)
mechanisms. Variants of these two mechanisms have been developed for the CMAQ system to
provide the necessary linkages teethe aerosol and aqueous chemistry processes. Current plans
also call for adding a third mechanism — the SAPRC-97 mechanism (Carter, 1997). Although
these mechanisms should be adequate for many air pollution applications, it may be necessary to
Modify or even replace these mechanisms to address some issues. To facilitate changing
mechanisms and adding new ones, the CMAQ system has been equipped with a generalized
chemical mechanism processor. It must be emphasized, however, that supplemental data for other
CMAQ processors may be required when one of the predefined mechanisms is modified or
replaced. This is addressed in more detail section 8.2.5.
The remainder of this chapter addresses different aspects of the representation of gas-phase
chemistry hi the CMAQ system. The first section includes background information on chemical
mechanisms and provides the rationale for including the predefined chemical mechanisms in the
CMAQ system. The subsequent section describes the predefined chemical mechanisms as
implemented in the CMAQ system and addresses adding new mechanisms or changing existing
ones. This is followed by a description of reaction kinetics as it relates to the CCTM
representation, and the final section describes the mathematical procedures used internally in the
CCTM to solve the equations that arise from the mathematical representation of gas-phase
chemistry.
8-2
-------
EPA/600/R-99/030
8.1 Background
A chemical mechanism is a collection of reactions that transforms reactants into products,
including all important intermediates. Chemical mechanisms developed for air quality modeling
are highly condensed, parameterized representations of a true chemical mechanism. They include
artificial species and operators, and many of the mechanism reactions are parameterizations of a
large set of true atmospheric reactions. In some cases, mechanism reactions may include elements
which have no physical significance (e.g., products with negative stoichiometry). While it would
be difficult to design a generalized mechanism processor to handle all possible parameterizations
used in various condensed mechanisms, some degree of generalization in the CMAQ system is
achieved by using special conventions for entering chemical mechanisms. First, all reactions in the
mechanism are treated as if they are elementary, and the stoichiometric coefficients for all
reactants must be one. Since all reactions are assumed to be elementary, a reaction can have no
more than three reactants. These conventions permit the reaction rate to be derived directly from
the stoichiometric equation, thereby simplifying the mathematical representation of the reactions.
Other conventions adopted for the CMAQ generalized mechanism processor are included in
Section 8.3 and in Chapter 15.
Mechanism species can be divided into two categories: inorganic and organic. The number of
important inorganic species is relatively small, and they are almost always represented explicitly in
chemical mechanisms. The important inorganic species included in these mechanisms are ozone,
nitric oxide, nitrogen dioxide, nitric acid, nitrous acid, hydrogen peroxide, sulfur dioxide, and
several radicals formed through their interactions with other species. Although most of the
chemical reactions involving these species are common to all mechanisms, some differences do
exist. For example, some of the mechanisms omit a few reactions because they are normally
minor pathways and thus do not affect modeling results significantly. Also, different rate
constants may be used for some reactions, especially those that are photolytic. The
representation of organic species usually differs more substantially, however. Some species in the
mechanism represent real organic compounds, but others represent a mixture of several different
compounds. The manner in which the grouping of organic compounds is carried out typically
distinguishes one mechanism from another, and that is described next. In this chapter, the phrase
mechanism species is used to refer to any species in the gas-phase mechanism, regardless of
whether it is an explicit species or not.
Although explicit mechanisms have been developed for many organic compounds, the resultant
number of reactions and species needed to represent their atmospheric chemistry is too large to
model efficiently in photochemical grid models such as the CCTM. In addition, explicit
mechanisms have not yet been developed for most organic compounds, thereby requiring that
some reaction pathways be postulated. Thus, both compression and generalization are necessary
when depicting organic reactions. Although chemical mechanisms differ in the manner in which
organic species are represented, the mechanism developer usually chooses some distinguishable
organic properly to group similar organics into classes that reduce both the number of mechanism
species and reactions. The three most common representations include the lumped structure
8-3
-------
EPA/600/R-99/030
technique, the surrogate species approach, and the lumped species method. In the lumped
structure approach, organic compounds are apportioned to one or more mechanism species on the
basis of chemical bond type associated with individual carbon atoms (Whitten et al., 1980). In the
surrogate species method, the chemistry of a single species is used to represent compounds of the
same class (e.g., Lurmann et al., 1987). Generalized reactions are then written based on the
hypothetical model species. The lumped species method is very similar to the surrogate species
approach, but various mechanism parameters associated with a particular surrogate are adjusted
to account for variations in the composition of the compounds being represented by the surrogate
species (e.g., Carter, 1990, and Stockwell et al,, 1990).
The construction of a compact chemical mechanism necessarily introduces varying levels of
distortion, generalization, and omission in the representation of atmospheric chemistry (Jeffries,
1995). Although mechanisms are routinely tested using results obtained from environmental
chamber experiments, the data are often insufficient to resolve uncertainties associated with some
of the chemical representations. For example, Carter (1990) noted that much is unknown about
several important reaction types, and that their representations "... continue to be largely
speculative or are based on empirical models derived from fits to environmental chamber data."
Further, rate constants for some reactions are either unknown completely, or significant
disagreement exists as to their accuracy. Several studies have been conducted to compare
different chemical mechanisms (e.g., Leone and Seinfeld, 1985; Hough, 1988; and Dodge, 1989).
These comparisons revealed that the mechanisms often yield results that are similar for some
species. This could indicate that the fundamental atmospheric chemistry is fairly well understood
for these species, or that the mechanisms were derived from the same experimental kinetic or
mechanistic data, which may or may not be accurate. Larger differences tend to occur for those
species for which the atmospheric chemistry is more uncertain. Thus, it is often difficult to assess
the relative merits of different mechanisms when applied to any one situation. Therefore, the
CMAQ system includes the capability to use more than just one chemical mechanism.
Given the inherent uncertainties in existing chemical mechanisms, alternate approaches for
representing gas-phase chemistry are needed. One approach being explored involves decreasing
uncertainties associated with the simplifications that are introduced to reduce mechanism size.
This approach is based on the concept that much of the information needed for an expanded
chemical representation does not necessarily have to be included in the mechanism explicitly, but
rather can be maintained in auxiliary variables linked to a relatively small number of core species
that are included in the mechanism (Jeffries et al., 1993). Thus, it may be possible to expand
chemical representations without greatly increasing the size of the basic mechanism, and this
could be a future enhancement to the treatment of gas-phase chemistry in the CMAQ system.
8.2 Chemical Mechanisms in the CMAQ System
This section includes summary descriptions of the two basic chemical mechanisms included in the
CMAQ system -- the CB4 (Gery et al., 1989) and the RADM2 (Stockwell et al., 1990). Since the
SAPRC-97 mechanism (Carter, 1997) is to be added in the near future, some discussion of it is
. . . 8-4
-------
EPA/600/R-99/030
also included. These mechanisms require that information be supplied to the CCTM in a form
that is unique for each mechanism, and this is carried out in several sub-systems incorporated in
the CMAQ system. These include the emissions processing system which generates emissions for
key mechanism species; the initial conditions/boundary conditions processor that generates
ambient starting and boundary concentrations for mechanism species; and the photolysis rate
processor that produces mechanism specific photolysis rates. The reader is referred to the
chapters describing those sub-systems for a description of the treatment of mechanistic data, and
to the mechanism references for a more detailed description of each chemical mechanism.
In addition the base mechanisms, both the base CB4 and RADM2 mechanisms have been
modified in the CMAQ system to provide necessary linkages for aerosol and aqueous chemistry
processes, and the RADM2 mechanism has also been modified to create two new mechanism
variants that include enhanced isoprene chemistry representations. Note that the existing sub-
systems provide all of the necessary information for the extensions to the base mechanism. The
modifications to the base mechanisms are discussed below in the section on mechanism
extensions. Complete listings of all mechanisms currently available in the CMAQ system are
included in Appendix 8A. The last portion of this section briefly discusses changing the base
mechanism or their variants or adding new mechanisms to the CMAQ system.
8.2.1 CB4 Mechanism
The CB4 mechanism is a lumped structure type that is the fourth in a series of carbon-bond
mechanisms, and differs from its predecessors notably in the detail of the organic compound
representation. It has been used in models such as EPA's Empirical Kinetic Modeling Approach
(EPA, 1989) and Regional Oxidant Model (Lamb, 1983), and in versions IV and V of the Urban
Airshed Model (EPA, 1991 and SAI, 1993). The CMAQ implementation of the basic CB4
mechanism includes 36 species and 93 reactions, including 11 photolytic reactions.
The CB4 uses nine primary organic species (i.e., species emitted directly to the atmosphere as
opposed to secondary organic species formed by chemical reaction in the atmosphere). Most of
the organic species in the mechanism represent carbon-carbon bond types, but ethene (ETH),
isoprene (ISOP) and formaldehyde (FORM) are represented explicitly. The carbon-bond types
include carbon atoms that contain only single bonds (PAR), double-bonded carbon atoms (OLE),
7-carbon ring structures represented by toluene (TOL), 8-carbon ring structures represented by
xylene (XYL), the carbonyl group and adjacent carbon atom in acetaldehyde and higher molecular
weight aldehydes represented by acetaldehyde (ALD2), and non-reactive carbon atoms (NR).
Many organic compounds are apportioned to the carbon-bond species based simply on the basis
of molecular structure. For example, propane is represented by three PARs since all three carbon
atoms have only single bonds, and propene is represented as one OLE (for the one carbon-carbon
double bond) and one PAR (for the carbon atom with all single bonds). Some apportionments are
based on reactivity considerations, however. For example, olefins with internal double bonds are
represented as ALD2s and PARs rather than OLEs and PARs. Further, the reactivity of some
compounds may be lowered by apportioning some of the carbon atoms to the non-reactive class
8-5
-------
EPA/600/R-99/030
NR (e.g., ethane is represented as 0,4 PAR and 1.6 NR). Apportioning rules have been
established for hundreds of organic compounds, and are built into the emissions processing sub-
systems to produce the appropriate emission rates for the CB4 mechanism species.
The CB4 mechanism described by Gery et al. (1989) has undergone several changes since its
publication. In 1991, the PAN rate constants were changed and a termination reaction between
the XO2 operator and the HO2 radical were added. Subsequently, terminal reactions for the
XO2N operator were also added. An updated CB4 isoprene chemistry mechanism based on the
work of Carter (1996) was developed in 1996. All of these changes have been incorporated in the
CMAQ version. It should also be noted that the original CB4 mechanism incorporated simple
Arrhenius type rate constant expressions that were derived from more complex expressions for
temperature and pressure dependent rate constants. Since the top of the CCTM domain may
extend to heights that makes pressure dependencies important, the CMAQ version incorporates
the original expressions rather than the derived ones.
8.2.2 RADM2 Mechanism
' " ' ' ' • ' f ; • .
The RADM2 mechanism is a lumped species type that uses a reactivity based weighting scheme to
adjust for lumping (Stockwell et al., 1990). It has evolved from the original RADM1 mechanism
(Stockwell, 1986), and is employed in version 2 of the Regional Acid Deposition Model (Chang
et al., 1987). The base mechanism as implemented in the CMAQ system contains 57 model
species and 158 reactions, of which 21 are photolytic.
In RADM2, the primary organics are represented by 15 mechanism species, five of which are
explicit because of their high emission rates or because of special reactivity considerations
(methane, ethane, ethene, isoprene, and formaldehyde ). The other ten represent groups of
organic compounds aggregated on the basis of their reactivity with the hydroxyl radical (HO)
and/or their molecular weights. To account for varying reactivities of the different organics that
are lumped into a single group, emissions of each organic within a group are weighted by a
reactivity factor (F) that is computed as the ratio of the fraction of emitted organic compound that
reacts to the fraction of the mechanism species that reacts:
F = — /[H°] d° . (8_i)
The integral term is estimated from a daily average integrated HO radical concentration of 110
ppt min that was derived from RADM simulations (Stockwell et al,, 1990). Note that F
approaches unity if the reactivity of the emitted organic nearly equals that of the mechanism
species or if the reactivities of both are very large. As with CB4, the RADM2 lumping and
weighting rules have been built into the CMAQ emission processing system.
8-6
-------
EPA/600/R-99/030
The implementation of the RADM2 mechanism in the CMAQ system is almost identical to that
described by Stockwell et al, (1990), with only two minor modifications. First, the reaction of HO
with cresol (CSL) was reformulated as follows to eliminate negative stoichiometry in the
mechanism:
From: HO + CSL - 0.1 HO2 + 0.9 XO2 + 0.9 TCO3 - 0.9 HO kHO+CSL;
To: HO + CSL -. 0.1 HO2 + 0.9 XO2 + 0.9 TC03 WcsiJ
HO + CSL-CSL 0.9kH0.CSL.
(Note that negative stoichiometry is permitted in the CMAQ system, but was removed here for
consistency with previous implementations of RADM2.) Second, the concentration of methane in
the CCTM is assumed to be constant at 4.5xl013 molecules/cc. Thus, methane was removed as a
reactant in the reaction of OH with methane and the corresponding rate constant changed from
second-order to pseudo first-order using the assumed CCTM methane concentration.
8.2.3 SAPRC-97 Mechanism
The SAPRC-97 mechanism (Carter, 1997) employs the lumped surrogate species approach, but
offers the capability to incorporate semi-explicit chemistry of selected organics. The SAPRC
series of mechanisms evolved from the "ALW" mechanism of Atkinson et al. (1982). SAPRC-97
is similar to its predecessors SAPRC-90 and SAPRC-93, but incorporates improvements to
aromatic chemistry and updates to reactions of many individual organic compounds. Although
many of the reactions for organic compounds are generalized and incorporate non-explicit
species, product yield coefficients and rate constants are tabulated for over 100 individual organic
compounds. Thus, each of these organics can be modeled individually by including their semi-
explicit chemistry in the mechanism. Due to computational constraints, however, the full set of
organic compounds cannot be incorporated in an Eulerian model. For this situation, the
mechanism is condensed by lumping individual organic compounds into groups with
corresponding rate constants and product yield coefficients that have been weighted by mole
fractions of the individual organics. The mole fractions are typically derived from emission
inventory data used in the model simulation. Thus, unlike the previous mechanisms, the SAPRC-
97 mechanism can potentially change with each application since new rate constants and product
yield coefficients can be computed for each application.
The SAPRC-97 mechanism has been developed with supplemental software to facilitate
constructing mechanisms of varying levels of condensation. Documentation of the procedures
include three distinct levels of detail, differing primarily in the number of organic species that are
included in the mechanism (Carter, 1988). Since no decisions have made on the form of the
mechanism that will be added to the CMAQ system, listings for this mechanism are not included
in Appendix 8 A. Documentation will be provided when the CMAQ version is made available
however.
8-7
-------
EPA/600/R-99/030
8.2.4 Extended Mechanisms
: • :• • ••• (
Each gas-phase chemical mechanism has been linked to aqueous chemistry and to aerosol
formation processes. Since these linkages required some modifications to the original gas-phase
mechanisms, different versions of the same mechanism were created for modeling gas-phase
chemistry alone or for modeling gas-phase chemistry with or without aerosols and/or aqueous
chemistry. Different versions of the mechanisms are distinguished by means of a special naming
convention. Gas-phase mechanisms that have not been modified are referred to by their base
name (e.g., CB4, RADM2, and SAPRC when the latter is available). Mechanisms that have been
modified to account for aerosol production have their names appended with "_AE", mechanisms
modified for aqueous chemistry are appended with "_AQ", and mechanisms modified for both are
appended with "_AE_AQ". Thus, CB4_AE_AQ refers to the CB4 gas-phase mechanism that has
been modified to include linkages to both aerosols and aqueous chemistry. A second set of
RADM2 gas-phase mechanisms that incorporates new isoprene chemistry has also been included
in the CMAQ system. These mechanisms include either "_CIS1" or "_CIS4" in their names to
denote that the mechanism incorporates enhanced isoprene chemistry. Methods used to develop
the extended mechanisms are described below according to the three types of extensions: aerosol,
aqueous chemistry, and isoprene chemistry.
8.2.4.1 Aerosol Extensions
A major pathway leading to the formation of aerosols is the oxidation of sulfur dioxide (SO2) to
sulfate, primarily by the gas-phase reaction of SO2 with the hydroxyl radical (OH). All
mechanisms in the CMAQ system incorporate this reaction. Because organics are represented
differently in the base mechanisms, however, aerosol formed from the reactions of organic
compounds must be handled somewhat differently. In the CCTM, organic aerosol formation is
quantified using aerosol yields, i.e., pigm'3 of aerosol produced per ppm of organic reacted with
OH, ozone or nitrate radical (NO3). The yields used in the CCTM are those reported by Bowman
et al. (1995) that were derived from the work of Pandis et al. (1992). These yields are given in
terms of the SAPRC-90 chemical mechanism species, so some adjustments were required to adapt
them to the CMAQ mechanisms. In the CMAQ system, aerosol production is assumed to occur
from reactions involving five different generic organic groupings. Individual mechanism species
are then mapped to these general groupings to obtain the aerosol yields. The five generic groups
are defined as: 1) long-chain alkanes; 2) alkyl-substituted benzenes such as toluene and xylene; 3)
cresol and phenols; 4) long-chain olefins; and 5) monoterpenes. Note that the aerosol yields vary
significantly among these five groups, so it is important to map the organic species in each
mechanism to the proper aerosol production group. The remainder of this section describes the
mapping that is used for the CMAQ base mechanisms and how the aerosol production rates are
determined from the gas-phase reactions. The derivation of the yields used in the CCTM and the
manner in which they are used in the aerosol module are described in Chapter 10.
To apply the aerosol yields, the amount of reactant consumed by reaction must be determined for
several mechanism species. In the CMAQ system, this is accomplished by using "counter" species
-
8-8
-------
EPA/600/R-99/030
that have been added as products to those reactions involving the mechanism species of interest
(e.g., Bowman et al., 1995). These counter species are essentially "dummy" species with no
physical significance, and are not subjected to any other model process such as advection or
diffusion. Thus, changes in their concentrations reflect the effect of chemical reaction alone.
Also, their inclusion in the mechanisms does not affect basic gas-phase chemistry since they do
not interact with any of the other species in the mechanism.
Special procedures are used in the CCTM to determine aerosol production from monoterpenes
since their aerosol yields are relatively large and they are either lumped with other organic
compounds into a single mechanism species or are apportioned among several mechanism species.
The approach involves tracking the rate of reaction of monoterpenes separately from the rates of
their mechanistic representation. The CMAQ emission processor generates emissions for
monoterpenes as a unique species in addition to lumping or apportioning the emissions into the
appropriate mechanism species. Whenever aerosols are being modeled, the unique monoterpene
species is included in the mechanisms and is modeled as a separate species. As with the counter
species, however, the monoterpene species is incorporated in the mechanisms such that it does
not affect the basic gas-phase chemistry. This is described in more detail below. A potential
future modification to the CMAQ system would involve incorporating a more explicit
representation of monoterpenes in the base mechanisms that would eliminate the need for this
special treatment and would also improve the chemical representation of these species in the gas-
phase mechanisms (e.g., Stockwell et al., 1997).
• RADM2_AE. Much of the linkage between the RADM2 mechanism and aerosol
formation is relatively straightforward. Aerosol production from SO2 and long-chain
alkanes is derived from the amount of SO2 and HC8 reacted with OH, respectively.
Similarly, aerosol production from alkyl-substituted benzenes is derived from the sum of
the TOL and XYL reactions with OH. The production from phenols and cresols is based
upon the sum of the CSL reactions with OH and NO3. Thus, special counter species
named SULAER, HC8AER, TOLAER, XYLAER, and CSLAER have been added to
track these reactions.
In the RADM2 mechanism, both monoterpenes and other olefinic compounds are lumped
into the mechanism species OLI. As noted above, however, monoterpenes are modeled
separately whenever aerosols are modeled. The monoterpenes are represented in the
RADM2 mechanism by the species TERP, and the following reactions are added:
TERP + HO - TERPAER + HO
TERP + NO3 - TERPAER + NO3
TERP + O3 - TERPAER + O3
8-9
-------
EPA/600/R-99/030
These reactions use the same rate constants as the reaction of OLI with these species, and
have the TERPAER counter species added to track the throughput of this reaction. Note
that concentrations of the reactants OH, NO3 and ozone are unaffected by these reactions
since their production equals their loss.
i
The final pathway for aerosol production in the RADM2 mechanism is via reaction of
long-chain olefins with OH, NO3 and ozone. The RADM2 mechanism species OLI is used
as the surrogate for long-chain olefins, and thus a counter species product named
OLIAER is added to each of the reactions of OLI with OH, NO3 and ozone. Since OLI
includes both monoterpenes and other olefins, however, OLIAER tracks the reaction rate
of both. The amount of long-chain olefins reacted is determined by subtracting the
concentration of the counter species TERPAER from that of OLIAER. The yield of
aerosols from long-chain olefins is then applied to this difference to obtain aerosol
production by this pathway.
...... ..
• CB4_AE. Since the CB4 gas-phase mechanism is structure-based, individual organic
molecules are often disaggregated and assigned to more than one mechanism species. For
example, long-chain alkenes are apportioned to both the PAR and OLE mechanism
species. Thus, many of the organic mechanism species contain fragments of molecules,
and the identity of the original contributing organic compound is lost. As a result, it is not
possible to ascertain with certainty the amount of long-chain alkanes and alkenes reacting
hi the CB4 mechanism, and thus aerosol production via these pathways is omitted. The
production of aerosols from the reactions of toluene, xylene, and cresol is included,
however, by tracking the amounts of TOL, XYL, and CRES that react using the counter
species TOLAER, XYLAER, and CSLAER. The manner in which aerosol production
from monoterpenes is modeled is identical to that used in the RADM2 mechanism.
Monoterpenes are modeled independently as the mechanism species TERP, with rate
constants for the reactions of TERP with OH, O3, and NO3 set to the same values as those
used in the RADM2 mechanism extension.
I
8.2.4.2 Isoprene Extensions
Over the past few years, the importance of isoprene in ozone formation has become a major
concern. Its representation in the original gas-phase mechanisms was substantially condensed,
partially because of computational resource considerations and partially due to significant
uncertainties about the pathways of its reaction products. Recent mechanistic and environmental
chamber studies have led to a greater understanding of its atmospheric chemistry and thus
improved mechanistic representations (Carter and Atkinson, 1996). In the CMAQ system, two
different levels of more detailed isoprene chemistry have been included in the RADM2
mechanism, and these are referred to as the one-product and the four-product Carter isoprene
mechanisms (Carter, 1996). Both are condensed forms of the more detailed mechanism
developed by Carter and Atkinson (1996). Since this detailed mechanism may be too large to use
in full-scale Eulerian modeling studies, Carter condensed the detailed mechanism to two levels of
":. > '• . • • . ' I • . .-
8-10
-------
EPA/600/R-99/030
detail: one in which isoprene products are represented by four products, and one in which only
one product is used. The four-product mechanism is the lesser condensed of the two, and
includes the explicit representation of many of the isoprene's unique products (e.g., methacrolein,
methyl vinyl ketone, and methacrolein's PAN analogue). The one-product form lumps the major
products into a single species, thereby yielding a more compact albeit less explicit mechanism. As
noted above, these two mechanisms are named RADM2_CIS4 and RADM2_CIS1, and both have
been linked to aerosols and aqueous chemistry as well. It should also be noted that the isoprene
chemistry incorporated in the CMAQ CB4 mechanism corresponds to the Carter 1-product form,
but the 4-product form is not available for the CB4 mechanism in the CMAQ system
8.2.4.3 Aqueous Chemistry Extensions
The base RADM2 mechanism does not have to be modified to link it to the aqueous chemistry
processes since the aqueous processes in the CMAQ system are similar to those incorporated in
the original RADM model. As described in Chapter 15, other aspects of the linkages require a
separate mechanism with a unique name. The linkages to aqueous chemistry do require some
minor changes to the CB4 mechanism however. These changes were based on a variant of the
CB4 mechanism developed for acid deposition modeling by Gery et al. (1987). In this version,
the following product species that were omitted in the base CB4 mechanism are included: formic
acid, acetic acid, peroxyacetic acid, and methylhydroperoxide (MHP). Since these species are
products only, their inclusion in the mechanism does not affect the concentrations of any of the
other mechanism species. It should be noted, however, that the concentration of MHP in this
modified mechanism represents an upper limit for two reasons. First, known decomposition
pathways for it are not included in the mechanism. Second, the production of MHP will be
overstated since it is produced by an operator that includes radicals other than the methylperoxy
radical (Gery et al., 1987).
8.2.5 Changing or Adding Mechanisms in CMAQ
As noted in the introduction to this chapter, the CMAQ system has been instrumented with a
generalized chemical mechanism processor to facilitate making changes to existing mechanisms or
adding new mechanisms. The procedures for altering or adding a new mechanism are described in
EPA (1998), and will not be repeated here. It should be emphasized, however, that the addition
of a new mechanism will likely require modifications to the previously mentioned subsystems that
provide key mechanism-specific information, i.e., emissions, initial/boundary condition, and
photolytic rate processors. Changes to an existing mechanism would also likely require
modifications to these processing subsystems if new organic species are added or if an alternative
organic grouping scheme is implemented. If changes are limited such that they affect only the
reactions of intermediate and/or product species, however, these subsystems may not need to be
changed at all, and the modifications can then be implemented solely within the generalized
chemical mechanism processor. For example, the modifications to the base mechanisms to
provide linkages to aerosol and aqueous chemistry and to expand isoprene chemistry that were
described previously did not require any major changes to the other processors except to add a
8-11
-------
EPA/600/R-99/030
photolysis rate for acrolein in the photolytic rate processor for the CIS 1 and CIS4 versions of
RADM2.
I
I < II . . ' ' . |
Although the CMAQ system provides a convenient tool for making mechanism changes, some
caution should be exercised in modifying existing CMAQ mechanisms. The mechanisms currently
in the CMAQ system have been evaluated outside of the CMAQ system using environmental
chamber data and/or more detailed chemical mechanism representations. Any proposed changes
to reactions or reactions rates that significantly affect model predictions should normally be
subjected to similar independent testing before being introduced into the CMAQ system and
subsequently used in modeling applications. Thus, it would be expected that the introduction of
most changes to a mechanism in the CMAQ system would only be performed by a researcher who
is experienced in atmospheric chemistry and is familiar with the base mechanism.. Finally, it
should be noted that the existing CMAQ mechanisms are fully specified. In most instances, it will
only be necessary for a user to choose one of the existing mechanisms for their application, and it
will not be necessary to make any changes to that mechanism.
, ,: -::, . . , . . • . . |
8.3 Reaction Kinetics
The rates of chemical reaction determine whether a species is formed or destroyed by gas-phase
chemistry. Since the CMAQ system treats all reactions as if they are elementary, the laws of
reaction kinetics can be used directly to develop mathematical expressions for the rates of each
chemical reaction. This section describes the rate expressions and the forms of the rate constants
that are used in those expressions, with special emphasis placed on the conventions used in the
CMAQ system. The reader may also wish to refer to Chapter 15 and EPA (1998) for details on
how mechanism data are entered in the CMAQ system.
•
•
8.3.1 Reaction Rates
The rate of a chemical reaction / (r,) can be expressed as the product of a rate constant (k,) and a
term that is dependent on the concentrations of the reactants:
, ; " ' *. - i i"",1 .in1 . " . i. . r' ,' . . '• ' • • . • .it" ' ! | ... "
>; = ^/(concentration) . (8-2)
For elementary reactions, the concentration dependent term is simply the product of reactant
concentrations, and the rate of reaction takes one of the following forms:
i
.... ., . i .
for first-order reactions
i i
r. = \ ktC^C2 for second-order reactions (8-3)
12C3 for third-order reactions
wr^ere C,, C2,and C? refer to the concentration of reactants 1, 2 and 3, respectively. Note that
when a species reacts with itself, the concentration dependent term includes the species
8-12
-------
EPA/600/R-99/030
concentration squared. Thus, the rate for the reaction NO + NO + O2 - 2NO2 is equal to
A[NO][NO][O2] and not A[NO][OJ.
Several important ter-molecular reactions involve O2 and/or N2 which mediate those reactions by
absorbing energy from exothermic bi-molecular reactions. When either N2 or O2 serves this role,
the third body is usually referred to as "M", where M = N2 + O2. Since their concentrations are
relatively stable in the atmosphere, some mechanism developers convert second- or third-order
reactions that include these species to a reaction one order lower by multiplying the higher-order
rate coefficient by the concentration of M, O2, or N2. The CMAQ convention is to include third-
body reactants in the reaction rate calculations if they are explicitly shown in the reaction, and to
omit them if they are not shown or included only as comments. For example, consider the
bimolecular reaction: O'D + O2 - O3P + O2. If the reaction is written in this form, the reaction is
assumed to be second-order and the CCTM will use the appropriate concentration for O2 in the
rate computation. If the reaction is written as O'D - O3P (or as O'D {+ O2} - O3P, where here
the braces denote a comment), the reaction rate will be assumed to be first-order and the CCTM
will not include the O2 concentration in the reaction rate calculation. In the latter case, the
mechanism developer must specify a pseudo first-order rate constant for the reaction. The same
convention also applies to H2O.
8.3.2 Rate Constant Expressions
As shown in Equation 8-3, the rate of reaction is related to the concentration term by a constant
of proportionality k,. The rate constant k, can take many forms depending upon the characteristics
of the reaction. One important class of uni-molecular reactions involves the absorption of radiant
energy and subsequent dissociation of the reactant into product species. The rate constants for
these types of reactions are functions of the incident radiant energy and properties of the
absorbing molecule, such as the absorption cross section and the quantum yield. In the CMAQ
system, these rate constants are calculated by the photolytic rate processor, and the details of
these calculations are described in Chapter 14. The remainder of the reactions are classified as
thermal, and their rate constants are typically functions of temperature and sometimes pressure.
The calculation of these rate constants is discussed below.
To facilitate incorporating rate constant information for thermal reactions, the CMAQ generalized
mechanism processor (discussed in Chapter 15) has been designed to accept the standard rate
constant expressions used in NASA (1997). Rate constant information is most often supplied in
cms units (i.e., gas concentrations in molecules/cc and time in seconds), but some mechanisms use
mixing ratio units (i.e, gas concentrations converted to mixing ratios in parts per million and time
in minutes). The CMAQ generalized mechanism reader is designed to accept either, but they
must be consistent throughout the mechanism (i.e., the same units must be used for all rate
constant forms that can be expressed in either set of units). Some rate constant expressions (e.g.,
falloff expressions and other special forms discussed below) can be expressed only in cms units,
however, and must always be in these units even when mixing ratio units are being used for all
other types of rate constants. The CCTM will automatically perform the necessary units
8-13
-------
EPA/600/R-99/030
i
i
conversions during the model simulation. Nevertheless, since the CMAQ domain typically extends
through the entire troposphere, cms units are usually preferred because differences in number
density differences with height are explicitly accounted for with those units.
Descriptions of the forms of rate constant expressions currently used in the CMAQ system are
presented next.
• Arrhenius Equation. Many rate constants exhibit a temperature dependence that
corresponds to the Arrhenius equation:
'kf*A4-*ir> \ (8-4)
where A is the pre-exponential factor, E is the activation energy divided by the gas
constant .ft, and T is the temperature in degrees Kelvin. For this form of reaction, either
cms or mixing ratio units may be used, and only A and E need to be specified.
.
: i
• Temperature Dependent A-factors. For some reactions, the temperature dependence of
the pre-exponential factor can become significant, and the Arrhenius equation does not
hold. These rate constant expressions can often be put in the following form that is
available in the CMAQ system:
k = A (T/300f e<-£/r> (8-5)
where A, E, and Tare defined as above, and B is an empirically derived constant that
provides a best fit to the data (Pitts-Finlayson and Pitts, 1986). For this form, either set of
units can be used, and only A, E and B need to be specified.
Falloff Expressions. Several ter-moleeular reactions exhibit pressure dependencies that
can be significant when modeling atmospheric chemistry. These can be especially
important when modeling from the troposphere through the stratosphere. In these cases,
the rate constant increases with increasing pressure. In effect, the behavior of these
reactions approaches second-order at high pressure and third-order at low pressure.
Equation 8-6 gives an effective second-order rate constant for the falloff region between
these two limits.
_ {1 * [AT'log(*0[M]/*J]2r'
In Equation 8-6, &0and &„ are the low- and high-pressure limiting rate constants,
respectively, and are calculated using the temperature dependent .4-factor form described
above. The parameters F^ and N are also reaction specific, but for atmospheric conditions
are very often 0.6 and 1.0 respectively (Finlayson-Pitts and Pitts, 1986). For this type of
rate constant expression, A, E, and B must be specified for both k0and k» , and only cms
8-14
-------
EPA/600/R-99/030
units are allowed. If the parameters Wand Fc are not specified, the values listed above will
be used,
• Special Forms. The following special rate const-ant forms are also in general use and
have been included in the CMAQ system:
* = *, + *j[M] (8-7)
kJM]
<8-8>
h = A (1.0 + 0.6P) (g-9)
Equation 8-7 is used for the rate constants of the reactions forming hydrogen peroxide
from hydroperoxy radicals (HO2 + HO2 - H2O2 and HO2 + HO2 + H2O - H2O^. Equation
8-8 is used for the reaction of the hydroxyl radical with nitric acid (HO + HNO3 - NO3 +
H2O), and Equation 8-9 is used for the reaction of the hydroxyl radical with carbon
monoxide (HO + CO -> HO2 + CO2). In these equations, k0, kh k2, and k3 are calculated
using the Arrhenius equation, and A and E must be specified for each, with A given in cms
units. In Equation 8-9, P is the atmospheric pressure in atmospheres and A can be
specified in either set of units.
• Reverse Equilibria Forms. The CMAQ system also includes a special reverse
equilibrium form for first-order decomposition reactions. With these types of reactions,
the equilibrium constant is input in a form similar to the Arrhenius equation. Thus, the
rate constant can be expressed as follows:
* = kflA4~EIT) (8-10)
In Equation 8-10, ^-is the rate constant for the forward reaction forming a species, and the '
denominator is an Arrhenius-like form for the equilibrium constant. These reaction rate
coefficient types are used, for example, for the decomposition of pernitric acid and
nitrogen pentoxide (i.e., HNO4 - HO2 + NO2 and N2OS - NO2 + NO3). In the CMAQ
system, A, E, and the corresponding forward reaction must be specified. Either set of
units may be used with this form.
8.4 Mathematical Modeling
This section describes the mathematical modeling concepts used in the CCTM to simulate gas-
phase chemical reactions. The first sub-section describes the fundamental equations that must be
solved and some of the difficulties encountered in obtaining solutions to them. The next two sub-
sections describe the two gas-phase chemistry solvers that are currently available in the CCTM.
The last sub-section summarizes some of the important solver characteristics.
8-15
-------
EPA/600/R-99/030
8.4.1 Governing Equations
j
As described in Chapter 5, operator splitting allows gas-phase chemistry to be de-coupled from
physical processes such as advection, diffusion and deposition, and, as noted in the introduction to
this chapter, gas-phase chemistry is modeled separately from aerosol formation and aqueous
chemistry. As a consequence, continuity equations for each gas-phase mechanism species can be
formulated and solved independently on a cell-by-cell basis. By using the kinetics laws for
elementary reactions and applying a mass balance to each species, the following equation for the
rate of change of each species concentration can be derived for a single cell:
dt
where
dC.
• = />, -L,C, (8-11)
, = £ vfi/r, (8-12)
/=!
and
L,C= £ r, (8-13)
. . /=i .. . .
In Equations 8-11 through 8-13, C, is the concentration of species /, vu is the stoichiometric
coefficient for species / in reaction /, and r, is the rate of reaction /. The sum l=\...m, is over all
reactions in which species z appears as a product, and the sum /=!...«, is over all reactions in
which species / appears as a reactant.
I
•
' Equation 8-11 states that the change in species concentration is equal to the chemical production
of that species minus its chemical loss, and it is the fundamental species continuity equation for
gas-phase chemistry that is solved in the CCTM. If the concentration of species / is known at
some particular time, its concentration can be computed at a later time by solving Equation 8-11.
Since the production and loss terms contain references to other species concentrations, however,
Equation 8-11 must be solved as part of a coupled set of ordinary differential equations.
i .
It should also be noted that the CCTM contains an option for including emissions in either the
vertical diffusion process or in gas-phase chemistry. When emissions are included in gas-phase
chemistry, the fundamental form of the Equation 8-11 is not altered since an emission source term
is simply a zeroth-order production rate. Thus, for the discussions that follow, the production
term P, is assumed to include an emission source term if species / is emitted and emissions are
included in gas-phase chemistry.
8-16
-------
EPA/600/R-99/030
The system of non-linear, ordinary differential equations (ODEs) arising from Equation 8-1 1 for JV
species can be expressed as follows:
dC.
-i = P ,(«,() - L^fyC. = fi(t,t) i = 1,2,... JV (g-14a)
at
with the initial conditions:
c('o) = co (8-14b)
where c is the vector of species concentrations and N is the total number of species in the
chemical mechanism. Numerical "marching" methods are typically employed to obtain
approximate solutions for this class of problem. In these methods, the concentrations of all
species are given at the starting point and a solution is computed at selected time intervals (i.e.,
time steps) using the right hand side of Equation 8-14a. Two sources of difficulty arise in
obtaining numerical solutions to these equations as they apply to atmospheric chemistry problems.
First, the system is nonlinear because the production and loss terms include second- and third-
order reactions. Second, the system of equations is "stiff" because of the widely varying time
scales of the chemical reactions and complex interactions among species. A stiff system can be
described mathematically as one in which all the eigenvalues of the Jacobian matrix of Equation 8-
14a are negative, and the ratio of the absolute values of the largest-to-smallest real parts of the
eigenvalues is much greater than one. Systems are typically termed stiff if the latter ratio is
greater than 104. For atmospheric chemistry problems, the ratio is often greater than 1010, making
the system very stiff (Gong and Cho, 1993).
The stiffness problem coupled with the fact that these equations must be solved for tens of
thousands of cells in a typical modeling application require that special numerical methods be
employed. The use of standard explicit methods is often precluded because relatively small time
steps are required to maintain numerical stability and obtain accurate solutions. On the other
hand, classical implicit methods that are both accurate and stable have not often been used
because of high computational demands. As a result, several special techniques have been
developed to obtain reasonably accurate solutions in a computationally efficient manner. At
present, two solution techniques are available in the CCTM: the implicit Sparse-Matrix
Vectorized Gear algorithm (SMVGEAR) and a variant of the explicit Quasi-Steady State
Approximation (QSSA) method. Each of these is described in detail below. Although each of
these techniques, as well as others that have been used in atmospheric chemistry models, have
been designed to be computationally efficient, they still consume 50 to 90% of the total CPU time
used in a model simulation. Thus, obtaining a numerical solution to Equation 8-14a,b is normally
the most computationally intensive portion of the CCTM.
8-17
-------
EPA/600/R-99/030
8.4.2 SMVGEAR
Numerical solvers based on the algorithm developed by Gear (1971b) have traditionally been used
to obtain accurate solutions to stiff ODE problems. The technique is implicit in method, does not
amplify errors from one step to another and incorporates automatic step size and error control. In
fact., solvers based on this method have often been used to evaluate other faster solution methods
for accuracy (e.g., Odman et al., 1992; Gong and Cho, 1993; Dabdub and Seinfeld, 1995; and
Say lor and Ford, 1995). Past versions of this code have rarely been installed in Eulerian models,
however, because of the high computational cost. Jacobson and Turco (1994') have modified the
Gear algorithm to obtain considerable speedups on vector computers. The SMVGEAR algorithm
is highly vectorized to improve computational performance on vector computers and it
incorporates special sparse matrix techniques to increase computational efficiency. Further
enhancements have been obtained by ordering the cells for processing. Each of these are
described further below. Since the technique is based on Gear's original algorithm, it is briefly
described first. For more details on the Gear method, the reader is referred to Gear (197 la) and
Gear(1971b).
8.4.2.1 Standard Gear Algorithm
The Gear algorithm is one of a class of methods referred to as backward differentiation formulae
(BDF). The generalized BDF that forms the basis for Gear's method can be expressed as follows:
Cn =
E «j*.-j (8-15)
where n refers to the time step, h is the size of the time step,/? is the assumed order, /30 and #• are
scalar quantities that are functions of the order, and f(cn/J is the vector of production and loss
terms defined by the right hand side of Equation 8-14a. The method is implicit since
concentrations at the desired time step n depend on values of the first derivatives contained in
f(cn,/n) that are functions of the concentrations at the same time. The order of the method
corresponds to the number of concentrations at previous time steps that are incorporated in the
summation on the right hand side of Equation 8-15.
To facilitate changing step size and estimating errors, the multi-step method in Equation 8-15 is
transformed to a multi-value form in which information from only the previous step is retained,
but information on higher order derivatives is now used (Gear, 1971b). In this formulation, the
solution to Equation 8-14a,b is first approximated by predicting concentrations and higher order
derivatives at the end of a time step for each species using the following matrix equation:
,'
z,,n.«» = Bz/n_, (8-16)
8-18
-------
EPA/600/R-99/030
where zf = [c,, /zc',,,.., hpc/p)/p\]T, the subscript n,(Q) refers to the prediction at the end of time step
n, and the subscript n-l refers to values obtained at the end of previous time step (or the initial
conditions when n = 1). B is the Pascal triangle matrix, the columns of which contain the
binomial coefficients:
B =
'l
1 1 1 •
123-
1 3 •
1 •
0
•• 1 1
•• p-l p
I p
1
(8-17)
The prediction obtained from Equation 8-16 is then corrected by solving for z/n such that the
following relations hold for all species:
= Z
/,n,(0)
(8-18)
In Equation 8-18, r is a vector of coefficients that is dependent on the order, but r2, the element
corresponding to the first derivative location in z, is always equal to one. Thus, the correct value
of cin is obtained when the calculated value of c '„ equals/(cn,?n) in Equation 8-18. An
approximate solution for cin is obtained by applying Newton's method to the system of equations
that correspond to the first equation in 8-18 for all species. This leads to the following corrector
iteration equation:
m)] (8-19)
where m refers to the Newton iteration number, the vector f(cn,4) is calculated using
concentrations computed for the m-th iteration, 6c is a vector containing the most recent
estimates of first derivatives, I is the identity matrix, and J is the Jacobian matrix whose entries
are defined as:
Be,
ij = 1.2....AT
(8-20)
At the end of each iteration, the vector containing the first derivatives ( 8c) is updated, but higher
order derivatives in z need not be computed until convergence is achieved.
After convergence is achieved, the local truncation error for each species, e,, is given by:
8-19
-------
EPA/600/R-99/030
(8-21)
The error is estimated in the algorithm by neglecting the O(/z^2) term and approximating
from the backward difference ofhfc^/pl which can be calculated using the last components of the
vectors z^m) arid zin.t defined above. These error estimates are used to control accuracy and to
change both the time step size and the order of the method when warranted.
: ; • • . •,:'.•..!
Although several variants of the basic Gear algorithm have been developed, the fundamental
coRiputational scheme can be described generically as follows. At the beginning of any
integration interval, the order is set to one and the starting time step is either calculated or
selected by trie user. Each time step is initiated by predicting concentrations at the end of the time
step using Equation 8-16. Corrector iterations are then carried out using Equation 8-19 until
prescribed convergence criteria are achieved or non-convergence is deemed to have occurred.
When convergence is achieved, the error is computed using the approximation for Equation 8-21 .
If the error is within prescribed limits, the solution for the step is accepted and the step size and
order to be used injhe next step are estimated. The size of the time step is estimated for the
current order, the next lowest order, and the next highest order using error estimates derived from
Equation 8-21 for the next step. From these, the largest time-step size and its corresponding
order are then selected for use in the next step. If either the convergence or error test fails, the
integration is restarted from the beginning of the failed time step after re-evaluating the Jacobian
matrix, reducing the size of the time step, and/or lowering the order.
The individual operations described above are normally handled automatically in Gear algorithms.
To reduce computational demands, the algorithms also utilize several empirically based rules. For
example, the Jacobian matrix is only updated after a prescribed number of successful steps have
been completed, if the order changes, or if a convergence or error test failure occurs. In the
Newton iterations, progress towards convergence is monitored and the iterations halted if the
progress is judged insufficient or if three complete iterations have been performed without
convergence being achieved. To maintain numerical stability, changes to the size of the time step
and the order are allowed no more than once every p+l steps for a/j-th order method.
8.4.2.2 Vectorized Gear Algorithm
i
Jacobson and Turco (1994) have modified the Gear algorithm to incorporate additional
computational efficiencies that can achieve speedups on the order of 100 on vector computers.
About half of the improvement is attributed to enhanced vectorization, and half to improved
matrix operations. Because of the improved matrix operations, SMVGEAR also runs faster than
traditional Gear solvers on non-vector machines, but the greatest benefit will be obtained with
vector machines. The major enhancements incorporated in SMVGEAR are now described in
more detail.
8-20
-------
EPA/600/R-99/030
In the conventional application of the Gear-type algorithm, the method is applied to each grid cell
individually. With this implementation, the length of the innermost loops in most computations
corresponds to the number of species, which is typically on the order of 30 to 100. In
SMVGEAR, the modeling domain is divided into blocks of cells, and the Gear algorithm is
applied to each cell within a block simultaneously. With this structure, the length of the innermost
loops for most calculations is equal to the number of cells in a block. Substantial improvements in
vectorization can therefore be obtained if the block size is larger than the number of species.
Jacobson and Turco (1994) found that a block size on the order of 500 cells achieves about 90%
of the maximum vectorization speed on a Cray C-90 computer. The use of larger block sizes may
not substantially increase computational speed and may incur some additional penalties. For
example, memory requirements increase with increasing block size. Furthermore, the size of a
time step in SMVGEAR is the same for each cell within a block and is based on the time-step
estimate for the stiffest cell in the block. Limiting the block size can therefore reduce excess
calculations that need to be performed for the less stiff cells. Jacobson (1995) also achieved
computational savings by ordering the cells by stiffness before dividing them into blocks. Each
block then tends to contain cells of similar stiffness, thereby reducing excess computations for
some cells. Jacobson found excess computations were reduced by about a factor of two, and that
these reductions more than offset the additional work incurred with calculating and sorting the
cells by stiffness.
Much of the computational intensity associated with the Gear method arises from the matrix
operations that are needed to perform the Newton iterations in the corrector step. Jacobson and
Turco have introduced two techniques to improve the efficiency of these operations. First, all
known cases of multiplication by zero in matrix multiplication and decomposition are eliminated.
This is particularly beneficial for atmospheric chemistry problems since the Jacobian matrices are
almost always sparse (i.e., they contain a large number of zero entries). However, decomposition
techniques that are applied to these matrices often result in substantial fill-in, thereby reducing the
benefits of employing sparse-matrix techniques. To maintain maximum sparsity after the
decomposition operation, SMVGEAR orders the species in the Jacobian such that those with the
fewest partial derivative terms are located in the top rows, and those with the most are in the
bottom rows. At the very beginning of the program, the ordering is done and a symbolic
decomposition is performed to identify multiplies by zero. Since the computations associated
with the matrix operations are determined entirely by the structure of the chemical mechanism, it
is necessary to do this only once and the results can then be applied to every cell uniformly.
The SMVGEAR algorithm has been implemented in the CCTM with minor changes to the
original algorithm but extensive changes to the original computer code. The code changes arose
from linking the algorithm to the generalized chemistry processor used in the CMAQ system, and
developing a driver routine specific to the CCTM structure. The only significant change to the
algorithm involved modifying the code to eliminate the possibility of obtaining negative
concentrations. With the standard Gear algorithm, negative concentrations can occur when a
species is rapidly depleted, although the magnitudes of these concentrations are extremely small.
In the CCTM implementation, a lower bound on allowable concentrations is applied, and the rates
8-21
-------
EPA/600/R-99/030
of change and the Jacobian matrix are modified to reflect that no changes in concentration are
occurring when the lower bound is reached. Comparisons with the standard Gear algorithm show
virtually no differences in species concentrations above the lower bound, but a small penalty in
computational performance is incurred. Nevertheless, the approach insures a positive-definite
solution.
The computational performance of SMVGEAR is also affected by the error tolerances used for
the Newton iteration convergence tests and the local truncation error tests. Error control in
SMVGEAR is similar to that used in LSODE (Hindmarsh, 1980), Both a relative and an absolute
tolerance must be specified. In their discussion of ODE solvers, Byrne and Hindmarsh (1987)
relate the relative error tolerance to the number of accurate digits and the absolute error tolerance
to the noise level (i.e., the size of the largest concentration that can be neglected). If r is the
number of accurate digits required, then Byrne and Hindmarsh suggest setting the relative
tolerance to lQ"(r*13. The absolute error tolerances cannot be specified as generically because
particular model applications may require different accuracies for the mechanismn species. In the
CCTM implementation of SMVGEAR, the relative tolerance and absolute tolerances have been
preset to 10"3 and 10"9 ppm, respectively. However, these values can be changed by the user
relatively easily in the CCTM as described in EPA (1998).
8.4.3 QSSA Solver
The QSSA solver is a low order, explicit solver that exhibits good stability for stiff systems.
Although less accurate than the Gear solver, it is still a reasonably accurate, fast solver that is
especially suitable for large scale grid models. There are actually many versions of solvers that go
by the name "QSSA" (e.g., Mathur et al., 1998). The solver developed for the CCTM is a
predictor/corrector version based on the one developed by Lamb and used in the Regional
Qxidant Model (Lamb, 1983, and Young et al, 1993).
i "A '.i f - ;
The QSSA method originates from assuming integration time steps sufficiently small such that in
Equation 8-11, the production and loss rate terms Pt and Lt can be considered constant. If the
Jacobian is diagonally dominant, this assumption may be valid as At - 0, and the time step
solution at ^, = tn + At can be written formally as:
C, = C, +(C, - C, )e~L'" (8-22)
•ft R 0" s v
** *~ ' ' '~' " \
where C( = Pl/Liand Ct is the solution at tn.
\ ", •'. ' .. . • , ••;,'
The CCTM QSSA makes no a priori assumptions about reaction time scales. For example, there
are no assumed steady states. However, the algorithm separates the numerical computation into
either an Euler step, a fully explicit integration, or an asymptotic evaluation based on
photochemical lifetimes estimated from an initial, predictor calculation off, and Lt. The cut-offs
and equations for each predictor step are:
8-22 '
-------
EPA/600/R-99/030
Eulerstep: Z,,A/ < 0.01
Explicit: 0.01 < L^t < 10.0
Asymptotic: Z,,A/ t: 10.0
C, = C, + (Ps + L.C, )
C,. = C, + (C^ - C,. ) eCle,, (e= 0.01):
•-L In
L.
1 - A-
(8-27)
lfLa=0,
AC
Sttt = —
P.
(8-28)
The tolerance parameter A is controlled by the rate at which the key NO species concentrations
are changing; if they are changing too rapidly, the tolerance is tightened, otherwise it is relaxed:
8-23
-------
EPA/600/R-99/030
_ , 0.001, if - I [NO] fc 0.5% per minute
A — ^ dt (8-29)
0.005, otherwise
After determining a time scale for each species a, At is set to mm{dttt}. For computational
efficiency, At is further constrained to be no less than one second and of course is also constrained
to be no greater than the total integration time.
In the predictor step, the species concentrations are updated (C, - C*) with the optimal time step
u||ng the Euler-step, explicit or asymptotic calculations described above. Once C* is calculated,
QSSA computes new production and loss rate coefficients P* and L', respectively.
• i
In the corrector step, the final production rate is set as the average of the initial and predictor
values, P, = (P' + P,) / , and the new concentration Ct for time /„+, =ttt + At is computed using
the same cut-offs based on L, that were determined in the predictor step:
Eulerstep: C, = C^ + (Pt * L* Ct) Af
Explicit: C, = Pt/L' + (C, - PJL^e'1'^ (8-30)
Asymptotic: C, = Ptl L'
The algorithm has been optimized for vector computers by moving the grid cell loops into the
innermost position (Young et al., 1993) as is done in SMVGEAR described above. To minimize
storage requirements, grid cell blocking has been implemented wherein blocks of cells are handed
off to the solver in sequence. The CMAQ QSSA has also been optimized for the Cray T3D by
utilizing various coding techniques aimed specifically at that architecture. Some of these
optimizations are described in Chapter 19.
8.4.4 Summary
Gear type solvers have generally been considered the most accurate for gas chemistry,
representing "exact solutions" (provided the controlling numerical tolerances are sufficiently
tight). Until the advent of SMVGEAR, however, it has not been feasible to use these solvers in
Eulerian models. SMVGEAR is designed to run optimally on high end vector computers such as
the Cray C90, but its use on scalar machines may be impractical. Although less accurate, the
QSSA solver may be more suitable for those types of computers.
The issue of accuracy versus computational speed is a continuing concern (Mathur et al., 1998),
particularly since the availability of high-end machines like the Cray C90 is limited, the CMAQ
QSSA solver presents a reasonable, numerically efficient alternative and, although not considered
8-24
-------
EPA/600/R-99/030
as accurate as a Gear-type solver, may be sufficiently accurate for modeling, taking into
consideration the uncertainties of the other numerical modeling components. Accuracy can be
somewhat improved by using shorter integration steps in the solver, but then computational work
mounts, defeating the purpose. For the CMAQ QSSA, accuracy will be compromised when the
ODE system is very stiff and the system Jacobian strays from diagonal dominance. Nevertheless,
the trade-off between solution efficiency and accuracy may still warrant its use in these cases.
Finally, both the SMVGEAR and the QSSA solvers have been incorporated in the CMAQ system
with the predefined accuracy controls that were described in the previous two sections. Of
course, these error controls can be changed by the user if desired, but that will of course affect
both the prediction accuracy and the efficiency of obtaining a solution. Note also, that either
solver can be used with any of the CMAQ chemical mechanisms that were described in section
8.2.
8.5 References
Atkinson R., Lloyd A. C. and Winges L. (1982) An updated chemical mechanism for
hydrocarbon/NOx/SO2 photooxidations suitable for inclusion in atmospheric simulation models.
Atmos, Environ., 16, 1,341-1,355.
Bowman F. M., Pilinis C., and Seinfeld J. (1995) Ozone and aerosol productivity of reactive.
Atmos. Environ., 29, 579-589.
Byrne G. D. and Hindmarsh A. C. (1987) Stiff ODE solvers: a review of current and coming
attractions../, Comput. Phys. 70, 1-62.
Carter W. P. L. (1988) Appendix C Documentation of the SAPRC Atmospheric Photochemical
Mechanism Preparation and Emissions Processing Programs for Implementation in Airshed
Models. Final Report for California Air Resources Board Contract No. A5-122-32.
Carter W. P. L. (1990) A detailed mechanism for the gas-phase atmospheric reactions of organic
compounds. Atmos. Environ. 24A, 481-515.
Carter W. P. L. (1996) Condensed atmospheric photooxidation mechanisms for isoprene. Atmos.
Environ. 24, 4,275-4,290.
Carter W. P. L. and Atkinson R. (1996) Development and evaluation of a detailed mechanism for
the atmospheric reactions of isoprene and NOx. Int. J. Chem. Kinet. 28,497-530.
Carter W. P. L., Luo D. and Malkina I. L. (1997) Environmental Chamber Studies for
Development of an Updated Photochemical Mechanism for Reactivity Assessment. Final Report
for California Air Resources Board Contract No. 92-345, Coordinating Research Council, Inc.,
Project M-9 and National Renewable Energy Laboratory, Contract ZF-2-12252-07.
8-25
-------
EPA/60Q/R-99/03Q
•l!"' ' ': ' ' ' !'
Chang J. S., Brost R. A., Isaksen I. S. A., Madronich S., Middleton P., Stockwell W.R, and
Waleck CJ. (1987) A three-dimensional Eulerian acid deposition model: physical concepts and
formulation, J. geophys. Res., 92,14,681 - 14,700.
-. - - ^ ^ . . .. . ., . . ..,)...
Dabdub D. and Seinfeld J. H. (1995) Extrapolation techniques used in the solution of stiff ODEs
associated with chemical kinetics of air quality models. Atmos. Environ. 29,403-410.
Dodge M.C. (1989) A comparison of three photochemical oxidant mechanisms, J. geophys. Res.,
94, 5,121-5,136.
EPA (1989) Procedures for Applying City-specific EKMA. EPA-450/4-89-012.
EPA (1991) User's Guide for the Urban Airshed Model, Volume I: User's Manual for UAM
(CB4). EPA-450/4-90-007a.
EPA (1998) EPA Third-Generation Air Quality Modeling System, ModeIs-3 Volume 9b User
Manual. EPA-600/R-98/069b.
i
Gear C. W. (197la) Numerical Initial Value Problems in Ordinary Differential Equations.
Prentice-Hall, Englewood Cliffs, NJ.
Gear C. W. (1971b) The automatic integration of ordinary differential equations. Comm. ACM
14, 176-179. . '
Gery M. W., Morris R. E., Greenfield S. M,, Liu M. K., Whitten G. Z., and Fieber J. L. (1987)
Development of a Comprehensive Chemistry Acid Deposition Model (CCADM), Final Report for
Interagency Agreement DW 14931498, U. S. Environmental Protection Agency and U. S.
Department of Interior.
1 ' • '• ,'•• • . " i •
Gery M. W., Whitten G. Z,, Killus J. P. and Dodge M. C. (1989) A photochemical kinetics
mechanism for urban and regional scale computer modeling. J, geophys. Res. 94, 12,925-12,956.
Gong W. and Cho H. R. (1993) A numerical scheme for the integration of the gas-phase chemical
rate equations in three-dimensional atmospheric models. Atmos. Environ. 27A, 2,147-2,160.
Hindmarsh A. C. (1980) LSODE and LSODI, two new initial value ordinary differential equation
solvers. ACMNewsl. 15,10-11.
Hough A. (1988) An intercomparison of mechanisms for the production of photochemical
oxidants. J. geophys. Res., 93, 3,789-3,812.
Jacobson M. and Turco R. P. (1994) SMVGEAR: A sparse-matrix, vectorized Gear code for
atmospheric models. Atmos. Environ. 28, 273-284.
., "i . 8-26 '
-------
EPA/600/R-99/030
Jacobson M. (1995) Computation of global photochemistry with SMVGEARII. Atmos. Environ.
29,2,541-2,546.
Jeffries H. E. (1995) Photochemical Air Pollution. Chapter 5 in Composition, Chemistry, and
Climate of the Atmosphere. Ed. H. B. Singh, Von Nostand-Reinhold, New York.
Jeffries H. E., Gery M. and Murphy K. (1993) Advanced Chemical Reaction Mechanisms and
Solvers for Models-3. Progress Report for EPA Cooperative Agreement CR-820425.
Lamb, R.G. (1983) A Regional Scale (1000 km) Model of Photochemical Air Pollution. Part I -
Theoretical Formulation. EPA-600/3-83-035, U. S. Environmental Protection Agency, Research
Triangle Park, NC.
Leone J.A. and Seinfeld J.H. (1985) Comparative analysis of chemical reaction mechanisms for
photochemical smog. Atmos. Environ., 19, 437-464.
Lurmann F. W., Carter W. P. L. and Coyner L. A. (1987) A Surrogate Species Chemical
Reaction Mechanism for Urban Scale Air Quality Simulation Models Volume 1. EPA-600/3-87-
014a.
Mathur R., Young J. O., Schere K. L., and Gipson G. L. (1998) A Comparison of Numerical
Techniques for Solution of Atmospheric Kinetic Equations. Atmos. Environ., 32, 1,535-1,553.
NASA (1997). Chemical Kinetics and Photochemical Data for Use in Stratospheric Modeling
Evaluation No. 12. JPL Publication 97-4, Jet Propulsion Laboratory, Pasedena, CA.
Odman, M. T., Kumar N. and Russell A. G. (1992) A comparison of fast chemical kinetic solvers
for air quality modeling. Atmos. Environ. 26A, 1,783-1,789.
Pandis S. N., Harley R. A., Cass G. R. and Seinfeld J. H.(1992) secondary aerosol formation and
transport. Atmos. Environ. 26A, 2,269-2,282.
Pitts-Finlayson B.J. and Pitts J.N. (1986) Atmospheric Chemistry: Fundamentals and
Experimental Techniques, J. Wiley and Sons, New York.
SAI (1993) Systems Guide to the Urban Airshed Model (UAM-V). Systems Applications
International, San Rafael, CA.
Saylor, R. D. and Ford G. D. (1995) On the comparison of numerical methods for the integration
of kinetic equations in atmospheric chemistry and transport models. Atmos. Environ.29, 2,585-
2,593.
8-27
-------
EPA/600/R-99/030
Stockwell W. R_, Middleton P, and Chang J. S. and Tang X. (1990) The second generation
regional acid deposition model chemical mechanism for regional air quality modeling. J. geophys.
Res. 95 (dlO), 16,343-16,367.
Stockwell W.R. (1986) A homogeneous gas phase mechanism for use in a regional acid
deposition model, Atmos. Environ., 20, 1,615-1,632,
• • — 'I
Stockwell W. R., Kirchner F., Kuhn M., and S. Seefeld (1997) A new mechanism for regional
afinospheric chemistry modeling. J. geophys. Res. 102 (D22), 25,847-25,879.
Whitten G. Z., Hogo H. and Killus J. P. (1980) The carbon bond mechanism: a condensed kinetic
mechanism for photochemical smog. Envir. Sci. Technol. 14, 690-701,
Young J. O., Sills E., Jorge D. (1993) Optimization of the Regional Oxidant Model for the Cray
YrMOP. EPA/600/R-94-065, U. S, Environmental protection Agency, Research Triangle Park,
NC.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
Appendix 8A Chemical Mechanisms Included in the CMAQ System
Table 8 A-l, CB4 Mechanism Species List
Table 8A-2. RADM2 Mechanism Species List
Table8A-3. CB4 Mechanism
Table 8A-4. CB4_AE Mechanism
Table 8A-5. CB4_AQ Mechanism
T|ble 8A-6, CB4_AE_AQ Mechanism
T|ble8A-7. RADM2 and RADM2_AQ Mechanisms
">•* : ' '-'".• ^ ' ' "" ' J^ L
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
Table8A-9. RADM2_CISl and RADM2__CIS1_AQ Mechanisms
J 8-28
-------
BPA/600M-99/030
Table 8A-10. RADM2_CIS1__AE and RADM2_CIS1_AE_AQ Mechanisms
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
Notes to Tables 8A-3 through 8A-12:
a)
The mechanism listings are divided into two parts. The first lists the reactions and the
second lists the rate constant expressions. > ,
b) The parameters for the rate constants are given in cms units. Rate constants
calculated in cms units for T=298 °K and P= 1 atm are shown in the rightmost column
of these listings.
c) For photolytic reactions, photo table refers to the photolysis rates described in
Chapter 14. The rate constant for all photolytic contains a zero entry in these tables,
but is calculated in the CCTM as the product of the scale factor and the photolysis
rate that is calculated by the CMAQ photolysis rate processor.
d) The falloff rate expression referred to in these tables is Equation 8-6 in Section 8.3.2.
Table 8A-1
CB4 Mechanism Species List
Nitrogen Species
NO
NO2
HONO
NO3
N2O5
HNO3
PNA
Oxidants
03
H2O2
Sulflir Species
S02
SULF
Atomic Species
O
O1D
Nitric oxide
Nitrogen dioxide
Nitrous acid
Nitrogen trioxide
Nitrogen pentoxidc
Nitric acid
Peroxynitric acid
Ozone
Hydrogen peroxide
Sulfur dioxide
Sulfliric acid
Oxygen atom (triplet)
Oxygen atom (singlet)
Odd Hydrogen Species
OH
HO2
Carbon oxides
Hydroxyl radical
Hydroperoxy radical
CO
Hydrocarbons
PAR
ETH
OLE
TOL
XYL
ISOP
Carbon monoxide
Paraffin carbon bond (C-C)
Ethene (CHj=CHj)
Olefmic carbon bond (C=C)
Toluene (Q,H4-CH5)
Isoprenc
Carbonvls and phenols
FORM
ALD2
MOLY
GRES
Organic nitrogen
PAN
NTR
Organic Radicals
C2O3
ROR
CRO
Formaldehyde
Acetaldehyde and higher aldehydes
Methyl glyoxal (CH,C(O)C(O)H)
Cresol and higher molecular weight phenols
Peroxyacyl nitrate (CH3C(O)OONO2)
Organic nitrate
Peroxyacyi radical (CH,C(O)OO-)
Secondary organic oxy radical
Melhylphenoxy radical
8-29
-------
EPA/600/R-99/030
Qperttort
XO2 NO-to-NO2 Operation
XO2N NO-to-nitrate operation
Product? oforganic;
TO2 Toluenc-hydroxyl radical adduct
OPEN High molecular weight aromatic
oxidation ring fragment
ISPD Products of isoprene reactions
gpecies idded for aerosols
SOLAER Counter species for H2SO4 production
TOLAER Co™16* species for toluene reaction
XYLAER Counter species for xyiene reaction
CSLAER Counter species for cresol reaction
TERPAER Counter species for terpene reaction
TBRP "" Monoterpenes
Species added for aqueous chemistry
FACD Formic acid
AACD Acetic and higer acids
PACD Pcroxy acetic acid
UMHP Upperlimitofmethylhydroperoxide
8-30
-------
EPA/600/R-99/030
Table 8A-2
RADM2 Mechanism Species List
Nitrogen Species
NO
NO2
HONO
NO3
N2O5
HNO3
HNO4
Oxidants
O3
H2O2
Sulfar Species
S02
SULF
Atomic Species
O3P
O1D
Nitric oxide
Nitrogen dioxide
Nitrous acid
Nitrogen trioxide
Nitrogen pentoxide
Nitric acid
Peroxynitric acid
Ozone
Hydrogen peroxide
Sulfur dioxide
Sulfuric acid
Oxygen atom (triplet)
Oxygen atom (singlet)
Odd Hydrogen Species
HO
HO2
Carbon oxides
CO
Alkanes
ETH
HC3
HC5
HC8
Alkenes
OL2
OLT
OLI
ISO
Arotnatics
TOL
XYL
CSL
Hydroxyl radical
Hydroperoxy radical
Carbon monoxide
Ethane
Alkanes w/ 2.7x10-" > *„„ < 3.4xlO'12
Alkanes w/ 3,4x 1 Q~" > km < 6.8x 10'12
Alkanes vi/koa>6.Sx\Q"a
Ethene
Terminal olefins
Internal olefins
Isoprene
Toluene and less reactive aromatics
Xylene and more reactive aromatics
Cresol and other hydroxy substituted
aromatics
OP1
OP2
PAA
Organic acids
ORA1
ORA2
Methyl hydrogen peroxide
Higher organic peroxides
Peroxyacetic acid
Formic acid
Acetic and higher acids
Peroxy radicals from alkanes
MO2 Methyl peroxy radical
ETHP Peroxy radical formed from ETH
HC3P Peroxy radical formed from HC3
HCSP Peroxy radical formed from HC5
HC8P Peroxy radical formed from HC8
Peroxv radicals from alkcncs
OL2P Peroxy radical formed from OL2P
OLTP Peroxy radical formed from OLTP
OLIP Peroxy radical formed from OLIP
Peroxv radicals from aromatics
TOLP Peroxy radical formed from TOL
XYLP Peroxy radical formed from XYL
Peroxy radicals with carbonvl groups
ACO3 Acetylperoxy radical
KETP Peroxy radical formed from KET
TCO3 H(CO)CH=CHCOj
Peroxy. radicals involving nitrogen
XO2 NO-to-NOj Operator
XNO2 NO-to-nitrate operator
OLN NO3-alken adduet
Species added for aerosols
SULAER Counter species for H2SO, production
HC8AER Counter species for HC8 reaction
OLIAER Counter species for OLI reaction
TOLAER Counter species for toluene reaction
XYLAER Counter species for xylene reaction
CSLAER Counter species for cresol reaction
TERPAER Counter species for terpcnc reaction
TERP Monoterpenes
Organic nitrogen
PAN
TPAN
ON1T
Formaldehyde
Acetaldehyde and higher aldehydes
Ketones
Glyoxal
Methyl glyoxal
Unsaturated dicarbonyl
Peroxyacetyl nitrate and higher PANs
H(CO)CH=CHCOjNO2
Organic nitrate
Organic peroxides
8-31
-------
EPA/600/R-99/030
Table8A-3. CB4 Mechanism
Reaction List
{
{
{
I
{
{
{
{
{
1
{
{
{
{
1
T
C
{
1
I
{
I
{
{
{
{
{
{
{
{
{
{
{
{
{
f
{
{
{
{
{
{
t
{
{
{
1}
2}
3}
4}
5}
6}
7}
8}
9}
10}
11}
12}
13}
14}
15}
16}
17}
18}
19)
20}
21}
22}
23}
24}
25}
26}
27}
28}
29}
30}
31}
32}
33}
34}
35}
36}
37}
38}
39}
40}
41}
42}
43}
44}
45}
46}
47}
48}
49}
so}
51}
52}
53}
N02
0
03
0
0
0
03
03
03
01D
01D
01D
03
03
N03
NO3
NO3
NO3
N205
N205
NO
NO
OH
HONO
HONO
HONO
OH
OH
H02
HO2
PNA
PNA
HO2
HO2
H2O2
H202
CO
FORM
FORM
FORM
FORM
FORM
ALD2
ALD2
A1D2
ALD2
C203
C2O3
PAN
C203
C203
OH
PAR
4-
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
+
4-
+
4-
4-
4-
+
4"
4-
+
+
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
+
4-
+
+
+
4-
+
4-
4-
+
4-
4-
"-'•
+
+
4-
4-
4-
hv -->
102] -->
NO "" --»
BO2 -->
N02 " -->
NO -->
N02 -->
hv -->
hv -->
tN2l -->
C021 —>
[H2O] -->
OH -->
HO2 -->
hv -->
NO -->
NO2 -->
NO2 -->
[H20] -->
-->
NO 4- [02] — >
NO2 4- tH20] -->
NO -->
hv -->
OH -->
HONO -->
N02 -->
HNO3 -->
NO -->
NO2 -->
-->
OH -->
HO2 -->
HO2 + [H2O1 -->
hv -->
OH -->
OH -->
OH -->
hv -->
hv -->
O -->
NO3 -->
O -->
OH -->
N03 -->
hv -->
NO -->
NO2 -->
-->
C203 -->
H02 - - >
-->
OH -->
NO
O3
N02
NO
NO3
N02
N03
0
O1D
0
O
2.000*OH
H02
OH
0.890*NO2
2 . 000*NO2
NO
N2OS
2,000*HN03
N03
2.000*N02
2.000*HONO
HONO
OH
NO2
NO
HMOS
NO3
OH
PNA
HO2
NO2
H202
H2O2
2 . ,000*OH
HO2
H02
H02
2.000*HO2
CO
OH
HHO3
C203
C203
C2O3
XQ2
4- FORM
NO2
4- HO2
PAN
C2O3
2.000*X02
0.790*FORM
+ Q,790*OH
X02
0.870*X02
4- 0.110*ALD2
4- O
+ 0.890*0
+ NO2
+ N02
4- NO
4- N02
4- N02
4- NO2
4- CO
4- CO
+ HO2
4- HO2
4- OH
+ HN03
+ 2,000*HO2
4- X02
4- NO2
4- 2.000*PORM
4- 0.790*XO2
+ FORM
+ 0,130*X02N
+ Q.760*ROR
4- 0,110*NO
+ CO
4- CO
•f CO
4- FORM
4- 2,000*HO2
4- 0.790*HO2
+ HO2
+ 0 . 110*HO2
- 0.110*PAR
8-32
-------
EPA/ffOO/K-99/030
Table8A-3. CB4 Mechanism
{ 54}
{ 55}
{ 56}
{ 57}
{ 58}
{ 59}
{ 60}
{ 61)
{ 62}
{ 63}
{ 64}
{ 65}
{ 66J
{ 67}
{ 68}
{ 69}
{ 70}
{ 71}
{ 72}
{ 73}
{ 74}
{ 75}
{ 76}
{ 77}
{ 78}
{ 79}
{ 80}
{ 8!}
{ 82}
{ 83}
{ 84}
{ 85}
{ 86}
{ 87}
{ 88}
{ 89}
{ 90}
ROR
ROR
ROR
OLE
OLE
OLE
OLE
ETH
ITH
ETH
TOL
TO2
TO2
CRES
CRES
CRO
XYL
OPEN
OPEN
OPEN
MGLY
MGLY
I SOP
I SOP
ISOP
I SOP
XO2
XO2
X02N
SO2
SO2
XO2
X02N
XO2N
XO2H
ISPD
ISPD
4-
4-
4-
4-
+•
4-
4-
4-
+
4-
4-
+
4-
4-
4-
4-
4-
4-
4-
4-
4-
+
4-
+
+
+
+
4-
+
4-
4-
4-
4-
NO2
O
OH
O3
NO3
O
OH
03
OH
NO
OH
NO3
N02
OH
OH
hv
O3
OH
hv
O
OH
O3
NO3
NO
XO2
NO
OH
HO2
HO2
XO2N
XO2
OH
03
--> 1
- 2
-->
-->
--> 0
4- 0
4- 0
-->
4-
--> 0
4- 0
_
--> 0
4-
-->
4- 1
-->
4- 0
-->
--> 0
4- 0
--> 0
4- 0
-->
--> 0
4" 0
-->
-->
--> 0
4- 0
-->
+
-->
--> 0
4- 0
4- 0
-->
-->
--> 0
4- 0
--> 0
+ 0
--> 0
4- 0
4- 0
--> 0
+ 0
+ 2
-->
-->
-->
-->
-->
-~>
-->
-->
,_>
--> 1
4- 0
4- 0
--> 0
.100*ALD2
.100*PAR
HO2
NTR
.630*ALD2
.300*CO
.220*PAR
FORM
• HO2
.500»ALD2
.440*H02
PAR
.910*XO2
ALD2
FORM
.700*HO2
X02
.220*ALD2
FORM
.080*XO2
,560*TO2
.900*NO2
.100»NTR
CRES
,400*CRO
.300*OPEN
CRO
NTR
.700*HO2
,80Q*MGLY
X02
C203
C2O3
,030*ALD2
,030*XO2
.760*H02
X02
C2O3
.750*ISPD
.250*HO2
.912*ISPD
.912»HO2
.650*ISPD
.066*H02
,150*ALD2
.200*ISPD
.800*HO2
.400*PAR
NO2
NTR
SU1F
SULF
,565*PAR
.503*HO2
.273*ALD2
.114*C2O3
+
4-
4-
4-
4-
4-
-
4-
4-
4-
-
4-
4.
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
+
4-
4"
4-
4-
4.
4-
4-
4-
4"
4-
*
4-
4-
4-
4-
4-
4-
4-
4-
4-
0
0
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.960*X02
.040*X02N
.380*HO2
.200*FORM
.200*OH
ALD2
PAR
.740*FORM
.220*XO2
.090*X02N
PAR
.700*XO2
,300*OH
.560*FORM
.420*CO
.360*CRES
.900*HO2
HO2
. 600*X02
HNO3
.500*XO2
.100*PAR
.000*CO
FORM
HO2
. 620*C2O3
. 690*CO
.200*MGLY
C2O3
HO2
.500*FORM
.250*C2O3
.629*FORM
.088*X02N
.600*FORM
.266*OH
.350*PAR
.•SOO*NTR
.200*NO2
HO2
,167*FORM
,334*CO' ;
,498*C2O3
.150*FORM
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4-
4-
4-
0
0
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
.940*H02
,020*ROR
.280*XO2
,020*XO2N
X02
.330*CO
.100*OH
FORM
NO2
CO
HO2
.120*HO2
.440*HO2
.900*OPEN
.600*HO2
,200*CRES
.300*TO2
.000*HO2
CO
.700* FORM
.080*OH
CO
.250*XO2
.250*PAR
.991*XO2
,200*XO2
.200*C2O3
.066*CO
XO2
.800*ALD2
.713*XO2
.168*MGLY
. 850*MGLY
8-33
-------
EPA/600/R-99/030
Table8A-3. CB4 Mechanism
••••••' . • . 4- 0.154*H02 4- 0.268*OH 4- 0.
4- 0.020*ALD2 4- 0.360*PAR 4- 0.
{ 91} ISPD 4- NO3 --> 0.357*ALD2 4- 0.282*FORM 4- 1.
4- Q.925*HO2 4- 0.643*CO 4- 0.
4- 0.075*C203 4- 0.075*X02 4-0.
{ 92} ISPD + hv --> 0.333*CO 4- 0.067*ALD2 4- 0.
"! - 'I ' 4- 0.832*PAR 4- 1.033*HO2 • 4- 0 .
, 4- 0.967*C203
{ 93} ISOP 4- NO2 --> 0.200*ISPD 4- 0.800*NTR 4-
. 4- 0.800*HO2 4- 0.200*NO 4- 0.
4- 2.400*PAR
Rate Expression
k( 1) uses photo table NO2_CBIV88 , scaled fay l.OOOOOE+00
k( 2) is a falloff expression using:
kO * 6.000QB-34 * (T/300)**(-2.30)
kinf • 2.8000E-12 * ST/3005**t 0.00)
' F - 0.60," n » 1.00
Jc( 3) « 1.8000E-12 * exp( -1370. 0/T)
k< 4! - 9.3000E-12
k< 5) is a falioff expression using:
kO ..« 9.0000E-32 * (T/300)**(-2.00)
kinf » 2.2000S-11 * (T/300)**S 0.00! . • -
- F - 0,60, n « 1.00
k( 6) is a falloff expression using:
kO . 9.0000E-32 * (T/300) **(-!. 50!
kinf • 3-OOOOE-ii * ST/300!**( o.oo)
P » 0.60, n - 1.00
k{ 7) . 1.2000E-13 * exp( -2450. 0/T)
k( 8) uses photo table NO2_CBIV88 , scaled by 5.30000B-02
kt 9) uses photo table O3O1D_CBIV88 , scaled by i.OOOOOE+00
k( 10) . 1.8000E-11 * expt 107. 0/T)
k( 11) - 3.2000E-11 * expt 67. 0/T)
kC 12) - 2.2000E-10
k( 13) - 1.6000E-12 * exp( -940. 0/T)
k( 14) - 1.4000E-14 * expt -580. 0/T)
kS 15) uaes photo table HO2_CBIV88 , scaled by 3.39QOOE4-01
k( 16) - 1.3000E-11 * expt 250. 0/T)
k( 17) . 2.5000E-14 * expt -1230. 0/T)
k( 18) is a falloff expression using:
kO « 2.2000E-30 * (T/300!**(-4.30!
kinf - l.SOOOE-12 * (T/300)** (-0.50)
F m 0.60, n m 1.00
k( 19) - 1.3000E-21
k( 20) - k( 18) / Keq, where Keg = 2.700B-27 * exp( 11000. 0/T)
k( 21) ~ 3.3000E-39 * exp ( 530. 0/T)
k( 22) m 4.4000E-4Q ' '." '
k! 23! is a falloff expression using:
kO » 6.7000E-31 * (T/300! ** (-3 .30)
kinf m 3.0000E-11 * (T/300) **{-!. 00)
F - 0.60, n = 1.00
k( 24) uses photo table N02_CBIV88 , scaled by 1.97500E-01
kt 25) . 6.6000E-12
k< 26) • l.OOOOE-20
k( 27) is a falloff expression using:
kO - 2.6000E-30 * (T/300)** (-3. 20)
kinf • 2.4000E-11 * (T/300)**(-1.30)
064*X02
225*CO
282*PWR
850*NTR
075*HN03
900*FORM
700*XO2
f
X02
800*ALD2
" ' ' • i
Rate Constant
{O.OOOOOE+00}
{1.37387E-14J
{1.81419E-14}
(9.30QOOE-12)
(1.57527E-12)
{1.66375B-12}
{3.22581E-17}
{O.OOOOOE4-00}
IO.OOOOOE+OO)
{2.57757E-11}
{4.00676E-11}
{2.200001-10}
(6. 826501-14}
{1.99920E-15}
{O.OOOOOB+00}
{3.0080SE-11}
{4.03072B-16}
{1.26440E-12}
{1.30000E-21}
{4.360291-02}
{1.95397E-38}
{4.39999E-40}
{6.697011-12}
f
{O.OOOOOE+OO}
{6.60000E-12}
{l.OOOOOE-20}
{1.1488SE-11J
0.60,
1.00
8-34
-------
EPA/600/R-99/030
Table8A-3. CB4 Mechanism
k(
k!
k(
k!
kC
k(
k(
k(
k(
k(
k(
k!
k(
k(
k{
k{
k(
k(
k(
. k(
k(
k(
k(
k(
k(
kC
k(
k{
k(
k(
k(
Jc<
k(
k(
k!
k!
k{
k(
kC
k(
k(
k!
k(
kC
k{
k(
k(
k(
k(
k(
k(
kC
k(
k(
285
V —
ko =
k2 =
k3 =
29)
30)
kO
kinf
F m
31)
32)
33)
34)
35)
36)
37)
38)
395
40)
41)
42)
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)
is a special rate expression of the form:
kO + {k3£M] / (1 + k3[M]/k2)}, where
7.2000E-15 * exp( 785. 0/T) - •
4.10001-16 * exp( 1440. 0/T!
1.9000E-33 * exp ( 725. 0/T)
= 3.7000E-12 * exp( 240. 0/T)
is a falloff expression using:
= 2.3000E-31 * (T/300)**(-4.60)
= 4.2000S-12 * CT/300)**C 0.20)
0.60, n = 1.00
= k( 30) / Keq, where Keg = 2.100E-27 * expC 10900. 0/T)
= 1.30001-12 * exp( 380. 0/T)
= 5.9000E-14 * exp( 1150. 0/T)
= 2.2000E-38 * exp( 5800. 0/T)
uses photo table HCHOmol_CBIV88 . , scaled by 2.55000E-01
•> 3.1000E-12 * exp( -187. 0/T) . • . • . -
= 1.SOOOE-13 * Sl.O + 0.6*Pressure)
= 1.00001-11
uses photo table HCHOrad_CBIV88 , scaled by l.OOOOOE+00
uses photo table HCHOmol_CBIV88 , scaled by l.OOOOOE+00 .
= 3.0000E-11 * exp{ -1550. 0/T)
= 6.3000E-16
= 1.2000E-11 * exp( -986. 0/T)
= 7.0000E-12 * exp( 250. 0/T)
= 2.5000E-15
uses photo table ALD_CBIV88 , scaled by l.OOOOOE+00
= 3.4900E-11 * exp( -180. 0/T)
= 2.6300E-12 * exp! 380. 0/T)
= 2.00001+16 * exp (-13500. 0/T)
= 2.5000E-12
= 6.50001-12
= l.lOOOE-t-02 * exp( -1710. 0/T)
= 8.10001-13 . . •-
•= 1. 00001+15 * exp( -8000. 0/T)
= 1.6000E+03
= 1.50001-11
= 1.2000E-11 * expf -324. 0/T)
= 5.2000E-12 * exp( 504. 0/T)
= 1.40001-14 * expf -2105. 0/T)
= 7.7000E-15 • ' • • •
m 1.00001-11 *. exp( -792. 0/T)
= 2.0000E-12 * exp{ 411. 0/T)
= 1.3000E-14 * exp{ -2633. 0/T)
= 2.10001-12 * exp! 322. 0/T)
= 8.10001-12
= 4.2000E+00
- 4.10001-11 ,
= 2.2000E-11 - . •
* 1.40001-11
= 1.70001-11 * exp( 116. 0/T)
= 3.0000E-11-
uses photo table HCHOrad_CBIV88 , scaled by 9.Q4000E+00.
= 5.40001-17 * exp! -500. 0/T) ;
= 1.7000E-11
uses photo table HCHOrad_CBIV88 , scaled by 9.6400014-00
= 3.6000E-11
= 2.54001-11 * exp! 407. 6/T)
= 7.8600B-15 * exp! -1912. 0/T)
= 3.0300B-12 * exp! -448. 0/T)
= 8.1000E-12 . .
= 1.70001-14 * exp( 1300. 0/T)
{1.47236E-13}
{8.27883E-12}
{1.48014E-12}
(9.17943E-02)
{4 .65309E-12}
{2.79783E-12}
{6.23927E-30}
{O.OOOOOE+OO}
{1.65514E-12}
. {2.40000E-13} .
{l.OOOOOE-ll}
{O.OOOOOE+OO}
{O.OOOOOE+OO}
{1.65275E-13}
{6.30000E-16}
{4.38753E-13}
{l. 619721-11}
{2.50000E-15}
{O.OOOOOE+OO}
{1.90766E-11}
{9.41356E-12} :
{4.23268E-04J
{2.50000E-12}
{6.50000E-12}
{3.54242E-015
{8.100001-13}
•{2.19325E+03}
{l.60000E-t-03}
{l. 500001-11} •
{4.04572E-12}
{2.821731-llj
{1.19778E-17}
{7.70000E-15} . ,
{7.01080E-13}
{7.94340E-12}
{1.89105E-18} •
{6.18715E-12}
{8.10000E-12}
{4.200001+00}
{4.100001-11}.
{2.20000E-11J
{1.40000E-11}
{2.50901E-11}
{3.00000E-11} •
{O.OOOOOE+OO}
{1.00858E-17}
{1.700001-11}
{O.OOOOOE+OO}. •
{3.60000E-11}
{9.97368E-11} . .
{1.28512E-17}
{6.738191-13}
{8.100001-12}
{1.33359E-12}
8-35
-------
EPAA500/R-99/030
Table 8A-3. CB4 Mechanism
k{
k{
k(
k<
k<
k{
k(
k<
k{
k(
k(
k<
82}
83)
84!
85)
86)
87)
as)
89)
90)
91)
92)
93)
M
-
«
m
m
m
m
m
m
m
8
4
1
7
7
1
3
3
7
1
uses
»
1
.1000E-12
.3900E-13 *
.3600B-06
,67005-14 *
.6700E-14 *
.7300E-14 *
.45001-14 *
.""ieoOE-ll
.1100E-18
.OOOOS-15
photo table
.49QOE-19
{8.10000E-12}
exp(
exp{
exp(
exp{
exp(
ISO
1300,
1300,
1300,
1300.
.Q/T)
.0/T)
.0/T)
,0/T)
.0/T)
ACROI.EIN
{7,
{1.
.5100SE-13}
,350001-06}
{6. 016841-12}
{6.
U.
{2,
-•{3,
{7.
(1.
, scaled by 3.60000B-03 {0.
{1.
•01684E-12}
.35712E-12}
.706401-12}
.3SOOOE-11}
.11000E-18}
.000001-15}
.OOOOOft-t-00}
,49000E;19}
8-36
-------
EPA/600/R-99/030
Table 8A-4. CB4 AE Mechanism
Reaction List
{ 1}
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
{ 8}
{ 9}
{ 10}
{ 11}
{ «}
{ 13}
{ "}
{ 15}
{ 16}
{ «}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40}
{ 41}
{ 42}
{ 43}
{ 44}
{ «}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
{ 54}
N02
O
O3
O
O
O
O3
O3
O3
O1D
O1D
DID
O3
O3
NO3
N03
N03
NO3
N2O5
N205
NO
NO
OH
HONO
HONO
HONO
OH
OH
HO2
HO2
PNA
PNA
H02
HO2
H2O2
H2O2
CO
FORM
FORM
FORM
FORM
FORM
ALD2
ALD2
ALD2
ALD2
C2O3
C2O3
PAN
C2O3
C203
OH
PAR
ROR
+ hv
+ [02]
+ NO
+ NO2
4- NO2
+ NO
+ NO2
+ hv
4- hv
+ [N2]
+• [02]
+ [H2O]
4- OH
4- HO2
•f- hv
4- NO
4- NO2
4- NO2
+• [H20]
+ NO + [O21
+ NO2 4- [H2O]
4- NO
+ hv
+ OH
-1- HONO
4- NO2
4- HNO3
4- NO
+ N02
+• OH
+ H02
+ HO2 + [H2O]
+ hv
4- OH
4- OH
+ OH
+ hv
4- hv
+ O
4- NO3
+ O
+ OH
+ NO3
+ hv
+ NO
+ NO2
+ C2O3
+ HO2
+ OH
--> HO
--> 03
--> N02
--> . NO
— > NO3
--> NO2
--> N03
--> 0
--> O1D
--> O
--> 0
--> 2.000*OH
--> HO2
--> OH
--> 0.890*NO2
--> 2.000*N02
--> NO
--> H2O5
--> 2.000*HNO3
- - > N03
--> 2.000*N02
--> 2.000*HONO
--> HONO
--> OH
--> NO2
--> NO
-<•> HN03
--> NO3
--> OH.
--> PNA
- - > HO2
- - > NO2
--> H2O2
- - > H2O2
— > 2.000*OH
- - > HO2
--> H02
- - > HO2
--> 2.000*HO2
--> CO
--> OH
- - > HNO3
--» C203
--> C2O3
— > C2O3
- - > XO2
+ FORM
--> N02
+ HO2
--> PAN
--> C2O3
--> 2.000*XO2
— > 0.790*FORM
+ 0.790*OH
--> XO2
--> 0.870*XO2
+ 0.110*M,D2
--> 1.100*MJD2
- 2.100*PAR
+ O
+ 0.890*O
+ NO2
•f NO2
•f NO
+ NO2
•f NO2
+ NO2
+ CO
+ CO
4- HO2
+ HO2
+ OH
+ HNO3
+ 2.000*H02
+ XO2
+ N02
+ 2.000*FORM
+ 0,790*XO2
+ FORM
+ 0.130*XO2N
+ 0.760*ROR
+ 0.960*XO2
+ 0,040*XO2N
+ 0.110*NO
-t- CO
+ CO
+ CO
+ FORM
+ 2.000*HO2
+ 0.790*HO2
+ HO2
+ 0.110*HO2
- 0,110*PAR
+ 0.940*HO2
4- 0.020*ROR
8-37
-------
EPA/600/R-99/030
Table 8A-4. CB4 AE Mechanism
{ 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}
{ 85}
{ 86}
j{ 87}
i{ 88}
' { 8S}
{ 90}
ROR
ROR
OI.S
OLE
OLE
ots
ETH
ETH
ETH
TOL
T02
TO2
CRES
CRES
CRO
XYL
OPEN
OPEN
OPEN
MGI.Y
MOW
ISOP
ISOP
ISOP
ISOP
X02
XO2
X02N
SO2
S02
X02
XO2N
X02N
X02N
ISPD
ISPD
4-
4-
4-
4-
4-
4-
4-
4.
4-
+
+
4.
4.
4-
4-
4-
4-
4-
4-
4-
1 .
+
•f
4-
4-
4-
4-
4-
+
4.
+.
+
4-
4-
KO2
O
OH
03
NO3
O
OH
O3
OH
NO
OH
NO3
NO2
OH
"OH
hv
O3
OH
hv
O
OH
03
NO3
NO
XO2
NO
OH
HO2
HO2
XO2N
X02
OH
03
-->
-->
--> 0
4- 0
+ 0
-->
4.
--> 0
4- 0
-
--> 0
+
. -->
4- 1
.->
4- 0
-->
--> 0
4- 0
--> 0
4- 0
-->
--> 0
4- 0
-->
-->
--> 0
4- 0
+
-->
4-
-->
--> 0
4- 0
.+ 0
-->
-->
--> 0
4. 0
--> 0
4- 0
--> 0
4- 0
4* 0
--> 0
+• 0
4- 2
-->
__>
-->
-->
— >
-->
-->
— >
-->
** •• :* X
4- 0
4- 0
--> 0
4- 0
HO2
NTR
.630*AIJ32
.300*CO
,220*PAR
FORM
H02
.SOO*ALD2
.440«HO2
PAR
.910*XO2
ALD2
FORM
.700*HO2
XO2
.220*ALD2
FORM
.080*XO2
.S60*TO2
.900*N02
,XOO*NTR
CRES
.400*CRO
.3QO*OPEN
CRO
NTR
.700*HO2
.800*MG1Y
XYIAER
X02
C203
'C2O3
.030*ALD2
.030*X02
.76Q*H02
X02
C2O3
,750*ISPO
.250*H02
.912*ISPD
.912*H02
.650*ISPD
,066*HO2
150*R7,ng
,200*ISPD
.800*HO2
-400*PAR
NO2
NTR
SUI.F
SULF
.565*PAR
.503*HO2
.273*ALD2
,H4*C2O3
.154*H02
4-
+•
+
4-
-
4-
+
+
-
+
4-
+.
4-
4-
4-
4-
-i-
4-
4.
-4.
4-
4-
4-
+
4-
4"
4-
4-
4.
4-
•*•
+
4-
4.
4-
4-
4.
+ '
4-
4-
4-
4-
+
4-
+
+
0
0
0
0
0
0
0
0
1
0
0
0
0
0
1
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.380*HO2
.200*FORM
,200*OH
ALD2
PAR
.740*FORM
,220*XO2
.OSO*XO2N
PAR
.700*X02
.300*OH
.S60*FORM
,420*CO
.360*CRES
TOLAER
.900*HO2 "
HO2
.600*X02
CSLAER
HNO3
.SQO*XO2
.100*PAR
.000*CO"
FORM
HO2
.620*C2O3
.6?0*CO
.200*MGI.Y'
C2O3
HO2
,SOO*FORM
,25Q*C2O3
.629*FORM
.088*XO2N
.600*FORM
.266*OH
,350*PAR ' '
.800*NTR
.200*NO2
HO2
SDLXER
,167*FORM
.334*CO
.498*C2O3
.1SO*FORM
.268*OH
+
+
4-
4-
4>
4-
+
+•
4-
4.
4-
4-
4-
4-
4-
4-
4-
4-
+•
4-
+
4-
+
4-
4-
4-
4,
4-
4-
4-
4-
4>
4-
+
0
0
0
0
0
0
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
6
0
.280*XO2
.020*XO2N
' XO2
.330*CO
. 100*OH
FORM
N02
CO
H02
,120*HO2
.440*HO2
.900*OPEN
.600*H02
CSUiER
.200*CRES
.300*TO2
.000*HO2
CO
,700*FORM
.080*OH
CO
.250*X02
,2SO*PAR
.991*XO2
,200*XO2
,20p*C2O3
, 066*CO
XO2
.800*ALD2
SULAER
. 713*XO2
.168*MGLY
.8SO»MGIiY
.064*XO2
8-38
-------
EPA/600/R-99/030
Table 8A-4. CB4 AE Mechanism
+
{ 91} ISPD + NO3 -->
+
+
{ 92} ISPD + hv -->
* +
+
{ 93} ISO? + NO2 -->
*
{ 94} TERP + OH -->
{ 95} TERP + N03 -->
{ 96} TERP +03 -->
Rate Expression
k( 1) uses photo table N02_CBIV88
k( 2) is a falloff expression using;
kO = 6.0000E-34 * CT/300}**(-2.30)
kinf = 2.8000E-12 * (T/300) »*( 0.00)
F = 0.60, n = 1.00
k( 3) = 1.8000E-12 * exp( -1370. 0/T)
k( 4) = 9.3000E-12
fc( 5) is a falloff expression using:
kO = S.OOOOE-32 « (T/300)**(-2.00)
kinf = 2.2000E-11 * (T/300)**( O.OOS
F = 0.60, n = 1.00
k( 6) is a falloff expression using:
kO = 9.0000E-32 * (T/300) **(-!. SO)
kinf = 3.0000E-11 * (T/300) **( 0.00)
F = 0.60, n = 1.00
k( 7) = 1.2000E-13 » exp( -2450. 0/T)
k{ 8) uses photo table NO2_CBIV88
k( 9) uses photo table 03O1D_CBIV88
k( 10) m 1.8000E-11 * exp( 107. 0/T)
k( 11) = 3.2000E-11 * exp( „ 67. 0/T)
k( 12) = 2.200QE-10
k( 13) = 1.6000E-12 * exp( -940. 0/T)
k( 14! = 1.4000E-14 *'exp( -580. 0/T) '
k( 15) uses photo table HO2_CBIV88
k{ 16! = 1.3000E-11 * exp(~ 250. 0/T)
k( 17) = 2.SOOOE-14 * exp( -1230. 0/T)
k( 18) is a falloff expression using:
kO = 2.2000E-30 * (T/300)**(~4.30)
kinf - 1.5000E-12 * (T/300) ** (-0 .50)
F = 0.60, n •= 1.00
k( 19) = 1*3000E-21
Q.020*AIJ}2
0.3S7*ALD2
Q.925*HO2
0.075*C203
0.333*CO
0.832*PAR
0.967*C2O3
0.2QO»ISPD
0.800*HQ2
TERPAER
TERPAER
TERPAER
, scaled by 1.
, scaled by 5 .
, scaled by 1.
, scaled by" 3.
k( 20) = k! 18) / Keg, where Keg = 2.700E-21 * expS
k( 21) = 3.3000E-39 * exp ( 530. 0/T)
k( 22) = 4.4000E-40
k( 23) is a falloff expression using:
kO = 6.7000E-31 * (T/300)** (-3. 30)
kinf = 3.0000E-11 * (T/300) *»(-!. 00!
F = 0.60, , n = 1.00
k( 24) uses photo table NO2_CBIV88
k( 25) = 6.6000E-12
k( 26) = l.OOOOE-20
k( 27) is a falloff expression using:
kO > 2.6000E-30 * (T/300) ** (-3 .20)
, scaled by 1.
+ 0.360*PAR ' + 0.22S*CO
+ 0.282«FORM + 1.282*PAR
+ 0.643*CO + 0.850*NTR
+ 0.07S*X02 + Q.075*HN03
+ 0.067*ALD2 + 0.900*FORM
+ 1.033*HO2 + 0.700*XO2
+ 0.800*NTR + X02
+ 0.200*NO + 0.800*ALD2
+ OH
+ NO3
+ O3
Rate Constant
OOOOOE+00 {O.OOOOOE+00}
{1.37387E-14}
{1.81419E-14}
{9.30000E-12}
{1.57527E-12}
{1.66375E-12}
{3.225S1E-17}
30000E-02 {0.000001+00}
000001+00 {0. 000001+00}
{2.57757E-11}
{4.00676E-11}
{2.20000E-10}
{6.826SOE-14}
J1.99920E-15)
390001+01 JO. OOOOOE+00}
{3. 008051-11}
{4.03072E-1S}
{1.26440E-12}
{1.30000E-21}
11000. 0/T) {4.36029E-08}
{1.95397E-38}
{4.39999E-40}
{6.69701E-12}
97500E-01 {0. OOOOOE+00}
{6.60000E-12}
{l.OOOOOB-20}
J1.148B5E-11}
8-39
-------
EPA/600/R-99/030
Table.8A-4. CB4 AE Mechanism
k{
k(
k(
k(
k(
k(
k(
let
k(
k!
k(
kC
k(
k(
k(
k{
k(
k(
k{
k(
k(
k!
k(
k(
k(
k!
k(
k{
kC
k(
k(
k(
k!
k{
k(
k(
k{
M
k(
k(
k(
k{
k(
k{
k(
k(
kS
k{
k(
k(
k(
k(
kinf = 2.4000E-11 * (T/300) **{-l,.30)
F = O.SO, n m 1.00
28) is a special rate expression of the form:
k = ko + {k3 [M] / (1 + k3[M]/k2)}, where
kO = 7.2000E-15 * exp ( 785. 0/T)
k2 = 4.1000E-16 * expC' 1440, 0/T)
k3 =
29)
30)
ko
kinf
F =
31)
32)
33)
34)
35)
36)
37)
38)
39)
40)
41)
42)
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)
• 1.
= 3
is a
=
=
9000S-33 * exp( 725. 0/T)
.7000E-12 * exp! 240. 0/T)
falloff expression using:
2.30001-31 * (T/300)** (-4. SO)
4.2000E-12 * ST/300)**S 0.20)
{1
fa
{i
.47236E-13}
l- • ' . 'i
« < > a
i , i '' •
T,
.27883E-12}
.48014E-12}
0.60, n = ' 1.00 • " .
= k(
= 1
= 5
= 2
uses
= 3
= l
= 1
uses
uses
= 3
= g
= 1
m 7
= 2
uses
= 3
= 2
» 2
= 2
= 6
= 1
= 8
= 1
• 1
•» 1
• 1
«= 5
= 1
= 7
«• 1
= 2
= 1
= 2
= 8
= 4
= 4
= 2
= 1
= 1
- 3
uses
- 5
0 1
uses
= 3
m 2
= 7
<= 3
30) / Keq, where Keq = 2.100E-27 * exp ( 10900. 0/T)
.3000E-12 * exp( 380. 0/T)
.9000E-14 * exp( 1150. 0/T)
.2000E-38 * expC 5800, 0/T)
photo table HCHOmol_CBIV88 , scaled by 2.55000E-01
.1000E-12 * exp( -187. 0/T)
.5000E-13 * (1.0 + 0.6*Pressure)
.-OOOOE-11
photo table HCHOrad_CBIV88 , scaled by l.OOOOOB+00
photo table HCHOtnol_CBIV88 , scaled by 1. 000001+00
.OOOOE-11 * exp( -1550. 0/T)
.3000E-16
.2000E-11 * exp! -986, 0/T)
.OOOOE-12 * expt 250, 0/T)
.5000E-15
photo table ALD_CBIV88 , scaled by l.OOOOOE+00
.4900E-11 * expC -180. 0/T)
.6300E-12 * expS 380.0/TS
.OOOOE+16 * exp (-13500. 0/T)
.5000E-12
.SOOOE-12
.10001+02 * expC -1710. 0/T)
.1000E-13
.OOOOE+15 * exp( -8000. 0/T)
.6000E+03
.5000E-11
.2000B-11 * exp! -324, 0/T)
.2000E-12 * exp( 504. 0/T)
.4000E-14 * exp( -2105. 0/T)
.7000E-1S
, OOOOE-11 * exp! -792. 0/T)
.OOOOE-12 * exp( 411. 0/T)
.3000E-14 * exp( -2633. 0/T)
. 1000B-12 * exp( 322. 0/T)
.1000E-12
.2000E+00
.1000E-11
.2000E-11
.4000E-11
.7000E-11 * exp! 116, 0/T)
.OOOOE-3L1
photo table HCHOrad_CBIV88 , scaled by 9.04000E+00
.4000E-17 * exp{ -500. 0/T)
.7000E-J.1
photo table HCHOrad_CBIV88 , scaled by 9.64000E+00
.6000E-11
.5400E-11 * exp( 407. 6/T)
.8600E-15 * expC -1912. 0/T)
.0300E-12 * expC -448. 0/T)
{9
{4
{2
{6
{0
{1
{2
{1
{0
{0
{1
{6
{«
{1
{2
{0
{1
{9
{4
{2
(6
{3
{8
{2
(1
{1
(4
{2
{1
{•?
if
{7
{1
{6
{8
{4
{4
{2
{1
{2
{3
{0
{1
.17943E-02}
.65309E-12J
.79783B-12} ''
.23927B-30*}
.OOOOOB+OO}
.65514E-12}
.40000E-13}
.OOOOOE-ll}
.OOOOOE+OO}
.OOOOOE+00}
.65275E-13}
.30000B-16} r;
.38753E-13} '
.61972E-11J
.50000E-1S}
.OOOOOB+00}
.90766E-11}
.41356E-12}
-232S8E-04}
.50000E-12}
.SOOOOE-12) •••
.54242S-01)
.10000E-13}
.19325E+03)
.GOOOOE+03}
.sooooE-iij :;•
.04572E-12)
.82173E-11J
.19778E-17) --
.70000E-1B) '.-
.01080E-13)
.94340E-12}
-89105E-18}
.18715E-12)
.10000E-12}
.20000E+00} . :
.10000E-11} ;.
.20000E-11) |
.40000E-11} ,
.50901B-11) .'
.OOOOOE-ll}
.OOOOOE+00}
.Q0858E-17}
{1.70000E-11} •-.
(O.OOOOOE+'OO)
{3
{9
{1
{s
. SOOOOE-ll}
.97368E-11) -.
.285121-17} •
.738198-13}
8-40
-------
EPA/600/R-99/030
Table 8A-4. CB4 AE Mechanism
k(
k(
k{
kS
k(
k{
kC
k(
k(
k{
k(
k(
k(
k{
k<
k(
k{
80!
81)
82}
835
84)
85)
86!
87!
88)
89)
90!
91)
92)
93)
94)
95)
96)
=
=
=
=
=
=
=
=
=
=
=
=
8
1
8
4
1
7
7
1
3
3
7
1
uses
=
=
=
=
1
1
3
7
.1000E-12
, 7000E-14 *
.1000E-12
.39001-13 *
.3600E-06
.6700E-14 *
.67001-14 *
.73001-14 *
.4500E-14 *
.3600E-11
.1100E-18
•OOOOE-15
photo table
.49001-19
.07001-11 *
.2300S-11 *
.2900E-15 *
exp(
exp!
exp{
exp(
exp (
exp(
1300
160
1300
1300
1300
1300
• 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
ACROLEIH
exp(
exp ('
expC
549
-975
-1136
.0/T)
.0/T)
.0/T)
{8.100001-12}
{1.33359E-12}
{8.10000E-12}
{7.S1005B-13}
{1.36000E-06}
{S. 016841-12}
{6.01684E-12}
{1.35712E-12}
{2.70640E-12}
{3. 360001-11}
{7.11000E-18}
{l.OOOOOB-15}
, scaled by 3.60000E-03 {O.OOOOOE+00}
{1.49000E-19}
{6.752691-11}
{1.22539E-12}
{1.611251-16}
8-41
-------
EPA/600/R-99/030
Table 8A-5. CB4_AQ Mechanism
Reaci
{ 1)
{ 2}
{ 3}
{ 4}
{ 5}
{ 5}
{ ^}
{ 8}
{ 9)
{ 1°}
{ 11}
{ 12}
{ "}
{ 14}
{ is}
{ «}
{ 17}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40}
{ 41}
{ 42}
{ 43}
{ 44}
{ 45}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
tion. L
NO2
O
03
O
0
O
O3
O3
O3
DID
01D
O1D
03
O3
NO3
NO3
NO3
NO3
N2O5
N2O5
NO
NO
OH
HONO
HONO
HONO
OH
OH
HO2
H02
PNA
PNA
HO2
HO2
H202
H2O2
CO
FORM
FORM
FORM
FORM
FORM
ALD2
ALD2
ALD2
ALD2
C203
C2O3
PAN
C203
C203
OH
PAR
ist
4-
+
4-
4-
+
4-
+
4-
+
4-
4-
4-
4-
•f-
4-
4-
+
•f
4-
4-
+
4-
+
+
+
4-
+
4-
4-
+
4-
+
4-
4*
4-
4-
4-
4-
4-
4-
4-
+
4-
+•
4-
4-
4"
4-
4-
hv -->
[02] -->
NO , -->
NO2 . , — >
NO2 , -->
NO - - >
NO2 -->
hv , -->
hv , -->
[N2] -->
102] ' -->
[H20] -->
OH - - >
HO2 -->
hv -->
NO , , -->
N02( -->
NO2 -->
[H20] • -->
-->
NO 4- [O2] — >
NO2 4- [H2O] -->
NO '. -->
hv -->
OH -->
HONO -->
NO2 -->
HHO3 -->
NO — >
NO2 -->
-->
OH -->
HO2, -->
H02, 4- £H20] -->
hv. -->
OH . '.' -->
OH -->
OH . -->
hv : • — >
hv , -->
O , -->
NO3 -->
O -->
OH , -->
N03 -->
hv , . -->
NO, -->
NO2 -->
-->
C2O3 -->
H02 -->
-->
OH, . . -->
NO
O3
NO2
NO
NO3
NO2
NO3
O
O1D
O
O
2.000*OH
HO2
OH
0 . 890«N02
2.000*NO2
NO
N2O5
2.000-HN03
NO3
2.000*NO2
2.000*HONO
HONO
OH
NO2
NO
HNO3
NO3
OH
PNA
HO2
N02
' H2O2
H2O2
2.000*OH-
H02
H02
HO2
2.000*HO2
CO
OH
HKO3
C2O3
' C2O3
C203
XO2
4- FORM
N02
4- HO2
PAN '
C2O3
2.000*XO2
0.790*FORM
4- 0.790*OH
XO2
0.870*XO2
4- OvllO*AIiD2
4- O - .
4- 0.890*O 4- 0.110*NO . ; ::
4- NO2
4- NO2 .»-
4- NO
4- NO2
;
4- N02
4- NO2
'
4- CO •
4- CO
4- H02 4- co ' : '
4- HO2 4- CO
4- OH
+ HNO3
4- 2.000*HO2 4- CO
4- XO2 + FORM ,
4- NO2 '. ' ' ' • ' :
+ 2.000*FORM 4- 2,000*HO2
4- 0.790*XO2 4- 0,.790*HO2
+ 0.210*PACD
4- FORM 4- HO2
4- 0.130»XO2N 4- 0.110*H02
•4- 0.760*ROR - 0.110»PAR ',' , ' , . ,=
8-42
-------
EPA/600/R-99/030
Table 8A-5. CB4_AQ Mechanism
• {
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
54}
55}
56}
57}
58}
59}
60}
61}
62}
63}
64}
65}
66}
67}
68}
69}
70}
7!}
72}
73}
74}
75}
76}
77}
78}
79}
80}
81}
82}
83}
84}
85}
86}
87}
88}
89}
ROR
ROR
ROR
OLE
OLE
OLE
OLE
BTH
ITH
ETH
TOL
TO2
TO2
CRES
CRES
CRO
XYL
OPEN
OPEN
OPEN
MGLY
MGLY
ISOP
ISOP
ISOP
ISOP
XO2
X02
XO2N
S02
S02
XO2
X02N
XO2N
XO2N
ISPD
4- NO2
4- 0
4- OH
+ O3
+ N03
+ O
4- OH
4- O3
4- OH
+• NO
+ OH
+ NO3
-f NO2
+ OH
4- OH
•f hv
-f O3
4- OH
4- hv
4- O
+ OH
•f O3
+• NO3
4- NO
+ XO2
4- NO
+ OH
+• HO2
4- HO2
+• XO2N
+ X02
•f OH
--> 1
- 2
-->
-->
--> 0
4- 0
4- 0
-->
+
--> 0
+ 0
+- 0
--> 0
4"
-->
4- 1
- *->
•f 0
-->
+ 0
--> 0
4- 0
--> 0
+ 0
— >
--> 0
4- 0
-->
-->
--> 0
+ 0
-->
+
-->
--> 0
+ 0
4- 0
-->
-->
--> 0
4- 0
--> 0
4- 0
--> 0
4- 0
4- 0
--> 0
+ 0
4- 2
-->
-->
-->
-->
• — .,
-->
-->
— >
.-->
--> 1
+ 0
+ 0
.100*ALD2
..100*PAR
HO2
NTR
.630*ALD2
.300*CO
.220*PAR
FORM
H02
.500*ALD2
.440*H02
.200*FACD
.910*XO2
ALD2
FORM
.700*HO2
XO2
.220*ALD2
. FORM
.400*FACD
.080*XO2
-S60*TO2
.900*NO2
.100*NTR
CRES
.400*CRO
.300*OPEN
CRO
NTR
.700*HO2
.800*MGLY
XO2
C2O3
C203
.030*ALD2
.030*X02
.760*HO2
XO2
C2O3
.750*ISPD
.2SO*H02
.912*ISPD
,912*HO2
,650*ISPD
.066*HO2
.150*ALD2
.200*ISPD
.800*H02
.400*PAR
NO2
NTR
SULF
: SUPuF
-UMHP
.56S*PAR
.S03*HO2
.273*ALD2
+ 0
+ 0
+ 0
-f 0
•f 0
+
-
+ 0
•*- 0
+ 0
+ 0
-
+ 0
•»- 0
+ 1
+ 0
+ 0
+ 0
+
+ 0
4.
+ 0
+ 1
•f 2
+
+
+ 0
+ 0
+ 0
+
+
+ 0
+ 0
+ 0
+ 0
+ 0
4- 0
+ 0
+ 0
+ 0
•f
+ 0,
+ 0
+ 0
.S60*XO2
,040*X02N
.380*H02
.200*FORM
.200*OH
ALD2
PAR
.740*FORM
,220*X02
.200*AACD
.090*XO2N
PAR.
.700*X02
.300«OH
.560*FORM
.420*CO
.360*CRES
.900*HO2
HO2
,600»XO2
HNO3
.500*X02
.100»PAR
.ooo*co
FORM
H02
.620*C2O3
.690*CO
.200*MGLY
C2O3
HO2
.500*FORM
.250-*C203
.629*FORM
. 088*X02N
.600*FORM
.266*OH
.3SO*PAR
.800*NTR
.200*NO2
HO2
,167*FORH
.334*CO
.498*C2O3
4. 0
. + 0
+ 0
+ 0
4-
+ 0
4- .0
-
4-
4-
4-
4>
4- 0
+ 0
+ 0
4- 0
4. 0
4- 0
+ 2
4-
4- 0
4- 0
4-
4- 0
4- 0
4- 0
4- 0
+• 0
4- 0
4-
+ 0
4- 0
4- 0
.940*HO2
.020*ROR
.280*XO2
.020*XO2N
XO2
.330*CO
.100*OH
PAR
FORM
K02
CO
HO2
.120*HO2
.440*HO2
. 900*OPEN
.SOO*HO2
.200*CRES
.300*TO2
.000*HO2
CO
.700*FORM
.080*OH
CO
.250*XO2
.250*PAR
,991*X02
,200*X02
.200*C2O3
.066*CO
XO2 , ,
.800*ALD2
.713*XO2
.168*MGLY
8-43
-------
EPA/600/R-99/030
Table 8A-5. CB4_AQ
{ 90} ISPD + O3
{ 91} ISPD + NO3 ,
{ 92} ISPD + hV
{ 93} ISOP + NO2
Mechanism
--> 0
+ 0
+ 0
. ,--> - .°
+ 0
•f 0
--> 0
+ 0
+ 0
--> 0
+ 0
.114*C203
.154*HO2
.020*ALD2
.3S7*ALD2
.925*HO2
.075*0203
.333*CO
.832*PAR
,967*C2O3
.200*ISPD
. 800*HO2
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 1
+ 0
+ 0
,150*FORM
,268*OH
.3SO*PAR
.282*FORM
,643*00,
. 075*XO2
.067*ALD2
.033*H02
.800*NTR
,200*NO
+ q
+ 0
+ 0
+ 1
+ 0
+ 0
+ 0
+ 0
+
+ 0
. " -
.850*MGLY
.064*X02
.22S*CO
,282*PAR .•>....
.850*NTR _ '"_
.07S*HNO3" "' '' !' ' ~
.900*FORM
,700*XO2
X02
.800*ALD2
Rate Expression
k( l) uses photo table NO2_CBIV88
k! 2) is a falloff expression using:
kO = 6.0000E-34 * (T/300)**(-2.30)
2.8000E-12 * (T/300) *•'( 0.00)
0.60, n =! i.oo
1.8000E-12 * exp( -1370.0/T)
9.3000E-12
a falloff expression using:
9.0000E-32 » (T/300)**(-2.00)
2.2000E-11 * (T/300)**( 0.00)
60, n -' 1.00 '••
6) is a falloff expression using:*
kO = 9.0000E-32 * (T/300)**(-1.SO)
kinf = 3.0000E-11 * (T/300)*«( 0.00)
0.60, n - 1.00
= 1.2QOOE-13 * exp( -2450.0/T)
uses photo table NO2_CBIV88
uses photo table O3O1D_CBIV88
kinf =
F =
k( 3! «
k( 4) =
k( 5) is
kO
kinf =
F * 0.
k(
k(
k(
kS
k(
k(
k(
k!
k(
F =
7)
8)
9)
10)
11)
12)
13)
14)
k<
is
18?
kO
kinf =
F m
k{ 19)
k( 20)
k{ 21)
k( 22)
k( 23)
kO =
kinf =
F = 0.
k( 24)
k( 25! =
k( 26) =
scaled by l.OOOOOE+00
scaled by 5.
scaled by 1.
30000E-02
OOOOOE+00
1.8000E-11 * exp( 107.0/T)
3.2000E-11 * exp{ 67.0/T!
2.2000E-10
1.6000E-12 * exp{ -940.0/T5
1.4000E-14 * exp( -580.0/T)
k! 15) uses photo table NO2_CBIV88
k( 16) a 1.3000E-11 * exp{ 250.0/T)
k( 17) = 2.5000E-14 * exp! -1230.0/T)
a falloff expression using:
2.2000E-30 * (T/300)** (-4.30)
1.5000E-12 * (T/300) **(-0'.SO)
0.60, n = 1.00 "
1.3000B-21
k( 18) / Keq, where Keq = 2.700E-27
3.3000E-39 * exp( 530.0/T)
4.4000E-40
a falloff expression using:
6.7000E-31 * (T/300)««(-3.30)
3.0000E-11 * (T/300)**(-1.00)
60, n = *1.00" •'
uses photo table NO2_CBIV88
= 6.6000E-12
l.OOOOE-20
k! 27! is a falloff expression using:
kO = 2.6000E-30 * (T/300)**(-3.20)
kinf « 2.4000E-11 * (T/300)**(-1.30)
scaled by 3.390QQE+01
exp( 11000. 0/T)
is
scaled by 1.97500E-01
Rate Constant
{O.OOOOOE+00}
{1.37387E-14}
{1.81419E-14}
{9.30000E-12}
{1.57527E-12J
{1.66375E-12}
{3.225811-17}
{O.OOOOOE-I-OO}
{O.OOOOOE-hOO}
{2.57757E-11}
{4.00676E-11}
{2.20000E-10}
{6.826SOE-14}
{1.99920E-15}
{o.ooooos+oo}I
{3.008051-11}"
{4.03072E-16}
{1.26440E-12}
I'
{1.30000E-21}
{4.36029E-02}
{1.9S397B-3S}
{4.39999E-40}!
{6.69701E-12}'
{O.OOOOOE-t-00}
{6.60000E-12}
{l.OOOOOE-20}
J1.14885E-11}
8-44
-------
EPA/600/R-99/030
Table 8A-5. CB4_AQ Mechanism
F = 0.60, n = 1.00
k( 28) is a special rate expression o£ the form:
k = kO + {k3 [H] / (1 -i- k3[M]/k2)}, where
kO = 7,20001-15 * expS 785.0/T)
k2 = 4.1000E-16 * exp! 1440.0/T)
k3 = 1.9000E-33 * exp! 725.0/T)
kC 29) = 3.7000E-12 * exp( 240.0/TS
k{ 30) is a falloff expression using:
kO = 2.3000E-31 * (T/300)**(-4.60)
kinf = 4.2000E-12 * (T/300!**( 0.20)
F = 0.60, n = 1.00
k( 31) - k( 30) / Keq, where Keq = 2.100E-27 *
k( 32) = 1.3000E-12 * exp ( 380.0/T)
kS 33S - 5.9000E-14 * exp( 1150.0/T)
k( 34) = 2.20001-38 * exp( 5800.0/T)
k( 35) uses photo table HCHOntol_CBIV8S
kC 36) = 3.1000E-12 * expS -187.0/T)
k( 37) = 1.5000E-13 * {1.0 +• 0.6*Pressure)
k( 38) = l.OOOOE-11
k( 39) uses photo table HCHOrati_CBIV88 , scaled
k{ 40! uses photo table HCHOmol_CBIV8S , scaled
3.0600E-11 * exp( -1550.0/T)
exp( 10900.0/T)
scaled by 2.5SOOOE-01
k( 41)
k( 42)
kS 43)
k( 44)
6.3000E-16
1.20001-11 * expS -986.0/T)
7.0000E-12 * exp( 250.0/T)
k( 45) = 2.5000E-15
kC 46) uses photo table MiD_CBIV88
k{ 47) = 3.4900E-11 * exp! -180.0/T)
k( 48) = 2.6300E-12 * exp ( 380.0/T)
k( 49) = 2.0000E+16
k( 50) = 2.5000E-12
k( 51) a 6.5000E-12
k( 52) = 1.1000E+02
k( 53) = 8.1000E-13
k£ 54) = 1.0000E-H5
k( 55) = 1.6000E+03
k( 56) = 1.5000E-11
k( 57) M 1.2000E-11
k( 58) = 5.2000E-12
k( 595 = 1.4000E-14
k{ 60) m 7.7000E-15
k( 61) = l.OOOOE-11
k( 62) = 2.0000E-12
k{ 63) = 1.3000E-14
k{ 64) = 2.1000E-12
k( 65) = 8.1000E-12
k( 66) = 4.20QOE+00
k( 67) = 4.1000E-11
k( 68) « 2.2000E-11
k( 69S = 1.4000E-11
k{ 70) = 1.70001-11
k( 71) = 3.0000E-11
k( 72) uses photo table HCHOrad_CBIV88
k( 73) = 5.4000E-17 * exp ( -500.0/T)
k( 74) B 1.7000E-11
k( 75) uaes photo table HCHOrad_CBIV88
k( 76! = 3.6000E-11
k( 77) = 2.5400E-11 * exp! 407.6/T)
k( 78) = 7.8600E-15 * exp( -1912.0/T)
k( 79) = 3.0300E-12 * exp! -448.0/TS
k! 80) = 8.1000E-12
exp(-13500.0/T)
exp! -1710.0/T)
exp! -8000.0/T)
exp( -324.0/T)
exp( 504.0/T)
exp( -2105.0/T)
exp( -792.0/T)
exp! 411.0/T)
exp! -2633.0/T)
exp{ 322.0/T)
exp! 116.0/T)
by l,
by 1.
OOOOOS+00
OOOOQE+00
scaled by l.OOOOOE+00
scaled
scaled
by 9.04000E+00
by 9.64000E+00
{1.47236E-13}
{8.27883E-12}
{1.48014E-12}
{9.17943E-02}
{4.65309E-12}
{2.79783E-12}
{6.23927E-30}
{O.OQOOOE+OO}
(1.65514E-12)
{2.40000E-13}
{l.OOOOOE-11}
(O.OOOOOE+OO)
{O.OOOOOE+00}
{1.6S275E-13}
{6.30000E-16}
{4.38753E-13}
{1.61972E-11}
{2.500QOE-15}
{O.OOOOOE+OO}
{1.9076SE-11}
{9.41356E-12}
{4.23268E-04}
{2.50000E-12}
{6.50000E-12}
{3.54242E-01}
{8.10000E-13}
{2.19325E+03}
{1.60000E+03}
{1.50000E-11}
{4.04572E-12J
{2.82173E-11J
(1.19778E-17)
{7.70000E-1S}
{7.01080E-13}
{7.94340E-12}
{1.89105E-18}
{6.18715E-12}
{8.10000E-12}
{4.20000E+06}
{4.10000E-11}
{2.20000E-11J
{1.40000E-H}
{2.50901E-11}
{3.00000E-11}
{O.OOOOOE-I-OO}
{1.00858E-17}
{1.70000E-11}
{O.OOOOOE4-00}
{3.60000E-11}
{9.973e8E-ll}
{1.28512E-17}
{6.73B19E-13}
{8.10000E-12}
8-45
-------
EPA/600/R-99/030
Table 8A-5. CB4_AQ Mechanism
k{
k(
k!
kC
k(
kC
k{
k(
k(
k(
k(
kS
k(
81)
82)
83)
84)
85)
86)
87)
88)
89)
90)
91)
92)
93)
9
=
=
=
=
=
«=
m
Et
m
=
1
8
4
1
7
7
1
3
3
7
1
uses
»
1
.7000E-14 *
.1000E-12
.3900E-13 *
.3600E-06
.6700E-14 *
.6700E-14 *
.7300E-14 *
.4500E-14 *
.3600E-11
.1100E-18
.OOOOE-15
photo table
.4900E-19
expC
exp(
exp(
exp(
exp(
exp(
1300
ISO
1300
1300
1300
1300
ACROLEIN
.0/T)
.0/T)
.0/T)
.0/TS
.0/T)
.0/T)
, scaled by 3.'60000E-03
{1
(8
{7
{1
{6
{6
{1
{2
{3
{7
{1
(o
{1
.33359E-12)
.10000E-12}
.51005E-13)
.36000E-06}
.01684E-12} ,(
.01684E-12}
.35712E-12} .-.
.70640E-12}
.36000E-11}
.11000E-18}
.OOOOOE-15}
.OOOOOE+OO)
.49000E-19}
8-46
-------
EPA/600/R-99/030
Table 8A-6. CB4_AE_AQ Mechanism
Reaction List
{ 1)
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
{ 8}
{ 9}
{ 10}
{ 11}
{ 12}
{ "}
{ 14}
{ 15}
{ 16}
{ 17}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40}
{ 41}
{ 42}
{ 43}
{ 44}
{ 45}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
{ 54}
NO2
O
O3
O
O
O
03
03
O3
O1D
O1D
O1D
O3
03
NO3
NO3
NO3
N03
N2OS
N2O5
NO
NO
OH
HONO
HONO
HONO
OH
OH
H02
HO2
PNA
PNA
HO2
H02
H2O2
H2O2
CO
FORM
FORM
FORM
FORM
FORM
ALD2
ALD2
ALD2
ALD2
C2O3
C203
PAN
C2O3
C203
OH
PAR
ROR
+ hv
+ [02]
+ NO
+ NO2
+ NO2
+ NO
+ NO2
+ hv
+ hv
+ [N2]
+ [02]
+ [H20]
+ OH
+ HO2
+ hv
+ NO
+ NO2
+ NO2
+ [H20]
+ NO
+ NO2
'+ NO
+ hv
+ OH
+ HONO
+ NO2
+ HN03
+ NO
+ NO2
+ OH
+ HO2
+ HO2
+ hv
+ OH
+ OH
+ OH
+ hv
+ hv
+ O
+ NO3
+ O
+ OH
+ NO3
+ hv
+ NO
+ NO2
+ C2O3
+ HO2
+ OH
--> NO
--> O3
--> NO2
--> NO
--> NO3
- - > NO2
--> NO3
--> O
--> O1D
--> O
--> O
--> 2.000*OH
- - > HO2
--> OH
--> 0.890*NO2
--> 2.000*NO2
--> NO
--> N2OS
--> 2.000*HNO3
- - > NO3
+ [O2] --> 2.000*NO2
+ [H2O] --> 2.000*HONO
- - > HONO
--> OH
--> NO2
--> NO
•--> HNO3
- - > NO3
--> OH
- - > PNA
- - > HO2
--> NO2
- - > H2O2
+ [H20] --> H202
--> 2.000*OH
- - > HO2
- - > HO2
--> HO2
--> 2.000*HO2
--> CO
--> OH
- - > HNO3
--> C2O3
--> C2O3
--> C2O3
--> XO2
+ FORM
- - > NO2
+ HO2
--> PAN
--> C2O3
--> 2.000*XO2
--> 0.790*FORM
+ 0.790*OH
--> X02
--> 0.870*XO2
+ 0.110*ALD2
--> 1.100*ALD2
- 2.100*PAR
+ O
+ 0.890*0
+ NO2
+ NO2
+ NO
+ NO2
+ NO2
+ NO2
+ CO
+ CO
+ HO2
+ HO2
+ OH
+ HNO3
+ 2.000*HO2
+ XO2
+ NO2
+ 2.000*FORM
.+ 0.790*XO2
+ 0.210*PACD
+ FORM
+ 0.130*XO2N
+ 0.760*ROR
+ 0.960*XO2
+ 0.040*XO2N
+ 0.110*NO
+ CO
+ CO
+ CO
+ FORM
+ 2.000*HO2
+ 0.790*HO2
+ HO2
+ 0.110*HO2
- 0.110*PAR
+ 0.940*HO2
+ 0.020*ROR
8-47
-------
EPA/600/R-99/030
Table 8A-6. CB4_AE_AQ Mechanism
{
{
!
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
{
<
{
{
{
{
55}
56}
57}
58}
59}
60}
61}
62}
63}
54}
65}
66}
67}
68}
69}
70}
71}
72}
73}
74}
75}
76}
77}
78}
79}
80}
81}
82}
83}
84}
85}
86}
87}
88}
89}
90}
ROR
ROR
OLE
OLE
OLE
OLE
ETH
ETH
ETH
TOL
T02
TO2
CRES
CRES
CRO
XYL
OPEN
OPEN
OPEN
MGLY
MGLY
ISOP
ISOP
ISOP
ISOP
XO2
XO2
X02N
SO2
SO2
XO2
X02N
X02N
X02N
ISPD
ISPD
+ HO2
•f O
1
•f OH
+ O3
+ NO3
+ o
+ OH
+ O3
+ OH
+ NO
+ OH
+ NO3
+ NO2
+ OH
+ OH
+ hv
•<• 03
+ OH
+ hv
+ o!
+ OH
+ O3
•«• N03
+ NO
+ X02
+ NO
* OH
+ HO2
+ HO2
+ X02N
+ X02
+ OH
+ 03
-->
-->
--> 0
•f 0
+ 0
-->
+
-••> 0
+ 0
+ d
--» 0
+
--*>
+ i
-->
+ 0
-->
+ 0
--> 0
4- 0
--> 0
+ 0
-->
--> 0
+ 0
-->
-->
--> 0
+ 0
+
-->
+
-->
--> 0
+ 0
+ 0
-->
-->
--> 0
+ 0
--> 0
+ 0
--> 0
+ 0
+ 0
--> 0
+ 0
+ 2
-->
-->
-->
--»
-->
.->
-->
-->
«. a>
--> 1
+ 0
+ b
--> 0
HO2
NTR
.630*ALD2
.300*CO
.220*PftR
FORM
HO2
.500*ALD2
.440*HO2
.200*FACD
. 910*XO2
ALD2
FORM
.700*HO2
X02
.220*ALD2
FORM
.400*FACD
.080*X02
.560*T02
.900*NO2
.100*NTR
CRES
.400*CRO
.300*OPEN
CRO
HTR
.700*HO2
,800*MGLY
XYLAER
X02
C203
C203
. 030*ALD2
.030*X02
. 760*H02
XO2
C2O3
.756*ISPD
,250*HO2
.912* ISPD
.912*HO2
.650*ISPD
. 066*H02
. 150*ALD2
.200*ISPD
. 800*HO2
.400*PAR
N02
NTR
SULF
SULF
UMHP
.565*PAR
,503*HO2
.273*ALD2
. 114*C203
+ 0
•f- 0
+ 0
-
+ 0
+ 0
+ 0
+ 0
-
+ 0
+ 0
+ 1
•f 0
•f- 0
+
+ 0
+
+ 0
+
+
+ 0
•(- 1
+ 2
+
+
•*• 0
+ 0
+ 0
+
+
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
•*•
+
•*• 0
+ 0
+ 0
+ 0
.380*HO2
-200*FORM
.200*OH
ALD2
PAR
.740*FORM
.220*XO2
.200*AACD
.090«XO2N
PUR
.700*XO2
.300*OH
,560*FORM
.420*CO
.360*CRES
TOLAER
,900»HO2
HO2
,600*XO2
CSLRSR
HNO3
,500*X02
.100*PftR
.ooo*co
FORM
H02
,620*C203
.690*CO
.200*MGLY
C2O3
'HO2
.SOO*FORM
j250*C2O3
.629*FORM
,088*XO2N
.600*FORM
.266*OH
.3Sp*PAR
.800*NTR
.200*N02
HO2
SOIAER
, 167*FORM
.334*CO
.498*C203
.150*FORM
+ 0
+ 0
+
+ 0
+ 0
-
+
+
+
+
+ 0
+ 0
+ 0
+ 0
+
-I- 0
+ 0
+ 2
+
+ 0
+ 0
+
+ 0
+ 0
+ 0
+ 0
•*- 0
+ 0
+
+ 0
+
•f 0
+ 0
+ 0
.280*X02
.020*X02N
XO2
.330*CO '
.100*OH
PAR
FORM
NO2
CO
H02
.120*H02
•
.440*HO2
.900*OPEN
.600*HO2
CSIAER
;. . ! i
,200*CRBS
,300*T02
. 000*H02
CO
,700*FORM
,080*OH
, ' CO
.250*XO2
,250*PAR
. 991*XO2
,200*XO2
.200*C2O3
. 066*CO
.. X?2
.800*ALD2
SUIASR
.713*XO2
.168*MGLY
1 '
.850*MGLY
8-48
-------
EPA/600/R-99/030
Table 8A-6. CB4_AE_AQ Mechanism
{ 91}
{ 92}
{ 93}
{ 94}
{ 95}
{ 96}
ISPD
ISPD
ISOP
TERP
TERP
TERP
+ NO3
+ hv
+ N02
+ OH
+ NO3
+ O3
+ 0
+ 0
--> ' 0
+ 0
+ 0
--> 0
+ 0
+ 0
--> 0
+ 0
+ 2
-->
-->
-->
. 154*H02
.020*ALD2
.357»ALD2
.925*HO2 n
.075»C203 H
.333»CO H
. 832*PAR H
.967*C2O3
.200»ISPD H
.800*HO2 H
.400*PAR
TERPAER H
TERPAER H
TERPAER H
1- 0
1- 0
1- 0
K 0
h 0
K 0
t- 1
^ 0
K 0.
f-
^
K
.268*OH
.360»PAR
.282*FORM
.643*CO
.075»X02
.067*ALD2
.033*HO2
.800*NTR
.200*NO
OH
N03
03
+ 0.
+ 0.
+ 1.
+ 0.
+ 0.
+ 0.
+ 0.
+
+ 0.
064
225
282
850
075
900
700
800
*
*
*
*
*
*
*
*
X02
CO
PAR
NTR
HNO3
FORM
X02
XO2
ALD2
Rate Expression
k( 1) uses photo table NO2_CBIV88
k( 2) is a falloff expression using:
kO = 6.0000E-34 * (T/300)**(-2.30)
kinf = 2.8000E-12 * (T/300)**( 0.00)
F = 0.60, n = 1.00
k( 3) = 1.8000E-12 * exp( -1370.0/T)
k( 4) = 9.3000E-12
k( 5) is a falloff expression using:
kO = 9.0000E-32 * (T/300)*»(-2.00)
kinf = 2.2000E-11 * (T/300)**( 0.00)
F = 0.60, n =. 1.00
k( 6) is a falloff expression using:
kO = 9.0000E-32 » (T/300)»»(-1.50)
kinf = 3.0000E-11 * (T/300)**( 0.00)
0.60, n = 1.00
= 1.2000E-13 * exp( -2450.0/T)
uses photo table NO2_CBIV88
uses photo table O3O1D_CBIV88 ,
= 1.8000E-11 * exp( 107.0/T)
= 3.2000E-11 * exp( 67.0/T)
= 2.2000E-10
= 1.6000E-12
= 1.4000E-14
scaled by l.OOOOOE+00
F =
7)
8)
9)
k(
k(
k(
k( 10)
k( 11)
k( 12)
k( 13)
k( 14)
k( 15)
k( 16)
k( 17)
k( 18)
kO
kinf =
F =
k( 19)
k( 20)
k( 21)
k( 22)
k( 23)
kO
kinf =
exp( -940.0/T)
exp( -580.0/T)
uses photo table NO2_CBIV88
= 1.3000E-11 * exp( 250.0/T)
= 2.5000E-14 * exp( -1230.0/T)
is a falloff expression using:
= 2.2000E-30 * (T/300)**(-4.30)
* (T/300)**(-0.50)
00
1.5000E-12
0.60, n = 1
= 1.3000E-21
= k( 18) / Keq
= 3.3000E-39
= 4.4000E-40
is a falloff expression using:
= 6.7000E-31 * (T/300)**(-3.
3.0000E-11 * (T/300)**(-!.
F = 0.60, n = 1.00
k( 24) uses photo table NO2_CBIV88
k( 25) = 6.6000E-12
k( 26) = l.OOOOE-20
k( 27) is a falloff expression using:
scaled by 5
scaled by 1
30000E-02
OOOOOE+00
scaled by 3.39000E+01
Rate Constant
{O.OOOOOE+00}
{1.37387E-14}
{1.81419E-14}
{9.30000E-12}
{1.57527E-12}
{1.66375E-12}
{3.22581E-17}
{O.OOOOOE+00}
{O.OOOOOE+00}
{2.57757E-11}
{4.00676E-11}
{2.20000E-10}
{6.82650E-14}
{1.99920E-15}
{O.OOOOOE+00}
{3.00805E-11}
{4 .03072E-16}
{1.26440E-12}
where Keq = 2.700E-27 * exp( 11000.0/T)
exp( 530.0/T)
30)
00)
scaled by 1.97500E-01
{1.30000E-21}
{4.36029E-02}
J1.95397E-38)
{4.39999E-40}
{6.69701E-12}
{O.OOOOOE+00}
{6.60000E-12}
{l.OOOOOE-20}
{1.14885E-11}
8-49
-------
EPA/6QO/R-99/030
Table 8A-6. CB4_AE_AQ Mechanism
kO = 2.60001-30 * (T/300)**(-3.20)
kinf = 2.4000E-11 * (T/300)**{-1.30)
F = 0.60, n » 1.00
k( 28) is a special rate expression of the form:
k = kO + {k3 [M] / (1 + k3[M]/k2)}, where
kO = 7.200.0.E-15 .* exp! 785.0/T!
k2 = 4.1000E-16 * exp( 1440.0/T)
k3 = 1.9000E-33 * exp! 725.0/T)
k{ 29) = 3.70001-12 * exp! 240.0/T)
k{ 30) is a £allo£f expression using;
kO = 2,30001-31 * .(f/300)'** (-4.60) '
kinf = 4.2000E-12 * (T/300)**( 0.20!
F = 0.60, n = 'l.OO '""": ' '
k( 31) = k( 30) / Keq, where Keq = 2.1001-27
k! 32) = 1.30001-12 * exp! 380.0/T)
k! 33) = 5.9000E-14 * exp! 1150.0/T)
k( 34) = 2.2000E-38 * exp{ 5800.0/T)
k( 35) uses photo table HCHOmol_CBlV88
k! 36) = 3.1000E-12 * exp! -187.0/T)
k! 37) = l.SOOOE-13 * (i.o + 0.6*Pressure)
k( 38) = l.OOOOE-11
k! 39) uses photo table HCHOrad_CBIV88 , scaled
k( 40} uses photo table HCHOmol_CBIV88 , scaled
k( 41! = 3.0000E-11 * expC -1S50.0/T)
k( 42) » 6.3000E716
k{ 43) = 1.2000E-11 * exp! -986.0/T!
k! 44! = 7.0000E-12 * exp! 250.0/T)
k( 45! = 2.5000E-15
k( 46! uses photo table ALD_CBIV88
k! 47) = 3.4900E-11 * exp( -180.0/T)
exp{ 10900.0/T)
scaled by 2.55000E-01
k(
k!
k!
k(
k(
k!
kS
k(
k{
kC
k!
k!
k(
k(
k(
k!
k!
k!
k!
k{
kC
k!
k{
kS
48)
49)
50)
51)
52)
53!
54!
55)
56)
57)
58)
59!
60)
61)
62)
63!
64}
65}
66}
67}
68)
695
70)
71!
=
m
=
SJ
=
=
=
_
=
=
=
=
s
=
_
=
=
0
=
=
m
=
=
=
2
2
2
6
1
8
1
1
1
1
5
1
7
1
2
1
2
8
4
4
2
1
1
3
.6300E-12
.OOOOE+16
.SOOOE-12
.5000E-12
.1000E+02
.1000E-13
.OOOOE+15
.6000E+03
.5000E-11
.2000E-11
.2000E-12
.4000E-14
.7000E-15
.OOOOE-11
.OOOOE-12
.3000E-14
.10001-12
.10001-12
.2000E+00
.1000E-11
.2000E-11
.4000E-11
.70001-11
.OOOOE-11
* exp(
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp(
* exp!
* exp!
* exp(
* expC
380
-13500
-1710
-8000.
-324,
504,
-2105,
-792,
411,
-2633 ,
322,
116,
.0/T)
.0/T)
.0/T)
.0/T}
. 0/T)
. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
k( 72! uses photo table HCHQrad_GBIV88
k( 73!
exp! -500. 0/T)
= 5.4000E-17
k( 74) = 1.7000E-11
k( 75) uses photo table HCHOrad_CBIV88
= 3.6000E-11
= 2.5400E-11 * exp( 407.6/T)
k( 78! = 7.8600E-15 * exp! -1912.0/T)
k! 76)
k! 77)
by 1,
by 1.
OOOOOE+00
OOOOOE+00
scaled by 1.OOOOOE+00
scaled
scaled
by 9.04000E+00
by 9.64000E+-00
{1.47236E-13}
{8.27883E-12}
{l.48014E-12}:*-
{9.17943E-02} -
{4.6S309E-12}
{2.79783E-12} ...
{6.23927E-30}
{0.OOOOOE+00}
{1.6S514E-12}
{2.40000E-13}
{l.OOOOOE-ll}
{O.OOOOOE+OO}
{O.OOOOOE+OO}
{1.6S275E-13}'! i
{6.30000E-16} J
{4.387538-13}-•'
{1.61972E-11}-
{2.50000E-15}
{0.OOOOOE+00}.
{1.90766E-11}
{9.41356E-12}
{4.232S8E-04}
{2.SOOOOE-12}
{6.50000E-12}
{3.54242E-01} j
{8.10000E-13} ,
{2.1932SE+03}
{l.SOOOOE+03}
{l.SOOOOE-11}
{4.04572E-12} "
{2.82173E-11J
{1.19778E-17}
{7.70000E-15}
{7.01080E-13}
{7.94340E-12}
{1.89105E-18}
{6.18715E-12}:J
{8.10000E-12}
{4.20000E+00}
{4.10000E-11}
{2.20000S-11JH,
{1.40000B-11}-'
{2.50901E-11}..
{3.00000E-11}
{O.OOOOOE+OO}
{1.00858E-17}
{1.70000E-11} :
{O.OOOOOE+OO}
{3.60000E-11}:!,
{9.97368E-11}
{1.28512E-17}
8-50
-------
EPA/600/R-99/030
Table 8A-6. CB4_AE_AQ Mechanism
k{
k(
k(
k(
k(
k(
k(
k<
k!
k(
k(
k(
k{
k(
kS
kS
k(
k(
79)
80)
81)
82)
83)
84)
85)
86)
87)
88)
89)
90)
91)
92)
93)
94)
95)
96)
=
«=
ss
=
=
=
=
=
=
=
at
=
=
3
8
1
8
4
1
7
7
1
3
3
7
1
uses
ss
=
=
=
1
1
3
7
.03002-12 *
.1000B-12
.7000E-14 *
.1QOOE-12
.3900E-13 *
.3600E-OS
.6700E-14 *
.6700E-14 *
.7300E-14 *
.4500E-14 *
.3SOOE-11
.1100E-18
.OOOOE-15
photo table
.49001-19
. 0700E-11 *
.2300E-11 *
.2900E-15 *
exp(
expC
exp(
-448
1300
ISO
.0/T)
.0/T)
.0/T)
(6
(8
(1
{8
{7
.73819E-13}
. 100QOE-12J
.33359E-12}
.10000E-12}
.51005E-13}
{1.3SOOOE-06}
exp(
exp{
exp(
exp(
1300
1300
1300
1300
RCROLSIN
exp!
exp(
exp(
549
-975
-1136
.0/T)
.0/T)
.0/T)
.0/T)
» scaled by 3.60000S-03
.0/T)
.0/T)
.0/T)
{6
{6
(1
{2
(3
{7
{1
{0
(1
(S
{1
{1
.01684E-12)
.01684E-12}
.35712E-12}
.70640E-12}
.36000E-11}
.11000E-18}
,QQQQOE-1S}
.OOOOOE+OO}
.49000E-19)
.7S2S9E-11)
.22539E-12}
.61125E-16}
8-51
-------
EPA/600/R-99/030
Table 8A-7. RADM2 and RADM2_AQ Mechanisms
Reac
{ 1}
{ 2}
{ 3}
{ 5}
{ 6}
{ 71
{ 8}
{ 9}
{ «}
{ "I
{ "}
{ «}
{ "}
{ is}
{ 16}
{ 17}
{ 18}
{ 19}
{ 20}
< 21}
{ 22}
'{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 4°}
{ 41}
{ «}
{ 43}
{ 44}
{ 45}
{ 46}
{ 47}
{ «}
{ 49}
{ 50}
{ 51)
! 52}
{ 53}
{ 54}
{ 55}
{ 56}
{ 57}
tion L
N02
O3
O3
HONO
HNO3
HNQ4
NO3
NO3
H2O2
HCHO
HCHO
ALD
OP1
OP2
PAA
KET
GLY
GLY
MGLY
0CB
ONIT
O3P
O3P
O1D
O1D
O1D
O3
O3
03
H02
H02
HNO4
HO2
HO2
H202
NO
NO
O3
NO3
N03
N03
N03
N20S
N20S
HO
HO
HO
HO
HO
CO
HO
ETH
HC3
HC5
HC8
OL2
OLT
+ hv
+ hv '
+ hv (
+ hv '
+ hv '
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
. + hv '
+ hv
+ hv
+ hv
+ hv
4- hv
4- [M] \ 4- [02]
+ HO2'
- 4- [N2]
+• E02)
4- [H2O]
+ MO
4- HO
+ H02
+ NO
+ N02j
+ H02
+ HO2' + (H20J
+ HO . i
+ HO
'+ NO + [O2]
+ NO2
+ NO
+ N02
+ HO2
+ N02
+ [H2O]
+ NO2
+ HNO3
+ HNO4
+ HO2J
+ S021
••• HO' i
+ HO \
+ HO
+ HO
+ HO
+ HO
+ HO
--> 03P
- - > O1D
--> 03P
--> HO
--> ' HO
--> HO2
--> NO
--> NO2
--> 2,000*HO
--> CO
--> HO2
--> MO2
- - > HCHO
--> ALD
- - > M02
--> ACO3
--> 0.130*HCHO
--> 0.4SO*HCHO
--> ACO3
--> 0.980*H02
--> 0.200*ALD
+ N02
--> O3
--> NO
--> O3P
--> O3P
--> 2.000*HO
- - > MO2
- - > HO2
--> HO
--> NO2
--> HN04
- - > H02
--> H2O2
--> H2O2
.--> HO2
--> " HONO
--> 2,000*NO2
- - > NO3
--> 2.000*N02
--> NO
- - > HNO3
--> N205
--> NO2
--> °2.000*HN03
--> HNO3
--> N03
--> N02
-->
--> SULP
--> H02
--> M02
--> ETHP
--> 0.830*HC3P
+ Q.Q75*ALD
--> HC5P
--> HC8P
- - > OL2P
- - > OLTP
+ NO . , . , . .
.!.'' ".,., . .'.
+ NO . '". ' ' "
+ NO2
+ NO2
-t- O3P " :
+ HO2 + CO
-(• HO2 + CO
+ HO2 + HO
* HO2 + HO
+ HO . -,! ,
+ ETHP " '- " '. : :
+ 1.870*CO
+ 1.5SO*CO + 0,800*HO2
+ HO2 + CO
-1- 0.020*ACO3 4- TCO3
•»• 0.800*KET + HO2
+ HO ••
4- N02
i .
'i-f ' /
i
+ K02
+ N03
+ HO2
+ 0.170*HO2 + O.OQ9*HCHO
+ 0.025*KET
+• 0.2SO*X02
+• 0.7SO*X02
8-52
-------
EPA/6QO/R-99/030
Table 8A-7. RADM2 and RADM2_AQ Mechanisms
{ 58}
{ 59}
{ 60}
{ 61}
{ 52}
{ 63}
{ 64}
{ 65}
{ 66}
{ 67}
{ 68}
{ 59}
{ 70}
{ 71}
{ 72}
{ 73}
{ 74}
{ 75}
{ 76}
{ 77}
{ 78}
{ 79}
{ 80}
{ 81}
{ 82}
{ S3}
! 84}
{ 85}
{ 86}
{ 87}
{ 88}
{ 89}
{ 90}
{ 91}
{ 92}
{ 93}
{ 94}
{ 95}
{ 96}
{ 97}
{ 98}
{ 99}
{100}
{101}
{102}
{103}
{104}
{105}
OLI
TOL
XYL
CSL
CSL
HCHO
ALD
KET
GLY
MGLY
DCB
OP1
OP2
PAA
PAW
ONIT
ISO
AC03
PAN
TC03
TPAN
M02
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
ACO3
TCO3
TOLP
XYLP
ETHP
KETP
OLN
HCHO
ALD
GLY
MGLY
DCB
CSL
OL2
OLT
OLI
ISO
OL2
OLT
OLI
+ HO
4 HO
+ HO
-1- HO
•f HO
4- HO
4 HO
+ HO
4- HO -
4- HO
4 HO
4 HO
4 HO
4 HO
4- HO
4- HO
4 HO
4 NO2
4- NO2
4 NO
+ NO
4 NO
4 NO
4 NO
4- NO
. + NO
4 NO
+ NO
4 NO
4 NO
+ NO
4- NO
4- NO
4 NO3
4 N03
4 NO3
4 NO3
4- NO3
4 NO3
4 NO3
4 NO3
4 N03
4 NO3
4 O3
4 O3
4- O3
-->
--> 0
--> 0
--> 0
-->
-->
-->
-->
— >
-->
-->
--> 0
--> 0
-->
-->
-'->
-->
-->
-->
-->
-->
-->
--> 0
•f 0
--> 0
+ 0
--> 0
+ 0
--> 1
•(• 0
-->
+
-->
4- 0
-->
-->
+• 0
+ 2
-->
+ 0
-->
+ 0
-->
-->
-_>
-->
-->
-->
-->
-->
-->
-->
'->
-->
-->
.->
+• 0
--> 0
+• 0
+• 0
--> 0
OLIP
.750*TOLP
.830*XYLP
.100*HO2
CSL
H02
ACO3
KETP
HO2
AGO 3
TCO3
.500«MO2
.500*HC3P
ACO3
. HCHO
HC3P
. OLTP
PAN
AC03
TPAN
TCO3
HCHO
. 750*ALD
,036*ONIT
.380*ALD
.920*NO2
.3SO*ALD
.240*ONIT
,600*HCHO
.200*ALD
ALD
NO2
HO 2
,100*KET
M02
NO2
.110*MGLY
.000*XO2
N02
,160*GLY
NO2
.806*DCB
. ALD
MGLY
HCHO
HO 2
AC03
HN03
HH03
HNO3
HN03
OLN
OLN
OLN
OLN
HCHO
.120*HO2
.530*HCHO
.200*ORA1
.220'*MO2
.180*HCHO
+
•t-
+
4-
+
+.
4
4
+
+
4-
-1-
+
4-
+
4-
•t-
4-
+
4-
4-
+
-t-
+
+
4-
+.
+
4-
+.
4.
+.
4
4
+
4-
+
4
+
4
+
4
4
0
0
0
2
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
.250*CSL
. 170*CSL
.900*XO2
CO
,ooo*co
CO
.500*HCHO
.500*ALD
NO3
NO2
N02
NO2
HO2
.250»KET
.964*NO2
.690*KET
,920*H02
.060*KET
.760*NO2 -
HO2
HCHO
.450*ALD
NO2
N02
, 920*HO2
,050*ACO3
HO2
,700*DCB
HO2
HO2
NO2
ALD
HNO3
HNO3
HO2
, ACO3
TCO3
XNO2
.400*ORA1
,. 500*ALD
.200*ORA2
.100*HO
,"720*ALD
4- 0.250*HO2
4- 0.170*HO2
+ 0.900*TCO3
4- 0.500*HO
4- 0.500*HO
4- X02
4- NO2
4- 0,090*HCHO
4- 0.964*HO2
4 0,080*ONIT
4- 0.040*HCHO
4- 0,760*HO2
4- NO2
4- HO2
4- 6,280*HCHO
4- 0,890*GLY
4- 0.950*CO
4 0.170*MGLY
4- 0.450*MGLY
4- NO2
4- HO2
4- 2.000*N02
4- CO
4- 2.000*CO
4- CO
4 0.500*CSL
4- 0.420*CO
4- 0.330*CO
4- 0.230*HO2
4- 0.100*KET
8-53
-------
EPA/600/R-99/030
Table 8A-7. RADM2 and RADM2_AQ Mechanisms
{106}
{107}
{108}
{109}
{110}
{111}
{112}
{113}
{114}
{115}
{116}
{117}
{118}
{119}
{120}
{121}
{122}
{123}
{124}
{125}
{126}
{127}
{128}
{129}
{130}
{131}
{132}
{133}
{134}
{135}
{136}
{137}
{138}
{139}
{140}
{141}
{142}
{143}
{144}
ISO
H02
HO2
HO2
HO2
HO2
H02
HO2
HO2
H02
HO2
HO2
HO2
HO2
HO2
MO2
MO2
MO2
MO2
MO2
MO2
M02
MO2
MO2
HO2
MO2
M02
MO2
MO2
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
CLIP
KETP
ACO3
ACO3
4-
4-
4-
4-
+
4-
+
4-
4.
4-
+
4-
4-
4-
4-
4-
4-
4-
+
4-
+
4-'
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
O3 ,
MO2
ETHiP
HC3P
HC5P
HC8P
OL2P
OLTP .
OLIP
KETP
ACO3
TOLP
XYLP
TC03
OLN
MO2
ETHP
HC3P
l
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
ACO3
TOLP
XYLP
TC03
OLK
ACO3
ACO3
ACO3
ACO3
t
AC03
ACO3
ACO3
ACO3
ACO3
TOLP
4- 0
4- 0
--> . 0
4- 0
4- 0
.->
-->
-->
-->
-->
-->
. . -->
-->
-->
-->
-->
-->
-->
-->
--> 1
--> 0
--> 0
4.
--> 0
4-
--> 0
+
~ ~ > 1
--> 1
--> 0
4- Q
--> 0
-->
4- 0
-->
4- 0
-->
4- 2
--> 0
+ 0
+ 0
--Si 1
4-
-->
4- 0
--> 0
+ 0
--> 0
4- Q
--> 0
4- 0
--> 0
4- 0
-->
4- 0
--> 0
4- 0
-->
4- Q
*" ** > 2
.230*CO
.260*HO2
,530*HCHO
.200*ORA1
.22'0*M02
DPI
OP2
OP2
OP2
OP2
OP2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
.500*HCHO
.750*HCHO
. 840*HCHO
HO2
. 770*HCHO
H02
.800*HCHO
HO2
.5SO*HCHO
.250*HCHO
,890*HCHO
. S50*KET'
,750*HCHO
HCHO
. 500*ORA2
HCHO
. 700*DCB
HCHO
, 000*HO2
,J50Q*HCHQ
,*500*ORA2
.47S*CO
.750*HCHO
NO2
ALD
. 500*ORA2
. 770*ALD
.SOO*MO2
.410*ALD
. 500*M02
,460*ALD
.500*MO2
, 800*HCHO
.500*MO2
ALD
. 500*MO2
. 725*ALD
.SOO*HO2
MGLY
. 500*ORA2
.000*MO2
MO2
+
4-
4-
4-
+
+
4-
+•
4-
4-
4-
4-
4-
4-
+
+
+
4-
4-
+
+
4-
4-
4-
+
4-
+
4-
4-
4-
4.
4-
4-
4-
+
4-
4-
8-5
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
0
Q
0
0
0
6
0
0
0
i
0
0
0
0
0
0
0
0
0
4
.060*ORA1
.140*HO
.500»ALD
. 200*ORA2
.1QO*HO
HO2
HO2
.770*ALD
.410*ALD
.460*ALD .
.350*ALD
, 750*ALD
.725*ALD
, 750*MGLY
.500*HO2
.170*MGLY
.000*HO2
.4SO*MGLY
.445*GLY
.025*ACO3
XO2
.500*H02
.500*HO2
.260*KET
.500*ORA2
.750*KET
,500*ORA2
.390*KET
.500*ORA2
.600*ALD
, 500*ORA2
, 500»HCHO
.500*ORA2
,55Q*KET
, 500*MO2
.500*HO2
.170*HGLY
4-
4-
4-
4-
+
4-
4-
+
4-
4-
4-
4-
4-
+
+
+
+
4-
+
4-
4-
4-
+
4-
4-
4-
4-
4-
0
0
0
0
0
0
0
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.290*ORA2
.310*M02
,330*CO
,230*HO2 '-' ' " *
'"
:.' • ..'•'"•
.750*ALD ,
.260*KET
.750*KST
*l, '! " --!„
.390*KET ' ~.'
H02
HO2
'HO2
, .
! ' , I '. -I'1'
H02 : •{ ' ;•'
.SOO*MO2
.160*GLY
.806*DCB
,055*MGLY
,460*HO2
ALD " "
• ',
."500*MO2" - '*
.SOO*HO2
.500*H02
,500*HO2
,500*H02 ,'• ' '"•' . -"
.500*H02
.140*HCHO
,SOO*ORR2 :
.500*M02
.160*GLY
-------
EPA/600/R-99/030
Table 8A-7. RADM2 and RADM2_AQ Mechanisms
{145}
{146}
{147}
{148}
{149}
{150}
{151}
{152}
{153}
{154}
{155}
{156}
{157}
{158}
AC03
AC03
AC03
OLN
XO2
XO2
XO2
X02
X02
XN02
XN02
XN02
XN02
XN02
+ XYLP
+ TC03
+ OLN
+ OLN
+ HO2
+ MO2
+ AC03
+ X02
+ NO
+ N02
+ H02
+ M02
+ AC03
+ XN02
+ 0.700'>DCB
--> M02
+ H02
- - > " M02
+ 0.110»MGLY
+ 2.000'*X02
- - > HCHO
+ N02
--> 2.000*HCHO
--> OP2
- - > HCHO
- - > M02
-->
- - > NO2
--> ONIT
--> OP2
--> HCHO
--> M02
-->
+ H02
+ 0.450*MGLY
+ 0.920*HO2
+ 0.05Q*AC03
+ . ALD
+ 0.500*MO2
+ 2.000*ALD
+ H02
+ H02
+ 0.806*DCB
+ 0.890*GLY
+ 0.950*CO
+ 0.500*ORA2
+ 2.000*N02
Rate Expression
Rate Constant
1)
2)
3)
4)
5)
6)
7)
8)
9)
k(
k(
k(
k(
k(
k(
k(
k(
k(
k( 10)
k( 11)
k( 12)
k( 13)
k( 14)
k( 15)
k( 16)
k( 17)
k( 18)
k( 19)
k( 20)
k( 21)
k( 22)
k( 23)
k( 24)
k( 25)
k( 26)
k( 27)
k( 28)
k( 29)
k( 30)
k( 31)
kO
kinf
F =
k( 32)
k( 33)
k =
uses photo table NO2_RADM88
uses photo table O3O1D_RADM88
uses photo table O3O3P_RADM88
uses photo table HONO_RADM88
uses photo table HNO3_RADM88
uses photo table HNO4_RADM88
uses photo table NO3NO_RADM88
uses photo table NO3NO2_RADM88
uses photo table H2O2_RADM88
uses photo table HCHOmol_RADM88
uses photo table HCHOrad_RADM88
uses photo table ALD_RADM88
uses photo table MHP_RADM88
uses photo table HOP_RADM88
uses photo table PAA_RADM88
uses photo table KETONE_RADM88
uses photo table GLYform_RADM88
uses photo table GLYmol_RADM88
uses photo table MGLY_RADM88
uses photo table UDC_RADM88
uses photo table ORGNIT_RADM88
= 6.0000E-34 * (T/300)**(-2.30)
6.5000E-12
1.8000E-11
3.2000E-11
2.2000E-10
2.0000E-12
1.6000E-12
1.1000E-14
3.7000E-12
exp (
exp(
exp(
120.0/T)
110.0/T)
70.0/T)
exp( -1400.0/T)
exp( -940.0/T)
exp( -500.0/T)
exp( 240.0/T)
is a falloff expression using:
= 1.8000E-31 * (T/300)**(-3.20)
4.7000E-12 * (T/300)**(-1.40)
0.60, n = 1.00
= k( 31) / Keq, where Keq = 2.100E-27 *
is a special rate expression of the form:
kl + k2[M], where
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
scaled by l.OOOOOE+00
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOB+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{O.OOOOOE+00}
{6.09302E-34}
{9.72293E-12}
{2.60365E-11}
{4.04730E-11}
{2.20000E-10}
{1.82272E-14}
{6.82650E-14}
{2.05452E-15}
{8.27883E-12}
{1-39058E-12}
exp( 10900.0/T)
{8.62399E-02}
{3.01634E-12}
8-55
-------
EPA/600/R-99/030
Table 8A-7, RADM2 and RADM2_AQ Mechanisms
kl « 2.200QE-13 * exp{ 620.0/TS
k2 = 1.900QE-33 * expf 980.0/T)
k( 34) is a special rate expression of the form:
k = kl + k2[Ml\ where
kl = 3.0800E-34 * expf 2820.0/T)
k2 = 2.66001-54 * exp( 3180.0/T)
k( 35) = 3.3000E-12 * exp( -200.0/TS
k( 36) is a falloff expression using:
kO = 7.0QOOE-31 * (T/300)**(-2.60)
kinf = 1.5QOOE-11 * (T/300)**(-0.50)
0.60, n = 1.00
3.3000E-39 * exp( 530.0/TJ
1.4000E-13 * expf -2500.0/T)
exp{ ISO. 0/T)
exp( -1230.0/T)
P =
k( 37}
kE 38)
k( 39!
k{ 40)
k( 41)
k( 42)
1.1001-27 * exp( 11200.0/T)
= 1.7000E-11
» 2.50001-14
= 2.5000E-12
is a falloff expression using:
kO = 2.2000E-30 * (t/300)**(-4.30)
kinf = l.SOOOE-12 * (T/300)**(-0.50!
p = 0.60, n = i.oo
k( 43! = kf 42) / Keq, where Keq
k( 44) = 2.0000E-21
k( 455 is a falloff expression using:
kO = 2.6000E-30 * (T/300)**(-3.20)
kinf = 2.4QQOE-11 * (T/300)**(-!.30)
F = 0.60, n = 1.00
k{ 46) is a special rate expression of the form:
k = kO + {k3[M] / fl + k3[M]/k2)}, where
kO = 7.2000E-15'* exp! 785.0/f!
4.1QOOE-16 * expf 1440.0/T)
1.9000E-33 * exp( 725.0/T)
1.3000E-12 * exp( 380.0/T)
4.60001-11 * exp( 230.0/T)
a falloff expression using;
3.0000E-31 * '(T/300)** (-3.30}
l.SOOOE-12 * (T/300)**( 0.00)
60, n - 1.00
1.5000E-13 * (1.0 -i- 0.6*Pressure)
(T/300)*»( 2.00) * expf -1280.0/T)
(T/300)**C 2.00)
expf -540.0/T)
exp( -380.0/T)
exp( -380.0/T)
expf 411.0/T)
expf 504.0/T)
expf 549.0/T)
exp( 322.0/T)
exp{ 116.0/T)
k2
k3
k< 47) =
k( 48! =
k( 49! is
kO
kinf =
F = 0
k( 50) =
k( 51) =
k( 525 =
k{ S3) =
k( 54) =
kf 55) =
k( 56) =
kf 57) =
k( 58! =
kS 59! =
k{ 60! =
k! 61) =
k( 62) =
k( 63) =
k( 64) =
k< 65) =
k( 66! =
kf 67) =
k{ 68) =
k( 69) =
k( 70) =•
k! 71) =
k{ 72) =
kf 73) =
2.8300E+01
1.2330E-12
1.5900E-11
1.7300E-11
3.6400E-11
2.1SOOE-12
5.3200E-12
.0700E-11
.1000E-12
.89001-11
.OOOOE-11
.OOOOE-01
.OOOOE-12
6.8700E-12
1.2000E-11
.1500E-11
.70001-11
.8000E-11
.OOOOE-11
.OOOOE-11
.OOOOE-11
6.16501-13
l.SSOOE-11
exp( -444.0/T)
* k{ 61)
* expf
* expf
256.0/T)
-745.0/T!
(T/300)**( 2,00!
expf -540.0/T)
expf -444.0/T!
{6.78905E-30}
{1.68671E-12}
{4.87144E-12}
{1.95397E-38}
{3,18213E-17}
{2.81225E-11}
{4.03072E-16}
{2.SOOOOS-12}
{1.26440E-12}
{S.47034E-02}
{2.00000B-21}
{1.14885E-11}
{1.47236S-13}
{4.65309E-12}
{9.95294E-11}
{8.88848E-13}
{2.40000E-13}
{3.80672E-01}
{2.74210E-13}
(2.596691-12}
{4.83334E-12}
{1.01696E-11}
{8.53916E-12}
{2.88684E-11}
{6.75269E-11}
{6.18715E-12}
{2.78943E-11}
{4.00000E-11}
{3.60000E-11}
{9.00000E-12}
{1.62197E-11}
{9.85020E-13}
{1.1SOOOB-11}
{1.70000E-11}..
{2.80000E-11}
{l.OOOOOE-ll}
{l.OOOOOE-ll}
{l.OOOOOE-ll}
{1.3710SE-13}
{2.53137E-12}
8-56
-------
EPA/6QO/R-99/030
Table 8A-7, RADM2 and RADM2_AQ Mechanisms
M 74)
kC 75}
fc! 76}
k( 77)
k( 78)
k( 79)
k( 80)
Jc( 81)
k{ 82)
k( 83)
k( 84)
k( 85)
kC 86}
k{ 87)
k! 88)
k{ 89)
k( 90)
k( 91)
k( 92)
k( 93}
k( 94)
k( 95)
k{ 96)
k{ 97)
k{ 98)
k( 99)
kClOOS
kflQl!
k{102)
k{103)
k(104)
k(105)
k(l06)
k(107)
k(108)
kf!Q9)
kSHO)
Mill)
M112)
k(H3)
k(114)
k(H5)
k(ll6)
k(H7)
k(118)
k(119)
k{120)
M121)
kf!22!
k{123)
k(124)
k{125)
k{126)
k(1275
k{128)
k(129)
k(130)
M131)
k(132)
k(133)
k(X34)
_
=
.
=
=
_
=
=
m
«=
=
SI
=
=
=
=
=
a
=
=
=
=
ss
ss
s
=
=
=
B
-
=
=
=
=
=
0
-
SS
B
=
=
B
=
SS
B
=
=
=
=
=
=
=
ss
ss
ss
B
=
=
ss
=
=
2
2
1
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
1
6
1
1
2
2
1
3
5
1
1
7
1
7
7
7
7
7
7
7
7
7
7
7
7
7
7
1
1
4
3
2
1
1
1
1
9
1
1
9
1
.55QQE-11
.8000E-12
.9500E+16
.7000E-12
.9500E+16
•2000E-12
.2QOOE-12
.2000E-12
.2QOOE-12
.2QQ0E-12
.2000E-12
.2QOOE-12
.2000E-12
.20QOE-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.OOOOE-13
.4000E-12
.OOOOE-13
.4000E-12
.4000E-12
.2000E-11
.0000E-12
.OOOOE-11
.2300E-11
.8100E-13
.2000E-14
.3200E-14
.2900E-15
.2300E-14
.7000S-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7QQOE-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.9000E-13
.4000E-13
.2000E-14
.4000E-14
.9000E-14
.4000E-13
.4000E-13
.7000E-14
.7000E-14
.6000E-13
.7000E-14
.7000E-14
.6000E-13
.7000E-14
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
it
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
ir
it
it
*
*
*
it
*
*
*
*
*
*
if
expf
exp(
exp(
expf
exp (
exp!
exp(
expf
expf
exp(
expf
expC
expf
exp(
exp(
exp (
exp(
exp(
exp{
exp(
exp(
exp(
exp(
exp{
expS
exp!
expf
exp(
exp(
exp(
exp(
exp!
exp(
exp(
exp{
exp(
exp{
expC
expE
exp{
exp(
exp (
expf
expf
exp f
expf
expf
expf
expf
expf
expf
expf
expf
expf
expf
expf
expf
expf
409
181
-13543
-13543
180
180
180
180
180
180
180
180
180
180
180
180
180
180
-2058
-1900
-2058
-1900
-1900
-2923
-1895
-975
-2633
-2105
-1136
-2013
1300
1300
1300
1300
1300
1300
1300
13QO
1300
1300
1300
1300
1300
1300
220
220
220
220
220
220
220
220
220
220
220
220
220
220
,0/T)
,0/T)
,0/T)
.0/T)
,0/T)
,0/T)
,0/T)
,0/T)
-0/T)
.0/TS
.0/T!
,0/T)
,0/T)
,0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
.0/T)
.0/T)
.0/T)
,0/T)
,0/TS
,0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
,0/T!
,0/T)
,0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{1.
{5.
{3.
{4.
{3-
{7.
{7.
{7.
{7.
{7.
{7.
{7.
{7.
(7.
(7.
{7.
{7.
{7.
{7.
{6.
{2-
{6.
{2.
{2.
{2.
{1.
(1.
(1.
{5.
{1-
{!•
{!•
{1.
{6-
{6.
{6.
(6.
{6.
{6.
{6.
(6.
(6.
{6.
{6.
{6-
{6-
{6.
{3.
(2-
{8.
{7,
{6.
{2.
{2.
{3.
{3.
{2.
{3.
{3.
{2.
{3.
00601E-10} '
13974E-12}
57235E-04}
70000E-12}
57235E-04}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12)
68378E-12}
68378E-12)
68378E-12)
68378E-12)
68378E-12}
68378E-12}
68378E-12}
01030E-16}
38307E-15} :
01030E-16)
38307E-1S}
38307E-15J
20000E-11}
09940E-16}
73099E-14}
225391-12}
81000E-13}
74559E-18}
12933E-17}
61125E-16}
43295E-17}
04038E-12)
04038E-12}
04038E-12}
04038E-12J
04038E-12J
04038E-12}
04038E-12)
04038E-12}
04038E-12}
04038E-12}
04038E-12}
04038E-12J
04038E-12}
04038E-12}
97533E-13}
92919E-13)
78758E-14)
11376E-14}
06762E-14}
92919E-13}
92919E-13}
55688E-14}
55688E-14}
00859E-12}
55688E-14}
55688E-14}
00859E-12}
55688E-14}
8-57
-------
EPA/60G/R-99/030
Table 8A-7. RADM2 and RADM2_AQ Mechanisms
k(13S) m
k(136) =
k(137) .
k(138) »
k(139) -
k(140) =
k(141) =
k{142! m
k(143) =
k!144) «
kC14S) =
k(146) =
k{147) =
k(148) =
k{149) =
k(150) =
k(151) =
k(152) m
k(153) o
k(154) -
k(155! =
k(156) =
k(157) D
k(158) =
3
1
8
7
3
3
4
4
1
4
4
1
4
3
7
1
4
3
4
4
7
1
4
3
.400QE-13
.OOOOE-13
.4000E-14
.2000E-14
.4000E-13
.4000E-13
.2000E-14
.2000E-14
.1900E-12
.2000E-14
.2000E-14
.1900E-12
.2000E-14
.6000E-16
.7000E-14
.7000E-14
.2000E-14
.6000E-16
.2000E-12
.2000E-12
.7000E-14
.7000E-14
.2000E-14
.6000E-16
*
it
*
*
if
ic
*
ik
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
exp(
exp (
exp!
exp!
exp!
exp (
exp!
exp!
exp!
expC
exp(
exp(
expC
exp{
exp!
exp!
exp!
exp!
exp!
exp!
exp !
exp!
exp!
exp!_
220
220
220
220
220
220
220
220
220
220
220
220
220
220
1300
220
220
220
180
180
1300
220
220
220
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{7
{2
{1
{1
{7
{7
{8
{8
{2
{8
{8
(2
{8
{7
{6
{3
{8
{7
{7
(7
{6
{3
{8
{'
.11376E-13} "-
.092288-13} '•'. ' .:
.7S7S2E-13} (:,r . _ , ,;•
.506446-13} •'-'' ' " . '""'
. 11376E-13} !'=-> • , ••
.H376E-13} ::;
.78758E-14} "
.78758E-14} •"
.48981E-12) '•"•
.787581-14}™
.787581-14}
.489811-12}
.787S8E-14}
.53221E-16}
.04038E-12}
.556888-14}
.787S8E-14}
.532211-16} ||M| , ",,
.68378E-12}
.68378E-12}
.04038E-12}
,556881-14}
.787S8E-14}
.532211-16} ;
8-58
-------
EPA/600/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisriis
Reaction List
{ 1}
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
{ 8}
{ 9}
{ 10}
{ 11}
{ 12}
{ 13}
{ 14}
{ 15}
{ 16}
{ 17}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40}
{ 41}
{ 42}
{ 43}
{ 44}
{ 45}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
{ 54}
{ 55}
{ 56}
{ 57}
NO2
03
O3
HONO
HN03
HN04
N03
N03
H202
HCHO
HCHO
ALD
OP1
OP2
PAA
KET
GLY
GLY
MGLY
DCS
ONIT
03 P
03 P
O1D
O1D
01D
03
03
03
H02
H02
HN04
HO2
H02
H202
NO
NO
03
NO3
NO3
N03
N03
N2O5
N2O5
HO
HO
HO
HO
HO
CO
HO
ETH
HC3
HC5
HC8
OL2
OLT
+ hv
+ hv
+ hv
+ hv '
+ hv
+ hv
+ hv
+ hv
+ hv • •
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ [M]
+ N02
+ [N2]
+ [02]
+ [H20]
+ NO
+ HO
+ HO2
+ NO
+ NO2
+ HO2
+ H02
+ HO
+ HO
+ NO
+ N02
+ NO
+ NO2
+ HO2
+ N02
+ [H20]
+ NO2
+ HN03
+ HNO4
+ HO2
+ S02
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
--> 03P
--> 01D
--> 03P
- - > HO
- - > HO
- - > HO2
--> NO
- - > N02
--> 2.000*HO
- - > CO
- - > H02
- - > MO2
--> HCHO
- - > ALD
- - > M02
- - > AC03
--> 0.130*HCHO
--> 0.450*HCHO
• • • --> AC03
--> 0.980*H02
--> 0.200*ALD
+ NO2
•+• [02] --> 03
--> NO
--> 03P
--> 03P
--> 2.000*HO
--> NO2
- - > H02
--> HO
- - > N02
- - > HNO4
- - > H02
- - > H202
+ [H2O] --> H2O2
- - > H02
- - > HONO
+ [O2] --> 2.000*NO2
- - > NO3
--> 2.000*N02
--> NO
- - > HNO3
--> N2O5
- - > N02
--> 2.000*HNO3
- - > HN03
- - > N03
- - > NO2
-->
--> SULF
- - > H02
--> M02
- - > ETHP
--> 0.830*HC3P
+ 0.075*ALD
--> HC5P
--> HC8P
--> OL2P
- - > OLTP
+ NO ' ••..-•
+ ' NO ' •
+ NO2
+ N02
+ 03P
+ HO2 + CO
+ HO2 + CO
+ H02 + HO
+ H02 + HO
+ HO '
+ ' ETHP
+ l."870*CO
+ 1.550*CO + 0.800*H02
+ H02 + CO
+ 0.020*AC03 + TC03
+ 0.800*KET + HO2
+ HO
+ N02
+ NO2
+ N03
+ HO2 + SOLAER
+ 0.170*HO2 + 0.009*HCHO
+ 0.025*KET
+ 0.250*XO2
+ 0.750*X02 '+ HC8AER
8-59
-------
EPA/600/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
{ 58}
{ 39}
{ 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}
{ 85}
{ 86}
{ 87}
{ 88}
{ 89}
{ 90}
{ 91}
{ 92}
{ 93}
{ 94}
{ 95}
{ 96}
{ 97}
{ 98}
{ 99}
{100}
{101}
{102}
{103}
OLI
TOL
XYL
CSL
CSL
HCHO
ALD
KET
GLY
MGLY
DCB
DPI
OP2
PAA
PAN
ONIT
ISO
ACO3
PAN
TCO3
TPAN
M02
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
ACO3
TC03
TOLP
XYLP
ETHP
KETP
OLN
HCHO
ALD
GLY
MGLY
DCB
CSL
OL2
OLT
OLI
ISO
OL2
+ HO
+ HO
+ HO
+• HO
+ HO
+ .HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ Hp
+ HO
+ 4> :
+ NO2
4- NO2
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
f NO , ,
+ NO
+ NO
+ NO
+ NO
•f NO3
+ NO3
+ N03
+ NO3
+ NO3
+ NO3
r
-1- lfO3
+ NO3
+ IJO3'
+ NO3
+ O3
__>
,,. --». °
+
--> o
. +
--> o
+
-->
-->
-->
-->
-->
-->
-->
--_> 0
--> 0
-->
-->
-->
— >
- - >
-->
-->
-->
-->
--> 0
+ 0
--> 0
•f 0
--> ' 0
+ 0
--> 1
+ . 0
-->.
+
--> '
, -f- 0
- - >.
-->
. . + 0
, . ' -' * 2
.'v ""> ',
+* 0
-->
+ 0
-->
-->
-->
. -->
-->
-->
-->
-->
, -,->
+ 0
.... ... •">
-->
-->
/..,...--»
• '•-. ---->
+ 0
.OLIP
.750*TOLP
TOLAER
.830*XYLP
XYLAER
.100*H02
CSLAER
CSL
H02
ACO3
KETP
, HO2
ACO3
TC03
.500«MO2
.SOO«HC3P
AC03
HCHO
,HC3P
OLTP
.PAN
ACO3
' TPAN
TC03
HCHO
.7SO*ALD
.036*ONIT
,380*ALD
,920»NO2
,350*ALD
.240*ONIT
.600*HCHO
.20,0*ALD
ALD
N02
HO2
.100*KET
M02
N02
.110*MGLY
,000*XO2
N02
,160*GLY
N02
.806*DCB
ALD
MGLY
HCHO
HO2
ACQ3
HNO3
HN03
HN03
HNO3
.500*CSLA1R
OLN
OLN
OLN
OLN
HCHO
,120*HO2
+
.+ o
+ 0
+ 0
+
•«• 2
+
+ 0
+ 0
+
+
•f
+
+
4- 0
+ 0
+ 0
•f 0
+ 1
+ 0
+
+
+ ^
+
+
+ 0
+ 0
-t-
+ 0
+
+
+
+
+
+
+
+
+
+
+
-i- 0
OLIAEE
.250*CSL
.170*CSL
.900*XO2
CO
.ooo*co
CO
.500*HCHO
.SOO*ALD
NO3
NO2
NO2
N02
H02
.250*KET
.9S4*NO2
.690*KET
.920*HO2
.060*KET
.760*NO2
H02
HCHO
.450*ALD
NO2
N02
.920*HO2
.050*ACO3
HO2
.70.0*DCB
HO2
HO2
NO2
ALD
HHO3
HNO3
HO2
ACO3
TCO3
XNO2
OLIAER
,400*ORA1
"i'\ :ti -*ii
+ 0.250*HO2
M • :' -i
+ 0.170*HO2 • ;-
::, 1 " .!: J
+ 0.900*TCO3
+ O.SOO*HO
+ o.sbo*Hb
+ X02
?.- ^1 .' ' _•
•1 ' '
«- . -1 , • •
+ N02
+ 0.090*HCHO
+ 0.964*HO2
+ 0.080*ONIT
-t- 0.040*HCHO
•f 0.760*HQ2
+ - NQ2
" ' ' » !
+ H02
i • . . , :
-f- 0.280*HCHO
+ 0.890*GLY
•f 0.950*CO
+ 0. 1,7,0 *(^8LY • v,
1 . '.'.
+ 0.450*MGLY
+ NO2
+ HO2
+ 2.000*NO2
+ CO
+ 2.000*CO
+ CO
+ 0,5QO*CSL
+ 0,420*CO
8-6Q
-------
EPA/600/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
{104}
{105}
{106}
{107}
{108}
{109}
{110}
{111}
{112}
{113}
{114}
{115}
{116}
{117}
{118}
{119}
{120}
{121}
{122}
{123}
{124}
{125}
{126}
{127}
{128}
{129}
{130}
{131}
{132}
{133}
{134}
{135}
{136}
{137}
{138}
{139}
{140}
{141}
OLT
OLI
ISO
H02
HO2
H02
HO2
H02
HO2
HO2
HO2
H02
HO2
H02
HO2
HO2
HO2
MO2
M02
MO2
MO2
MO2
MO2
M02
M02
MO2
MO2
M02
MO2
MO2
MO2
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
+ O3
+ O3
+ 03
+ MO2
+ ETHP
+ HC3P
+ HC5P
+ HC8P
+ OL2P
+ OLTP
+ OLIP
+ KETP
+ ACO3
+ TOLP
+ XYLP
+ TC03
+ OLN
+ M02
+ ETHP
+ HC3P
+ HC5P
+ HC8P
+ OL2P
+ OLTP
+ OLIP
+ KETP
+ AC03
+ TOLP
+ XYLP
+ TC03
+ OLN
+ AC03
+ AC03
+ AC03
+ AC03
+ AC03
+ ACO3
+ AC03
--> 0
+ 0
.+ 0
--> 0
+ 0
+ 0
+
--> 0
+ 0
+ 0
-->
-->
-->
-->
-->
-->
-->
-->
-->
-->
^ ->
-->
-->
-->
--> 1
--> 0
--> 0
+
--> 0
+
--> 0
+
--> 1
--> 1
--> 0
+ 0
--> 0
-->
+ 0
-->
+ 0
-->
+ 2
--> 0
+ 0
+ 0
--> 1
+
-->
+ 0
--> 0
+ 0
--> 0
+ 0
--> 0
+ 0
--> 0
+ 0
-->
+ 0
--> 0
.530*HCHO
.200*ORA1
.220*M02
.180*HCHO
.230*CO
.260*H02
OLIAER
.S30*HCHO
.200*ORA1
.220*M02
OP1
OP2
OP2
OP2
OP2
OP 2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
.500*HCHO
,750*HCHO
.840*HCHO.
HO2.
.770*HCHO
H02
,800*HCHO
H02
. 550*HCHO
,250*HCHO
,890*HCHO
.550*KET
.750*HCHO
HCHO
.500*ORA2
HCHO
,700*DCB
.HCHO
,000*H02
,500*HCHO
.500*ORA2
.475*CO
,750*HCHO
NO2
ALD
.500*ORA2
.770*ALD
,500*M02
.410*ALD
.500*M02
.460*ALD
.500*M02
.800*HCHO
.500*MO2
ALD
.500*M02
.725*ALD
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+
+
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 2
+ 0
+ 0
+ 0
+
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
+ 1
+ 0
+ 0
+ 0
+ 0
+ 0
+ 0
-500*ALD
,200*ORA2
-100*HO
.72Q*ALD
.060*ORA1
.140*HO
,500*ALD
,200*ORA2
.100*HO
HO2
HO2
,770*ALD
,410*ALD
.460*ALD
,350*ALD
.75Q*ALP
.725*ALD
. 750*MGLY
.500*HO2
.0,70*MGLY
. 000*HO2
,450*MGLY
,445*GLY
.025*ACQ3
XQ2
,500*HO2
.500*HQ2
,260*KET
.500*ORA2
.750*KET
,500*QRA2
,390*KET
,500*ORA2
,600*AI
-------
EPA/600/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
+
{142} KETP + ACO3 -->
+
{143} AC03 + AC03 -->
{144} ACO3 + TOLP -->
, +
{145} ACO3 + XYIiP -->
+
{146} AC03 + TCG3 -->
*
, +
{147} ACQ3 + OLN -->
+
{148} Ot|N + qtH,s . -->
{149} XO2 + IfO2 7 -->
{150} XO2 + l»|O2 -:>
{151} XO2 + ACO3 -->
{152} X02 + XO2 -->
{153} XO2 + NO -<•>
{154} XNO2 + NO2 -->
{155} XNO2 + HO2 -->
{156} XNO2 + MO2 -->
{157} XN02 + AC03 -->
{158} XNO2 + XNO2 -->
{159} TERP + HO -->
{160} TERP + N03 -->
{161} TERP + O3 -->
Rate Expression
k( 1) uses photo table N02JRADM88
k( 2) uses photo table O3olD_RADM88
k( 3) uses photo table O3O3P_RADM88
k( 4) uses photo table HONO_RADM88
k( 5) uses photo table HNO3_RADM88
kt 6) uses photo table HN04_RADM88
k( 7} uses photo table NO3NO_RADM88
kS 8) uses photo table NO3NO2_RADMS8
k( 9S uses photo table H2O2_RADM88
k( 10! uses photo table HCHOmol_RADM88
k{ 11) uses photo table HCHOrad_RADM88
k( 12) uses photo table AtD_RADM88
k( 13) uses photo table MHP_RADM88
k( 14) uses photo table HOP~RADM88
kS 15! uses photo table PAA_RADMaa
kS 16! uses photo table KETONE_RADM88
k( 17) uses photo table GtYfonnJRADMBS
k( 18) uses photo table GIiYmol_RADM88
k( 19) uses photo table MGLY_RADMB8
k( 20) uses photo table UDC_RADM88
k! 21) uses photo table QRGNITJRADMBB
k( 22) = 6.0000E-34 * (T/300) ** S-2 .30!
k( 23! » 6.SOOOB-12 * exp { 120. 0/T!
k( 24) m 1.8000E-11 * exp( 110. 0/T)
k( 25) « 3.2000E-11 * exp( 70. 0/T)
k( 26) = 2.2000E-10
k( 27) m 2.0000E-12 * exp ( -1400. 0/T)
k( 28) = 1.6000E-12 * expS -940. 0/T!
kS 29! = 1.1000E-14 * expS -500. 0/T)
0.500*H02 + 0.500*MO2 + 0.500*ORA2
MGLY + 0.500*HO2 + 0.500*MO2
0.500*ORA2 - ;;. :
2.000*M02
MO2 + 0.170*MGLY + 0.160*GLY
0.700»DCB + H02
MO2 + 0.450*MGLY + 0.806*DCB •
H02 ' "''. '..'.
M02 + 0.920*HO2 + 0.890*GtY
0.110*MGI.Y + 0.050*AC03 + 0.950*CO
2.0PO*XO2
HCHO + ALD + O.SOO*ORA2
NO2 + O.SOO*MO2
2.000*HCHO + 2.000*ALD 4- 2.000*HO2 "
CJP2 ' ' \''~
HCHO + H02 '
MO2 '- - ' -
N02
ONIT
OP2 v
HCHO + HO2
MO2
. ' ! ' - ' -;i ' '::
TERPAER + HO
TERPAER + H03
TERPAER + O3 r ..
Rate Constant
, scaled by l.OOOOOE+00 {O.OOOOOE+00};
, scaled by l.OOOOOE+00 {O.OOOOOE+OO} [> ,
, scaled by LOOOOOE+OO {O.OOOOOE+00}: :
, scaled by l.OOOOOE+00 {0,OOOOOE+00}
, scaled by l.OOOOOE+00 {0 . OOOOOE+00}
, scaled by l.OOOOOE+00 {0 . OOOOOE+OOJ
, scaled by l.OOOOOE+00 {O.OOOOOE+00}
, scaled by l.OOOOOE+00 {O.OOOOOE+00}
, scaled by l.OOOOOE+00 {O.OOOOOB+00}4 ,
, scaled by l.OOOOOE+00 {O.OOOOOB+00}
, scaled by l.OOOOOE+00 { 0. OOOOOE+00 }-
, scaled by l.OOOOOE+00 {0. OOOOOE+00 }
, scaled by l.OOOOOE+00 { 0 . OOOOOE+00}
, scaled by l.OOOOOE+00 {0. OOOOOE+00}
, scaled by l.OOOOOE+00 {O.OOOOOE+00}*. »"
, scaled by l.OOOOOE+00 {O.OOOOOS+00}
, scaled by l.OOOOOE+00 {O.OOOOOB+00}
, scaled by l.OOOOOE+00 {0 . OOOOOE+00 }
, scaled by l.OOOOOE+00 {O.OOOOOE+00}
, scaled by l.OOOOOE+00 {O.OOOOOB+00}
, scaled by l.OOOOOE+00 {0,OOOOOB+OOJ
{6. 093021-34}-
{9.72293B-12}
{2.6036SE-11}
{4.04730E-11}
{2.20000E-10}
{1.82272E-14}
{6.82650B-14}
{2.05452E-1S}
8-62
-------
EPA/600/R-99/03Q
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
k{ 30) = 3.70001-12 * exp( 240,0/TS
k( 31) is a falloff expression using:
kO = 1.80001-31 * (T/300!**(-3.20)
kinf = 4.7000E-12 * (T/300)**(-1.40!
F = 0.60, n = 1.00
k{ 32) = k{ 31) / Keq, where Keq - 2.100E-27 * exp{ 10900.0/T)
k( 33) is a special rate expression of the form:
k = kl + k2 [M] , where
kl = 2.2000E-13 * exp! ' 620.0/T)
k2 = 1.9000E-33 * exp( 980.0/T)
k( 34) is a special rate expression of the form:
k = kl + k2 [M] , where
kl = 3.0800E-34 * exp( 2820.0/T)
k2 = 2.6600E-54 * exp ( 3180.0/T)
k( 35) = 3.3000E-12 * exp( -200.0/T)
k( 36) is a falloff expression using:
kO
kinf =
F
k{ 37) =
k( 38) =
k( 39) =
k( 40) =
k( 41) =
k! 42) is
kO
kinf =
7.0000E-31 * (T/30Q!**(-2.60!
1.5000E-11 * (T/300)**(-0.50)
0.60, n = 1.00
3.3000E-39 * exp! 530.0/T)
1.4000E-13 * exp! -2500.0/T)
exp( 150.0/T)
exp( -1230.0/T)
1.100E-27 * exp! 11200.0/T)
1.7000S-11
2.5000E-14
2.SOOOE-12
a falloff expression using: -•
2.2000E-30 * (T/300)**(-4.30)
1.5000E-12 * (T/300)*»(-0.50)
F = 0.60, n = 1.00
k( 43) = k( 42) / Keq, where Keq =
k( 44) = 2.00001-21
k( 45) is a falloff expression using:
kO =. 2.6000E-30 * (T/300)**(-3.20}
kinf = 2.4000S-11 * (T/300!**(-1.30)
F = 0,60, n = 1.00
k( 46) is a special rate expression of the form:
k = kO 4 {k3 [M] / (1 + k3[M]/k2)}, where
kO = 7.2000E-15 * exp! 785.0/T)
k2 = 4.1000E-16 * exp! 1440.0/T)
k3 = 1.9000E-33 * exp! 725.0/T)
k! 47) = 1.3000E-12 * exp( 380.0/T)
4.60001-11 * exp! 230.0/T)
is a falloff expression using:
= 3.0000E-31 * (T/300)**(-3.30)
1.5000E-12 * {T/300)**( 0.00)
0.60, n.= 1.00
» 1.5000E-13 * (1.0 4 0.6*Pressure)
(T/300)**( 2.00) * exp( -1280.0/T)
(T/300)**( 2.00)
exp( , -540.0/T!
k( 48! =
k( 49)
kO
kinf =
F =
k( 50!
k( 51)
k( 52)
k! S3!
k( 54)
k( 55)
k( 56)
k! 57)
k( 58!
k! 59)
k! 60!
k( 61!
k( 62)
k( 63)
k( 64)
fc( 65)
= 9
2.8300E+01
1.2330E-12
1.S900E-11
1.7300E-11
3.6400E-11
2.1500E-12
5.3200E-12
1.0700E-11
2.1000E-12
1.8900E-11
4.0000E-11
9.0000E-01
OOOOE-12
6.8700E-12
1.2000E-11
exp! -444.0/T)
* exp!
* exp(
* exp!
* exp!
* exp!
* exp (
* exp!
* k( 61)
* exp!
* exp(
-380.0/T)
-380.0/T)
411.0/T)
504.0/T)
549.0/T)
322.0/T)
116.0/T)
256.0/T)
-745.0/T)
{8.27883E-12}
{1.390S8E-12}
{8.62399E-02}
{3.01634E-12}
{6.78905E-30}
{1.68671E-12}
{4.87144E-12}
{1.35397E-38}
{3.18213E-17}
{2.812251-11}
{4.03072E-16}
{2.50000E-12}
{1.264401-12}
{5.47034E-02}
{2.000001-21}
{1.148851-11}
{1.47236E-13}
{4.65309E-12}
{9.952941-11}
{8.888481-13}
{2.400001-13}
{3.806721-01}
{2.74210E-13J
{2.596691-12}
{4.83334E-12}
{1.016961-11}
{8.53916E-12J
{2.88684E-11}
{6.75269E-11}
{6.187151-12}
{2.78943E-11}
{4.0000QE-11}
{3.600001-11}
{9.000001-12}
{1,621971-11}
{9.850201-13}
8-63
-------
EPA/600/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
kt 66)
k( 67}
k( 68)
kt 69)
k( 70!
k{ 71)
k{ 72)
k{ 73)
k{ 74)
k( 75!
k( 76)
k( 77)
k{ 78)
k! 79)
k( 80)
k! 81)
k( 82)
k! 83)
k( 84)
kC 85!
k( 86)
kC 87!
k( 88)
kC 89!
k( 90)
k{ 913
k! 92)
k( 93)
k( 94)
k( 95)
k{ 96)
k( 97)
k{ 98)
k( 99)
kClOO)
kdoi)
k(102!
k(103)
k(104)
k(105)
ksioe)
k(107)
kSlOS!
k(109)
k(110)
k(lll)
k(112)
k(H3)
k(H4)
k(llS)
k(116)
k(U7)
ktllS)
k(119!
k(120)
k!12l!
k(122)
kd23!
k(124)
k(125)
k(126)
S3
SB
-s,
BS
=
S£
=
ss
C=
ss
=
ss
SS
at
-c
ss
m
m
&
XX
m
as
=
ss,
ss
=
m
=
=
0
=
m
zx
K
=
s
=
s
m
=
=
=
=
*
=
=
.
=
=
m.
.
=
M
=
=
K
=
=
1
1
2
1
1
1
6
1
2
2
1
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
1
6
1
1
2
2
1
3
5
1
1
7
1
7
7
7
7
7
7
7
7
7
7
7
7
7
7
1
1
4
3
2
1
.1500S-11
.7000E-11
.8000E-11
.OOOOE-11
.OOOOE-11
.OOOOE-11
.16SOE-13
.5500E-11
.5500E-11
.8000E-12
.9500E+16
.70QQE-12
.9500E+16
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.200QE-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.OOOOE-13
.4000E-12
.OOOOE-13
.4000E-12
.4000E-12
.2000E-11
.OOOOE-12
.OOOOE-11
.2300E-11
.8100E-13
.2000E-14
.3200E-14
.2900E-15
.2300E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.70QQE-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000S-14
.7000E-14
.900QE-13
.4000E-13
.2000E-14
.4000E-14
.9000E-14
.4000E-13
!
* (T/300)**(
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*•
*
*
*
*
*
*
*
*
*
*
*
«
if
*
*
*
*
*
#
#
*
exp!
exp(
exp{
exp(
exp(
exp{
exp(
exp(
exp(
expS
exp(
exp(
exp(
expS
expt
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp{
expS
exp(
exp!
exp(
expt
exp<
exp (
expS
exp (
expC
exp (
expC
exp(
exp{
exp {
expE
exp!
exp!
exp (
expt
expt
expf
expt
expt
exp!
exp!
expt
exp!
-540
409
181
-13543
-13543
180
180
180
180
180
180
180
180
180
180
180
180
180
180
-2058
-1900
-2058
-1900
-1900
-2923
-1895
-975
-2633
-2105
-1136
-2013
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
220
220
220
220
220
220
2.00! * exp! -444
.0/1)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
.0/T)
.0/T)
.0/T)
.0/T!
• 0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
{1
{1
{2
{1
{1
{1
.0/T! {1
{2
(1
{5
(3
{4
{3
{7
{7
{7
{7
{7
{7
{7
{7
{7
{7
{7
{7
{7
{7
(6
{2
(6
{2
{2
(2
{1
{1
{1
{5
{1
{1
{1
{1
{6
{6
{6
{«
{6
{6
{6
{e
{6
(6
{6
{6
{6
{6
(3
{2
(8
{7
{6
{2
.15000E-11)
.70000E-11}
.800008-11}
.OOOOOE-ll}
.OOOOOE-llJ
.OOOQOE-ilj "
.371051-13} ,
.53137E-12}
.006018-10}
.13974E-12}
.57235E-04}
.700003-12}
.57235E-04} '
.68378E-12}
,6a378E-12}
.583781-12} ::r
.68378E-12}
.68378E-12}
.68378E-12}
.68378E-12) ,
.68378E-12} ; ' . „'
.683781-12} • . . . :
.68378E-12} i :;
.683781-12}
.68378E-12}
.68378E-12}
.68378E-12}
.01030E-16}
.38307E-1S}
.01030E-16),
.38307E-15}
.38307E-1S}
.20000E-11} ,
.09940E-16} '!••' •••(
.730995-14} *;
.22539E-12} •
.810008-13}. .
.74559E-18}
.12933E-17} ;
.6112SE-16}
.43295E-17} -
.04038E-12}
.04038E-12} V .
.040388-12}
.04038E-12J
.04038E-12}
.04038S-12}
.04038E-12}
.04038E-12}
.04038E-12}
.04038E-12} •
.04038E-12}
.04038E-12}
.04038E-12} -
.04038E-12}
.97533E-13} . ,
.92919E-13}
.78758E-14)
.11376E-14}
.OS762S-14}
.92919E-13}
8-64
-------
EPA/6QO/R-99/030
Table 8A-8. RADM2_AE and RADM2_AE_AQ Mechanisms
k(127)
k(128)
k{129)
k(130)
k(131)
k(132)
k<133)
k<134)
k(135)
k{136)
k(137)
k<138)
k(139)
k(140)
k<141)
k(142)
k(1435
k(144)
k(145)
k{146)
kS147)
k(148)
k(149)
k{150!
k{151)
k(1525
k(153)
k(154)
k(15S)
k{156)
k<157)
k{158)
k{159)
k{160)
k{161)
=
e
a
m
«
8=
a
=
=
a
sa
a
B
US.
m
a
SB
as
as.
a
a
a
as
=
a
=
a
a
a
•
=
=
a
a
a
1
1
1
9
1
1
9
1
3
1
8
7
3
' 3
4
4
i
4
4
1
4
3
7
1
4
3
4
4
7
1
4
3
1
1
1
.4000E-13
.7000.E-14
.70001-14
.600QE-13
.7000E-14
.7000E-14
.6000E-13
.70001-14
.4000E-13
.OOOOE-13
.40001-14
.2000E-14
.4000E-13
.4000E-13
.2000E-14
.20001-14
.1900E-12
.2000E-14
.2000E-14
.1900E-12
.2000E-14
.600QE-16
.7000E-14
.7000E-14
.20001-14
.60001-16
.2000E-12
.20QQE-12
.7000E-14
.70001-14
.2000E-14
.60QOE-16
.OOOOE+00
.OOOOE+00
.OOOOE+00
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
A
*
*
*
*
exp(
exp{
exp(
exp (
exp(
exp{
exp(
exp{
exp(
exp(
exp(
exp{
exp(
exp(
exp( .
exp{
exp(
exp(
exp (
exp{
exp(
exp(
exp(
exp{
exp(
exp (
exp(
exp{.
exp(
exp{
exp{
exp(
k( 58)
k<101)
k(105)
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
1300
220
220
220
180
180
1300
220
220
220
.0/T)
.0/T)
• 0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{2
{3
{3
{2
{3
{3
{2
(3
{7
{2
{1
{1
{'
{7.
{8
{8
{2
{8
{8
(2
{8
{7
(6
{3
{8
(7
{7
{7
{6
{3
{8
{7
{6
{1
{1
.92919B-13)
.556881-14}
.556881-14}
.008591-12}
.55688E-14}
,556881-14}
.00859E-12}
.555881-14}
.113761-13}
.09228E-13}
.75752E-13}
.506441-13}
.11376E-13}
.113761-13}
.787581-14}
.787581-14}
.48981E-12} .
.787581-14}
.78758E-14}
.489811-12}
.787581-14}
.53221E-16}
.040381-12}
.556881-14}
.78758E-14}
.532211-16}
.683781-12}
.683781-12}
.04038E-12}
.556881-14}
.78758E-14}
.53221E-1S}
.752691-11}
.225391-12}
.611251-16} .
8-65
-------
EPA/600/R-99/030
Table 8A-9. RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
Reac'
{ 1}
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
{ 8}
{ 9}
{ 10}
{ 11}
{ 12}
{ 13}
{ 14}
{ 15}
{ 16}
{ "}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40)
{ 41}
{ 42}
{ 43}
{ 44}
{ «}
{ 46}
{ 47}
< 48}
{ 49}
{ so}
{ 51}
f 52}
{ 53}
{ 54}
{ 55}
{ 56}
{ 57}
tion Li!
NO2
O3
O3
HONO
HNO3
HKO4
N03
NO3
H2O2
HCHO
HCHO
ALD
OP1
OP2
PAA
KET
QLY
GLY
MGLY
DCS
ONIT
O3P
O3P
O1D
O1D
O1D
O3
03
03
HO2
HO2
HNO4
HO2
HO2
H2O2
NO
NO
O3
NO3
NO3
NO3
NO3
N2O5
N2O5
HO
HO
HO
HO
HO
CO
HO
ETH
HC3
HC5
HC8
OL2
OLT
st ; •• ••
+ hv
+ hv
+ hv .
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
«• hv
+ hv
+ hv
+ hv
+• hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ [M] + [02]
+ N02
+ [N2]
+ [02]
+ [H20]
+ NO
+ HO
+• HO2
+ NO
+ NO2
,
+ H02
'+ HO2 + [H20]
+ HO
+• HO
+ NO f [O2]
t N02
+ NO
+ NO2
+ H02
+ N02
+ [H20]
+ NO2 . , ,
+ ,HNO3
+ HNO4
+ HO2
+ S02
+ HO
•f HO
+ HO
+ HO
+ SHO
+ ,HO . '
+ HO
--> < 03P
--> ' 6lD
--> O3P
— > HO
--> HO
. --> H02
--> NO
- - > NO2
--> 2.000*HO
- - > ."CO
--> HO2
--> • MO2
--> HCHO
--> ALD
--> M02
--> ACO3
--> 0.130*HCHO
--> 0.450*HCHO
--> . ACQ3
--> 0.980*HO2
--> 0.200*ALD
+ HO2
--> O3
--> NO
--> •• O3P
--> ' O3P
--> 2.000*HO
--> NO2
- - > HO2
--> • HO
--> N02
--> HNO4
--> H02
-->" H202
--> H202
— > HO2
--> HONO
--> 2.000*NO2 • •
--> N03
--> 2.000*N02
--> NO
- - > HN03
--> N2O5
--> N02
--> 2.000*HNO3
- - > HNO3
--> NO3
--> N02
_->
--> SULF
--> HO2
--> MO2
--> ETHP
--> 0.830*HC3P
+ 0.075*ALD
--> • HC5P
--> HC8P
--> ' OL2P
--> OLTP
•
- • .1
"
+ * NO' '
+ N02 '~ •
+ NO2 . ..
. . . . . . ..' . . ;..
+ O3P
. ' • • -f • ' f
• ";'
+ HO2 + I CO
* HO2 - + •»«" -CO'
•f \ . HO2 ' ft *'*. HO . |»
+ . " HO2 + ' "' HO -
+. ' . HO. . . ' • ... ,^' . , . :
+ ETHP
t .1.870*CO ...'..
+ 1,550*CO +"o.8pO*H02
+ HO2 . + ' CO • •
~ . . • i *f . «J§! *« „ * 4
+ 0.020*&CO3 t ' TC03 '
•4- 0.800*KET •*• H02
: -. . fi% • -K1
: ' • •
+ HO
+ NO2
i •£•• , , P»
. . m.
' • i'- i u. . • .-;*
r . :': . • -:•
+ NO2
+ NO3
• ; '• -i
•" \ i
+ H02
; •• • . ••;„
, . ,; ., , ,
+ 0.170*HO2 -f 0.009*HCHO
+ 0.02S*KST
+ 0.250*X02
+ 0.7SO*XO2
8-66
-------
EPA/600/R-99/030
Table 8A-9. RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
{ 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}
{ 85}
{ 86}
{ 87}
{ 88}
( 89}
{ 90}
{ 91}
{ 92}
{ 93}
{ 94}
{ 95}
{ 96}
{ 97}
{ 98}
{ 99}
{100}
{101}
{102}
{103}
OLI
TOL
XYL
CSL
CSL
HCHO
ALD
KIT
GLY
MGLY
DCS
OP1
OP2
PAA
PAN
ONIT
ACO3
PAN
TC03
TRMJ .
MO2
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
AC03
TCO3
TOLP
XYLP
ETHP
KETP
OLN
HCHO
ALD
GLY
MGLY
DCB
CSL
OL2
OLT
OLI
OL2
OLT
OLI
+
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4-
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
4
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO :
HO
HO
HO
HO
HO
HO
NO2
NO2
NO
MO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
N03
NO3
N03
NO3
N03
NO3
N03
N03
NO3
03
03
O3
-->
--> 0
--> 0
--> 0
-->
-->
--> .
-->
--> •
--> •
_-> .
--> 0
--> 0
-->
-->
-->
--> •
-->
-->
__;>
--> .
-.„> 0
4 0
--> 0
4 0
--> 0
4- 0
--> 1
+ 0
-->
+
*->
4- 0
-->- •
-->
4- 0
4- 2
-->
4- 0
-->
+• 0
— >
-->
--•>..
.,-.
-->
-->
-->
-->
-->
-->
— » ;•
-->
-->
4- 0
--> 0
4- 0
4- 0
--> 0
•4- 0
4- 0
OLIP
.750*TOLP
.830*XYLP
.100*HO2
CSL
HO2
ACO3 .
KETP
HO2
ACO3
• TCO3
.SOO*M02
.500*HC3P • -
ACO3
HCHO
HC3P
PAN
ACO3
TPAN
TC03
HCHO
.750*ALD
.036*ONIT
.380*ALD
.920*N02
.350*ALD
.240*ONIT
,600*HCHO
,200*ALD
ALD
N02 .
H02
.100*KBT
M02
NO2
.110*MGLY
,000*XO2
NO2
.160*GLY
N02
,806*DCB
ALD
MGLY
HCHO
H02
ACO3
HN03
HNO3 •
HN03
HN03
OLN
OLN
OLN
HCHO
.120*HO2
.530*HCHO
.200*ORA1
.220*MO2 •
.180*HCHO
.230*CQ
.260*HO2
4-.
4-
4-
4-
+
+
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4"
4-
4-
4"
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
4-
0
0
0
2
0
0
0
0
0
0
1
0
1
0
0
0
0
0
0
0
0
0
0
.250*CSL
.170*CSL
;'900*X02
CO
.ooo*co
CO
,SOO*HCHO
,SOO*ALD
NO3
NO2
NO2
NO2
H02
.250*K1T
.964*N02
.690*KET
.920*HO2
.OSO*KET
.760*N02
HO2
HCHO
,4SO*ALD
N02
N02
.920*HO2
.050*AC03
HO2
.700*DCB
HO2
HO2
N02
ALD
HN03
HMO3
HO2
ACO3
TCO3
XNO2
.400*ORA1
,500*AL0
.200*O8A2
.100*HO
.720*ALD
,060*ORA1
.140*HO
4-
+
4-
+
4-
4-
4-
-4-
4-
4-
4
4-
4-
4
4-
4
4
.4-
4
4
4
4
4
4
4
4
4
4
4
4
. 4
4
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
2
2
0
0
0
0
0
0
0
.250*HO2
. 170*HO2
,900*TCO3
.500*HO
.500*HO
X02
N02
.090*HCHO
.964*H02
.080*ONIT
.040*HCHO
.760*HO2
NO2
H02
,280*HCHO
. 890*GLY
.950*CO
.170*MGLY
.45Q*MGLY
MO2
HO2
.000*NO2
CO
.ooo*co
CO
.500*CSL
.420*CO
.330*CO
.230*HO2
.100*KET
.290*ORA2
.310*M02
8-67
-------
EPA/60Q/R-99/030
Table 8A-9. RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
{104}
{105}
{106}
{107}
{108}
{109}
{110}
{111}
{112}
{113}
{114}
{115}
{116}
{117}
{118}
{119}
{120}
{121}
{122}
{123}
{124}
{125}
{126}
{127}
{128}
{129}
{130}
{131}
{132}
{133}
{134}
{135}
{136}
{137}
{138}
{139}
{140}
{141}
{142}
{143}
H02
H02
HO2
H02
HO2
H02
HO2
H02
H02
H02
H02
H02
HO2
H02
MO2
M02
M02
M02
M02
M02
M02
M02
M02
M02
M02
M02
M02
M02
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
ACO3
AC03
AC03
ACO3
4"
4-
4-
4"
•f
4-
4-
4-
+
4-
+•
4-
4-
4-
4-
4.
4-
4-
4.
4-
+
4-
4.
4-
4-
4-
4-
4-
4-
MO2
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP . .
OLIP
KETP
ACO3
TOLP
XYLP
TCO3
OLN
MO2
ETHP
HC3P
HCSP
HC8P
OL2P
OLTP
OLIP
KETP
ACQ3
TOLP
XYLP
TCO3
OLN
AC03
ACO3
ACO3
ACO3
AC03
ACO3
AC03
ACO3
s
ACO3
TOLP
XYLP
TCO3
•i
--> 1
--> 0
--> 0
--> 0
+
— > 0
+
--> .1
--> 1
--> 0
4- 0
--> 0
-->
4- 0
-->
+ 0
— >
4- 2
--> 0
4- 0
4- 0
--> 1
4.
.„>
4- 0
--> 0
4- 0
— > 0
4. 0
— > 0
+ 0
--> 0
+ Q
-->
+ 0
--> 0
4. 0
-->
+ 0
--> 2
-->
+• 0
-->
4.
-->
+ 0
OP1
OP2
OP2
OP2
OP2
OP2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
,5QQ*HCHO
. 750*HCHO
.840*HCHO
HO2
,77Q*HCHO
H02
.800*HCHO
HO2
.S50*HCHO
,250*HCHO
. 890*HCHO
.S50*KET
,750*HCHO
HCHO
. 500*ORA2
HCWO
,700*DCB
HCHO
, 000*HO2
,SOO*HCHO
.500*ORA2
,47S*CO
. 750*HCHO
N02
ALD
,SOO*ORA2
,770*ALD
.500*MO2
,410»ALD
,500*MO2
.460*ALD
,500*MO2
,800*HCHO
,500*M02
ALD
.500«M02
, 72S*ALD
,50Q*H02
HGLY
,500*ORA2
, 000*MO2
M02
.700*DCB
M02
HO2
M02
.110»MGLY
+ 0
•f- 0
+ 0
+• 0
+ 0
+ 0
4- 0
4- 0
+ 0
+ 2
+ 0
+ 0
+ 0
4.
4- 0
+ o
+ 0
4- 0
4- 0
4- 0
4- 1
+ 0
4- 0
•f 0
+ 0
4- 0
4- 0
4- 0
+• 0
4- 0
4-
4- 0
+ 0
4- 0
HO2
H02
,770*ALD
.410*ALD
,460*ALD
.3SO*ALD
.750*ALD
.725*ALD
. 750*MGLY
.SOO*HO2
.170*MGLY
.000*H02
,450*MGLY
,445»GIiY
.025*AC03
XO2
.500*H02
-
.500*H02
.260*KET
. SOO*ORA2
.750*KET
.500*ORA2
.390*KET
.500*ORA2
,600*ALD
.5QO*ORA2
.SOO*HCHO
.SOO*ORA2
,S50*KET
.500*M02
. 500*H02
.170*MGLY
HO2
.450*MGLY
,920*H02
.05p*AC03
fr
V"o
4- 0
+J'o
_
4-
4-
4-
4-
4-
+
4-
+
t
4-
+
* .
+
+
i
+
in
4>
+
4-
4-
4-
_+
4-
,
4-
+
4-
: i
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1 *"! : . ' .;
. 750*ALD
.2SO*KST
.7SO*KBT-; • •• ]'
ii • ' ' j»
.390*KET
H02
'. HO2 ":'
HO2
HO2
.500*MO2
if
,160*QLY - v
i . ,1
,806*DCB
.055*MOI>y i
,460*H02 - '.'
ALD
'!',: ••• •• «
. 500"*MO2 , .»
'*
.5PO*H02
•f •' . • "••; ' r, • • ,*'
.500*HO2
i i1' i ' ' l!'
.SOO*H02
.SOO*HO2
.500*H02 , I
,140*HCHO
.500*ORA2
. 5QO*M03; . .; '•
. 160*GLY
is n
.806*DCB
.890*GLY
.9SO»CO
8-68
-------
EPA/600/R-99/030
Table 8A-9. RADM2__CIS1 and RADM2_CIS1_AQ Mechanisms
{144} ACO3 4 OLN
{145} OLN 4 OLN
{146} XO2 4 HO2
{147} XO2 4 MO2
{148} XO2 4 ACO3
{149} XO2 4 XO2
{150} XO2 4 NO
{151} XNO2 4 MO2
{152} XNO2 4 HO2
{153} XNO2 4 M02
{154} XNO2 4 ACO3
{155} XNO2 4 XNO2
{155} ISO 4 HO
{157} ISO_RO2 4 NO
{158} ISO_R02 4 HO2
{159} ISO_R02 4 AC03
{160} ISO_RO2 4 MO2
{151} ISO 4 O3
{162} ISO 4 O3P
{163} ISO 4- NO3
{164} ISON_RO2 4 NO
{165} ISON_RO2 4 HO2
{166} ISON_RO2 4 ACO3. •
{167} ISON_RO2 4 MO2
{168} ISOPROD 4 HO ' '
{169} IP_RO2 4 NO
{170} IP_RO2 4- HO2
{171} IP_RQ2 + ACO3
{172} IP_RO2 + MO2
{173} ISOPROD •(• O3 •
{174} ISOPROD + hv
{175} ISOPROD +' NO3
Rate Expression
k{ IS uses photo table NO2_RADM88
Jc( 2} uses photo table O3O1D_RADM88
k( 3) uses photo table O303P_RADM88
+ 2 . 000*XO2
- - > HCHO
4- NO2
--> 2.000*HCHO
--> OP2
--> HCHO
- - > MO2
— >
--> N02
--> ONIT
--> OP2
- - > HCHO
--> ' MO2
— >
--> • - ISO_R02
--> 0:088*ONIT
+• 0.912*ISOPROD
--> OP2
--> ' 0.500*H02
+• ISOPROD
--> 0.500*HCHO
--> 0.600*HCHO
4- 0.270*HO
+ 0.200*XO2
--> 0.750*ISOPROD
4-' 0.250*MO2
--> ' ' ISON_RO2
--> • N02
+ 0.800*HO2
--> ONIT
--> 0.500*HO2
+ ALD
--> 0.500*HCHO
* OKIT
-->• 0.500*ACO3
--> • NO2
4- 0;550*ALD
+ 0.340*MGLY
— > OP2
--> 0.500*HO2
4 0.500*ALD
--> 0.500*HCHO
4- 0.500*KET
--> 0.26S*HO
+• 0.054*MO2
4 0.146*HCHO
+ 0.850*MGLY
--> 0.970*AC03
4- 0.200*HCHO
4- 0.033*KET
--> 0.075*ACO3
4- 0'.282*HCHO
•¥ 0.925*HO2
4 ALD 4 0.500*ORA2
4 0.500*MO2
4 2.000*ALD + 2.000*NO2
+ HO2
4- HO2
+ 0.079*X02
4 0.912*NO2 4 0.912*HO2
4 0.629*HCHO
-1- 0,500*MO2 4 0.500*ORA2
4 0.500*HO2 4 ISOPROD
4 0,S50*ISOPROD 4 0.390*ORA1
4 0.070*HO2 4 0.070*CO
+ 0.200»ACO3 + 0.150*AIiD
4 0.250*AC03 4 0.250*HCHO
4 0.800*ALD 4 0.800*ONIT
+ 0.200*ISOPROD 4 0.200*NO2
4 O.SOO*MO2 4 0.500*ORA2
4 OWIT
4 0.500*HO2 4 ALD
4 0.500*IP_RO2 4 0.200*XO2
4 HO2 • 4 0*. 590*CO
4 0.250*HCHO '4 0.080*GLY
4 0.630*KET
4 0.500*MO2 4 0.500*ORA2
4 0.500*KET
4 0.500*HO2 4 0.500*ALD
4 0.100*H02 4 0,114*ACO3
4 0.070*XO2 4 0.155*CO
4 0.020*ALD 4 0.010*GLY
4 0.090*KET 4 0.462*ORA1
4 0.333*H02 4 0.700*MO2
4 0.333*CO 4 0.067*ALD
4 0.075*HNO3 4 0.643*CO
4 0.925*OHIT 4 0.282*ALD
4 0.925*XO2
Rate Constant
, scaled by l.OQOOQE+00 {O.OOOOOE+00}
, scaled by 1.00000E400 {0 .000001+00}
,' scaled by l.OOOOOE+00 {0 . OOOOOE+00}
8-69
-------
EPA/600/R-99/030
Table8A-9. RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
k(
k(
k{
k(
k(
k(
k(
k(
k{
kS
k{
k{
k(
k{
k(
k(
kC
k{
k(
k(
k(
k(
kS
k(
k{
k(
k(
k{
kS
k(
4)
5)
6!
7)
8)
9)
10)
11)
12)
13)
14)
15!
16)
17)
18)
19)
20)
21)
22)
23)
24!
25)
26)
27)
28!
29)
30)
31)
ko
kinf
F m
32)
33)
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
= 6.
= 6.
a 1.
— *>
» 2.
— o
= 1.
= 1.
= 3.
is a
= i
= 4
0.60
= k(
is a
photo table HONO_RADM88 ,
photo table HNO3_RADM88 ,
photo table HNO4_RM>M88 ,
photo table NO3NO_RADM88 ,
photo table NO3NO2_RM3M88 ,
photo table H2O2_RADM88 ,
photo table HCHOmol^RRDMSS ,
photo table HCHOrad_RADM88
photo table ALD_RADMBS
photo table MHP_RKDM88
photo table HOP_RADM88 ,
photo table PAA_RADM88 ,
photo table KETONE_RM3M88 ,
photo table GLY£orm_RADMS8 ,
photo table GLYmol_RADM88 ,
photo table MGL>Y_RADM88 ,
photo table 0DC_RADM88
photo table ORGNIT_RADM88 ,
OOOOE-34 * (T/3QO)**{-2.30)
5000E-12 * exp( 120, 0/T)
8000E-11 * expC 110. 0/T)
2000E-11 * exp{ 70. 0/T)
2000E-10
OOOOE-12 * exp( -1400. 0/T)
6000E-12 * exp{ -940. 0/T)
1000E-14 * exp( -500. 0/T)
7000E-12 * exp! 240. 0/T)
falloff expression using:
.8000E-31 * (T/300) ** (-3.20)
.7000E-12 * (T/300)**(-1.40)
, n = ' " 1 . 00
sealed
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
31) / Keq, where Keq = 2.100E-27 *
special rate expression of the
form:
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
expS
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.000001+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+QO
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
10900. 0/T)
{o
{o
{0
{0
{0
(o
{0
(o
{o
{0
{0
.OOOOOE+00} " -: ' !
. OOQQOE+OO}
.ooooos+oo)
.OOOOOE+00}
. OOOOOE+00 }
.OOOOOE+00)
.OOOOOE+00}
, OOOOOE+00} ;'"
.OOOQOE+OO}
.OOOOOE+00}
. OOOOOE+00}
{O.OOQQOE+OO}
{o
{0
{0
{0
{o
{o
{6
{9
{2
{4
{2
{1
{6
{2
(8
(1
{8
{3
.OOOOOE+00}
. OOOOOE+00 }
.OOOOOE+00}
.OOOOOE+00}
.OOOOOE+00}
.OOOQOE+OO}
.09302E-34}
.72293E-12}
.60365E-11}
.04730E-11}
.2QOOOE-10}
.82272E-14}
.82S50E-14J •'
.05452E-15}
.27883E-12}
.39058E-12}
•;-: •_; v
.62399E-02}
.01634E-12}
k = kl + K2 [M] , , where
fcl = 2.2000E-13 * exp{ 620.0/T)
k2 = 1.90001-33 * 'exp! 980,0/T)
k( 34) is a special rate expression of the form:
k m kl + k2 [M] ,! where
kl = 3.0800E-34 * expt 2820.0/T)
k2 = 2.6600E-54 * exp( 3180.0/T)
k{ 35) = 3.3000E-12 * exp{ -200.0/T)
k( 36) is a falloff expression using:
kO = 7.0000E-31 * (T/300)**(-2.fiO)
kinf = llsOOOE-11 * (f/300)**(-0.50)
3.60, n = 1.00
F =
k( 37)
k{ 38)
k( 39!
k{ 40)
k{ 41)
fc{ 42)
expC 530.0/T)
expf -2500.0/T!
exp ( 150 . 0/T)
exp{ -1230.0/T)
kO
kinf =
1? ~~
k( 43)
k{ 44)
k( 45)
fcO
kinf -
= 3.3000E-39
= 1.4000E-13
=• 1.7000E-11
= 2.5000E-14
= 2.5000E-12
is a falloff expression using:
= 2.2000E-30 * (T/3005**(-4.30)
1.5000E-12 * (T/300)**(-0.50)
0.60, n = 1.00
= k( 42) / Keq, where Keq =
= 2.0000E-21
is a falloff expression using
= 2.60QOE-30 * (T/300)**(-3
2.4000S-11 * (T/300)**(-!
1.100E-27 * exp( 11200.0/T)
.20)
.30)
F = .0.60, n = 1.00
k( 46) is a special rate expression of the forms
{6.7890SE-30}
{1.68671E-12}
{4.87144E-12}
{1.95397E-38}
{3.182138-17}
{2.81225E-11}
{4.03072E-16}
{2.500008-12}
{1.26440E-12}
{5.47034E-02}
{2.00000E-21}
{1.14885B-11J
{1.47236E-13}
8-70
-------
EPA/600/R-99/030
Table 8A-9. RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
k(
k(
k(
k =
kO =
k2 -.
k3 =
47)
48)
49)
kO
kO +
» 7.
« 4.
• 1.
^
0
is
1
4
a
{k3[M] /
2000E-15
1000E-16
9000E-33
.3000E-12
.6QQOE-11
falloff
( 1 + k3 [M] /k2 ) } , where
* expC 785. 0/T). - '
* expS 1440. 0/T! '
*
*
*
exp( 725.
exp ( 380
exp( 230
b/T)
.0/T)
.0/T)
expression using:
= 3.0000E-31
kinf =
k(
k(
k(
k{
kS
kC
kC
k{
k(
k{
k(
k(
kS
k<
k(
k(
k(
k£
kt
k(
k(
k(
k(
kC
kS
k(
k(
k(
k{
k{
kf
k(
k{
kC
k(
k(
k(
kC
k(
k(
k(
k(
k(
kC
kC
kC
k(
k(
k{
k{
F =
50)
51)
52)
53)
54)
55)
56)
57)
58)
59)
60S
61)
62)
S3)
64)
65)
66)
67)
68)
69)
70)
71)
72)
73)
74)
75)
76)
77)
78)
79)
80)
81)
82)
83)
84)
85)
85)
87)
88)
89)
90)
91)
92)
93)
94)
95!
96)
97)
98)
99)
0
B
=
5S
s;
ss
=.
«
=
=
ys
s
=
=
=
ss
£3
=
=
ss
=
KS
=
=
0
=
s;
=
=
ss
a
=
=
:&
=
55
=
=
=
=
SS
=
0
=
55
e
=£
=
=
=
=
k{100! =
1.5000S-12
.60, 'n =
1
2
1
1
1
3
2
5
1
2
1
4
9
9
6
1
1
1
2
1
1
1
6
1
2
1
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
1
6
1
1
2
2
1
3
.5000E-13
.8300E+01
.2330E-12
.5900B-11
,73002-11
.6400E-11
.1500E-12
.3200E-12
.0700E-11
.1000E-12
.8900E-11
.OOOOE-11
.OOQOE-01
.OOOOE-12
.8700E-12
.2000E-11
,1500E-11
.7000E-11
.800QE-11
.OOOOE-11
.OOOOE-11
.OOOOE-11
.1650E-13
.5500E-11
.8000E-12
.9500E+16
.7000E-12
.9500E+16
.2000E-12
.2000E-12
.2000E-12
.20001-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.OOOOE-13
.4000E-12
, OOOOE-13
.4000E-12
.4000E-12
.2000E-11
.OOOOE-12
.OOOOE-11
.2300E-11
1.
* ST/300)**
* ST/300)**
00
(-3.30)
t 0.00)
* (1.0 + 0.6*Pressure)
*
*
*
*
*
*
*
*
*-
*
*
*
*
*
*
*
*
<*•
*
*
*
*
*
*
*
*
*
*
<*
*
*
*
*
<*•
*
#
*
*
*
*
(T/300)**(
{T/300)**(
exp( -540
expC -380
exp( -380
exp ( 4 11
expt 504
.exp( 549
expC 322
expS 116
M 61)
,exp( 256
expt -745
(T/300)** (
exp( -540
expS 181
expS-13543
exp(-13543
exp( 180
expS 180
expS 180
expS 180
exp'( iso
exp( iso
exp( 180
exp( 180
exp( 180
expt 180
exp i 18-0
expt ISO
expt 180
expt 180
expt -2058
expt -1900
expt -2058
exp{ -1900
expt -1900
expt -2923
expt -1895
expt -975
2.00) * expt -1280. 0/T)
2.00) * expt -444. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
2.00) * expt -444. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
-0/T!
.0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
{4
{9
{8
{2
{3
(2
{2
{4
{1
. {8
{2
{6
{6
(2
{4
{3
(9
{1
{9
{1
(1
{2
{1
{1
{1
{1
{2
{5
{3
{4
{3
{7
{7
.65309E-12}
.95294E-11}
.aaa48E-i3}
.40000E-13)
.80672E-01}
.74210E-13}
.596691-12}
.83334E-12)
•01696E-11}
.53916E-12}
.88684E-11J
.75269E-11)
.18715E-12)
.78943E-11}
.000001-11}
.600001-11}
.OOOOOE-12}
.62197E-11}
.850201-13}
.ISOOOE-ll}
.70000E-11}
.800001-11}
.OOOOOE-ll}
.OOOOOE-ll}
.OOOOOE-ll}
.37105E-13}
.531371-12}
.13974E-12}
.57235E-04}
.700001-12}
.57235E-04)
.68378E-12}
.683781-12}
{7.683781-12}
{7
{7
{7
{7
{7
(7
{7
.68378E-12}
.683781-12}
.683781-12}
.68378E-12}
.683781-12}
.6837SE-12}
.683781-12}
{7.68378E-12}
{7
{7
{7
(6
(2
{6
{2
{2
{2
{1
U
{1
.683781-12}
.68378E-12}
.683781-12}
.01030E-16}
.38307E-15}
.010301-16}
.38307E-15}
.383071-15}
.20000E-11}
.099401-16}
.730991-14}
.225391-12} "
8-71
-------
EPA/600/R-99/030
Table 8A-9.
k(lOl)
kf!02)
kf!03)
k(104)
k(105)
lc{106!
k(107)
kflOS)
k(109)
kfllO)
k(lll)
k(ll2)
k(H3)
k(114)
kflis)
k(116)
k(117)
k(118)
k(119)
k(120)
k(121)
k(122)
k(123)
k(124)
k{125)
k(126)
k(127)
k(128!
kf!29)
k(130)
k(l3l)
k(132)
k!133)
kf!34)
k(13S)
k(136)
k(137)
k(138)
kfl39)
k(140)
k(141)
k(142)
k«143)
k{144)
k(145)
k(146)
kf!47)
kf!48)
k{149)
kflSO)
k{151)
k(152)
k(153)
k(154)
kfiss)
k(156)
k{157)
k(158)
k{159)
k(160)
k(16l!
—
SS
=
m
=
ss
=
=
=
&*
=
SS
=
=
=
=
=
=
-
=
=
=
=
=
=
m
=
=
SS
m
=
m
=
1.
1.
7.
7.
7.
7.
7.
7,
7.
7.
7 .
7.
7.
7.
7.
7.
7.
1.
1.
4.
3.
2.
1.
1.
1.
1.
9.
1.
1.
9.
1.
3 .
1.
RADP
A2_
2000E-14
3200E-14
2900E-1S
it
.*
#
7000E-14 *
7000E-14
7000E-14
70001-14
*
*
*
7000S-14 *
7000E-14 *
7000E-14
7000E-14
7000E-14
*
*
#
7000E-14 *
7000E-14 *
7000E-14
7000E-14
*
*
7000E-14 *
9000E-13 *
4000E-13
2000E-14
*
*
4000E-14 *
9000E-14 *
4000E-13 *
4000E-13
70001-14
70001-14
*
*
*
60001-13 *
7000E-14 *
7000E-14 *
6000E-13 *
70001-14 *
40001-13 *
OOOOE-13
*
= 8.4000E-14 *
=
B
S3
SS
=
=
=
=
=
=
a
=
m
=
=
VS.
=
=
=
m
m
=
=
=
=
=
=
7.
3.
3.
4.
4.
1.
4.
4.
1.
4 .
3.
7.
1.
4 .
3.
4 .
4 .
7.
1.
4.
3.
2.
4.
7.
8.
3.
7.
20001-14 *
40001-13
4000E-13
2000E-14
2000E-14
1900E-12
2000E-14
2000E-14
*
*
*
*
*
*
*
19001-12 *
2000E-.14
*
6000E-16 *
70001-14 *
7000E-14 *
20001-14
*
60001-16 *
2000E-12
*
2000E-12 *
7000E-14 *
7000E-14
2000E-14
*
*
6000E-16 *
54001-11 *
2000S-12 *
70001-14
40001-14
40001-14
*
*
*
86001-15 *
CIS1 and RADM2.
expf -2633.
expf -2105.
expf -1136,
expf 1300.
expf, 1300.
exp! . 1300.
expf 1300.
expf 1300.
expf 1300.
expf 1300.
expf 1300.
expf 1300.
expf 1300.
expf 1300.
exp! 1300.
exp! 1300.
expf 1300.
expf 220.
expf 220.
exp! 220.
exp! 220,
expf 220,
expf 220.
expf 220.
exp! 220.
expf 220.
exp ( 220.
expf 220.
expf 220.
expf 220.
expf 220.
expf 220.
exp! 220.
exp{ 220.
expf 220.
expf 220.
.expf 220.
exp! 220.
exp! 220.
expf 220.
expf 220.
exp! 220.
exp( 220.
expf 220.
expf 220.
exp! 1300.
expf 220,
exp( 220.
expf 220.
exp! 180.
exp( 180.
exp! 1300.
exp( 220.
expf 220.
exp! 220.
(T/300)** (
(T/300)** !
(T/300)** (
(T/300)** (
(T/300)** (
(T/300) ** (
0/T)
0/T)
0/T)
0/T)
0/T)
0/T) .
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T!
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
1.00) *
1,00) *
1.00) *
1.00) *
i.oo) *
i.oo! *
_CIS1_AQ Mechanisms ;
{1.74559E-18}
(1.12333E-17)
(1.61125S-16)
{6.04Q38E-12}
{6.04038E-12}
{6.04038B-12}lA.,. , ,',
{6.04038E-12} '..
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12} • ••
{6.04038E-12}
{3.97533E-13} ! "
{2.92919E-13}
{8.787S8E-14} -•
{7.11376E-14}
{6.06762E-14} ."..
{2.92919E-13}
{2.92919E-13} x
{3.S5688E-14} "^
{3.55688E-14}
{2.00859E-12}
{3.55688E-14}
{3.55688E-14} '.
{2.00859E-12} ^
{3.55G88E-14} '-- '
{7.11376E-13} ;';
{2.09228E-13} J
{1.757S2B-13}
{1.5Q644E-13}
{7.11376E-13}
{7.11376E-13}
{8.7875SE-14}
{8.787S8E-14}
{2.48981E-12} ,t
{8.787S8E-14}
{8.78758E-14} '
{2.48981E-12}
{8.78758E-14 j
{7.S3221E-1GJ iii1' .1
{6.04038E-12} '*"
{3.5S688E-14} •••
{8.78758E-14J ~:
{7.S3221E-16}
{7.58378E-12}
{7.68378E-12}
{6.04038E-12}
{3.S5688E-14}
{8.787S8E-14}
{7.53221E-16}
expf 407. 6/T) {9. 907198-11}
exp! 181. 2/T! {7.66335E-12} -
exp! 1298. 3/T) {S.96598E-12} ."
expf 221. 4/T) {1.7S402E-13}
exp! 221.4/T) {7.09961E-14},; - , ~
exp! -1912. 2/T) {1.27569E-17} T '-
8-72
-------
EPA/600/R-99/030
Table 8A-9, RADM2_CIS1 and RADM2_CIS1_AQ Mechanisms
M162)
k(163)
k(164)
k{16S)
k(166)
k(167)
k(168)
k(169)
k(170)
k(171)
k{172)
k(173)
k(174)
k{17S!
= 3
= 3
» 4
= 7
= 8
» 3
= 3
= 4
B 7
= 8
= 3
» 7
uses
= 1
eoooE-ii
0300E-12
2000E-12
7000E-14
4000E-14
4000E-14
3600E-11
2000E-12
7000E-14
4000B-14
4000E-14
1100E-18
*
*
*
*
*
it
*
*
*
photo table
00001-15
(T/300) **( 1
(T/300)**
(T/300) **
(T/300)**
(T/300)**
(T/300)**
(T/300)**
(T/30QS**
1
1
•J.
1
1
1
1
(T/300!**! 1
ACR0LEIN
1-
.00)
.00)
.00!
.00)
.00)
.00)
.00)
.00)
.00)
* exp(
* exp(
* exp(
* exp!
* exp!
* exp(
* exp!
* exp(
* exp!
, scaled
-447.
181.
1298.
221.
221.
181.
1298.
221.
221.
by 3
9/T)
2/T)
3/T)
4/T)
4/T)
2/T)
3/T)
4/T)
4/T!
.600008-03
(3
{6
(7
{S
{1
{7
{3
{?
{S
{1
{?
(7
(o
{1
.60000E-11)
.69552E-13J
.6633SE-12)
.96S98E-12)
.75402E-13}
.09961E-14}
.36000E-11)
.66335E-12}
.96598E-12}
.754021-13}
.09961E-14}
.11000E-18}
.OOOOOE+00}
.OOOOOE-1S)
8-73
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms i,(-MP i
: • -. 1 'J. .... • V, • ' " • : • • ' '-'" ' • '
Reaction List i • • .. i«" ! .
{ 1}
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
f 8}
{ 9}
{ 10}
{ 11}
{ 12}
{ "}
{ 14}
{ 15}
{ 16}
{ 17}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 40}
{ 41}
{ «}
{ 43}
{ 44}
{ «}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
{ 54}
{ 55}
{ 56}
{ 57}
NO2
03
O3
HONO
HNO3
HNO4
N03
NO3
H2O2
HCHO
HCHO
ALD
OP1
OP2
PAA
KET
GLY
GLY
MGLY
DCB
ONIT
O3P
O3P
O1D
O1D
DID
O3
O3
O3
HO2
HO2
HNO4
HO2
HO2
H202
HO
NO
O3
NO3
NO3
N03
NO3
N205
N2O5
HO
HO
HO
HO
HO
CO "
HO
ETH
HC3
HC5
HC8
OL2
OLT
+ hv
•f hv
+ hy
+ hy
+ hv
•f hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
•f hv
+ hv
+ hv
+ hv
+ hv
+• hv
• + hy
i- CM]
+ NO2
+ {H2J
+ (02]
f [H2O]
+ NO
+• HO
+ HQ2
-1- NO
+ NO2
f
•f H02
+ HO2
+• HO
+ HO
+• NP
+ HO2
+ NO
+ NO2
+• H02
+ HO2
+ [H20]
+ M[O2
+ HNO3
+• HNO4
+ HO2
+ SO2
+ HO
,
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
--> - O3P
- - > O1D
--> O3P .
--> HO
--> HO
- - > ' HO2
--> NO "
--> NO2
--> 2.000*HO
--> CO
--> HO2
--> MO2
- - > HCHO
--> ALD
--> MO2
--> ACO3
--> 0.130*HCHO
--> 0.450*HCHO
--> ACO3
--> 0.980*HO2
--> 0.200*ALD
+ N02
•*• £O2] —> ' O3
--> NO
--> O3P
--> O3P
--> 2.000*HO
--> NO2
- - > H02
--> HO
--> NO2
--> HNO4
- - > HO2
— > H202
+• [H2O] --> H2O2 -
- - > HO2
— > HONO
+ [O2] --> 2.000*NO2
--> N03
--> 2.000*N02
.'...-.--> NO
....... --> HNO3
--> N2O5
— > N02
--> 2.00Q*HN03
--> HNO3
--> ' NO3
- - > N02
-->
--> SULP
--> H02
- - > MO2
--> ETHP
--> 0.830*HC3P
+ Q.075*MJ3'
--> HCSP
- - > HC8P
-->' ' OL2P
- - > OLTP
• '.' ."' •• .i\ ;!- ''./' .- 5
+ NO ;
™' — - • ' »
+ NO *
+ N02 . "• ' '
+ N02 - v '•' -• -'
a . . ,.,
+ O3P
> « *" •
+• H02 + " ' CO ' ' ' '*•
+• HO2 + CO '!
+ H02 + "HO
+ H02 + HO
+ HO
+ ETHP . . . -, : . ,-. " .;;;
+ 1.870*CO
+ 1.5SO*CO •*• 0.800*H02
+ HO2 4- CO
+ 0.020*AC03 -f TC03 .,• '
+ 0.800*KET +"'"' """HO2* ' J"'
i -•
%
HO ; /,. ;:
+ NO2 ..•--.
•'• i1 , .'
+ NO2
'":::
+ NO3
™
"• • • . . • •,;,
+ HO2 + SULABR
•J
+ 0,170*H02 -t O.Q09*HCHO
+ Q.02S*KET
+ 0.250*XO2
+ 0.7SO*XO2 +, "" HC8AER' '
8-74
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
{ 58}
{ 59}
{ so}
{ 61}
{ 62}
{ 63}
{ 64}
{ 65}
{ 66}
{ 67}
{ 68}
{ 69}
{ 70}
{ 71}
{ 72}
{ 73}
{ 74}
{ 75}
{ 76}
{ 77}
{ 78}
( 79}
{ 80}
{ 81}
{ 82}
{ 83}
{ 84}
{ 85}
{ 86}
{ 87}
{ 88}
{ 89}
{ 90}
{ 91}
{ 92}
{ 93}
{ 94}
{ 95}
{ 96}
{ 97}
{ 98}
{ 99}
{100}
{101}
{102}
OLI
TOL
XYL
CSL
CSL
HCHO
ALD
KET
GLY
MGLY
DCS
OP1
OP2
PAA
PAN
ONIT
AC03
PAN
TCO3
TPAN
MO2
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
ACO3
TC03
TOLP
XYLP
ETHP
KETP
OLN
HCHO
ALD
GLY
MGLY
DCB
CSL
OL2
OLT
OLI
OL2
OLT
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
-1- HO
+ HO
+ HO
+ HO
+ HO
+• NO2
+ NO2
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO3
+ NO3
+ NO3
+ NO3
+ NO3
+ NO3
+ NO3
+ NO3
+ N03
+ O3 :
+ 03
- - > OLIP
--> 0.750*TOLP
+ TOLAER
--> 0.830*XYLP
+ XYLAER
--> 0.100*HO2
+ CSLAHR
--> CSL
--> HO2
--> AC03
--> KETP
--> HO2
--> . AC03
- - > TC03
--> 0.500*MO2
. --> 0.500*HC3P
--> - - ACO3
--> HCHO
--> HC3P
--> • PAN
- - > AC03
--> . TPAN
. , - - > TCO3
--> HCHO
--> 0.750*ALD
+ Q.036*ONIT
--> Q.380*ALD
+ 0.920*N02
--> 0.3SO*ALD
+ 0.24Q*ONIT
--> 1,600'HCHO
+ 0.200*ALD
--> ALD
+ • N02
- - > H02
+ 0.100*KET
- - > M02
--> N02
+ 0.110*MGLY
+ 2,000*X02
--> N02
+ 0.160»GLY
--> NO2
+ 0.806»DCB
--> ALD
- - > MGLY
- - > HCHO
--> H02
- - > . ACO3
--> HN03
--> HN03
•>-> . HN03
--> HN03
+ 0.500*CSLAER
--> OLN
--> • - OLN
--> OLN
--> . • HCHO .
+ 0.120»HO2 ,
--> O.S30*HCHO
+ 0.200*ORA1
+ OLIJU3R
-i- 0.250*CSL
+ 0.170*CSL
+ 0.900*XO2
•f CO
+ 2.000*CO
+ CO
+ 0,500*HCHO
+ 0,SOO*ALD
-f NO3
+ NO2
+ NO2
+ NO2
+ HO2
+ 0.250*KET
•f 0.964*NO2
+ 0,690*KET
+ 0,920*HO2
+ 1.060«KET
+ 0,7SO*NO2
+ HO2
+ HCHO
+ 1.4SO*ALD
+ NO2
+ NO2
•f 0.920*H02
+ 0.050*ACO3
+ HO2
+ 0.700«DCB
+ HO2
+ HO2
+ N02
+ ALD
+ HNO3
+ HNO3
+ H02
+ ACO3
+ TCO3
+ XN02
+ OLIAER
+ 0.400*ORA1
+ 0.500*ALD
+ 0.200*ORA2
+ 0
+ 0
+ 0
+ 0
+ 0
+
+
-«• 0
+ 0
+ 0
+ 0
•f- 0
+
+
+ 0
+ 0
+ 0
+ 0
+ 0
+
+
+• 2
+
+ 2
+
+ 0
+ 0
+ 0
•»• o
.2SO*HO2
.170*HO2
,900*TCO3
.500*HO
.500*HO
XO2
NO2
.090*HCHO
.964»H02
.080*ONIT
.040»HCHO
.760*H02
N02
H02
.280*HCHO
,890*GLY
.950*CO
.170*MGLY
.450*MGLY
N02
HO2
,000*NO2
CO
.000*CO
CO
.500*CSL
.420*CO
.330*CO
.230*H02
8-75
-------
EPA/600/R-99/030-
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
{103}
{104}
{105}
{106}
{107}
{108}
{109}
{110}
{111}
(112)
{113}
{114}
{115}
{116}
{117}
{118}
{119}
{120}
{121}
{122}
{123}
{124}
{125}
{126}
{127}
{128}
{129}
{130}
{131}
{132}
{133}
{134}
{135}
{136}
{137}
{138}
{139}
{140}
(141}
OLI
H02
HO2
HO2
H02
H02
H02
H02
H02
H02
HO2
H02
H02
HO2
HO2
M02
MO2
MO2
M02
MO2
MQ2
MO2
M02
MO2
MO2
M02
MO2
MO2
MO2
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
AC03
ACO3
+
+
+•
+•
4-
4-
+
+
4"
4-
4-
+
+
4-
4.
+
4-
4-
4-
4.
+
+
4-
4-
4-
+•
4-
4-
+
4-
+
4-
4-
4-
4-
4-
4-
4.
+
O3
t
Mp2
ETHP
H£3P
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
ACO3
TOLP
XYLP
TCO3
OLN
M02.
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
AC03
TOLP
XYLP
TCO3
OLN
ACO3
AC03
ACO3
AC03
AC03
ACO3
ACO3
AC03
AC03
TOLP
4- 0
--> 0
4- 0
4- 0
4.
_„>
-->
-->
-.->
-->
— >
-->
— >
-->
-->
'. - ">
•*->
-->
-->
.--> 1
— > 0
--> 0
+
--> 0
+
--> 0
4.
--> 1
--> 1
--> 0
4- 0
--> 0
-->
+• 0
-->
+ 0
-->
4- 2
--> 0
4- 0
4- 0
— > 1
+
-->
4- 0
--> 0
•«• 0
--> 0
4- 0
--> 0
4- 0
--> Q
+ 0
-->
4- 0
--> 0
4- 0
-->
4- 0
--> 2
-->
. 220*M02
.180*HCHO
.230*CO
.260*H02
OLIAER
OP1
OP2
OP2
OP2
OP2
OP2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
.500*HCHO
. 7SO*HCHO
.840*HCHO
HO2
.770*HCHO
H02
.800*HCHO
HO2
.SSO»HCHO
,250*HCHO
.890*HCHO
.5SO*KET
.750*HCHO
HCHO
.500*ORA2
HCHO
.700*DCB
HCHO
.000*H02
.500*HCHO
.500*ORA2
.47S*CO
.750*HCHO
NO2
ALD
.500*ORA2
.770*ALD
. 500*MO2
.410*ALD
.500*MO2
.460*ALD
.SOO*MO2
.800*HCHO
,500*MO2
ALD
.500*MO2
,725*SLD
.500*HO2
MGLY
.500*ORA2
.000*M02
M02
4-
+
4-
. 4-
4-
+
+
4-
4-
+
4-
4-
4-
+
+
4-
4-
4-
+
4.
4-
4-
4-
+
+
+
4-
4-
4-
+
4-
+
4-
4-
4-
4^
0
0
p
q
0
0
0
0
Q
0
0
0
0
2
0
0
0
0
0
0
b
0
p
1
0
0
0
p
6
p
0
0
0
,100*HO
.720*ALD
.060*ORA1
.140*HO
H02
HO2
.770*ALD
.410*ALD
.460*ALD
.3SO*AUO
.7SO*ALD
.72S*ALD
.7SO*M6LY
.500*H02
.170*MGLY
.000*HO2
.450*MGLY
.44S*GLY
. 025*AC03
XO2
.SOO*HO2
.500*HO2
.260*KET
.SOO*ORA2
.750*KET
,500*ORA2
.390*KET
.500*ORA2
.600*ALD
.500*ORA2
.SOO'HCHO
.SOO*ORA2
.550»KET
.SOO*M02
.500*H02
. 170*MGLY
4- 0.100'KKT
4- 0 , 29,Q*ORA2
4- 0.310*M02
""
t • •':' , •«
.': : "_
'''•' '.?..
+ 0,750*ALD"
4-~0.260*KST
+' 6.7SO«KBT
4- 1,390*KET
4- HO2
4- ... HO2 .
. "t .' . '.. H02 '
4- HO2
4-. 0.500*KO2
:. ; _. .
4- 0.160*GLY
4- 0.806*DCB
4- 0.055*MGLY
+ 0.460*HO2
4- ALD
4- 0.500*K02
"' ' 'in,: '
4- 0.500*HO2
'" '
+~b.S6b*HO2
4-" 0.500*H02 '
,!,'
4- O.SOO*H02
4- 0.500*HO2
':£? '• S t • , 1. ",
4- 0.140*HCHO
+ O.SOO*ORA2
4- 0.500*MO2
4- 0,160*GLY
.;..:
1
-•«
?
i 'i
^*
... . _,
•"
' ,
s-.j
w .Ml
£ "'
8-76
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
{142}
{143}
{144}
{145}
{146}
{147}
{148}
{149}
{150}
{151}
{152}
{153}
{154}
{155}
{156}
{157}
{158}
{159}
{160}
{161}
{162}
{163}
{164}
{165}
{166}
{167}
{168}
{169}
{170}
{171}
{172}
{173}
{174}
{175}
{176}
{177}
{178}
ACO3
ACO3
ACO3
OLN
XO2
X02
X02
X02
XO2
XNO2 •
XN02
XNO2
XNO2
XNO2
TSRP
TERP
TERP
ISO
ISO_R02
ISO_RO2
ISO_R02
ISO R02
ISO
ISO
ISO
ISON_R02
ISON_R02
ISON_R02
ISON_RO2
ISOPROD
IP_RO2
IP_RO2
IP_R02
.IP_RO2
ISOPROD
ISOPROD
ISOPROD
4- XYLP
•f TC03
4- OLN
4- OLN
4- H02
4- MO2
4- SCO3
•f XO2
+ NO
+ NO2
+ HO2
4- MO2
+• AC03
4 XNO2
4- HO
4- NO3
4- O3
+• HO
+ NO
-t- HO2
+ ACO3
+ MO2
4- 03
4- O3P
+ NO3
4- NO
4- HO2
+ ACO3
4- MO2
+ HO
4- NO
4- HO2
4- ACO3
-f. MO2
4- O3
+• bv
4- N03
4-
-->
+
-->
4-
• 4-
-->
+
-->
-->
-->
— >
_->
— >
-->
-->
_->
-->
__>
-->
-->
-->
-->
-->
4-
-->
-->
4-
-->
-->
4-
4-
_->
+
-->
-->
4-
.->
-->
+
-->
4.
-->
-->
+
+
-->
-->
+•
-->
+
-->
4-
+•
4-
-->
4-
4-
-->
4-
4-
0
0
2
2
0
0
0
Q
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
. 700*DCB
MO2
HO2
M02
. 110*MGLY
.600*X02
HCHO
NO2
.000*HCHO
OP2
HCHO
M02
N02
ONIT
OP2
HCHO
M02
TERPAER
TERPAER
TERPAER
ISO_RO2
.088*ONIT
.912*ISOPROD
OP2
.500*HO2
ISOPROD
.500»HCHO
. 600*HCHO
.270*HO
.200*X02
. 750*ISOPROD
.250*M02
ISON_RO2
NO2
.800*HO2
ONIT
.500*HO2
ALD
.500*HCHO
ONIT
.500*ACO3
NO2
.550*ALD
,340*MGLY
OP2
.500*HO2
,500*ALD
.5QO*HCHO
. 500*KET
.268*HO
.054*M02
.146*HCHO
.850*MGLY
,970*ACO3
.200*HCHO
.033*KET
. 075*AC03
,282*HCHO
.925*HO2
+
HO2
+ 0.4SO*MGLY
4- 0.
4- 0.
4-
4- 0.
4- 2.
4.
4-
4.
4-
4-
4- 0.
4- 0.
+ 0.
4- 0.
4- 0.
4- 0,
4- 0.
4- 0.
4- 0,
4- 0.
+ 0.
4- 0,
+
4- 0.
4- 0.
4-
4- 0.
4- 0,
4- 0.
4- 0.
+ 0.
4- 0.
4- 0.
4- 0.
4- 0.
4- 0.
4- 0.
4- 0,
4- 0.
4- 0.
920*HO2
050*ACO3
ALD
SOO*MO2
000*ALD
H02
HO2
HO
NO3
03
079*XO2"
912*N02
629*HCHO
500*M02
500*H02
650* ISOPROD
070*HO2
200*AC03
250*ACO3
800*ALD
200*ISOPROD
500*MO2
ONIT
500*HO2
500*IP_RO2
H02
250*HCHO
630*KET
500*M02
500*KET
500*H02
100*HO2
070*XO2
020*ALD
090*KET
333*H02
333*CO
Q75*HN03
925*ONIT
925*X02
4-
4-
• 4-
4-
4-
4-
4-
+
+
4-
4-
4-
4-
4-
4-
+
4-
4-
4-
4-
4-
4-
+
+
4-
4-
+
4-
4-
0.806«DCB
0.
0.
0.
2.
0,
0.
0.
0.
0.
0.
0.
0,
0.
0.
0.
0.
0.
0.
0.
a.
0.
0.
0.
0.
0.
0.
890*GLY
950*CO.
500*ORA2
000*NO2
912*H02
500*ORA2
ISOPROD
390*ORA1
070*CO
150* ALD
250*HCHO
800*ONIT
200*NO2
500*ORA2
ALD
200*X02
590*CO
080*GLY
500*ORA2
500*ALD
114*ACO3
155*CO
010*GLY
4S2*ORA1
700*MO2
067*ALD
643*CO
282*ALD
8-77
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
Rate Expression
k(
kf
kf
kC
kf
kf
k(
kf
kf
kf
kf
kf
kf
kf
k(
kf
kC
kf
k(
k(
k(
kf
kf
kf
kC
kf
k(
k(
kf
k!
k(
kf
k.S
1)
2)
3!
4)
5)
6)
7)
8)
9)
10)
11)
12!
13)
14!
15)
16)
17)
18)
19)
20)
21)
22)
23)
24)
25)
26)
27)
28)
29)
30)
31)
kO
kinf
F =
32)
33)
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
uses
= 6,
s 6.
= 1.
= 3.
= 2.
s* 2 .
= 1.
• 1.
= 3,
is a
• 1
=» 4
0.60
= k(
is a
photo table NO2_RADM88 , scaled
photo table 0301D_RADM88 ,
photo table 03O3PJ*ADM88 ,
photo table HONO_RADM88 ,
photo table HNO3_RADM88 ,
photo table HN04_RADM88 ,
photo table NO3NO_RADM88 ,
photo table N03N02_RADM88 ,
photo table H2O2_RADM88 ,
photo table HCHOmol_RADM88 ,
photo table HCHOrad_RADM88 ,
photo table MJ3_RADM88 ,
photo table MHP_RM5M88
photo table HOP_RM>M88 ,
photo table PAA_RRDM88 ,
photo table KETONE_RADM88 ,
photo table GLYform_RMM88 ,
photo table GLYmol_RM>M88 ,
photo table MGLY_RM3M88 ,
photo table UDC_RADM88 ,
photo table ORGNIT_RADM8 8 ,
OOOOE-34 * fT/300) **{-2.30)
5000E-12 * expf 120. 0/T)
8000E-11 * expf 110. 0/T)
2000E-11 * expC 70. 0/T)
2000E-10
OOOOE-12 * expf -1400. 0/T)
6000E-12 * exp! -940.0/T)
1000E-14 * expf -500. 0/T)
7000E-12 * expf 240. 0/T)
falloff expression using:
.80008-31 * (T/300) ** (-3.20) ,.
.70008-12 * (T/300) **(-!. 40!
, n = l.'QO
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
scaled
31) / Keq, where Keq = 2.100E-27 *
special rate expression of the
"form:
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
by
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
1
expf
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOQE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.000001+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOOE+00
.OOOOQE+00
.OOOOOE+00
.OOOOOE+00
10900. 0/T)
Rate Constant
{0
{ 0
{ 0
{o
{o
{o
{o
{o
{o
{0
{0
i°
{o
!o
{0
{0
{0
(o
{o
{0
{0
{6
{9
{2
{4
(2
{1
{6
{2
{8
(1
{8
{3
. OOOOOE+00 }
.OOOOOE+00}
. OOOOOE+00 } ,-
.OOOOOE+00}
.OOOOOE+00}
.OOOOOE+00} ;' • •
.000008+00}
.OOOOOE+00}
.000008+00} ,
.000008+00}
.000008+00}
.000008+00} •'
.OOOOOE+00} * i
.OOOOOE+00}
,000008+00}
.OOOOOE+00}
.OOOOOE+00}
.OOOOOE+00}
.OOOOOE+00}
.000001+00} . :
.OOOOOE+00} -
.093028-34} ;
.722938-12} T
.60365E-11} :- '
.047308-11} "
.200001-10}
.822728-14} ' '
.826508-14}
.054528-15} •'
,278838-12}
.390588-12}
*".'!* -t
•I- : ;
.62399E-02}
.016348-12} •'• - -'.
k = kl + k2[M], where r
kl = 2.2000E-13 * expf 620.0/T)
k2 = 1.9000E-33 * expt 980.0/T)
k( 34) is a special rate expression of the form:
k = kl + k2 [M] ,; where " .-
kl = 3.0800E-34 * expf 2820.0/T)
k2 = 2.66QQE-54 * exp! 3180.0/T)
k( 35) » 3.3000E-12 * expC -200.0/T)
k( 36) is a falloff expression using:
kO m 7.0000B-31"* Sf/300)**f-2.60)
kinf « l.SOOOE-11 * '(T/300)**(-0.50)
F = 0.60, n = i.'OO ,
kf 37) a 3.3000E-39 * expS 530.0/T)
k( 38) = 1.4QOOE-13 * expf -2500.0/T)
k( 39) = 1.7000E-11 * expf 150.0/T)
k! 40) = 2.5000E-14 * expf -1230.0/T)
k( 41) = 2.SOOOE-12
kf 42) is a falloff expression using:
kO m 2.2000E-30 * (T/300)**(-4.30)
kinf a l.SOOOE-12 * (T/300)**(-0.50)
{6.789058-30}
{1.68671E-12}
{4.87144E-12}
{1.9S397B-38}
{3.182138-17}
{2.81225E-11}
{4.030721-16}
{2.SOOOOE-12}
{1.26440E-12}
8-78
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
k(
F = 0.60, n = 1.00
43) = k( 42) / Keg, where Keq = 1.100E-27 * exp ( 11200. 0/T) {5.
44) = 2.0000E-21 {2.
45) is a falloff expression using: {l.
kO = 2.6000E-30 * (T/300) ** (-3 . 20)
kinf = 2.4000E-11 * (T/300)**(-1.30)
F = 0.60, n = 1.00
46) is a special rate expression of the form: {l.
k = kO + {k3 [M] / (1 + k3[M]/k2)}, where
kO = 7.2000E-15 * exp ( 785. 0/T)
k2 = 4.1000E-16 * exp( 1440. 0/T)
k3 =
47)
48)
49)
kO
kinf
F =
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)
85)
86)
87)
88)
89)
90)
91)
92)
: 1.
=
=
is
a
=
0,
=
=
=
=
=
2:
=
=
=
=
=
=
=
=:
=
=
=:
=
=
=
=
=
=:
=
=
=
=
=
=
=
=
=
=
=
=
=
:r
=
=
=
=
=
=
1
4
a
9000E-33 *
.3000E-12
.6000E-11
*
*
exp ( 725 .
exp( 380
exp( 230
0/T)
.0/T)
.0/T)
falloff expression using:
3.0000E-31
1.5000E-12
.60, n = 1.
1
2
1
1
1
3
2
5
1
2
1
4
9
9
6
1
1
1
2
1
1
1
6
1
2
1
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
.5000E-13
.8300E+01
.2330E-12
.5900E-11
.7300E-11
.6400E-11
.1500E-12
.3200E-12
.0700E-11
.1000E-12
.8900E-11
.OOOOE-11
.OOOOE-01
.OOOOE-12
.8700E-12
.2000E-11
.1500E-11
.7000E-11
.8000E-11
.OOOOE-11
.OOOOE-11
.OOOOE-11
.1650E-13
.5500E-11
.8000E-12
.9500E+16
.7000E-12
.9500E+16
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.OOOOE-13
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
* (T/300)**
* (T/300)**
00
(-3.30)
( 0.00)
(1.0 + 0 .6*Pressure)
(T/300)**(
(T/300)**(
exp( -540
exp( -380
exp( -380
exp( 411
exp( 504
exp( 549
exp( 322
exp ( 116
k( 61)
exp( 256
exp( -745
(T/300) **(
exp( -540
exp( 181
exp (-13543
exp (-13543
exp( 180
exp ( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( 180
exp( -2058
2.00) * exp( -1280. 0/T)
2.00) * exp( -444. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
-0/T)
.0/T)
.0/T)
.0/T)
.0/T)
2.00) * exp( -444. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{4-
{9-
{8.
{2.
{3.
{2.
{2.
{4.
{1.
{8.
{2.
{e-
{6-
{2.
{4.
{3.
{9.
{1.
{9-
{!•
{!•
{2-
{!•
{1.
{!•
{1.
{2.
{5.
{3-
{4.
{3-
{7-
{7.
{7.
{7-
{7.
{7.
{7-
{7-
{7.
{7-
{7-
{7.
{7.
{7.
{«•
47034E-02)
OOOOOE-21)
14885E-11J
47236E-13}
65309E-12}
95294E-11}
88848E-13J
40000E-13)
80672E-01)
74210E-13)
59669E-12}
83334E-12)
01696E-11J
53916E-12)
88684E-11}
75269E-11}
18715E-12}
78943E-11}
OOOOOE-ll}
60000E-11)
OOOOOE-12J
62197E-11)
85020E-13J
15000E-11J
70000E-11}
80000E-11)
OOOOOE-llj
OOOOOE-ll}
OOOOOE-ll}
37105E-13)
53137E-12}
13974E-12)
57235E-04}
70000E-12)
57235E-04J
68378E-12}
68378E-12)
68378E-12J
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12J
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12)
68378E-12J
01030E-16}
8-79
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS 1_AE and RADM2_CIS 1_AE_AQ Mechanisms
kt 93)
kt 94)
kt 95)
k( 96)
k( 97)
kt 98)
k( 99)
k(100)
k(101)
k(1025
kt!03)
kt!04)
ktlOS)
M1065
k!107)
ktlOS)
k(109)
k(llO)
k(lll)
k(112)
k(113)
k(114)
ktllS)
ktllG)
ktl!7)
ktllS)
ktll9)
k{120)
k(121!
k(122!
k(123)
k(124)
k(125)
k{126)
k{127)
k(128!
kt!29)
kt!30)
kt!31)
kt!32S
kt!33S
k(134)
k(135)
k(136)
k(137)
k(138)
k(139)
k(140)
kt!41)
k!142)
kt!43)
ktl«4)
k(145)
k{146)
k(147)
k£148)
k(149)
k(150)
k(151)
k(152)
k{153)
.
m
=
=
—
=
=
m
=
=
a
=
m
_
—
=
_
=
=
=
—
K
=
=
=
a.
—
m
_
_
_
B
m
a
e
m
*
=
=
a
_
ss
m
—
-
=
=
XX
=
_
—
=
_
ss
—
a
9
_
-
m
=
1
6
1
1
2
2
1
3
1
1
7
7
7
7
7
7
7
7
7
7
7
7
7
7
7
1
1
4
3
2
1
1
1
1
9
1
1
9
1
3
1
8
7
3
3
4
4
1
4
4
1
4
3
7
1
4
3
4
4
7
1
I
.4000E-12
.OOOOE-13
.4000E-12
.4000E-12
.2000E-11
.OOOOE-12
.OOOOE-11
.2300E-11
.2000E-14
.3200E-14
.2900E-15
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.9000E-13
.4000E-13
.2000E-14
.4000E-14
.9000E-14
.4000E-13
.4000E-13
.7000E-14
.7000E-14
.6000E-13
.7000E-14
.7000E-14
.6000E-13
.70001-14
.4000E-13
.OOOOE-13
.4000E-14
.2000E-14
.4000E-13
.4000E-13
.2000E-14
.2000E-14
.1900E-12
.2000E-14
.2000E-14
.1900E-12
.2000E-14
.60001-16
.70001-14
.7000E-14
.2000E-14
.6000E-16
.2000E-12
.2000E-12
.70QOE-14
.7000E-14
*
*
*
*
*
if
ir
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
if
if
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
if
if
it
*
*
*
*
exp 5
expt
expt
expt
expt
expt
expt
exp(
expt
expt
expt
expt
expt
expt"
expt
expt
expt
expt
expt
expt
expt
expt
expt
exp(
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
expt
exp (
expt
expt
expt
exp(
expt
expt
expt
axp t
expt
expt
exp t
exp (
expt
expt
exp (
exp (
-1900
-2058
-1900
-1900
-2923
-1895
-975
-2633
-2105
-1136
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
220
1300
220
220
220
180
180
1300
220
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T!
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
• 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T!
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{2
{6
{2
{2
{2
{1
{1
{1
(1
(1
{1
{6
(6
{6
(e
(6
{6
{6
{6
{6
{6
{6
{6
{6
{6
{3
{2
{8
{7
{6
{2
{2
{3
{3
{a
{3
{3
(2
{3
{7
{2
{1
{1
{7
{7
{8
{8
{2
{8
{8
{2
{8
{7
{6
{3
{8
{7
{7
{7
{6
{3
.38307E-15} ; ;r
.01030S-16}
.38307E-15} ; ;
.38307E-15} : -
.20000E-11} ,, '• T'
.09940E-16)
.73099E-14)
.22539E-12} '• '^
.74559E-18}
.12933E-17}
.S1125E-16}
.04038E-12}
.04038E-12}
.04038E-12}
.04038E-12}
.04038E-12}
.04038E-12}'
.04038E-12}
.04038E-12)
.04038E-12}
.04038E-12}
.04038S-12}
.04038E-12}
.04038E-12}
.04038E-12}
.97S33E-13J
.92919E-13J r
.78758E-14} '-
.11376E-14}"
.06762E-14J '"" 7'
.92919E-13}
.92919E-13}
.55688E-14} ' ::
.55688E-14}
.00859E-12} 'L
.556883-14} \" ':
.S5688S-14} <*• *
.00859E-12}
.S5688B-14}
.11376K-13}
.09228B-13} .-
.757523-13}
.50644E-13}
.11376E-13}
.11376E-13}
.78758E-14)
.78758E-14}
.48981E-12}
.78758E-14}
.787588-14}
.48981E-12}
.78758E-14} .
.53221E-16} "
.04038E-12} „-, ;
.55688E-14} -• ;
.787S8E-14J
.53221E-16)
.68378E-12J
.68378E-12}
.04038E-12}
.55688E-14}
8-80
-------
EPA/600/R-99/030
Table 8A-10. RADM2_CIS1_AE and RADM2_CIS1_AE_AQ Mechanisms
k(154)
k(155)
k(156)
k(157)
k(158)
k(159)
k(160)
k(161)
k(162)
k(163)
k(164)
k(165)
k(166)
k(167)
k(168)
k(169)
k(170)
k(171)
k(172)
k(173)
k(174)
k(175)
k(176)
k(177)
k(178)
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
4
3
1
1
1
2
4
7
8
3
7
3
3
4
7
8
3
3
4
7
8
3
7
uses
=
1
.2000E-14
.6000E-16
.OOOOE+00
.OOOOE+00
.OOOOE+00
.5400E-11
.2000E-12
.7000E-14
.4000E-14
.4000E-14
.8600E-15
.6000E-11
.0300E-12
.2000E-12
.7000E-14
.4000E-14
.4000E-14
.3600E-11
.2000E-12
.7000E-14
.4000E-14
.4000E-14
.1100E-18
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
photo table
.OOOOE-15
exp( 220
exp( 220.
k( 58)
k(100)
k(103)
(T/300)** (
(T/300)**(
(T/300) **(
(T/300)**(
(T/300) **(
(T/300)**(
(T/300)** (
(T/300)**(
(T/300) **(
(T/300) ** (
(T/300) ** (
(T/300) **(
(T/300)**(
(T/300)**(
(T/300)** (
ACROLEIN
.0/T)
.0/T)
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
00)
* exp (
* exp(
* exp (
* exp(
* exp (
* exp( -
* exp(
* exp(
* exp(
* exp(
* exp(
* exp(
* exp(
* exp(
* exp(
, scaled
407.
181.
1298.
221.
221.
1912.
-447.
181.
1298.
221.
221.
181.
1298.
221.
221.
by 3
6/T)
2/T)
3/T)
4/T)
4/T)
2/T)
9/T)
2/T)
3/T)
4/T)
4/T)
2/T)
3/T)
4/T)
4/T)
.60000E-03
{8.78758E-14}
{7.53221E-16}
{6.75269E-11}
{1.22539E-12}
{1.61125E-16J
{9.90719E-11}
{7.66335E-12}
{5.96598E-12J
{1.75402E-13J
{7.09961E-14}
{1.27569E-17}
{3.60000E-11}
{6.69552E-13}
{7.66335E-12J
{5.96598E-12}
{1.75402E-13}
{7.09961E-14}
{3.36000E-11}
{7.66335E-12}
{5.96598E-12}
{1.75402E-13}
{7.09961E-14}
{7.11000E-18J
{O.OOOOOE+00}
{l.OOOOOE-15}
8-81
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
Reac
{ 1}
{ 2}
{ 3}
{ 4}
{ 5}
{ 6}
{ 7}
{ 8}
{ 9}
{ 10}
{ 11}
{ 12}
{ 13}
{ "}
{ 15}
{ 16}
{ 17}
{ 18}
{ 19}
{ 20}
{ 21}
{ 22}
{ 23}
{ 24}
{ 25}
{ 26}
{ 27}
{ 28}
{ 29}
{ 30}
{ 31}
{ 32}
{ 33}
{ 34}
{ 35}
{ 36}
{ 37}
{ 38}
{ 39}
{ 4°}
{ 41}
{ 42}
{ 43}
{ 44}
{ 45}
{ 46}
{ 47}
{ 48}
{ 49}
{ 50}
{ 51}
{ 52}
{ 53}
{54}
{ 55}
{ 56}
{ 57}
tion Lis
N02
03
O3
HONO
HNO3
HNO4
NO3
N03
H202
HCHO
HCHO
ALD
OP1
OP2
PAA
KET
GLY
GLY
MGLY
DCB
ONIT
03 P
03P
oib
O1D
O1D
O3
O3
03
H02
HO2
HNO4
HO2
HO2
H2O2.
NO
NO
O3
NO3
NO3
NO3 ,
NO3
N2O5
N205
HO
HO
HO
HO
HO
CO
HO
ETH
HC3
HC5
HC8
OL2
OLT
t
+ hv
+ h'v
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ h'v
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ h'v
+ hv
+ hv
+ [M]
+ NO2
+ [N2]
+ [02]
+ [H20]
+ NO
+ HO
+ HO2
+ NO
+ NO2
+ H02
+ HO2
+ HO
+ HO
+ NO
+ NO2
+ NO
+ N02
+ H02
+ N02
+ [H20]
+ NO2
+ HN03
+ HITO4
+ HO2
+ SO2
+ HO
+ Hp
+ HO
+ HO
+ HO
+ HO
+ Hp
.--> °3P
... "> °1D
--> 03 P
--> HO
--> HO
--> H02
--> . NO
- - > NO2
--> 2.000*HO
--> co
--> HO2
--> MO2
--> HCHO
--> ALD
--> M02
--> AC03
--> 0.130*HCHO
--> 0.450*HCiHO
— > AC03
--> 0.980*HO2
--> 0.200*ALD
+ NO2
+ [02] --> 03
--> NO
'"--> O3P
" '' " ,--> , ' P3P
--> 2.000*HO
-"> NO2
--> HO2
--> HO
--> N02
--> HN04
--> HO2
--> H202
+ [H2O] --> H2O2
--> HO2
--> HONO
+ [O2] --> 2.000*NO2
--> NO3
--> 2.000*NO2
."> . '. .NO
--> HNO3
--> N2O5 ,
--> NO2
--> 2.000»HNO3
--> HNO3
--> NO3
--> NO2
-*> • '
--> SULF
--> HO2
'-> MO2
--> ETHP
--> 0.830*HC3P
+ 0.075*ALD
* ' ' " . 'I', ..,'1 ' /
--> HC5P
--> HC8P
--> OL2P
--> OLTP
+ NO
+ NO
•f N02
+ NO2
+ O3P
+ H02
+ H02
+ HO2
+ HO2
•«• HO
+ ETHP
•«• 1.870*CO
+ i.550*CO
+ H02
+ 0.020*AC03
+ 0.800*KET
•t- ' HO
+ NO2
+ NO2
•
+ NO3
+ H02
+ 0.170*HO2
+ 0.02S*KET
+ 6.250»x62
+ 0.750*XO2
!""',„ "'! , ':::
-i • ., '; i - -1 ; : '.. i a
,.. ,-,
+ CO
+ CO-,, •
•f HO
+ HO
'
+ 0.800*H02
+ CO
+ TCO3
•f H02
1 1 -I.1.
V "' • if ' ;"il
.•! ," , ! .;'!
,"''
•• :
" > •'
1
....
.' •. ., ". , , ;
5 • i,i; •
.11, r ' „ ,'
'.' '.
".,'" ,.: ''"'.i-i ' • . " 'i
.Hi • • ••
,, '. '•" . ' •".;
+ 0.009*HCHO
TJI
. ii . "" i" i! ' i
8-82
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
{ 58} OLI
{ 59} TOL
{ 60} XYL
{ 61} CSL
{ 62} CSL
{ 63} HCHO
{ 64} ALD
{ 65} KET
{ 66} GLY
{ 67} MGLY
{ 68} DCB
{ 69} OP1
{70} OP2
{ 71} PAA
{72} PAN
{ 73} ONIT
{ 74} AC03
{ 75} PAN
{ 76} TCO3
{ 77} TPAN
{ 78} MO2
{ 79} HC3P
{ 80} HC5P
{ 81} HC8P
{ 82} OL2P
{ 83} OLTP
{ 84} OLIP
{ 85} ACO3
{ 86} TCO3
{ 87} TOLP
{88} XYLP
{89} ETHP
{ 90} KETP
{ 91} OLN
{ 92 } HCHO
{ 93} ALD
{ 94} GLY
{ 95} MGLY
{ 96} DCB
{ 97} CSL
{ 98} OL2
{ 99} OLT
{100} OLI
{101} OL2
{102} OLT
{103} OLI
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO '
•f HO
•f HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ HO
+ N02
+ NO2
+ NO
+ NO
•f NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ NO
+ N03
•f N03
+ N03
+ N03
+ NO3
+ NO3
+ N03
+ NO3
+ NO3
+ O3
+ 03
+ O3
--> OLIP
--> 0.750*TOLP
--> 0.830*XYLP
--> 0.100*HO2
--> CSL
- - > HO2
--:> AC03
--> ' KETP
- - > HO2
--> AC03
--> TC03
--> 0.500*M02
--> 0.500*HC3P
--> ACO3
--> HCHO
--> HC3P
--> PAN
--> ACO3
--> TPAN
--:• TCO3
--> HCHO
--> 0.750*ALD
+ 0.036*ONIT
--> '0.380*ALD
+ 0.920*NO2
--> 0.350*ALD
+ 0.240*ONIT
--> 1.600*HCHO
+ 0.200*ALD
--» ALD
+ N02
--> H02
+ 0.100*KET
--> ' M02
- - > NO2
+ 0.110*MGLY
+ 2.000*X02
--> NO2
+ 0.160*GLY
- - > N02
+ 0.806*DCB
--> ALD
; --> MGLY
- - * HCHO
- - > H02
--> ACO3
--> HNO3
--> HNO3
--> HNO3
--> HN03
- - > OLN
--> OLN
--> OLN
--> HCHO
•f 0.120*H02
--> 01530*HCHO
+ 0.200*ORA1
+ 0.220*M02
' --> 0.180*HCHO
+• 0.230*CO
+ 0.260*HO2
+ 0.250*CSL
•f 0.170*CSL
+ 0.900*XO2
+ CO
•f 2.000*CO
+ CO
+ 0.500*HCHO
+ 0.500*ALD
+ NO3
•f N02
+ NO2
+ N02
+ HO2
•f 0.250*KET
+ 0.964*N02
+ 0.690*KET
•f 0.920*HO2
+ 1.060*KET
+ 0.760*N02
+ HO2
+ HCHO
+ 1.450*ALD
+ NO2
+ N02
•f 0.920*HO2
+ 0.050*ACO3
+ H02
+ 0.700*DCB
+ HO2
+ H02
+ . N02
+ ALD
+ HNO3
+ HNO3
+ HO2
+ AC03
+ TCO3
+ XNO2
+ 0.400*ORA1
+ 0.500*ALD
+ 0.200*ORA2
+ 0.100*HO
+ 0.720*ALD
•f 0.060*ORA1
+ 0.140*HO
+ 0.250*HO2
+ 0.170*H02
+ 0.900*TCO3
+ 0.500*HO
+ 0.500*HO
+ XO2
+ NO2
+ 0.090*HCHO
+ 0.964*HO2
+ 0.080*ONIT
+ 0.040*HCHO
+ 0.760*HO2
+ NO2
+ HO2
+ 0.280*HCHO
+ 0.890*GLY
+ 0.950*CO
+ 0.170*MGLY
+ 0.450*MGLY
+ NO2
+ HO2
+ 2.000*NO2
+ CO
+ 2.000*CO
+ CO
+ 0.500*CSL
+ 0.420*CO
+ 0.330*CO
+ 0.230*HO2
+ 0.100*KET
+ 0.290*ORA2
+ 0.310*MO2
8-83
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
{104}
{105}
{106}
{107}
{108}
{109}
{110}
{111}
{112}
{113}
{114}
{115}
{116}
{117}
{118}
{119}
{120}
{121}
{122}
{123}
{124}
{125}
{126}
'{127}
{128}
{129}
{130}
{131}
{132}
{133}
{134}
{135}
{136}
{137}
{138}
{139}
{140}
{141}
{142}
{143}
H02
HO2
H02
H02
H02
H02
HO2
H02
H02
H02
H02
H02
H02
HO2
MO2
M02
MO2
MO2
M02
M02
M02
MO2
MO2
MO2
M02
M02
M02
MO2
ETHP
HC3P
HC5P
HC8P
OL2P
OLTP
OLIP
KETP
ACO3
ACO3
ACO3
ACO3
1
+ MO2
+ 'ETHP
•*• HC3P
+ HC5P
+ HC8P
+ OL2P
+ OLTP
+ OLIP
+ KETP
+ ACO3
':» " • 1-
•*• TOLP
+ XYLP
+ TCO3
•*• OLN
+ M02
•*• ETHP
•*• HC3P
+ HC5P
•*• HC8P
•*• OL2P
•*• OLTP
+ OLIP
•*• KETP
•*• ACO3
•*• TOLP
•*• XYLP
•*• TCO3
•*• OLN
•*• ACO3
•*• AC03
•*• ACO3
•*• ACO3
+ ACO3
•*• AC03
•*• ^C03
•*• ACO3
•*• AC03
+ TOLP
•*• XYLP
+ TC03
::>
--> i
--> 0
--> 0
+
--> 0
+
--> 0
+
--> 1
--> 1
--> 0
•*• 0
--> 0
-->
+ 0
-->
•*• 0
-->
•*• 2
--> 0
•*• 0
+ 0
--> 1
+
-->
•*• 0
--> 0
•*• °
- - > 0
•*• 0
--> 0
+ 0
--> 0
t o
-->
+ o
--> 0
+ 0
-->
•*• 0
--> 2
-->
•*• 0
-->
+
-->
+ 0
OP1
OP2
OP2
OP2
OP2
OP2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
.500*HCHO
.750*HCHO
.840*HCHO
HO2
.770*HCHO
HO2
.800*HCHO
HO2
.550*HCHO
.250*HCHO
.890*HCHO
.550*KET
.750*HCHO
HCHO
.500*ORA2
HCHO
.700*DCB
HCHO
.000*HO2
.500*HCHO
.SOO*ORA2
.47S*CO
.750*HCHO
NO2
ALD
.SOO*ORA2
.770*ALD
.SOO*MO2
.410*ALD
.SOO*MO2
.460*ALD
.500*M02
,800*HCHO
.500*MO2
ALD
.SOO*MO2
.725*ALD
.500*HO2
MGLY
.SOO*ORA2
.000*M02
MO2
.700*DCB
MO2
H02
MO2
.110*MGLY
+
+
•*• 0
+ 0
+ 0
+ 0
+ 0
•*• 0
•*• 0
+ 0
•*• 0
+ 2
+ 0
+ 0
+ 0
+
+ 0
•*• 0
+ 0
+ 0
•*• 0
+ 0
+ 1
•*• 0
+ °
•*• 0
+ 0
+ 0
• •*• o
+ 0
•*• 0
+ 0
+
•*• 0
+ 0
+ 0
HO2
HO2
.770*ALD
.410*ALD
.460*ALD
.350*ALD
.750*ALD
.725*ALD
. 750*MGLY
.500*HO2
.170*MGLY
. 000*HQ2
.450*MGLY
.445*GLY
.02S*AC03
XO2
.500*HO2
.500*HO2
-260*KET
.500*ORA2
.7SO*KET
.500*ORA2
.390*KET
.SOO*ORA2
.600*ALD
.SOO*ORA2
.SOO*HCHP
.SOO*ORA2
.55p*KET
.500*M02
.500*H02
. 170*MGLY
HO2
.4SO*MGLY
.920*H02
.OSO*AC03
: U
+ 0
+ 0
+ 0
f"
+ 1
+
+
•*•
+
.+ . °
"ill.
+ 0
+ 0
+ 0
+ 0
+
-liKi.
+ 0
+ 0
•*• 0
+ 0
t °
= .,.:
t o
.+_.' o
+ 0
+ 0
+ 0
•*• 0
+ 0
t °
'.'i
.750*ALD
.260*KET
.750*KET
,,, '!' " " ":
.390*KKT
H02
HO2
H02
H02
.500*HO2
T'f".! ' "'" ''"' :!l!
in" ; i.
.160*GLY
.806*DCB
.OS5*MGLY I
.460*H02
ALD
tit: ' • )
.500*M02
.50Q*HO2
i ' „ *
.500*HO2 ';
.500*H02
.500*H02
.SpO*HQ2
.14Q*HCHO '"'•': ^
."5"00*ORA2 ':"
.500*M02
.160*GLY
.806*DCB
:... ' '",::
.890*GLY
.9Sp*CO
8-84
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
{144}
{145}
{146}
{147}
{148}
{149}
{150}
{151}
{152}
{153}
{154}
{155}
{156}
{157}
{158}
{159}
{160}
{161}
{162}
{163}
{164}
{165}
{166}
{167}
{168}
{169}
{170}
{171}
{172}
{173}
{174}
{175}
{176}
{177}
{178}
{179}
ACO3
OLN
XO2
X02
X02
XO2
XO2
XNO2
XNO2
XN02
XNO2
XNO2
ISO
ISO_R02 •
ISO_R02 •
ISO_R02 •
ISO_RO2 •
ISO
ISO
ISO
ISON_R02 •
ISON_R02 -
ISON_RO2 •
ISON_RO2 -
MACR
MACR_RO2 •
MACR_RO2 -
MACR_RO2 •
MACR_RO2 -
MACR
MACR
MACR
MVK
MVK_RO2 -
MVK_R02 -
MVKJR02 -
+ OLN
+ OLN
+ HO2
+ M02
+ AC03
+ X02
i- NO
+ NO2
+ HO2
i- M02'
+ ACO3
i- XNO2
+ HO
i- NO
<- H02
f AC03
f M02
f 03
f 03P
V NO3
V NO
V H02
V ACO3
V MO2
f HO
V NO
h HO2
V AC03
h MO2
h O3
t hv
h N03
h HO
t- NO
h H02
^ AC03
+ 2
+
--> 2
-->
-->
-->
-->
-->
-->
-->
-->
-->
-->
-->
--> 0
+ 0
+ 0
-->
--> 0
+ 0
+ 0
--> 0
+ 0
--> 0
+ 0
+ 0
+ 0
--> 0
+ 0
-->
-->
+ 0
-->
--> 0
+
--> 0
+ 0
--> 0
-->
+ o
-->
--> 0
+ 0
+ 0
--> 0
+ 0
--> 0
+ 0
+ 0
--> 0
+ 0
+ 0
--> 0
+ 0
-->
-->
+ o
+ 0
-->
--> 0
.000*XO2
HCHO
NO2
.000*HCHO
OP2
HCHO
MO2
N02
ONIT
OP2
HCHO
M02
ISO_R02
.088*ONIT
.362*ISOPROD
.629*HCHO
OP2
.500*HO2
.362*ISOPROD
.629*HCHO
.500*HCHO
.230*MACR
.600*HCHO
.390*ORA1
.070*CO
.150*ALD
.750*ISOPROD
.250*MO2
ISON_RO2
N02
.800*HO2
ONIT
.500*HO2
ALD
.500*HCHO
.500*ONIT
.500*MC03
NO2
.840*CO
OP2
.500*HO2
.840*KET
.150*MGLY
,650*HCHO
.840*CO
.630*ORA1
.110*CO
. 100*ACO3
. 660*HO2
.670*HCHO
.340*X02
.500*MCO3
.500*HO2
MVK_RO2
N02
. 700*ACO3
.300*HO2
OP2
. 500*H02
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
.+
+
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
ALD
.500*MO2
.000*ALD
HO2
HO2
.079*X02
.912*NO2
.230*MACR
.079*XO2
.500*MO2
.230*MACR
.500*HO2
.320*MVK
.390*MACR
.270*HO
.200*X02
.100*ISOPROD
.250*MC03 H
.800*ALD ^
.200*ISOPROD H
.500*MO2 H
ONIT
. 500*HO2 H
. 500*MACR_RO2
HO2 n
. 150*HCHO n
.500*MO2 n
. 840*CO H
. 500*HO2 n
.150*MGLY
.210*HO n
.200*HCHO n
.330*MCO3 n
.670*ACO3 n
.500*HNO3 n
.500*ONIT n
.700*ALD n
.300*HCHO n
.200*MO2 n
f 0
f 2
h 0
h 0
h 0
h 0
h 0
h 0
h 0
h 0
h 0
h 0
^ 0
h 0
h 0
1- 0
K 0
K 0
K 0
1- 0
K 0
K 0
^ 0
K 0
K 0
K 0
K 0
K 0
K 0
v 0
.500*ORA2
.000*NO2
.912*H02
.320*MVK
.500*ORA2
.320*MVK
.362*ISOPROD
.629*HCHO
.160*MVK
.070*HO2
.200*MC03
.250*HCHO
.800*ONIT
.200*NO2
.500*ORA2
.500*ALD
. 840*KET
. 150*MGLY
. 500*ORA2
.150*HCHO
. 840*KET
.110*HO2
.100*XO2
.670*CO
.340*HO
.500*CO
.500*XO2
.700*XO2
.300*MGLY
.500*ORA2
8-85
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
{180} MVK_R02 + M02 -->
{181} MVK + O3 -->
{182} MVK + hv -->
{183} MPAN -->
{184} MdO3 + NO -->
{185} MCO3 + N02 -->
{186} MCO3 + ko2 -->
{187} MCO3 + MO2 -->
{188} MCO3 + ACO3 -->
{189} MCO3 + MC03 -->
{190} ISOPROD + HO -->
{191} IP_RO2 + NO -->
{192} IP_RO2 + HO2 -->
{193} IP_R02 + ACO3 -->
{194} IP_RO2 + MO2 -->
{195} ISOPROD + 63 -->
{196} ISOPROD + hv -->
{197} ISOPROD + NO3 -->
1
1 ' "
Rate Expression
k( 1) uses photo table NO2_RADM88
k( 2) uses photo table O3O1D_RADM88 ,
k( 3) uses photo table O3O3P_RADM88 ,
k( 4) uses photo table HONO_RADM88 ,
k( 5) uses photo table HNO3_RADM88 ,
k( 6) uses photo table HNO4_RADM88 ,
k( 7) uses photo table NO3NO_RADM88 ,
k( 8) uses photo table NO3NO2_RADM88 ,
k( 9) uses photo table H2O2_RADM88 ,
k( 10) uses photo table HCHOmol_RADM88 ,
k( 11) uses photo table HCHOrad_RADM88 ,
k( 12) uses photo table ALD_RADM88 ,
k( 13) uses photo table MHP_RADM88 ,
k( 14) uses photo table HOP_RADM88 ,
k( 15) uses photo "table PAA_RADM88
k( 16) uses photo 'table KETONE_RADM8 8 ,
k( 17) uses photo table GLYf orm_RADM8 8 ,
k( 18) uses photo table GLYmol_RADM8 8 ,
k( 19) uses photo table MGLY_RADM88 ,
k( 20) uses photo table UDC_RADM88 ,
k( 21) uses photo table ORGNIT_RADM88 ,
k( 22) •= 6.0000E-34 * (T/300) ** (-2 . 30)
k( 23) = 6.5000E-12 * exp ( 120. 0/T)
k( 24) = 1.8000E-11 * exp( 110. 0/T)
+ 0.700*ALD + 0.300*HCHO + 0.300*MQLY
0.800*HCHO + 0.500*HO2 + 0.700*ALD
+ 0.700*MO2 + 0.300*MGLY
0.670*ORA1 + 0.160*HO + 0 . 110*HO2
+ 0.110*CO + 0.950*MGLY + 0 . 100*HCHO
+ 0.050*XO2 + 0.050*ACO3
0.700*ISOPROD + 0.700*CO + 0.300*MO2
+ 0.30p*MCO3
MC03 + NO2
NO2 + " ' HCHO '+ " ': ACO3 I'H'-i"
MPAN
PAA
2.250*HCHO + 0.500*HO2 + 0.500*MO2
2.000*MO2 + HCHO
2.000*M02 + 2.000*HCHO
0.313*ACO3 + 0.687*IP_RO2
NO2 + HO2 + 0.610*CO
+ 0.270*ALD + 0.030*HCHO + 0.180*6LY
+ 0.210*MGLY + 0.700*KET
OP2
0.500*HO2 + 0.509*Mp2 + 0.500*ORA2
+ 0. 500*ALD +.0.500*KET *j , "";:"'' : ;
0.500*HCHO + 0. 500*HO2 + 0.5QO*ALD '
+ 0.500*KET
0.476*HO + 0.072*HO2 + 0.168*MO2
+ 0.237*AC03 + 0.100*X02 + 0.243*CO
+ 0.218*HCHO + 0.062*ALD + 0.278*KBT
+ 0.031*GLY + 0.653*MGLY + 0.044*ORA1
1.216*CO + 0.434*ALD + 0.350*HCHO
+ 0.216*KET + 1.216*H02 + 0.784*ACO3
0.668*CO + 0.332*HCHO + 0.332*ALp
+ ONIT . + HO2 +' ,; XO2 . .'(
. . _^ < ,.
Rate Constant
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {0 . OOOOOE+00} '"I-
scaled by l.OOOOOE+00 {0 . OOOOOE+00}
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {0 . OOOOOE+00}
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {O.OOOOOE+OOJ
scaled by l.OOOOOE+00 {0 .OOOOOE+00} ",: ,::;
scaled by l.OOOOOE+00 {0 .OOOOOE+OOJ
scaled by l.OOOOOE+00 {0 .OOOOOE+00}
scaled by l.OOOOOE+00 {0 . OOOOOE+00} .: ;
scaled by l.OOOOOE+00 JO . OOOOOE+00} "
scaled by i. OOOOOE+00 JO .OOOOOE+00}
scaled by l.OOOOOE+00 {0 . OOOOOE+00 }
scaled by l.OOOOOE+00 {0. OOOOOE+00}
scaled by l.OOOOOE+00 {0 . OOOOOE+00} "".
scaled by l.OOOOOE+00 {0. OOOOOE+00} ' '
scaled by l.OOOOOE+00 {0. OOOOOE+00 }
scaled by l.OOOOOE+00 {0 . OOOOOE+00 } I'1:"
{6.09302E-34}
{9.72293E-12}
{2.60365E-11}
8-86
-------
EPA/600/R-99/03G
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
kC
fc(
k(
k<
k(
k(
k(
25!
26!
27)
28)
29)
30)
31!
kO
= 3.
-5
= ' 2.
= 1.
= 1.
= 3.
is a
= i
2000E-11
2000E-10
OOOOE-12
6000E-12
1000E-14
7000E-12
falloff
* exp { 70
* exp { - 14 0 0
* exp( -940
* exp( -500
* exp( 240
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
expression using:
.8000E-31 * (T/300!**
C-3.20)
{4
{2
{1
{6
{2
{8
{1
.04730E-11}
.200001-10}
. 82272E-14}
.82650E-14}
.054S2E-15}
.27883E-12}
.33058E-12}
kinf = 4.7000E-12 * (T/300)**(-1.40!
F = 0.60, n = 1.00
k{ 32) = k( 31) / Keq, where Keg = 2.100E-27 * exp( 10900.0/T)
k( 33) is a special rate expression of the form:
k = kl + k2 (MI , where " . .
kl = 2.2000E-13 * exp( 620.0/T!
k2 = 1.9000B-33 * expS 980.0/T)
k( 34) is a special rate expression*of the form:
k = kl + k2 [M] , where
kl = 3.0800E-34 * exp( 2820.0/T)
k2 = 2.6600E-54 * exp( 3180.0/T)
M 35) = 3.3000E-12 * expC -200.0/T)
k( 36! is a falloff expression using:
kO = 7.0000E-31 * CT/300)**(-2.60)
kinf = l.SOOOE-11 * ST/300) ** (-0.50)
F = 0.60, n = 1.00
3.3000E-39 * exp( 530.0/T)
exp( -2500.0/T)
exp( 150.0/T)
exp( -1230,. 0/T!
k( 37)
k( 38)
k( 39)
k( 40!
1.4000E-13
1.7000E-11
2.5000E-14
k( 41! = 2.5000E-12
k( 42) is a falloff expression using:
kO m 2.2000E-30 * (T/300)**(-4.30)
kinf = 1.5000E-12 * (T/300)**(-0.50)
F = 0.60, n = 1.00
k( 43) = k( 42) / Keq, where Keq = 1.100E-27 * exp( 11200.0/T)
k( 44! = 2.0000E-21
k(
k(
k(
k(
k(
k(
k(
k(
k{
k{
k(
k(
k(
k(
k(
kC
45! is
kO
kinf =
F = 0
46) is
k = ko
kO =
k2 «
k3 =
47) =
48) =
49) is
kO
kinf =
F = 0
50) =
51) =
52) =
S3) =
54} =
55) =
56) =
57) =
58) =
59) =
60! =
a falloff expression using:
2.6000E-30 * (T/300)**(-3 .20)
2.4000E-11 * (T/300) ** (-1.30)
.60, n = 1.00
a special rate expression of the form:
+ {k3[M] / (1 + k3[M]/k2!}, where.
7.2000E-15 * exp( 785, 0/T!
4.1000E-16 * exp( 1440. 0/T)
1.9000E-33 * exp< 725.. 0/T)
1.3000E-12 * exp( 380. 0/T)
4.6000E-11 * exp( 230. 0/T)
a falloff expression using:
3.0000E-31 * (T/300)**(-3.30)
l.SQOQE-12 * (T/300!**( 0,00)
.60, n = 1.00
1.5000E-13 * (1.0 •+• 0.6*Pressure)
2.8300E+01 * (T/300)**( 2.00) * exp ( -1280. 0/T)
1.2330E-12 * (T/300)**! 2.00) * exp ( -444. 0/T)
1.S900E-11 * exp( -540. 0/T)
1.7300E-11 * expC -380. 0/T)
3.6400E-11 * exp( -380. 0/T)
2.1500E-12 * exp( 411. 0/T)
5.3200E-12 * exp( 504. 0/T)
1.0700E-11 * exp( 549. 0/T)
2.1000E-12 * exp( 322.0/T)
1.8900E-11 * exp( 116. 0/T!
{8.62399E-02}
{3.01634E-12}
{6.78905E-30}
{1.68671E-12}
{4.871441-12}
{1.95397E-38}
{3.182131-17}
{2.81225E-11}
{4.030721-16}
{2.SOOOOE-12}
{1.264401-12}
{5.47034E-02}
{2.00000E-21}
{1.14885E-11}
{1.472361-13}
{4.65309E-12}
{9.952941-11}
{8.88848E-13}
{2.40000E-13}
{3 .80672E-01}
{2.742101-13}
{2.59669E-12}
{4.833341-12}
{1.016961-11}
{8.53916E-12}
{2.886841-11}
{6.752691-11}
{6.187151-12}
{2.789431-11}
8-87
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
k( 61)
k( 62)
k( 63)
k( 64)
k( 65)
k( 66)
k( 67)
k( 68)
k( 69)
k( 70)
k( 71)
k( 72)
k( 73)
k( 74)
k( 75)
k( 76)
k( 77)
k( 78)
k( 79)
k( 80)
k( 81)
k( 82)
k( 83)
k( 84)
k( 85)
k( 86)
k( 87)
k( 88)
k( 89)
k( 90)
k( 91)
k( 92)
k( 93)
k( 94)
k( 95)
k( 96)
k( 97)
k( 98)
k( 99)
k(100)
k(101)
k(102)
k(103)
k(104)
k(105)
k(106)
k(107)
k(108)
k(109)
k(110)
k(lll)
k(112)
k(113)
k(114)
k(115)
k(116)
k(117)
k(118)
k(119)
k(120)
k(121)
=
=
=
=
=
=
=
=
=;
=
=
=
=
=
=
=
=
=:
=
=
=
=:
=
=
=
=:
=
=
=
=
:;
=:
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
:=
=
=
=
=
=
=
=
=
=
=
=
4
9
9
6
1
1
1
2
1
1.
1.
6.
1.
2.
1.
4.
1.
4.
4 .
4 .
4.
4.
4.
4.
4.
6.
1.
6.
1.
1.
2.
2.
1.
3.
1.
1.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
7.
1.
1.
4.
3 .
.,
OOOOE-11
OOOOE-01
OOOOE-12
8700E-12
2000E-11
1500E-11
7000E-11
8000E-11
OOOOE-11
OOOOE-11
OOOOE-11
1650E-13
5500E-11
8000E-12
9500E+16
7000E-12
9500E+16
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
OOOOE-13
4000E-12
OOOOE-13
4000E-12
4000E-12
2000E-11
OOOOE-12
OOOOE-11
2300E-11
2000E-14
3200E-14
2900E-15
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
7000E-14
9000E-13
4000E-13
2000E-14
4000E-14
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
k( 61)
exp(
exp(
256
-745
(T/300)**(
exp (
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exP(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp (
exp(
exp (
exp(
exp(
exp(
exp(
exp (
exp(
exp(
exp(
exp(
exp(
exp(
exp(
exp(
-540
181
-13543
-13543
180
180
180
180
180
180
180
180
180
180
180
180
180
180
-2058
-1900
-2058
-1900
-1900
-2923
-1895
-975
-2633
-2105
-1136
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
1300
220
220
220
220
.0/T)
.0/T)
2.00) * exp( -444. 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
{4-
{3.
{9.
{I-
{9-
{1-
{1.
{2.
{1.
{1.
{1-
{1.
{2.
{5-
{3-
{4.
{3.
{7-
{7.
{7.
{7.
{7.
{7.
{7.
{7.
{7.
{7-
{7.
{7.
{7.
{7.
{6.
{2-
{6.
{2.
{2.
{2-
{1-
{1.
{1.
{1.
{1-
{1.
{6.
{6.
{6.
{6.
{6.
{6.
{6.
{6-
{6-
{6-
{6.
{6.
{6-
{6.
{3-
{2-
{8.
{7.
OOOOOE-ll) ::
60000E-11}
OOOOOE-12)
62197E-11}
85020E-13}
isoooE-ii} " ' ;
70000E-11)
80000E-11}
OOOOOE-ll}
OOOOOE-ll}
OOOOOE-ll}, ,
37105E-13}
53137E-12}
13974E-12} ;:;;
57235E-04}
70000E-12}
57235E-04} :
68378E-12}
68378E-12}
68378E-12}',!,,: ..,;,
68378E-12}
68378E-12}"
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12}
68378E-12 } ' .:,!
68378E-12}
01030E-16}
38307E-15}
01030E-16}
38307E-15}
38307E-15}
20000E-11} "
09940E-16} , "•;
73099E-i4}
22539E-12}
74559E-18}
12933E-17}
61125E-16}
04038E-12}
04038E-12}
04038E-12}
04038E-12})? "•
04038E-12}
04038E-12}
04038E-12}
04038E-12J
04038E-12}
04038E-12}
04038E-12}
04038E-12}
04038E-12}
04038E-12}
97533E-13}
92919E-13}
78758E-14}
11376E-14}
8-88
-------
EPA/600/R-99/030
Table 8A-11, RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
k(122)
It (123)
k(124)
k(125)
It (126)
k(127)
k!128)
k(129)
k(i3o)
k(131)
It (132)
k(133)
k(134)
k(135)
k(136)
k(137)
k(138)
k(139)
k(140)
k(i4l)
k(142)
k(143)
k(144)
k(14S)
k(146)
k(147)
k(148)
k(149)
k(lSO)
k(151)
k<152)
k(153)
k(154)
k(15S)
k(156)
MJ.57)
k(158)
k(159)
k'(160)
kSlSl)
k(162)
k(163)
k(164)
k{165)
k(166)
k(167)
k(168)
k(169)
k(170)
fcCl71)
k(172)
k{173)
k(174)
M175)
k(176)
k(177)
k(178)
k(179)
k(180)
k(181)
k(182)
=
=;
=
=
=
B:
=
s=
=
=:
=:
=
=
=
=
=.
=
=
3=
=
=
=
=
=
=
=
=
=
E=
=
2
1
1
1
1
9
1
1
9
1
3
1
8
7
3
3
4
4
1
4
4
1
4
3
7
1
4
3
4
4
7
1
4
3
2
4
7
8
3
7
3
3
4
7
8
3
1
4
7
8
3
1
uses
=
=
=
=
=
=
D
1
4
4
7
a
3
7
uses
.9000E-14
.4QQQE-13
.40001-13
.70001-14
.7000E-14
.6000E-13
.70001-14
-7000E-14
.6000E-13
.70001-14
.4000E-13
.OOOOE-13
.4000E-14
.2000E-14
.4000E-13
.4000E-13
.2000E-14
.200QE-14
.19001-12
.20001-14
.2000E-14
.1900E-12
.20001-14
.6000E-16
.700QE-14
.7000E-14
.2000E-14
.6000E-16
.2000E-12
.2000E-12
.7000E-14
.70001-14
.20001-14
.6000E-16
,54001-11
.2000E-12
.7000E-14
.4000E-14
.40001-14
.86001-15
.6000E-11
.03001-12
.20001-12
.7000E-14
.4000E-14
.4000E-14
.86001-11
.2000E-12
.7000E-14
.4000E-14
.4000E-14
.3600E-15
* exp( 220
* exp ( 220
* exp(, 220
* exp( 220
* exp ( 220
* exp( 220
* exp( 220
* exp( 220
* exp ( 220
* expC 220
* exp( 220
* exp( 220
* expC 220
* exp( 220
* exp( 220
* exp( 220
* expC 220
* exp( 220
* exp( 220
* expC 220
* exp{ 220
* exp( 220
* expC 220
* expt 220
* exp( 1300
* exp( 220
* expC 220,
* expt 220.
* exp( 180.
* exp( 180.
* exp{ 1300
* exp( 220,
* exp{ 220,
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
w
*
photo table
.SOOOE-12
-1400E-12
.20001-12
.7000E-14
.4000E-14
.4000E-14
,51001-16
*
*
*
•*
*
*
*
photo table
exp( 220,
(T/300)**(
CT/300)**t
{T/300)**(
(T/300)**(
(T/300>**{
(T/300)**(
(T/300)**(
(T/300)** (
(T/300)**t
(T/300) **(
(T/3005** t
tT/300)**(
(T/300)**(
(T/300) **{
(T/300) **(
(T/300) **(
(T/300) ** (
ACROL1IN
(T/300) **(
(T/300) **(
(T/300) **(
(T/300)**(
(T/300) **(
(T/300)** (
(T/3005 **(
aCROLEIN
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
• 0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
.0/T)
,0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
1.00)
1,00)
1.00)
1.00)
1.00)
1.00)
1,00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1,00)
1.00)
1,00)
1.00)
1.00)
1.00)
1.00)
1.00)
*
-*
*
*
*
exp (
exp<
exp(
expt
expt
407
181
1298
221
221
« exp{ -1912
#
*
•*
*
*
*
*
*
*
*
*
,
*
*
*
*
*
*
*
,
expt
expC
exp (
expt
expC
expt
exp (
expt
expE
exp(
expt -
scaled
exp( -
exp(
expS
expE
expt
exp(
expS -
scaled
-447
181
1298
221
221
176
181
1298
221
221
2113
by
1726
452
181
1298
221
221
1519
by
.6/T)
,2/T)
.3/T)
.4/T)
.4/T)
,2/T)
.9/T)
.2/T)
.3/T)
.4/T)
.4/T)
.1/T)
.2/T)
.3/T!
.4/T)
.4/T)
.5/T)
3.600001-03
.0/T)
.9/T)
.2/T)
.3/T)
.4/T)
.4/T)
.7/T)
1.11000E-02
{6.06762E-14}
{2.92919E-13}
{2.92919E-13}
{3.55688E-14}
{3.556881-14}
{2. 008591-12}
{3.55S88E-14}
{3.556881-14}
{2.008591-12}
{3.55688E-14}
{7.11376E-13}
{2.092281-13}
{1.757521-13}
{1,506441-13}
{7.113761-13}
{7.11376E-13}
{8.78758E-14}
{8.78758E-14}
{2.489811-12}
{8.787581-14}
{8.787581-14}
{2.489811-12}
{8.787581-14}
{7.53221E-16}
{6.04038E-12}
{3.556881-14}
{8.787581-14}
{7.53221E-16}
{7.68378E-12}
{7.68378E-12}
{6.04038E-12}
{3.55688E-14}
{8.787581-14}
{7.S3221E-16}
{9.90719E-11}
{7.6633SE-12}
{5.965981-12}
{1.75402E-13}
{7.09961E-14}
{1.27569E-17}
{3,600001-11}
{6.69552E-13}
{7.663351-12}
{5.965981-12}
{1.75402E-13}
{7.099S1E-14}
{3.33618E-11}
{7.66335E-12}
{S. 965981-12}
{1.75402E-13}
{7.09961E-14}
{1.123301-18}
{0.000001+00}
{4,547531-15}
{1. 879901-11}
{7.66335E-12}
{5.9659BE-12}
{1.754021-13}
{7.099S1E-14}
{4.54966E-18} '
{o. 000001+00}
8-89
-------
EPA/600/R-99/030
Table 8A-11. RADM2_CIS4 and RADM2_CIS4_AQ Mechanisms
k(183!
k(184)
k(185)
k(186)
k{187)
k(188)
kS189)
k(190J
k(!91)
k!i92!
k{193)
k(194)
k(19S)
k(196)
kE197)
m
=
IS
IS
3E
=
•
=
a
•
•
•
m
1
4
2
7
9
1
1
3
4
7
8
3
7
uses
=
1
« ;',, i
.6000E+16
.20001-12
.8000E-12
.7000E-14
.6000E-13
.19001-12
.1900E-12
.3SOOE-11
.2000E-12
.7000E-14
.4000E-14
.4000E-14
.1100E-18
*
*
*
*
*
*
*
*
*
*
*
photo table
.00001-13
(T/3QO)**(
(T/3QO!**(
(T/300)**C
(T/300)**(
(T/300)**{
CT/300)**C
(T/300)**(
(T/300)**(
(T/300)**(
CT/3QO)**C
(T/300)** (
ACROLEIN
1
1
1
1
1
1
1
1
1
1
1
.00}
.00!
.00)
.00)
.00)
,00)
.00)
.00)
.00)
.00)
.00)
* eXp(-1348S.O/T)
* expC
*.exp(
* exp(
* exp{
* exp(
* expE
* exp(
* expS
* exp{
* expC
, scaled
j
181. 2/T)
181. 2/T)
1298. 3/T)
221. 4/T)
221. 4/T)
221. 4/T)
181. 2/T)
1298. 3/T)
221. 4/T)
221. 4/T)
by 3.SOOOOB-03
(3. 52 53 SB- 04} '"
{7.66335E-12}
{S.10890E-12 )•-•» • '• •
{S.96598E-12}
{2.00460E-12} . ,
{2.48486E-12}
{2.484SSE-12}
{3.36000E-11}
{7.6633SE-12}"- "
{5.96S98E-12}""
(l. 754028-13} . .
{7.09961E-14}
{7.11000E-18} J;|' ',, -
{O.OOOOOE+OO}!1-
|l.OOOOOE-i3} ":
"ftf
.,.
If *
... •'' 1
: '*
-'= €
8-90
-------
EPA/600/R-99/030
Table 8A-12. RADM2_C1S4_AE and RADM2_CIS4_AE_AQ Mechanisms
Reaction List
{ 1} NO2
{ 2} O3
{ 3} 03
{ 4} HONO
{ 5} HN03
{ 6} HN04
{ 7} NO3
{ 8} N03
| 9} H202
{ 10} HCHO
{ 11} HCHO
{ 12} AU>
{ 13} OP1
{ 14} OP2
{ 15} PAA
{ 16} KBT
{17} QLY
{ 18} SLY
{ 19} MQLY
{ 20} DCS
{ 21} ONIT
{22} O3P
{ 23} O3P
{ 24} 01D
{ 25} O1D
{ 26} O1D
{ 27} 03
{ 28} O3
{ 29} 03
{ 30} HO2
{ 31} H02
{ 32} HN04
{ 33} H02
{ 34} HO2
{ 35} H2O2
{ 36} NO
{ 37} NO
{ 38} O3
{ 39} N03
| 40} NO3
{ 41} M03
{ 42} NQ3
{ 43} N2OS
{ 44} N2O5
{ 43} HO
{ 46} HO
{ 47} HO
{ 48} HO
{ 49} HO
{ 50} CO
{ 51} HO
{ 52} ETH
{ 53} HC3
{ 54} HC5
| 55} HC8
{ 56} OL2
{ 57} OLT
+ hv •
+ hv
4- hv
+ hv
+ hv
•<• hv
+ hv
+ hv
+ hv
+ hv
+ hv
+ • hv
+ hv
+ hv •
+ hv
+ hv
+ hv
+ hv
+ hv
+ hv
•»• hv
+ (Ml + [02 J
+ N02
+ CH2]
+ [02]
-+ [H2O]
+ NO
+ HO
+ H02
+ NO
+ NO2
+ HO2
+ H02 + [H20J
+ HO
+ HO
+ NO + [O2J
+ NO2
+ NO
+ NO2
+ HO2
+ N02
+ [H20]
+ NO2
+ HNO3
+ HN04
•f HO2
+ S02
+ HO
+• HO
+• HO
+ HO
+ HO
+• HO
+ HO
•<•» O3P
--> " O1D
— > O3P
••>-> HO
- --> - • HO
--> H02
" --> NO
-•-> N02
'-•>' 2.000*HO
'-»> CO
--> H02
'--> MO2
--> HCHO
...» • ALD
--> MO2
--> AC03
--> 0.130*HCHO
--* 0.450*HCHO
-«> AC03
—?. 0.980*HO2
--> 0.200*AIJJ
+ NO2
--> O3
-•> NO
--> O3P
--> 03P
--> 2.000*HO
--?. NO2
--> HO2
--> HO
— •> . NO2
- - > HN04
--> H02
•»-> H202
--> ' ' H202
--> HO2
--> HONO
--> 2.000*NO2
--> NO3
--> 2.000*NO2
--> NO
--> HN03
--> N205
--> NO2
-r> 2.000*HN03
--> - HN03
--> NO3
--> NO2
-,>
--> SUM1
--? HO2
- - > MO2
--> ETHP
--> 0.83Q*HC3P
4- 0,07S*ftLD
--> HCSP
-,-•> HC8P
--* OL2P
— > OLTP
+• ' NO
+ NO
+ N02
+ 'N02
+ ' " 03P
4- HO2 4- CO
4- " ' HO2 ' ' + " ' ' CO
4- H02 + '•' HO
' 4- H02 + HO
4- HO
4- ETHP
4- 1.870*CO
4. 1,550*CO 4- 0.800*HO2
4- H02 4- CO
4- 0.020*ACO3 4- IC03
4- 0.800*KET 4- HO2
+ HO
4- NO2
+ NO2
4- NO3
4- HO2 4- SULA1R
4- 0.170*HO2 + 0,009*HCHO
+ 0.025*KET -
4- 0.250*XO2
4- 0.750*XO2 4- HC8AER
8-91
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
{58} OLI
{ 59} TOL
{ 60} XYL
{ 61} CSL
{ 62} CSL
{ 63} HCHO
{ 64} ALD
{ 65} KET
{ 66} GLY
{ 67} MGLY
{68} DCB
{ 69} OP1
{ 70} OP2
{ 71} PAA
{ 72} PAN
{ 73} ONIT
{74} ACO3
| 75} PAN
{76} TCO3
{ 77} TPAN ,
{78} M02
{79} HC3P
{ 80} HCSP
{ 81} HCSP
{ 82} OL2P
{83} OLTP
{ 84} OLIP
{ 85} ACO3
{86} TCO3
{ 87} TOLP
{ 88} XYLP
{ 89} ETHP
{90} KETP
{ 91} OLN
{92} HCHO
{ 93} ALD ,
{ 94} GLY
{ 95} MGLY
{ 96} DCB
{ 97} CSL
{ 98} OL2
{99} OLT
{100} OLI
{101} OL2
{102} OLT
+ HO
-i- HO
+ HO
+ .HO
-t- HO
+ HO
+ 'HO
•f HO
•*• po
+ HO
+ JHO •
•f HO
+ HO
+ HO
+ HO
+ HO
+ NO2
i
+ NO2
.1
+ NO .
•f NO .
+ NO
+ NO,
r
+ NO
+ NO
+ NO
+ NO
-1- NO
•f NO
+ NO
+ NO
+ NO
+ NO
+ N03
•f N03
+ N03
+ NO3
+ NO3
+ N03
•t- ,N03
+ NO3
+ NO3
+ ,03
•*• O3
--> • OLIP
--> 0,750*TOLP
+ TOLAER
--> 0.830*XYLP
• . •(- XYLAER
--> 0.100*HO2
'-(• CSLAER
.--> CSL
--> HO2
- - > AC03
--> KETP
--> HO2
--> AC03
--> TC03
--> 0.500*MO2
--> 0.500*HC3P
--> AC03
--> HCHO
--> HCSP
--> PAN
-->... ACO3
--> TPAN. .
- ,,, . .. , . --> . TCO3
--> , HCHO
--> 0.750*aLD
+ 0.036*OSIT
--> 0.380*ALD
.+ 0.920*N02'
--> ' 0.350*ALD
+ 0.240*ONIT
--> - 1.600*HCHO
•1- 0.200*ALD
— > KLD
+ .'. . ' NO2
--> H02
+ 0.100*KET
- - > MO2
--> NO2
+ 0.110*MGLY
+ 2.000*XO2
--> N02
+ 0.160»GLY
--> N02
+ 0.806*DCB
--> ALD
--> - MGLY
--> HCHO
- - > HO2
--> ACO3
--> HNO3.
- - > HNO3
--> HN03
--> HNO3
+ 0.500*CSLAER
--> _ OLN
--> ' OLN
--> _ _ OLN
--> HCHO
+ 0,120*HO2
--> 0.530*HCHO
+ 0.200*ORA1
+ OLIABR
+ 0,250*CSL
+ 0.170*CSL
+ 0,900*XO2
+ CO
+ 2.000*CO
+ CO
+ 0.500*HCHO
-(• , 0 . 500+ALD
+ N03
+ NO2
+ NO2
+ . NO2
+ HO2
+ 0.250*K1T
+ 0.964*NO2
+ 0.690*KET
+. 0.920*H02
+ 1.060*KET
+ 0,760*NO2
-1- HO2
+ HCHO
•f 1,450*ALD
+ N02
•f NO2
+ 0.920*HO2
+ 0.050*AC03
+ HO2
+ 0.700*DCB
+ HO2
+ H02
+ ' N02
+ ALD
+ HNO3
+ HN03
+ H02
+ ACO3
+ TCO3
+ XNO2
+ OLIABR
+ 0.400*ORA1
+ O.SOO*ALD
-t- 0.200*ORA2
+ 0.2SO*HO2
+ 0.170*HO2
- " '•:' . i
+ 0,900*TCO3
. i , .,
t: ' •„ ' •• niiiiin
+ 0.500*HO
+ O.SOO*HO
+ X02
. . .:
+ NO2 " !
+ 0.090*HCHO
+ 0,964*HO2
+ 0.080*ONIT s
+ 0.040*HCHO ,
-t- 0.760*H02
+ N02
+ HO2
+ 0.280*HCHO
+ 0.890*0LY
+ 0.950*CO»- -. •«
* • : t ' . , \
'+ 0.170*MGLY
+ 0.450*MSLY
"+ " NO2 • ' '
+ H02
+ 2.000*K02
+ CO
'V 2.obo*co
+ CO
+ 0,SOO*CSL
+ 0.420*CO
+ 0.330*CO
+ 0.230*HO2
8-92
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
{103} OLI
{104} H02
{105} H02
{106} HO2
{107} HO2
{108} HO2
{109} HO2
{110} HO2
{ill} HO2
{112} HO2
{113} HO2
{114} H02
{115} HO2
{116} HO2
{117} HO2
{118} MO2
{119} MO2
{120} MO2
{121} M02
{122} MO2
{123} MO2
{124} MO2
{125} MO2
{126} M02
{127} MO2
{128} MO2
{129} MO2
{130} M02
{131} MO2
{132} ETHP
{133} HC3P
{134} HC5P
{135} HC8P
{136} OL2P
{137} OLTP
{138} OLIP
{139} KETP
{140} ACO3
{141} AC03
+ 03
+ MO2
+ ETHP
+ HC3P
+ HC5P
+ HC8P
+ OL2P
+ OLTP
+ OLIP
+ KETP
+ AC03
+ TOLP
+ XYLP
+ TC03
+ OLN
+ MO2
+ ETHP
-1- HC3P
+ HC5P
+ HC8P
+ OL2P
+ OLTP
4- OLIP
+ KETP
+ ACO3
+ TOLP
+ XYLP
+ TC03
+ OLN
+ ACO3
+ ACO3
+ ACO3
+ JVCO3
+ ACO3
+ ACO3
+ ACO3
+ ACO3
+ ACO3
+ TOLP
+ 0.220*M02
--> 0,180*HCHO
+ 0.230*CO
+ 0.2SO*HQ2
+ OLIAER
--> OP1
--> OP2
--> OP2
--> OP2
--> OP2
--> OP2
--> OP2
--> OP2 '
--> OP2
--> PAA
--> OP2
--> OP2
--> OP2
--> OMIT
— > 1.500*HCHO
--> 0.7SO*HCHO
--> 0.840*HCHO
+ HO2
--> 0,770*HCHO
+ H02
--> 0.800*HCHO
+ H02
--> 1,5SO*HCHO
— > 1.2SO*HCHO
--> 0.890*HCHO
+ 0,550*KET
--> 0.750*HCHO
--> HCHO
+ 0.500*ORA2
--> HCHO
+ 0.700*DCB
--> HCHO
+ 2.000*H02
--> 0.500*HCHO
+ 0.500*ORA2
+ 0.475*CO
--> • 1.750*HCHO
+ NO2
- - > ftLD
+ 0.500*ORA2
---> 0.770*ALD
+ 0.500*MO2
--> 0,410*ALD
+ O.SOO*MO2
•-> 0.460*ALD
+ 0.500*MO2
— > 0.800*HCHO
+ 0.500*M02
. --> ALD
+ 0.500*MO2
--> 0.725*ALD
+ 0.500*HO2
- - > MGLY
•f • 0 . 500*ORA2
--> 2,000*M02
-->' M02
+ 0.100*HO
+ 0.72Q*AL0
+ 0.060*ORA1
+ 0.140*HO
+ HO2
+ H02
+ 0.770*ALD
+ 0.410*ALD
+ 0.460*ALD
+ 0.350*ALD
+ 0.7SO*ALD
+ 0.72S*AL0
+ 0.750*MGLY
+ 0,500*HO2
+ 0.170*MGLY
+ 2,000*HO2
+ 0,450*MGLY
+ 0.445*OLY
+ 0.025*ACO3
+ X02
+ 0.500*HO2
+ 0.500*H02
+ 0.260*KET
+ 0.500*ORA2
+ 0.750*KET
+ 0.500*ORA2
+ 1,390*KET
+ 0.500*ORA2
+ 0.600*ALD
+ 0,500*ORA2
+ 0.500*HCHO
+ 0.500*ORA2
+ 0.550*KET
+ 0.5QO*MO2
+ 0.500*HO2
+ 0.170*MGLY
•+ 0.100*KET
+ 0.290*ORA2
-1- 0.310*MO2
f
+ 0.7SO*ALD
+ 0.260*KET
+ 0.750*KET
+ 1.390*KET
+ H02
+ H02
+ HO2
+ H02
+ 0,SOO*MO2
+ 0.160*GLY
+ 0.806*DCB
+ 0.05S*MGLY
+ 0.460*HO2
+ ALD
+ O.SOO*MO2
+ 0.500*HO2
+ 0.500*H02
+ O.SOO*HO2
+ 0.500*HO2
+ 0.500*HO2
+ 0.140*HCHO
•f 0,500*ORA2
+ 0.500*MO2
+ 0.160*GLY
8-93
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
{142}
{143}
{144}
{145}
{146}
{147}
{148}
{149}
{150}
{151}
{152}
{153}
{154}
{155}
{156}
{157}
{158}
{159}
{160}
{161}
{162}
{163}
{164}
{165}
{166}
{167}
{168}
{169}
{170}
{171}
{172}
{173}
{174}
{175}
{176}
{177}
ACO3
ACO3
ACO3
OLN
XO2
XO2
X02
X02
X02
XMO2
XNO2
XNO2
XNO2
XSO2
TSRP
TERP
TERP
ISO
ISO_RO2
ISO_RO2
ISO_RO2
ISO_R02
ISO
ISO
ISO
ISONJR02
ISON_RO2
ISON_RO2
ISON_RO2
MACR
MACR_R02
MACR_RO2
MACR_RO2
MACR_R02
MACR
MACR
4-
4-
4-
4-
4-
+
4-
+
4-
+
4-
4-
4-
4-
+
+
+
+
4-
+
4-
4-
+
+•
4-
4-
+
4-
4-
4.
4-
4-
+
4.
+
4-
XYLP
TC03
OLH
QLN
BQ2
MO2
ACO3
X02
NO
NO2
HO2
HO2
ACO3
XNO2
HO
NO3
03
HO
NO
HO2
AC03
I
MO2
,
03
O3P
NO3
NO
HO2
ACO3
M02
HO
NO
HO2
ACO3
HO2
03
I
hv
4- 0
-->
+
-->
+ 0
+ 2
-->
+
--> 2
— >
-->
' -->
-->
-->
— >
-->
-->
-->
-->
— >
-->
-->
.->
--> 0
+ 0
4- 0
-->
--> 0
4- 0
4- 0
--> 0
4- 0
--> 0
4- 0
4- 0
4- 0
--> 0
4- 0
-->
-->
4- 0
-->
--> 0
4-
--> 0
+ 0
--> 0
-->
4- 0
— >
--> 0
4- 0
4- 0
--> 0
+ 0
--> 0
4- 0
+ 0
--> 0
+ 0
4- 0
.700*DCB
MO2
HO2
M02
,110*MGLY
.000*XO2
HCHO
NO2
.000*HCHO
OP2
' 'HCHQ
MO2
NO2
ONIT
OP2
HCHO
MO2
T1RPAER
TERPAER
TERPAER
ISO_RO2
.088*ONIT
.362*ISOPROD
,629'HCHO
OP2
.500»HO2
.362*ISOPROD
.629*HCHO
.SOO'HCHO
,230*MACR
.600*HCHO
.390*ORA1
.070*CO
.150'ALD
,750*ISOPROD
.250*MO2 '
ISON_RO2
NO2
.800*HO2
ONIT
.500*HO2
ALD
.500*HCHO
.SOO'ONIT
.500*MC03
N02
.840*CO
OP2
.500*HO2
.840*KET
.150*MGLY
.650*HCHO
,840*CO
.630*ORA1
.110*CO
.100*ACO3
.660*HO2
.670*HCHO
.340*X02
+
4-
4-
4-
4-
4-
4-
+
4-
+
+
4-
4-
4-
4-
4.
4-
4-
4-
4-
4-
+
4-
4-
+
+
4-
4-
4-
+
4-
4-
4-
4-
+
4-
4-
+
4-
4^
4-
0
0
0
0
2
0
Q
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
HO2
,450*MGIiY
.920*H02
~.050*AC03
• • -
ALD
.500*MO2
.000*ALD
H02
HO2
HO
N03
03
. 079*X02
.912*NO2
.230*MACR
.079*XO2
. SOO*MO2
,230*MRCR
.500*HO2
.320*KVK
.390*MACR
.270*HO
.200*X02
.100*ISOPROD
-250*MCO3
.800*ALD
.200*ISOPROD
.500*MO2
ONIT
.500*HO2
,500*MACR_RO2
HO2
.150*HCHO
.SOO*MO2
.840*CO
.500*HO2
.150*MGLY
.210*HO
.200*HCHO
.330*MCO3
. 670*ACO3
4-
4-
4-
j
4-
4-
/
o
4-
4-
L,
4>
4-
4-
4-
4-
4-
4-
4-
f
4-
•""
+
4-
i-
4-
+
4-
4-
4-
4-
4-
4-
4-
0
0
0
0
2
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.806*DCB
. 890*GLY
. 950*CO '^
t i ' « :i
,500*ORA2
.000*NO2
'.I'
4
. - .
. 912*HO2
.320*HVK
i, !
,500*ORA2
.320*MVK
.362*ISOPROD "
.629*HCHO
. 160*MVKY ; "t '
. 070*H02
.200*MCO3
,250*HCHO
;l~ . -
.8QO*ONIT
.200*N02
. .,
,500*ORA2
.500*AIJ3
3J1
.840*KBT
.150*MGLY
::.
.500*ORA2
.150«HCHO
.840*KBT
.110*H02
.100*XO2
,670*CO 'i
,340*HO
8-94
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
{178} MACR + N03
{179} MVK + HO
{180} MVK_R02 + NO .
{181} MVK_RO2 + H02
{182} MVK_RO2 + RCO3
{183} MVK_RO2 + MO2
{184} MVK + O3
{185} MVK + hv
{186} MSAN '
{187} MC03 + NO
{188} MCO3 + NO2
{189} MCO3 + HO2
{190} MCO3 + M02
{191} MCO3 + ACO3
{192} MCO3 + MCO3
{193} ISOPROD + HO
{194} IP_RO2 + NO
{195} IP_R02 + HO2
{196} IP_RO2 + ACO3
{197} IP_RO2 + MO2 •
{198} ISOPROD + O3
{199} ISOPROD + hv
. {200} ISOPROD + N03
Rate Expression
k( 1} uses photo table
k( 2) uses photo table
k( 3) uses photo table
k( 4} uses photo table
k( 5) uses photo table
kC 6S uses photo table
k( 7) uses photo table
k( 8) uses photo table
kS 9) uses photo table
k( 10) uses photo table
k( 11! uses photo table
k( 12) uses photo table
k( 13) uses photo table
k( 14) uses photo table
k( 15) uses photo table
k{ 16) uses photo table
— > 0.500*MCO3 + 0.500*HN03 + 0.500*CO
+ 0.500*HO2 + 0.500*ONIT + 0.500*XO2
- - > MVK_RO2
: --> NO2 + 0,700*ALD + 0.700*XO2
+ 0.700*AC03 + 0.300*HCHO + 0.300*MGLY
+ 0.300*HO2
--> OP2
--> 0.500*HO2 -1- 1,200*MO2 + 0,500*ORA2
•1- 0.7QO*ALD + 0.300*HCHO + 0.300*MGLY
--> 0,800*HCHO + 0.500*HO2 + 0.700*ALD
+ 0.700*MO2 + 0.300*MGLY
--> 0.670*ORA1 + 0.160*HO + 0.110*H02
+ 0.. 110*CO + 0.950*MGLY + 0.100*HCHO
-1- 0.050*XO2 + O.OSO*ACX)3
' --> 0.700*ISOPROD + 0.700*CO •(• 0.300*MO2
+ 0.300*MCO3
--> MCO3 + NO2
--> NO2 + HCHO + ACO3
- - > MPAN
--> PRA
--> 2,250*HCHO + 0,500*HO2 + 0.500*MO2
--> 2.000*MO2 + HCHO
--> 2.000*MO2 •)- 2.000*HCHO
--> 0.313*ACO3 + 0.687*IP_RO2
--> NO2 + HO2 + 0,610*CO
+ 0.270*ALD + 0.030*HCHO + 0.180*GLY
+ 0.210*MGLY -(- 0.700*KET
- - > OP2
--> 0.500*HO2 + 0.500*MO2 + O.SOO*ORA2
+ O.SOO*ALD -I- O.SOO*KET
--> 0.500*HCHO + 0.500*HO2 + O.SOO*ALD
+ 0.500*KET ,
--> 0.476*HO -I- 0.072*H02 + 0.168*MO2
+ 0.237*ACO3 + 0.100*XO2 + 0.243*CO
+ 0.218*HCHO -1- O.OS2*ALD + 0.278*KET
+ 0.031*GLY + 0.653*MGI,Y + 0.044*ORA1
--> 1.216*CO + 0.434*ALD + 0.350*HCHO
•(- 0.216*KET -1- 1.216*HO2 + 0.784*ACO3
--> 0.668*CO + 0.332*HCHO + 0.332*ALD
+ ONIT + HO2 + XO2
Rate Constant
N02_RADM88 , scaled by 1,000001+00 {0 .OOOOOB+00}
03O1D_RADM88 , scaled by l.OOOOOE+00 {0 . OOOOOE+00}
03O3P_RADM88 , scaled by l.OOOOOE+00 {0. OOOOOE+00}
HONO_RADM88 , scaled by l.OOOOOE+00 {0 .000001+00}
HN03_RADM88 , scaled by l.OOOOOE+00 {0 . OOOOOE+00}
HHO4_RADM88 , scaled by 1, OOOOOE+00 {0. OOOOOE+00 }
NO3NO_RADM88 , scaled by l.OOOOOE+00 {0 . OOOOOE+00 }
NO3N02_RADM88 , scaled by l.OOOOOE+00 {O.OOOOOE+OOJ
H2O2_RADM88 , scaled by l.OOOOOE+00 {0 .OOOOOS+00}
HCHOmol_RADM88 , scaled by l.OOOOOE+00 {0 . OOOOOE+00}
HCHOrad_RADM88 , scaled by 1. OOOOOS+00 {O.OOOOOE+00}
ALD_RADM88 , scaled by 1. OOOOOS+00 {0 .OOOOOE+00}
MHP_RADM88 , scaled by l.OOOOOE+00 {0. OOOOOE+00 }
HOP_RADM88 , scaled -by l.OOOOOE+00 {O.OOOOOE+00}
PAA_RADM88 , Scaled by l.OOOOOE+00 {0 . OOOOOE+00}
KETONE_RADM88 , scaled by l.OOOOOE+00 {O.OOOOOE+00}
8-95
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2__CIS4_AE_AQ Mechanism! .
k( 17)
k( 18)
kC 19)
k( 20)
k( 215
k{ 22)
k( 23)
k( 24)
k{ 25)
k( 26)
kC 27)
k{ 28)
k( 29)
kS 30)
k( 31)
kO
kinf
F =
kC 32)
k( 33)
uses photo table GLYform__RM>M88 , scaled by l.OOOOOE+OO
uses photo table GLYmol_RADM88 , scaled by l.OOOOOE+00
uses photo, table MGLY_RADM88 , scaled by 1.00000I+00
uses photo table ODC_RADM88 , scaled by l.OOOOOE+00
uses photo table ORGNIT_RRDM88 , scaled by l.OOOOOE+QO
= 6.0000E-34 * (T/300)**(-2.30)
= 6.SOOOE-12 * exp( 120. 0/T)
= 1.80QOE-11 * exp( 110. 0/T)
- 3.2000E-11 * exp( 70. 0/T)
= 2.2000E-10
= 2.0000S-12 « expS -1400. 0/T)
= 1.6000E-12 * exp( -940. 0/T)
» 1.1000E-14 * expS -500. 0/T)
= 3.7000S-12 * exp( 240. 0/T)
is a falloff expression using:
m 1.8000S-31 * (T/300)** (-3.20)
«= 4.70001-12 * (T/300) ** (-1.40)
0.60, n « 1.00
= k( 31) / Keg, where Keg = 2.100E-27 * exp( 10900. 0/T)
is a special rate expression of the form:
!o
{o
{0
{0
{0
[6
{9
(2
{4
{2
{1
{6
{2
{8
{1
{8
{3
».'!»*• 1
.OOOOOE+00} , ;
.OOOOOE-i-OO)
.QQOOOE-t-OO}*" ."
. OOOOOE+00}
.000001+00} .
,093021-34}
.72293B-12)
.6036SE-11} ,
.04730E-11}
.20000E-10} :
-82272E-14}"'
.82650E-14}
.05452E-15}
.27883E-12},, ,
.39058E-12}
.623991-02}jj . ,• <:!
. 01634E-12} ;•• .:, •
k = kl + k2 [M] f where '• ••-" • ' • '
kl = 2.2000E-13 * exp! 620.0/T)
k2 = 1.9000E-33 * exp( 980.0/T)
k( 34) is a special rate expression of the form
k = kl + k2[M]t, where ••-.«:
kl = 3.08QOE-34 * exp( 2820.0/T)
k2 m 2.6600E-S4 * exp( 3180.0/T)
k( 35) = 3.3000E-12 * exp( -200.0/T)
is a falloff expression using:
= 7.0000E-31 * (T/300)**(-2.60)
- 1.5000E-11 * (T/300)**(-0.50)
0.60, n = 1.00
3.3000S-39 * exp( 530.0/T)
1.4000E-13 * exp( -2500.0/T)
1.7000E-11 * exp( ISO.0/T)
2.5000E-14 * exp( -1230.0/T)
2.5000E-12
s a falloff expression using:
- 2.2000E-30 * (T/300)**(-4.30)
1.5000E-12 * (T/300!**(-0.50!
0.60, n '= 1.00
k( 42) / Keq, where Keq =
1.100E-27 * exp( 11200. 0/T)
k( 36)
kO
kinf
F o
k( 37) =
k! 38) -
k( 39! m
k( 40) =
k( 41) =
k( 42) is
kO
kinf =
F
k( 43) =
k( 44) = 2.0000E-21
k( 45) is a falloff expression using:
kO = 2.6000E-30 * (T/300)**(-3.20)
kinf = 2,4000^-11 * (T/300)**(-1.30)
F = 0.60, n m 1.00 "
k( 46) is a special rate expression of the forms
k = ko + (k3 [M] / (1 + k3[M]/k2S}, where
kO = 7.2000E-15 * exp( 785.0/T)
k2 = 4.1000E-16 * exp! 1440.0/T!
k3 = 1.9000E-33 * expV 725.0/T)
k( 47) = 1.3000E-12 * exp( 380.0/T)
k( 48! = 4.60001-11 * expt 230.0/T)
k( 49) is a falloff expression using:
kO = 3.00001-31 * (T/300!**(-3.30r
kinf - 1.5000E-12 * (T/300)**( 0.00)
F = 0.60, n '= 1.00 '
k( 50) = 1.5000E-13 * (1.0 + O.S*Pressure!
k( 51) = 2.8300E+01 * (T/300)**( 2.00) * exp( -1280.0/T)
k( 52) = 1.2330E-12 * (T/300)**( 2.00) * exp! -444.0/T)
{6.789051-30}
{1.68671E-12}
{4.87144E-12}
{1.95397E-38}
{3.18213E-17}
{2.812251-11}
{4.03072E-16} f
{•2.50000E-12}
{1.264401-12}
{5.47034B-02}
{2.000008-21}
{1.14885E-11}
{l.47236E-13}^
{4.65309E-12}
{9.95294E-11},
{8.88848E-13}
{2.40000E-13}
{3.80672E-01}
{2.742108-13}
8-96
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
k( S3)
kf 54)
k( 55)
kf 56)
k( 57)
k( SB)
k( 59)
k( 60)
k( 61)
k( 62)
k! 63)
k( 64)
k( 65)
k( 66)
k( 67!
fc( 68)
k( 69)
k( 70!
k( 71!
k( 72)
k{ 73)
k( 74)
k( 75)
k( 76)
kC 77)
k( 78)
k! 79)
k( 80)
k( 81)
k( 82)
k( 83)
k( 84}
k( 85)
k( 86!
k( 87)
k( 88!
k( 89)
k( 90)
k( 91)
k( 92)
k! 93)
kC 94!
k( 95)
k( 96)
k( 97)
k( 98)
k( 99)
k(100!
k(101)
k{102)
k(103!
k{104)
kUQS!
k(106)
k(107)
k(108)
k{109)
k(110)
kCUi)
k(112)
k(113!
-2
=
=
=
si
ES
US
=
=
•xx
=
_
=
=
=
=
=
=
=
=
=
=
S2
=
=
=
=
=
is
=
=
m
=
ES
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
=
-
=
=
K5
=
=
=
*
1
1
3
2
5
1
2
1
4
9
9
6
1
1
1
2
1
1
1
6
1
2
1
4
1
4
4
4
4
4
4
4
4
4
4
4
4
4
4
6
1
6
1
1
2
2
1
3
1
1
7
7
7
7
7
7
7
7
7
7
7
.5900E-11
.73QOE-11
.6400E-11
.1500E-12
.3200E-12
.0700E-11
.1000E-12
.8900E-11
.OOOOE-11
.OOOOE-01
.OOOOE-12
.8700E-12
.2000E-11
.1500E-11
.7000E-11
.800QE-11
.OOOOE-11
.OOOOE-11
.OOOOE-11
.16SOE-13
.S500E-11
.8000E-12
.9500E+16
.7000E-12
.9500E+16
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
•2QQOE-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.2000E-12
.OOOOE-13
.4000E-12
, OOOOE-13
.4000E-12
.4000E-12
.2000E-11
.OOOOE-12
.OOOOE-11
.2300E-11
.2000E-14
.3200E-14
.2900E-15
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
.7000E-14
-*
*
*
*
*
w
*
*
exp (
exp(
exp(
exp(
exp (
exp(
exp (
exp(
-540.
-380.
-380.
411.
504.
549.
322.
116.
0/1)
Q/T)
0/T)
0/TS
0/T)
0/T)
0/T)
0/T)
* kC 61)
*
*
exp {
exp (
256.
. -745.
* (T/300)**{
*
*
*
*
*
#
*
*•
*
*
*
*
* -
«•
*
*
*
•*
*
«
*
•*
*
*
it
*
*
*
*
*
*
*
*
*
*
*
*
*
*
exp(
exp (
exp(-
exp(-
exp(
exp {
exp (
expC
exp (
exp(
exp (
exp(
exp (
exp(
exp(
exp(
exp(
exp (
exp(
exp (
exp(
expC
expS
exp!
exp(
exp(
exp(
expC
exp(
exp!
expf
exp (
exp(
exp (
exp{
exp (
exp (
exp (
expf
-540.
181.
13543.
13543 .
180.
180.
180.
180.
180.
180.
180.
180.
180.
180.
180.
180 .
180.
180.
-2058.
-1900.
-2058.
-1900.
-1900.
-2323 .
-1895.
-975.
-2633.
-2105.
-1136.
1300.
1300.
1300.
1300.
1300.
1300.
1300.
1300.
1300.
1300.
0/T)
0/T!
2.00) * expf -444. 0/T)
0/T!
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T!
0/T)
0/T!
0/T)
0/T!
0/T)
0/T)
0/T)
0/T)
0/T}
O/TJ
0/T!
0/T!
0/T)
0/T)
0/T)
0/T)
0/T)
0/T!
0/T)
0/T!
0/T)
0/T!
{2.59669E-12}
{4.83334E-12}
{1.01696E-11}"
{8.53916B-12}
{2.88684E-11}
{6.75269E-11}
{S.18715E-12}
{2.78943E-11}
{4..00000E-11}
{3.60000E-11}
{9.00000E-12}
{1.62197E-11}
{9.85020E-13J
{l.lSOOOE-ll}
{1.70000E-11J
{2.80000E-11}
{l.OOOOOE-ll}
{l.OOOOOE-11}
{l.OOOOOE-ll}
{1.37105E-13}
{2.S3137E-12}
{5.13974E-12}
{3.57235E-04}
{4.70000E-12}
{3.57235E-04}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12}
{7.68378S-12}
{7.68378E-12}
{7.68378E-12}
{7.68378E-12J
{6.01030E-16}
{2.38307E-15}
{6.01030E-16}
{2.38307E-15}
{2.38307E-15}
{2.20000E-11}
{1.09940E-16}
{1.73099E-14J
{1.22539E-12}
{1.74559E-18}
{1.12933E-17}
{1.61125E-16}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12J
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
{6.04038E-12}
8-97
-------
EPA/600/R-99/030
laoies/v-iz. KAuivjujwic>^_At ana KAiJMz_us^_Aii_Ay Mecnamsms '" - -
k(115) =
k!HS) =
k!119) m
k(120) =
k(iai) =
k(122) =
k!123) =
k(124) =
k(125) =
k(126) =
k(127) =
k(128) =
k(129) =
k(130) =
k(132) =
k!133) =
k(134) =
k(135) =
k!136) =
k(137) =
k{138) =
k(139) =
k(140) =
k!142) =
k!143) =
k(144) =
k(145) =
k<146) -
k(147) =
k(148) =
k(149) =
k(150) =
k(151) =
k(lS2) -
k(153) =
k(154) =
k(155) =
k(156) =
k(157) =
k(158) =
k(159) =
k(160) =
k(161) =
k{162) =
k(!S3) =
k(164) =
k[lS5) =
k!lS6) =
k(lS7) =
k!lS8) =
kC!69) =
k(170) =
k(171) =
k(1725 =
k(173) =
k(174) =
7.7000E-14
7.7000E-14
7.7000E-14
7.7000E-14
1.9000E-13
1.4000E-13
4.2000E-14
3.4000E-14
2.90001-14
1.4000E-13
1.4000E-13
1.70001-14
1.70001-14
9.6000E-13
1.7000E-14
1.70001-14
9.6000E-13
1.7000E-14
3.4000E-13
l.OOOOE-13
8.4000E-14
7.20001-14
3.40001-13
3.4000E-13
4 .2000E-14
4.20001-14
1.1900E-12
4.20001-14
4.2000E-14
1.1900E-12
4.20001-14
3.60001-16
7.70001-14
1.70001-14
4 .2000E-14
3.6000E-16
4.2000E-12
4
7
1
4
3
1
1
1
2
4
7
8
3
7
3
3
4
7
8
3
1
4
7
8
.20001-12
.7000E-14
.7000E-14
.20001-14
,60001-16
.OOOOE+00
.OOOOE+00
.OOOOE+00
.54001-11
.20001-12
.7000E-14
.40001-14
.4000E-14
.8600E-15
.60001-11
.0300E-12
.2000E-12
.7000E-14
.40001-14
.40001-14
.8600E-11
.2000E-12
.70001-14
.4000E-14
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
*
it
it
*
*
*
*
*
*
*
*
*
*
*
*
*
it
*
it
*
*
*
*
*
*
it
it
*
*
exp! 1300
exp! 1300
exp ! 1300
exp! 1300
exp! 220
exp( 220
exp! 22o
exp! 220
exp! 220
exp! 220
exp! 220
exp! 220
exp! 220
exp! 220
exp! 22o
exp! 220
exp! 220
exp! 220
exp! 220
exp! 22o
exp! 220
exp! 220
exp ! 220
exp! 220
exp! 220
exp! 220
exp ( 220
exp! 220
exp! 220
exp! 220
exp! 220
exp! 220
exp! 1300
exp! 220
exp! 220
exp! 220
exp! ISO
exp! 180
exp! 1300
exp ! 220
exp! 220
exp! 220
k! 58)
k!100)
k(103)
!T/300)*»!
(T/300)**!
(T/300)**(
{T/300J**!
(T/300) **(
(T/300) **(
(T/300)**!
!T/300) ** !
(T/300) ** (
(T/300) **{
(T/3005 ** (
(T/300) **(
(T/300) ** !
(T/300)**!
(T/300)** (
.0/T)
.0/T)
,0/T)
-0/T)
.0/T)
.0/T)
.0/T)
.0/T)
,0/T)
.0/T)
-0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
. 0/TS
.0/T)
.0/T)
-0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/T)
.0/TS
.0/T)
0/T)
.0/T)
-0/T)
0/T)
0/T)
.0/T)
.0/T)
0/T)
0/T)
0/T)
0/T)
0/T)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00)
1.00!
1,00)
1.00)
{6.04038E-12}
{6.04038E-12} ...
{6.04038E-12}
{6.04038E-12}
{3.97S33E-13}
{2.92919S-13} •" .
{8.787S8E-14} ; ;• .;'
{7.11376B-14}
{6.06762E-14}
{2.92919E-13}
{2.92919E-13}ip
{3.SS688E-14}" *' ™
{3.SS688S-14}, ,
{2.00859E-12}
{3.55688E-14}
{3.55688E-14}
{2.00859E-12} ^ , , . , . ,
{3.55688E-14};{.;. . , . ;.:
{7.11376E-13} .;: . ;
{2. 092288-13}
{1.75752E-13}
{1.50644E-13}
{7.11376B-13};.-
{7.11376E-13}'j ,, • ; -t
{8.78758E-14} 1 ' . .. J
{8.78758E-14}.
{2.489818-12}
{8.78758E-14}
{8.78758E-14} - >,
{8.78758E-14};:.;
{7.S3221E-16}
{6.040388-12} _
{3.555888-14}
{8. 787588-14}
{7.S3221E-16}
{7.68378S-12} "
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
* exp!
407
181
1298
221
221
-1912
-447
181
1298
221
221
176
181
1298
221
-S/T)
.2/T)
.3/T)
.4/T)
.4/T)
.2/T)
.9/1)
.2/T)
.3/T)
.4/T)
.4/T)
.1/T) •
.2/T)
.3/T)
.4/T)
{7
{6
(3
{8
{7
(6
{1
{1
{9
{7
{5
.68378E-12}
.04038E-12}
.556888-14} t* ' ' :- ' &
.78758E-14}
.S3221E-16)
.752698-11} '
.22539E-12} "•
.611258-16} ••• '
.907198-11}
.66335E-12}
.96598E-12}si"« . ,:
{1.754028-13}
{?
{1
{3
{6
{7
{5
{1
{7
{3
(7
{5
{1
.099618-14} -'
.275698-17}'-; ' ' '*' ;
.60000E-11}-,: . -
.695528-13} "••'
.6633SE-12}
.965988-12}*': , < ;:
,75402B-13}_ . '~
.09961E>14}-- - -v
.33618E-11}
.66335E-12}
.965988-12}
.75402E-13}
8-98
-------
EPA/600/R-99/030
Table 8A-12. RADM2_CIS4_AE and RADM2_CIS4_AE_AQ Mechanisms
k(17SJ
k(176)
k(177)
k(178)
k(179)
k(180)
k(181)
k(182)
k(183)
k{184)
k{185)
k(186)
k(187)
k{188)
k!189)
k(190)
k(191)
k(192)
k(193)
k(194)
k(195)
k(196)
k(197)
k(198)
k{199)
k(200)
=
=
3
1
uses
=
sx
=
=
=
=
=
1
4
4
7
8
3
7
uses
=
KS
=
SS
S3
=
XX
S2
=
E
=
=
=
1
4
2
7
9
1
1
3
4
7
8
3
7
uses
=
1
.4000E-14
.3600E-15
* (T/300)**(
* (T/300)**(
1.
1.
00)
00)
*
*
photo table ACROLEIH ,
.5000E-12
.1400E-12
.2000E-12
.7000E-14
.4000E-14
•4000E-14
.5100E-1S
* CT/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
1.
1.
1.
1.
1.
1.
1.
00)
00)
00)
00)
00)
00)
00)
*
*
*
*
*
*
*
photo table ACROLEIN
.6000E+16
.2000E-12
.SOOOS-12
.7000E-14
.6000E-13
.19001-12
.1900E-12
.3SOOE-11
.2000E-12
.7000E-14
.4000E-14
.40001-14
.1100E-18
* CT/300)**{
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300)**(
* (T/300!**(
* (T/300)**(
* (T/300)**(
* (T/300!**(
* (T/3QO)**(
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
1.
00)
00)
oo!
00)
00)
00)
00)
00)
00)
00)
00)
exp(
exp( -
scaled
exp( -
exp(
exp(
exp (
exp(
exp(
exp( -
scaled
221
2113
by
1726
452
181
1298
221
221
1519
by
* exp (-13486
*
*
*
*
*
*
*
*
*
*
photo table ACROLEIN ,
.OOOOE-13
exp(
exp(
exp!
exp(
exp(
exp(
exp(
181
181
1298
221
221
221
181
exp( 1298
expS
exp (
scaled
221
221
by
,4/TS
-5/T)
3.60000E-03
.0/T) .
.9/T)
• 2/T)
.3/T)
.4/T)
.4/T)
.7/T)
1.11000E-02
.0/T)
.2/T)
.2/T)
.3/T)
.4/T)
.4/T)
.4/T)
.2/T)
.3/T)
.4/T)
.4/T)
3.60000E-03
{7.09961B-14}
{1.12330E-18}
{O.OOOOOE+00}
{4.547S3E-1S}
J1.87990E-11}
{7.66335E-12J
{5.96598E-12}
J1.75402E-13}
{7.09961E-14}
{4.S4966E-18}
{O.OOOOOE+OO}
{3.S2536E-04}
{7.66335E-12}
{S.10890E-12}
{S.96598E-12}
{2.00460E-12}
{2.48486E-12}
{2.48486E-12}
{3.36000E-11}
{7.S6335E-12}
{5.96598E-12}
(1.7S402E-13)
{7.09961E-14}
{7.11000E-18}
{O.OOOOOE+00}
{l.OOOOOE-13}
8-99
-------
EPA/600/R-99/030
Chapter 9
PLUME-IN-GMD TREATMENT OF MAJOR POINT SOURCE EMISSIONS
N. V. Gillani
Earth System Science Laboratory
University of Alabama in Huntsville
Huntsville, Alabama 35899
James M. Godowitch*
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
A plume-in-grid (PinG) technique has been developed to more realistically treat the dynamic and
chemical processes impacting selected, major point source pollutant plumes in the Community
Multiscale Air Quality (CMAQ) modeling system. The principal science algorithms include a
Plume Dynamics Model (PDM) and a Lagrangian reactive plume code. The PDM processor
simulates plume rise, horizontal and vertical plume growth, and transport of each plume section
during the subgrid scale phase. It generates a data file of this information for use by the PinG
module. In contrast to the traditional Eulerian grid modeling method of instantly mixing the
emissions from each point source into an entire grid cell volume, the PinG module simulates the
relevant physical and chemical processes during a subgrid scale phase which allows each plume
section to expand in a realistic manner and to evolve chemically. The PinG module is iully
integrated into the CMAQ Chemical Transport Model (CCTM) in order to utilize the grid
concentrations as boundary conditions and it provides a feedback of the plume pollutants to the
grid model concentration field at the proper time and grid location. The technical approaches and
the model formulation of the relevant processes treated by this plume-in-grid approach are
described. The capabilities and limitations of the initial version of PinG are also discussed and
selected results from a test application for a single point source are briefly described.
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: James M. Godowitch, MD-80, Research Triangle Park, NC 27711. E-mail:
jug@hpcc.epa.gov
-------
EPA/600/R-99/030
9.0 PLUME-IN-GRTO TREATMENT OF MAJOR POINT SOURCE EMISSIONS
9.1 Introduction
Significant emissions of anthropogenic nitrogen oxides (NOJ and sulfur oxides (SO,^ are released
from individual elevated point sources into the atmosphere at various levels. These major point
source emissions are distributed throughout the U.S. The physical dimensions of their plumes are
relatively small initially and expand at finite growth rates. This diffusion-limited nature of plumes
is in sharp contrast to the traditional method applied in Eulerian photochemical grid modeling, -
which has been to uniformly and instantly mix point source emissions into the entire volume of a
model grid cell. Depending on the meteorological conditions, however, pollutant plumes may
require up to several hours to grow to the typical size of a regional model grid cell. Since the
horizontal grid resolution of regional scale domains for Eulerian models has generally been 20-30
km or greater, a real-world pollutant plume may remain a subgrid scale feature at current regional
model grid resolutions for a substantial time period and considerable distance after release. Thus,
the widely-used grid modeling approach inadequately treats subgrid scale plume transport and
diffusion.
There are also important implications on the chemical processes in models due to the inability of
large grid cells to adequately resolve a major point source plume. As a real-world plume
gradually grows, it concurrently evolves chemically as it entrains surrounding ambient air often
richer in volatile organic compounds (VOCs). However, these dynamic and chemical processes
of plumes are neglected when large point source emissions are instantly diluted into large grid
cells with other anthropogenic area emissions, which may actually be separated spatially from a ^
plume. The effect is that the simultaneous availability of NOX and VOC's in a large grid cell
prematurely initiates rapid photochemistry leading to distortion in the spatial and temporal
features of secondary species concentrations. This overdilution of point source emissions and the
subsequent distortion of chemical processes contribute to model uncertainty. Consequently, it has
been recognized that a realistic, subgrid scale modeling approach is needed which simulates the
relevant physical and chemical processes impacting this notable class of large point source
emissions, and in particular, that has the capability of properly resolving the spatial scale of
plumes and their growth.
There are approaches which have been applied in attempts to resolve fine scale emissions in
regional grid models. These have included uniform, nested grid modeling (Odman and Russell,-
1991; Chang et al., 1993), non-uniform grid modeling (Mathur et al., 1992)5 and the plume-in- ,
grid modeling (PinG) technique. The nested grid (telescoping) modeling approach employs the
same Eulerian model and simulations are performed with progressively finer-mesh grid sizes
within a domain to better resolve small scale emission features. While an advantage of this
approach is the use of a single model algorithm, a disadvantage is the successively higher
computational burden as the grid cell size is reduced in order to resolve major emission sources.
Efforts to implement and apply the plume-in-grid technique, in particular, have been performed by
Seigneur et al. (1983) in a version of the Urban Airshed Model (UAM-PARIS), by Morris et al.
9-2
-------
EPA/600/R-99/030
(1992) in the UAM-V model, by Kumar and Russell (1996) in the Urban-Regional Model (URM),
and by Myer et al. (1996) in the SAQM (SARMAP Air Quality Model) model. Briefly, each of
these PinG efforts simulate multiple plumes or puffs in a domain with no interaction between
individual plumes. Seigneur et al. (1983) employed a rectangular plume cross-section divided into
vertically well-mixed plume cells exhibiting varying widths to maintain equal pollutant mass
among the cells. It was applied to an urban domain with a 4 km grid cell size. The other efforts
employ an elliptical plume section divided into concentric rings, except for the Kumar and Russell
(1996) PinG. Plume material in an outermost elliptic ring or shell is sequentially transferred to the
grid model as it attains the grid size. In the treatment by Kumar and Russell, the plume was
returned to the grid model after one hour. The general result of applications from these PinG
implementations was that noticeable differences were found primarily within or near the grid cells
containing the major point sources and negligible impact was found far downwind. This finding is
to be expected since these formulations seemed to focus on the plume-core chemistry only in an
early phase of plume chemical evolution with the feedback of plume pollutants occurring rather
quickly to the grid model,
A cooperative research and development effort was undertaken to design and implement a plume-
in-grid (PinG) capability in the Models-3 CMAQ Chemical Transport Model (CCTM) in order to
address the need for an improved approach to treat major point source emissions. The current
PinG approach has been designed to be suitable for the largest, isolated point source emissions
with grid resolutions of regional modeling domains, where the subgrid scale error in the
representation of these sources is expected to be greatest without a PinG treatment. An
important consideration is to be able to spatially resolve a pollutant plume and to adequately
simulate physical expansion so that the chemical evolution can occur gradually.
Therefore, the objectives with this PinG technique are to provide an improved characterization of
the near-source pollutant field from the subgrid scale plume concentrations and to also generate
better regional pollutant concentrations downwind due to the feedback of the PinG concentrations
to the gridded domain. Consequently, when the PinG approach is applied to the largest point
source emissions in a regional, coarse-grid domain, improved initial and boundary conditions are
anticipated for use with subsequent simulations for a subdomain of the same grid resolution or for
fine-mesh nested domains that also need the regional gridded concentration fields.
In this chapter, the conceptual design of the PinG approach is discussed. The key modeling
components are a Plume Dynamics Model (PDM) and a Lagrangian reactive plume model .
designated as the PinG module since it is fully integrated and coupled with the CCTM Eulerian
grid model. The PDM has been developed to serve as a processor program to generate plume
dimensions, plume position information and related parameters needed by the PinG module. The
descriptions of the mathematical formulations and numerical techniques contained in these
modeling components are also presented. Their solutions form the basis of the physical-chemical
simulations of subgrid scale plumes as implemented in the principal algorithms of the PinG
modeling components.
9-3
-------
EPA/600/R-99/030
9.2 Overview of the Conceptual Framework of the Plume-in-Grid Treatment
The PinG technique is intended for the largest point sources, designated as major elevated point
source emissions (MEPSEs), which are isolated from notable area emission sources. A complete
description of the various criteria available to classify a point source as a MEPSE is provided in
the emission processing system in chapter 4. In the context of photochemical modeling, the NOX
emission rate is a useful criterion to classify a major point source as a MEPSE from the thousands
of individual major point sources in an inventory. Based on a NOX emission rate criterion of 50
tons/day as a lower limit, Figure 9-1 depicts the group of fewer than 100 MEPSEs in a 36 km
gridded domain covering the eastern US with emission rates greater than this criterion. The
MEPSE sources are situated away from major metropolitan emission source areas. It reveals that
numerous MEPSEs are distributed throughout the Ohio River and Tennessee River valley regions.
Since the current PinG technique also relies on linear superposition of MEPSE plumes for
determining the impact on grid concentration, the relatively isolated MEPSEs located in these
remote rural environments also provide for the lowest likelihood of plume-plume interactions.
An important feature of a PinG treatment is that it must be able to reproduce the various stages in
the chemical evolution found in a large, rural point source plume. Based on daytime experimental
field study plume data, Gillani and Wilson (1980) documented three distinct stages, as displayed
in Figure 9-2, in the chemical evolution of ozone (O3) in a pollutant plume from a high NOX
source. The O3 data measured across a plume at different downwind distances in Figure 9-2
illustrate ozone evolution in a large NOX point source plume. Although NOX was not obtained, it
would display a similar variation to SO2 shown in Figure 9-2. During the first stage, the relatively
fresh plume is dominated by primary NOX emissions and an O3 deficit exists in the plume due to
titration by very high NO concentrations. The chemistry in stage 1 is mostly inorganic and VOC-
limited since MEPSE plumes generally exhibit low VOC emissions. The second stage represents
a transition in the chemistry with rapid production of O3 along the plume edges and some O3
recovery in the plume core. Plume growth from dispersion processes allows for entrainment of
background air which generally contains a richer supply of ambient VOC concentrations.
Consequently, this promotes rapid photochemical O3 production which leads to the
characteristically higher O3 concentrations observed at the edges of a plume during this stage. ,
Figure 9-2 reveals that stage 2 certainly exhibits considerable subgrid scale plume structure in
species concentrations. Therefore, the proper simulation of plume chemistry throughout stage 2
is a key phase for a realistic, overall characterization of plume chemical evolution. In fact, foil
completion of stage 2 is an important requirement for a chemical criterion during the PinG
simulation before the plume pollutants are transferred to the grid system. In stage 3, the plume is
chemically mature and substantially diluted. In this mature stage, the broad plume would also
exhibit a high VOC/NOX ratio, low NOX concentrations, and O3 concentrations in excess of
background levels. In the modeling system, background (boundary) concentrations are provided
from a CCTM grid cell containing the subgrid scale plume section.
The chemical evolution, as described above, for a large MEPSE plume in the eastern US generally
may take up to several hours to reach full maturity even during the daytime period. The
9-4
-------
EPA/600/R-99/030
photochemical cycle is strongly influenced by the plume growth rate, particularly the rate of
horizontal spread, with stage 3 achieved when a typical MEPSE plume attains a width of about 30
km during the daytime period. In addition, the time period to reach chemical maturity is also a
function of the emission rate. There is a range of emission rates even within a MEPSE group.
For example, particular MEPSE plume with lower NOX emissions is expected to exhibit a faster
chemical evolution than a MEPSE with a higher emission rate, assuming other factors being the
same. Nevertheless, during the subgrid scale simulation period, the plume may travel a
considerable distance downwind (i.e. several model grid cells) of its source location in the daytime
planetary boundary layer (PBL) during the summer before reaching a width comparable to the
grid cell size.
The conceptual design of the Models-3 PinG is illustrated in Figure 9-3. PinG components
simulate the relevant processes impacting the hourly emission rate of all pollutant species during
the subgrid scale phase and it provides a feedback, or "handover", of the modeled plume species
concentrations at an appropriate handover time (t^o) to the CCTM grid framework. The PinG
module operates in a Lagrangian modeling framework to simulate the dynamics and kinetics in a
moving plume section during the subgrid scale phase whose duration depends primarily upon
source strength, plume growth rate, chemical composition of the background CCTM grid
concentrations, and sunlight. The PinG module simulates the plume processes concurrently
during the simulation of the CCTM grid model.
The PinG module contains a series of rectangular plume cross-sections, each of which are
composed of a contiguous, crosswind array of plume cells with the depth of each cell extending
vertically from plume bottom to a top height. As displayed in Figure 9-3 a, the vertical depth of
the single-layer plume model may be elevated initially, however, it eventually extends from the
surface to the PBL height (z;). This one-layer plume structure in the initial PinG module also
presents a limitation under high wind speed shear conditions since this situation cannot be
modeled adequately. While strong speed shears are more frequent during the nocturnal period,
the vertical extent of a plume is also generally confined, which lessens the possible impact of
speed shear. Nevertheless, this condition is identified and appropriately handled by the PinG
components.
9.3 Formulation of the Plume-in-Grid Modeling Components
9.3.1 Description of the Plume Dynamics Model
The PDM was developed to serve as a processor program in order to provide data needed for the
CCTM/PinG simulation. Therefore, the PDM and the PinG module are closely linked through
the PDM data file. The key processes simulated by the PDM include plume rise, plume transport,
and plume dispersion of each MEPSE plume cross-section. The PDM also specifies when a
particular plume section is to be transferred to the grid model based on an indicator flag which
communicates to the PinG module when this process is to be performed. The PDM is capable of
simultaneously treating up to 100 MEPSEs in a single simulation with the impact on
9-5
-------
EPA/600/R-99/030
computational time a function of the number of MEPSEs simulated. In addition, the maximum
simulation period is currently limited to 24 hours. Therefore, a single simulation of the CCTM
with PinG is also limited to a 24-hour time period. , ."_•' 7 "
9.3.1.1 Aspects of Plume Rise of MEPSE Sources
The plume rise treatment implemented in PDM contains the same algorithms applied in the
Emissions-Chemistry Interface Processor (ECDP). A detailed discussion of the plume rise ?
equations and treatment has been provided in section 4.4.2 of chapter 4 on the MEPPS system
and therefore is not repeated herein.
1 ' ' '• ' ' : " -I
The plume rise algorithms handle the entire diurnal cycle since plume sections are released hourly
throughout the 24 hour cycle. The temporal variation of the final effective plume height for - • -
hourly MEPSE plume releases is illustrated in Figure 9-4. In this case, a diurnal variation in the
final plume height is evident with lower plume rise occurring at night believed to be due in part to
stronger wind speeds. The plume height increases during the morning period as the PEL height
grows and the relatively high values found in the afternoon display little variation until the end of
the daytime period.
A notable feature of MEPSE plumes also revealed in Figure 9-4 is the rather significant height ^
often attained by this class of point sources. Statistics of the stack parameters from, MEPSE and
non-MEPSE groups of major point sources were determined. A comparison indicates notable
differences in certain stack parameters between these source groups. The average values from a
MEPSE group containing 84 stacks compared to the other major point source class containing
6539 stacks show a stack height of 207 m versus 45 m, a stack velocity of 20 jn/s versus 12.5
m/s , a stack diameter of 7.5 m versus 2.3 m, a stack exit temperature of 410° K versus 424° K,
and a stack exit flow rate of 908 m3/s versus 105 nrVs, respectively. Clearly, these results indicate
why MEPSE plume heights attain higher levels than most point sources. In general, MEPSE
stacks are considerably higher and their plumes exhibit greater buoyancy flux due primarily to
greater physical dimensions which contribute to higher plume rise heights. '.
9.3.1.2 Plume Dispersion Methods
A key aspect of modeling a subgrid scale reactive plume, as noted earlier, is the importance of
realistically specifying plume dimensions during travel time downwind. Practical methods have
been applied in the initial version of PDM to determine the horizontal and vertical dimensions of
each Lagrangian plume section throughout its life cycle.
Each rectangular plume cross-section has a width (Wp) and an overall height from plume bottom
to top denoted by Hp. The traditional dispersion parameters in the horizontal (oy) and vertical
(oj have been employed in order to derive these plume dimensions. In the current version, the
plume width and plume height values are determined according to
9-6
-------
EPA/600/R-99/030
WP = aay (9-la)
HP = hT ~ hb (
where a has been set to 3.545 which is obtained from 2(7t)H, an adjustment parameter from an
elliptical shape to the rectangular plume. The top and bottom heights of each plume cross-section
are denoted by hT and hb, respectively. The methods used to determine values for the parameters
in Equations 9-la and 9-lb are discussed next.
During the plume rise phase, a plume experiences initial growth due to buoyancy-induced
turbulence. To account for this process, practical methods have been used to determine initial
values for the dispersion parameters. The initial value of ay is computed according an expression
suggested in Irwin( 1979), as advanced by Pasquill (1976), given by
Qyo = M/3.5 (9-2)
In Equation 9-2, the initial value is a function of the plume rise (Ah) above the stack top.
Two methods have been installed to prescribe an initial vertical plume thickness. The widely used
approach advanced by Briggs (1975) is to define Hp to be equivalent to the amount of plume rise
(Ah). This method can result in a rather thick plume when there is considerable plume rise,
particularly, at night when experimental evidence suggests nocturnal plumes may be relatively
thin. Consequently, an optional approach provided as an experimental alternative has been
included. It consists of an empirical form developed by Gillani (1996) which is based on analyses
of observed plume data taken during a field study (Gillani et al., 1981). His regression analyses
produced an expression given by
(9-3)
where dT/dz is the vertical ambient temperature gradient at the plume centerline height and the
best-fit values for A and B are 15 and 117, respectively. A minimum azo value is set to 3 m.
Then, the initial Hp is determined from SZOFA times azo SZOFA is a user-specified input
parameter for a PDM simulation and it is currently set to 3.545.
Atmospheric turbulence and wind direction shear contribute to lateral plume growth during travel
downwind. During the daytime convective period, the turbulence component is often dominant,
although directional shear across the PBL also has an important role in contributing to plume
9-7
-------
EPA/600/R-99/030
spread with travel distance. In contrast, turbulence is generally weak at night, and horizontal
plume expansion may be dominated by wind direction shear over a plume's depth. Thus, the
composite horizontal (lateral) dispersion parameter (oy ) can be defined by ' ? i •
where the turbulence (a^) and direction shear (os) terms may both contribute to horizontal plume
growth after the initial plume spread.
A general form for o^ can be expressed by
°y, = vJATI^ (9-5)
where crv is the standard deviation of the lateral wind (v) component, T is travel time, and TL is the
Lagrangian time scale. Although various expressions for the function, f^/TjJ, have emerged (e.g.,
Draxler, 1976; Irwin, 1983), the following form from Weil (1985; 1988) and applied by others
(eg.,Venkatram, 1988) has been adopted.
(9"6)
This form is advantageous since it fits dispersion at both the short (t < TL) and also long travel
times (t»tj). Since the Lagrangian time scale is small relative to the plume section travel times
which are often several hours, the expressions above reveal the familiar relation that ajt increases
with the square root of travel time. Therefore, the following form according to Weil (1988) at
long travel times, which has been examined by Clarke et al. (1983), is given by
oj = 2al
-------
EPA/600/R-99/030
Ov = ii.(12 - O.Sz/L)113 (9-8)
= 0.15z/av (9_9)
where Zj is the PEL height and u» is the surface friction velocity. Under stable conditions defined
byL>0,
av = 1.3«.(1 - z/z) (
= 0.07z/av(z/z.)°-5 (9-11)
and z is the height of the plume centerline. For neutral conditions (L — 0), values are determined
by
a = l.3e(-2^ (9-12)
tL = 0.5z/oy(l + 15/S/ir.) (c
where fis the Coriolis parameter. Using ov and TL values, the contribution from the turbulence
component is computed incrementally from the derivative of Equation 9-7 and solved at each time
step. The turbulence term is accumulated with time. Other semi-empirical methods available in
PDM provide an alternative set of parameterizations for ov as provided by Weil (1988),
Nieuwstadt(1984), and Arya(1984) for the different stability regimes, which may be selected from
an input option when exercising PDM.
Wind direction shear due to turning of the wind over the vertical extent of a plume can also
provide an important contribution to horizontal plume growth. While turbulence often dominates
in the daytime PBL, direction shear during the nocturnal period is the principal mechanism for
lateral plume expansion since turbulence is generally weak. An expression by Pasquill (1979) has
been applied for all conditions to derive the direction shear term and is given by
oj, = aAelr2 (9-14)
9-9
-------
EPA/600/R-99/030
where a is 0.03, X is distance traveled by a plume section over a particular time interval, and A6
is the wind direction difference (radians) over the vertical extent of the plume. The shear and
turbulence terms are combined to solve for the total 0y2 in Equation 9-4.
After the initial plume thickness has been determined, the treatment for oz differs based on
whether the plume height is above or below the PBL height. When the plume centerline height is
above %, then the following expression from Gillani (1996) is applied to determine az,
2 2 (1 + bT°-5)
az = azo — (9-15)
(1 * £7f) ^ 1:V
where b is 2.3 and T is travel time. Equation 9-15 is applied for T < 4 hours. At longer travel
times, oz is set to zero for plumes existing above the PBL in the free atmosphere.
When a plume section is inside the PBL, the parameterization for the vertical plume dispersion
parameter can be expressed by the same general form as Oy,
(9-16)
As before, T is the travel time, and f has the same functional form as used for o^ although it now
contains the Lagrangian time scale in the vertical (TLZ). The standard deviation of the vertical
wind component (ow) is computed from the following expressions as given by Weil (1988) for
unstable conditions and from Venkatram et al (1984) for neutral/stable conditions, respectively.
ow = 0.6wm (convective case) (9-17)
ow = 1.3«,0 ~ zlz)m neutral'stable (9-18)
The MCDP data files provide values for the fraction velocity (u»), the convective velocity scale
(w.) and Zj. The plume centerline height is represented by z. The value of TLr is computed from
the following expressions by Hanna et al. (1982) for unstable and by Venkatram et al. (1984) for
neutral/stable conditions.
•c^ = 0.15— [1 - e~5z/z'] (unstable) (9.19)
9-10
-------
EPA/600/R-99/030
In the latter expression, the length scale (i) is derived according to Venkatram et al. (1984).
An example application of these methods from a PDM simulation for a plume released at night
from a MEPSE tall stack is shown in Figure 9-5. It shows the plume remains rather narrow and it
TLz = — (neutral!stable) (9-20)
stays elevated during the night. However, once the PBL grows beyond the plume height during
the morning period, the plume thickness expands to fill the entire PBL with the plume top height
matching Z; until it reaches a maximum height. The temporal variation of the plume width for the
same example case is displayed in Figure 9-6. It shows that since the plume has a limited vertical
extent at night, the plume width increases at a relatively slow rate as sufficient direction shear
existed over the plume depth to cause lateral spreading. During the daytime, however, horizontal
plume expansion occurs at a faster rate due to both greater turbulence and shear contributions.
For a regional grid cell size of 36 km, the plume width for this release reached the physical grid
size during the mid-morning hour. These results and other simulation results obtained from a
daytime release versus observed values (Godowitch et al., 1995) provide preliminary evidence
that these methods give a realistic depiction of the temporal behavior of the magnitudes of plume
width and depth. Additional cases have also been exercised, although results are not shown
herein, in order to assess the robustness and capability of the algorithms for different days. An
extensive evaluation of these dispersion methods in PDM is planned against field study data
obtained in the vicinity of major point sources in the Nashville, Tennessee area during the
Southern Oxidant Study experimental period in the summer of 1995.
9.3.1.3 Plume Transport
With the updated plume top and bottom heights, the mean wind components are determined by
averaging the winds over the layers spanned by the plume from hb to hT. This approach is
currently applicable because of the single vertical layer structure of the current PinG module. The
mean plume transport speed is used to determine an updated plume centerline position over the
time interval. In addition, the positions of the plume edges at the bottom and top are also found
in order to derive grid indices for later use in applying the proper grid cell concentrations for
boundary conditions of each plume section.
9.3.2 Formulation of the PIume-in-Grid Module
9.3.2.1 Overview of the Plume Conceptual Framework
9-11
-------
EPA/600/R-99/030 *
The Lagrangian plume model described originally by Gillani (1986) provided the conceptual basis
for the Models-3 PinG module. However, the computer algorithms in the CCTM/PinG have |
been rewritten to contain updated methods for the key processes to be described in a later section.
In a modeling framework, a PinG plume cross-section can be described as a semi-infinite vertical
slab moving along a Lagrangian trajectory with a mean wind flow. A plume cross-section is
considered rectangular with a vertical height (Hp ) and a width (Wp). Temporally, plume spread
in the vertical and horizontal is specified by growth rates dFL/dt and dWp/dt, respectively, A ..
plume cross-section is discretized laterally into an array of attached plume cells or pillars with the
same Hp at time t, as depicted in Figure 9-7. Each plume section consists of N plume cells
(currently, N - 10) with the width of each cell being equal. With respect to the plume centerline,
there are NL cells on the left side and NR cells to the right side, such that
Currently, NL= NR as L and R refer to the cells on the left and right sides, respectively. With
respect to the plume centerline, the right side of the plume section expands to the right and the.;
left side expands to the left. The width (Wj) of each plume cell is given by
(W)UR = (v/*i - y^uR (9-22)
where y values represent distances from the plume centerline position. Normalization of the
plume cell widths is performed such that as the plume expands laterally, each cell width increases
in the same proportion as the overall plume width. Thus, since the individual cell width is ;,;
normalized with respect to the total plume width, the result is a transformed moving crosswind
gridded array which is invariant with respect to time. The transformed crosswind
coordinate (T|) of the model is given by
li = r (9-23)
and dr|j/dt is zero. Therefore, although Wp and ys are time dependent, r]i remains constant during
a simulation. Additionally, although the fractional width of the cells across the plume section in
the PinG has been prescribed to be equal m the current setup of the module, the algorithms have
been generalized to allow for unequal plume cell widths and a different number of cells on the left
and right sides of the plume centerline as described in Gillani (1986).
9-12
-------
EPA/600/R-99/030
9.3.2.2 Formulation of the Plume Mass Balance Equation
The relevant processes, as noted in Figure 9-7, have been incorporated into the plume equation
for the mass balance of individual species. They include dilution and entrainment due to vertical
plume expansion, dilution and entrainment/detrainment due to horizontal plume expansion, lateral
diffusion, gas-phase chemistry, surface removal, and surface emission.
The subscript i denotes plume cell i (left or right of the plume centerline), subscript] denotes
species j, and superscript t denotes a particular time. Thus, Cj; is the concentration of species j in
cell i at time t, and m'^ is mass of species j in cell i at time t. In the following derivation of the
terms of the mass balance equation for any species j in cell i, the subscript] is omitted for
convenience; also, the equations and terms do not show "L" or "R" representing a left or right
plume cell, since the equation and terms apply similarly to both sides. Consequently, consider the
mass balance of (left or right) cell i (for species j) corresponding to the its expansion during a
small time interval dt.
dm,. = (dm)Disp + (3m ^ + (Bm)chem + (dm)Dep (9-24)
Now, the following expression is also applicable for the plume cell mass.
dmi = ml*dl - ml = Cl**W';*W';* - CJUt'W' (9-25)
Herein the alongwind dimension in the downwind direction (1 = Udt) is prescribed, which is
determined from the initial mean transport speed (U) over the time interval dt.
Mathematical expressions are presented in subsequent sections for the updated concentration of
species] in cell i due to the action of the individual processes during time interval dt. In the
current version of PinG, only surface emissions after plume touchdown are included in the
emission term since (non-MEPSE) point-source emissions into grid cells (above the lowest cell
layer) neighboring the MEPSE plume are released uniformly into the entire emission grid cell and
impact the MEPSE plume only through the boundary conditions. Surface removal is due to dry
deposition at the ground. Gas-phase chemistry is included with the full chemical mechanisms of
the CCTM in the chemistry term. In the future, the PinG will be adapted to employ the same
aerosol module used by the CCTM. No wet processes are treated in the initial PinG version. The
current approach to address limitations in the initial version of PinG is to transfer the plume over
to the grid solution whenever conditions exist (e.g., precipitation) which the PinG treatment
cannot adequately accommodate. This issue will be addressed in more detail in the section on
plume handover.
9.3.2.3 Treatment of Plume Expansion and Diffusion Processes
9-13
-------
EPA/600/R-99/030
i, „ . i jp. I • \ " n
Dispersion affects the mass balance of a plume cell as a result of dilution and
entrainment/detrainment processes related to the lateral and vertical expansion of a plume cross-
section during a time interval dt, as well as mass diffusion which impacts cell concentrations as a
result of concentration gradients between adjacent cells. The dispersion processes can be
expressed by
(8/w/W = (dm^DiUE/D + (dm)Dif ' ' " (9-26)
... " . '.'• " ' , ~ " . ' , ' . • • i'; i" \
where the first term on the right side of Equation 9-26 represents the dilution / entrainment /
detrainment processes and the last term is the eddy diffusion process.
The relationships developed herein apply to cells on either side of the plume centerline, since the
only difference occurs in their application when different boundary conditions occur at each edge,
'•.-•-- r . * ' «
* (9-27)
refers to the change of mass in plume cell i (for any transported species j) during dt as a result of
dilution and entrainment/detrainment. The changes due to just the horizontal and vertical
exchange processes can be expressed by
* ' (9-28)
'••> • , '•'•• : LJ -.r. •<
where Latdil refers to horizontal (lateral) dilution. In practice, the lateral dilution/
> f •* ./*•'"-• -m&m «*u -
entrainment/detrainment term in Equation 9-28 is determined and then the last term is solved.
The lateral dilution term in Equation 9-28 is given by
(9-29)
However, the following relationship also exists.
(9-30)
9-14
-------
EPA/600/R-99/030
since t| is invariant. Therefore, Equation 9-29 can be revised to the following form.
(9-31)
At this time, (i.e., after lateral dilution/entrainment/detrainment) the width of a plume cell is Wjt+dt.
With the cell widths remaining unchanged, plume vertical expansion occurs (i.e., in the up and
down direction for an elevated plume, and up only after plume touchdown). Let Cat+dt and Cbt+dt
denote the concentrations of species j above and below plume cell i, respectively, for the elevated
plume case. Ca and Cb represent the CCTM grid concentrations at t+dt. Once the plume becomes
surface-based, only Ca is relevant. The equation for the vertical dilution term is given by
l^d'(C';dtdHa + C'^'dH,) (9-32)
where the changes in Haand Hb denote plume vertical expansion of the upper and lower plume
boundaries, respectively. An assumption is made that
; BHa - dHb = O.SdHp (9-33)
for an elevated plume cross-section, and after a plume reaches the surface
are applicable. Equation 9-32 can be rewritten accordingly as
.. . j* * . j*
(9-35)
where
C';d'= (Ca + Q (elevated plume)
f^t^dt s^t+dt rt L j i \
C. = Ca (surface-based plume)
9-15
-------
EPA/600/R-99/030 t , ,
' . ' »" ' • • $. , s -,
. . • -r.ii--
for an elevated plume and a plume bottom at the surface, respectively. The combination of
Equations 9-31 and 9-35 produces the following form.
(9-37)
However, by substituting the next expression,
'AxaBD3 Cr*W?*»r* - C////>/ • '"' W"' -(9-381
* I ... i- ,
'fc- ' ft iV
r - • IP :;, jt" !'
liF" „ ' . '" i .1. . i • i . liiil" i '1 i
and rearranging, the concentration after the total dilution/entrainment/detrainment step is
'**
The first term on the right-side is related to the dilution effect and the next two terms are the
lateral and vertical entrainment/detrainment effects, respectively. The subsequent expressions are
also applicable.
y'
(9-40)
By substituting the above expressions into Equation 9-39, and rearranging gives
M* _ c,
wp
However, Equation 9-41 can be simplified and re- written as
9-16
-------
EPA/600/R-99/G30
where
a// dWD
-rf , ^ = -™ (9-43)
When applying Equation 9-42, certain boundary conditions must be specified. In particular, if
i = I ; T), = TI, = 0
\r • \r /-i /~it+dt (9-44}
1 = NL > ' = NR J C,>I = CbS
where Cbg!+<" and C.t+dt are the particular grid cell concentrations from the CCTM solution for the
current time step.
In the execution of the CCTM/PinG, both the grid-cell and plume-cell concentrations are updated
by numerical integration of the corresponding mass balance equations from time t to t+dt, where
dt is the CCTM advection time step (MSTEP). The sequence is that CCTM performs this
integration and updates the concentration field first, generating Ct+dt in each grid cell, before PinG
is called to do the same. PinG performs the integration in a fractional step approach; first for
plume dispersion (step 1) then for the surface emission and removal processes (step 2), and finally
for plume chemistry (step 3). The chemistry integration is actually performed in smaller chemistry
time steps, internally determined for the plume cells by the chemical solver, in sequence, until the
integration is completed for the time interval dt. The emission/removal step is implemented in a
single integration time step. The dispersion step is performed sequentially in subtime steps until dt
is reached. The PinG dispersion time step is constrained by a restriction when applying Equation
9-42 arising from the fact that, in a given time step, the location after expansion of the inner
interface of a given plume cell (i.e. y^df) cannot pass beyond the location of the outer interface of
that cell at the start of the ex expansion. Therefore, the following constraint applies.
y,<, y^ , 2 < i < NLfR (9-45)
This criterion requires that the following limit for the PinG time step must be satisfied.
9-17
-------
EPA/600/R-99/030
(9-46)
AM, = (9-47}
W dt
p
It should also be noted that AW* is related to AW according to
(9-48)
During the model simulation, Wp (t) and dWp/dt are provided to the PinG module from the PDM
data file. Thus, Aw* is determined directly, and Aw can also be computed using the plume widths
from time t and t+dt.
The dispersion step is not complete without inclusion of the diffusion term. The diffusion
equation of a species concentration for any plume cell i may be expressed by v. .,
dC, 3 dC,
i = _fjr l\ fa AQ\
i xv * \y-*ty)
dt dy y dy ...
where Ky is the horizontal eddy diffusion coefficient. It can be expressed by
*, = ~«# (9-50)
By applying the relationship between ay and Wp in Equation 9-1, the foEowtog form for Ky can
be derived.
Wl .
9-18
-------
EPA/6QO/R-99/030
Numerical integration of a form like Equation 9-49 is solved with the Crank-Nicholson method in
which centered differencing is used for both time and space derivatives to solve the diffusion
equation. This technique provided stable numerical solutions compared to an explicit method.
The resulting finite-difference equation is then solved by matrix decomposition of the coefficient-
matrix into upper and lower (LU) triangular matrices. For convenience in setting up the matrix
form of the finite difference diffusion equation for all plume cross-section cells, a set of
simultaneous equations is developed which are written in a vector-matrix form for solving by LU
decomposition.
The derivation of the set of equations is provided. By averaging Equation 9-49 in time to obtain
dC/dt at t+dt, and using center-differencing to solve for 32C/3y2, simplification leads to
t+t K»dt t*i
c*;1!- - y - ] + cf '[i+
*»i Kdt k KBt
*»ir_ y -i _ t y
, j_ _ _ —
+ C,*[l -
(9-52)
where small k is used as a time index for time t, and k+1 represents t+dt, i is the index to the left
edge of plume cell i, and y is the distance from the new reference position at the left
background location of the plume cross-section. By using the relationship that y-t— tl,Wp, the
following relations can be defined.
a, =
*!- 1,-,)
Kdt
9-19
-------
BPA/600/R-99/030
P, = «, + Y, (9-55)
By substituting the above expressions into Equation 9-52, the following formula is obtained.
(9-56)
Equation 9-56 represents a system of simultaneous equations in the form, . •:,
Ax = b (9-57)
where, x = (C;k+1 + ....Cj^*1 )' and N is the total number of plume cells. Additionally, the
expression, A = I - G, is defined where A is a square matrix of size (N+l)x(N+2), and b = *
(I+G)Ck . In these expressions, I is the identity matrix, and G is a tridiagonal matrix of the form,
G = tridiag (a, p, y ) , and is of size (N+2)x(N+2). Subsequently, the simultaneous system of
equations given by Equation 9-57 is solved using LU decomposition. - ,
The boundary conditions are also included in the above system of equations since N+2 is used.
Appropriate boundary conditions are provided by the CCTM grid concentrations for the right and
left edges of each plume cross-section. At left edge boundary, a, = p, = YI and b, — Cj = C^V
At the right edge boundary, aN+2 , pN+2, and yN+2 are zero which leads to bn+2 = C^2 = C^.
9.3,2.4 Surface Area Emissions and Dry Deposition
During the PinG simulation, surface emissions are injected into individual surface-based plume
cells. The surface emissions, contained in layer 1 values of the 3-D emissions file which is also
employed by the CCTM, are used in all plume cells of a plume cross-section. For each grid cell,
there is an emission rate for certain species (qj). In particular, a MEPSE plume cross-section
passes over such gridded surface emissions. Consequently, the change in concentration of plume
cell i during a time step from surface emissions is
3C/ =
where kj herein is an appropriate species-specific factor for the conversion of concentration from a
mass unit to the appropriate volume unit.
9-20
-------
EPA/600/R-99/030
Since the PinG formulation is composed of a single vertical layer, it was not designed to have the
capability to properly ingest other elevated point-source emissions into the MEPSE plume cross-
section. As long as such an along-path, point-source emission remain at a relatively low value,
the impact on the MEPSE concentration field is assumed to be felt through background
entrainment of the expanding MEPSE plume. If the other point source emission was ingested, of
course, such entrained mass is currently dispersed instantaneously throughout the vertical extent
of the plume. For this reason, when point-source emissions larger than a critical value are
encountered by the MEPSE plume, the current logic is to handover the MEPSE plume to the grid
solution.
Dry deposition is a sink term and occurs at the bottom of each plume cell based on the deposition
velocity concept.
dC/ Vi ,
— = - — C (9-59)
dt Hp ^ }
where C/ denotes the concentration of species j and the species-specific deposition velocity (Vdj)
values are available from a gridded data file. The same deposition velocity is applied to all cells of
a plume section using the Vd values from the grid cell in which the plume centerline is located.
9.3.2.5 Gas-phase Chemistry of Plumes
A gas-phase chemistry mechanism implemented in the CCTM is also invoked by the PinG module.
The current mechanisms include the RADM2 and carbon bond (CB-4) chemical mechanisms.
Details about these chemical mechanisms are provided in section 8. In addition, there is a
separate PinG module version for each chemical solver (i.e., SMVGEAR and QSSA). Minor
revisions were needed in the PinG versions of the solver codes to customize them to deal with the
plume concentration array whose dimensionality differs from the CCTM gridded concentration
array. Nevertheless, the PinG gas photochemistry treatment is identical to that of the CCTM.
Since only gas-phase plume chemistry has currently been implemented, when conditions
conducive to extensive aqueous chemistry are encountered, a plume section is transferred to the
CCTM grid system, as discussed in the handover section below. Future plans include
implementation of the existing CCTM aerosol module into PinG so that it also has the capability
to treat aerosol and particulate species.
9.3.2.6 Plume Initialization
The initialization of each plume cross-section is the first key procedure performed at the start of
its simulation. The PinG initialization of a new MEPSE release occurs when a plume cross-
section width reaches a minimum width. The width criterion is a user-specified variable and it is
specified for the PDM simulation. A flag indicator variable in the PDM file communicates to
PinG when a plume cross-section is ready to be initialized. The current minimum width for
9-21
-------
EPA/600/R-99/030
" " ' ',' • ..... f
5 i;: ',-. ; "',
' *
initialization of a MEPSE plume section has tentatively been set at .2 km. At initialization, it is
assumed that the lateral concentration distribution across the plume section of the primary
emission species exhibits a Gaussian shape. For all other species, initial concentrations have been
set to a machine minimum-value (ie. 10"30),
The lateral concentration distribution of the primary species in the plume cells (i.e., "initial
condition" of the plume concentration field) is given by
where, y* is (y; - y0), the distance of the outer edge of cell i from the plume centerline,
and U is the mean transport speed, qj is the MEPSE emission rate of species j, and
kj is the appropriate species specific mass-to-volume conversion factor. < w ,*
In PinG, it is assumed that the MEPSE emissions are at an hourly resolution, and that we are —
performing plume simulations of hourly releases. The assumption is made that the emission rate
(QJ) remains constant over the hour. Consequently, at the handover time, the transformed mass
impacted by the various plume dynamic and chemical processes corresponding to one hour of
emission is released to the grid solution. • ~* -.
9.3.2.7 Methodology for Plume Feedback to the Eulerian Grid ;v ;,
When the subgrid scale phase of a plume section simulation has been completed, the plume
material is ready for transfer to the CCTM grid. The total concentration of each species in any
plume cell is composed of a component equal to the background concentration, and an additional
component consisting of the plume concentration, which differs from the background grid level.
In performing the handover, the conceptual basis is that only the plume component is related to
the corresponding MEPSE emissions. Thus, the feedback is restricted to the plume contribution.
Additionally, since hourly plume releases are simulated from each MEPSE source in PinG, and
since the assumption has been made that each such hourly emission remains constant for the fiill
hour, the current practice is to handover the contents of a plume section corresponding to a full
hour release. Thus, for each plume species,
where average values for the plume and background grid concentrations are employed. The
average plume concentration (Cp) of each species is determined from all plumjs cejjs. The
background grid value (Cbg) represents the average of the CCTM grid-cell concentrations on
9-22
-------
EPA/600/R-99/030
each side of the plume for all appropriate layers over the vertical extent of the plume. The
handover plume mass is distributed into the column of cells up to Hp in which the plume centerline
is located at the handover time. The feedback of plume mass represents an adjustment to the
CCTM grid cell concentrations from an average concentration (Ctdj). From a mass balance
perspective,
Wo = (CP - c*>r, (9-62)
where VG is the CCTM grid cell volume. The plume volume (Vp) is determined from
V, = WpHp(mt) (9-63)
where U is the mean horizontal wind speed, At is a 1-hour interval and UAt corresponds to the
alongwind dimension (AXp) spanned by the plume section. In a similar fashion, the grid volume
impacted by the plume is given by
VG = AxotyjCZj - Z) (9-64)
where Ax<3 and AyG are the horizontal grid cell sizes, and Za' and ZGb are the heights of the top
and bottom of the model layers spanned by the plume.
Substituting the above expression into Equation 9-62 yields:
HWbx.
cadj =
-------
EPA/600/R-99/030
finer grid sizes are possible, however, test runs have not been performed but are needed in order
to assess the benefits of this approach at urban grid cell sizes. As the primary purpose of PinG is
to improve the subgrid-scale treatment of large point-source plumes, the key handover size • -<
criterion based on plume width relative to grid cell size has been prescribed according to
w -. • ..
-f * Ycr (9-66)
The default value for Yor is 1-0. Therefore, when a plume section width attains the grid cell size, a
flag indicator variable generated by PDM triggers PinG to perform the handover process.
Currently, the CCTM system is operated in a 1-way nesting mode, as multiple nested grids are not
performed in a single simulation. Consequently, PinG is not equipped to handle the movement of
a plume from a larger to a finer gridded domain within the same simulation.
Ideally, plume handover also occurs when the plume has also reached chemical maturation when
concentration differences within the plume become rather small. A chemical criterion also
incorporated into PinG involves a default condition recognizing chemical maturation. It is based
on an average plume concentration ratio of O3/OX ,where Ox= O3+NO2 denotes the principal *
oxidant species using concentrations from all cells in a plume section. This ratio is employed as a
surrogate indicator of plume chemical age. The specific chemical maturation criterion is:
"= * rchem (9-67)
where rchem represents the critical value. The default value has been set at 0.99. Tests have
indicated this value to be a good indicator corresponding to a chemically-mature stage 3 plume.
The use of O3 / Ox as a surrogate for plume age is not common compared to NOX / NOy where
NOy includes all the reactive oxides of nitrogen. However, NOj/NOy has not been applied as a
chemical criterion because injection of fresh emissions from large point sources has been observed
to suddenly decrease the chemical age of the background which leads to "premature" activation of
the chemical criterion. Experience from testing with the first chemical criterion has proved it to
be more satisfactory.
In the initial version of the Models-3 PinG, additional criteria are needed which can also trigger
plume handover to the Eulerian grid cells. As the PinG module is upgraded in the future, the ~
these criteria may be avoided.
9-24
-------
EPA/600/R-99/030
The next two criteria are related to non-MEPSE emissions in the immediate vicinity of a MEPSE
plume cross-section,
Urban Criterion
PinG currently does not have the facility to injest different surface emissions into individual plume
cells. Such a condition is likely to be encountered by a MEPSE plume after touch-down when it
passes over an urban-industrial area with a very inhomogeneous spatial distribution of emissions.
CKHani and Pleim (1996) have identified such "urban" areas based on the following criterion
pertaining to the emissions of NOX.
>1012 moleculeslcm2ls (9-68)
This condition denotes a fairly high NOX surface emission flux (fq). The critical value of 1012
chosen by Gillani and Pleim (1996) was based on an emission inventory with a horizontal spatial
resolution of about 20 km. For such a grid size, this condition corresponds to an emission rate of
about 0.3 kg/s. Thus, the selected "urban" emissions handover criterion is:
(9-69)
The above condition must be satisfied to trigger plume section handover, and the plume bottom
must also be at the ground for this criterion to activate.
Point-source Emissions Criterion
The initial PinG version does not possess multiple vertical layer resolution of the plume. Thus, if
a plume section entrains the elevated concentration of a primary species in a background grid cell
from a fresh NOX emission from a non-MEPSE major point source, such entrained mass is
instantaneously mixed vertically throughout the MEPSE plume. If the non-MEPSE point source
emission is large enough, the related error of its treatment in the MEPSE plume relative to its
treatment in the gridded solution becomes large. Thus, it is advantageous to handover the
MEPSE plume under this circumstance. A handover criterion of 3.33 x 10"6 ppm/s has been set
which corresponds to about 2 ppb/10 minutes. If the new non-MEPSE point-source emissions in
the immediate vicinity of the MEPSE plume are large enough to raise the average background
concentration (left/right and vertically-averaged over MEPSE plume height) by more than 2 ppb
in 10 minutes, then handover is also triggered.
9-25
-------
EPA/600/R-99/030
Precipitation Criterion ' -•_
PinG has no facility to handle wet removal of the plume. Consequently, if the total (convective
and non-convective) precipitation rate in the cell containing the MEPSE centroid exceeds a
critical value, a plume handover is also triggered. Currently, the default value of this critical
precipitation rate is 0.00008 cm/s or about 0.3 cm/hr. 7
Excessive Wind-Shear Criterion
'• • « T- ' ' '-I'- s , ......
Since the plume treatment is PinG is based on a Lagrangian simulation for a single vertical layer,
there is no facility to properly treat the effect of wind shear. When excessive speed wind shear
over the vertical extent of the plume exceeds a critical value, plume section handover will occur
on the basis that the treatment of such wind shear will be better treated in the multi-layer grid :
model.
Domain Boundary Criterion
Finally, a MEPSE plume section is transferred to grid cell in which the centroid of the MEPSE is
located if it is about to exit the gridded domain. Otherwise the plume concentration contribution
cannot be accounted for if the plume travels beyond the model domain boundary.
In all handover cases considered above, the entire MEPSE plume section is transferred. There is
a particular condition in which only a top portion of the plume may be handed over, while the ^
remaining part of the plume continues to be simulated. This type of premature partial handover is
forced because PinG does not possess adequate vertical resolution, and also because the plume
simulation in PinG is based on a Lagrangian approach. This condition arises when the plume is
well-mixed throughout the daytime PEL, and the PEL height decreases significantly along the
path of the plume, for whatever reason. In reality, in such a case, the plume would be split
between an upper portion above the PEL and a lower portion remaining within the PBL. The
two portions could then experience very diverse stability and flow conditions which cannot be~
accommodated in the current version of PinG.
The approach is that once the PBL height begins to decrease, and it decreases by more than 15%
of the previous Zj, the plume section is handed over. However, the lower portion continues to be
simulated. If the decrease in z5 is temporary and it later begins to increase again, the remaining
simulated plume would expand vertically just as if it was the full plume. When Zj continues to
decrease, as during the evening transition period, then the partial handover would continue
incrementally as the PBL decreases by each 15% segment until the nocturnal mixing height is
reached, at which time the remaining plume is also handed over. In this manner, PinG does not
continue to simulate the well-mixed daytime plume into the night. This approach avoids the
Lagrangian simulation of the deep night-time plume which frequently experiences substantial wind
shear conditions related to nocturnal jets and inertial oscillations.
9-26
-------
EPA/600/R-99/030
The situation is different for a fresh MEPSE release into the nocturnal stable layer above the
shallow mixing layer. Such a plume is likely to be thin in the vertical and, therefore, not very
likely experience excessive shearing conditions. In our Lagrangian approach, therefore, PinG will
generally simulate fresh nighttime plume sections which do not experience excessive shear
because of their limited vertical extent.
9.3.2.8 Test Application of the Flume-in-Grid Modeling Approach
The schematic diagram in Figure 9-8 shows how the PDM and PinG fit into the overall
Models-3 CMAQ system of science programs. Since PDM serves as a processor program, it is
exercised in advance of the CCTM/PinG simulation. PDM requires the MEPSE stack parameter
file generated by the MEPPS emissions system, and a set of meteorological data files prepared by
MCIP from an MM5 simulation. The details of these input files which are needed to perform a
PDM simulation are described in the Models-3 user guide (EPA, 1998). Input and output
parameters for the PDM processor are also defined in the user guide.
The PinG module is an integral part of the CCTM, as depicted in Figure 9-8, and it is invoked by
the CCTM driver program along with the other science processes. In the sequence of processes,
PinG is called just before the grid model performs the gas-chemistry step. Consequently, when
PinG completes a time step, the CCTM driver calls the chemistry to undertake gas-
photochemistry on the 3-D gridded array. When PinG is being exercised, it needs data files
generated by ECIP, MCIP, PDM and also the MEPSE emission file from MEPPS processing.
The PinG module also generates a plume concentration file containing the species concentrations
in each plume cell for each plume cross-section. Since the PinG concentration file is a Models-3
specific format, it can be viewed in a visualization software package designed for such data files.
In addition, an effort is planned to visualize both the plume and grid concentrations on the same
display.
The plume-in-grid algorithms were exercised for a single MEPSE within a 36 km gridded domain.
The plume cell O3 concentrations in Figure 9-9 are for various times from a selected plume
section released during the early afternoon. The PinG module simulated 10 plume cells in this
plume section, with the concentrations on each edge of the cross-section at various times being
the CCTM gridded values used as background conditions. It is evident that the modeled O3
concentrations displays the same chemical stages described earlier. A significant ozone deficit
exists in the narrow plume for 1-2 hours after release. As the plume expanded, the O3
concentration gradually recovers and exhibits higher O3 in some cells outside the plume core of
the cross-section at 15:30 than the grid value at each edge or in the plume core. This represents
stage 2 of plume chemical evolution. Eventually, stage 3 was attained as all plume cell O3
concentrations exceeded the CCTM concentrations on each side of the plume with the maximum
excess of about 10 ppb in this case. Shortly after 1730 LST, the last time displayed in Figure 9-9
for this plume section, it was transferred to the CCTM grid system. Therefore, this plume section
9-27
-------
EPA/600/R-99/030
was active in the subgrid phase about 5 hours in this case. Once the feedbeack of a plume section
occurs, PinG no longer simulates it which reduces computational time.
The results of this test case and others not described herein are encouraging since they compare
rather favorably in a qualitative sense with observed concentrations (not shown) across the same
MEPSE plume at about the same time. Nevertheless, an extensive evaluation of the PinG is
planned for several case study days with plume data for various species obtained during the
Southern Oxidant Study 1995 experimental intensive program conducted in the Nashville,
Tennessee area. The evaluation will assess the capability of the PinG components to treat
pollutant plumes and determine the impacts on the regional grid concentrations.
•' "• 4 ' > '.'• ' - ' ; , '1 . i '''
9.4 Summary
A plume-in-grid technique has been developed for use in the Models-3 Community Multiscale Air
Quality modeling system. The key algorithms include a plume dynamics Model (PDM) processor
designed to generate a data file for use in the PinG module simulation. The PinG module has „
been fully integrated into the CCTM Eulerian grid model to provide a more refined, realistic
treatment of the physical and chemical processes impacting selected major point source emissions
during a subgrid scale plume phase in a regional model application. The initialjreleiase version is
limited to performing gas-phase photochemistry within the plumes. A future PinG version is
expected to include an aerosol and paniculate modeling capability. Test simulation results have
been conducted and qualitative results are encouraging regarding the treatment of plume growth
and plume concentrations. However, a rigorous diagnostic evaluation of the various processes is
planned using the 1995 Southern Oxidant Study-Nashville experimental plume data. Future •.
advancements and refinements of the initial plume-in-grid algorithms are anticipated and upgraded
algorithms will be reflected in upcoming releases.
9.5 References
Arya, P., 1984: Parametric relations for the atmospheric boundary layer, Boundary-Layer
Meteorol.,30, 57-73.
Briggs, G.A., 1975: Plume rise predictions. In: Lectures on Air Pollution and Environmental
Impact Analyses, Workshop Proceeding, Boston, MA, 1975, pp 59-111.
Chang, J.S., K.H. Chang, and S. Jin, 1993: Two-way and one-way nested SARMAP air
quality model. International Conf. on Regional Photochemical Measurement & Modeling
Studies, November 8-12, San Diego, CA, A&WMA, Pittsburgh, PA. * "
Clarke, J.F., J. Ching, J.M. Godowitch, 1983: Lagrangian and eulerian time scales
relationships and plume dipsersion from the Tennessee Plume Study. Sixth Symp. on
Turbulence and Diffusion, March 22-25, 1983, Amer. Meteorol. Soc.3 Boston, MA.
9-28
-------
EPA/600/R-99/030
Draxler, R.R, 1976: Determination of atmospheric diffusion parameters, Atmos, Environ., 10,
99-105.
EPA, 1998: Models-3 Volume 9b: User Manual, EPA-600/R-98/069b, U.S. Environmental
Protection Agency, Research Triangle Park, NC 27711.
Gillani, N.V. and W.E. Wilson, 1980: Formation and transport of ozone and aerosols in
power plant plumes. Ann. N.Y. Acad. ScL, 338, 276-296.
Gillani, N.V., S. Kohli, and W.E. Wilson, 1981: Gas-to-particle conversion of sulfur in
power plant plumes: I: Parameterization of the conversion rate for moderately polluted
ambient conditions. Atmos. Environ,, 15,2293-2313.
Gillani, N.V. and J, E. Pleim, 1996: Subgrid scale features of anthropogenic emissions of
NOx and VOC in the context of regional eulerian models. Atmos. Environ., 30, 2043-
2059.
Gillani, N.V., 1986: Ozone Formation in pollutant plumes: A reactive plume model with
arbitrary crosswind resolution. U.S. Environmental Protection Agency, EPA-600/3-86-051,
Research Triangle Park, NC, 85 pp.
Gillani, N.V., 1996: Personal communication.
Godowitch, J.M., J. Ching, and N.V. Gillani, 1995: A treatment for Lagrangian transport and
diffusion of subgrid scale plumes in an eulerian grid framework. Eleventh Symp. on
Boundary Layers & Turb., March 27-31, Charlotte, NC, Amer. Meteorol. Soc., Boston,
MA, 86-89,
Hanna, S.R., G.A. Briggs, and R.P. Hosker, 1982: Handbook on atmospheric diffusion, U.S.
DOE, DOE/TIC-11223, DE82002045, National Technical Info. Center, Springfield, VA.
Hanna, S.R., 1984: Applications in air pollution modeling, Chap. 7, Atmos. Turb. & Air Poll.
Modeling, Ed. F.T.M. Nieuwstadt and H. van Dop, D. Reidel Publishing Co., Kluwer
Academic Publishers, Hingham, MA.
Hicks, B.B., 1985: Behavior of turbulence statistics in the convective boundary layer, J. of
Clim. and Applied Meteorology. 24, 607-614.
Irwin, J.S., 1979: Scheme for estimating dispersion parameters as a function of release height,
EPA-600/4-79-062, Research Triangle Park, NC., 56 pp.
Irwin, J.S,, 1983: Estimating plume dispersion - a comparison of several sigma schemes. J.
Clim. And Appl. Meteorol., 22, 92-114.
9-29
-------
EPA/600/R-99/030
Kumar, N. and A.G. Russell, 1996: Development of a computationally efficient, reactive sub-
grid scale plume model and the impact in the northeastern United State using increasing
levels of chemical detail, J. of Geophys. Res., 101, 16737-16744.
Mathur, R,, L.K. Peters, and R.D, Saylor, 1992; Sub-grid presentation of emission source
clusters in regional air quality modeling. Atmos. Environ., 26A, 3219-3238. .". . '
Morris, R.E, M.A. Yocke, T.C. Myers, and V. Mirabella, 1992: Overview of the variable-grid
Ujrban Airshed Model (UAM-V) 85th Annual Meeting of the A&WMA, June 21-26,
1992, Kansas City, MO., AWMA, Pittsburgh, PA.
* ; ^ • ^ •;• : "' • " •• • •' ..' :•* • i
Myer, T.C., P.O. Guthrie and S.Y. Wu, 1996: The implementation of a plume-in-grid module J,
in the SARMAP air quality model (SAQM). SYSAPP-96-06, Systems Applications
International, Inc., Available from Technical Support Div., California Air Resources '.'.
Board, Sacramento CA.
•' '* ' •' " • • •• ft" •-.*. • *
Niewstadt, F.T.M., 1984: Some aspects of the turbulent stable boundary layer, Boundary-
Layer Meteorol., 30, 31-55.
"' • v : ilV.,|if,, . , -
Odman, M. T. and A.G. Russell, 1991: Multiscale modeling of pollutant transport jnd
chemistry. J. of Geophys. Res., 96, D4, 7363-7370. ;".
Pasquill, F., 1976: Atmospheric dispersion parameters in Gaussian plum modeling: Part II,
Possible requirements for change in the Turner workbook values. EPA-600/4-76-030b, US
EPA, Research Triangle Park, NC , 53 pp. 1
Pasquill, F., 1979: Atmospheric dispersion modeling. J. of the Air Poll. Contrl. Assoc., 29,
117-119.
Seigneur, C., T.W. Tesche, P.M. Roth, and M.K. Liu, 1983: On the treatment of point source
emissions in urban air quality. Atmos. Environ., 17(9), 1655-1676.
Turner, D.B., T. Chico and J.A. Catalano, 1986: TUPOS - a multiple source gaussian ™
dispersion algorithm using on-site turbulence data. U.S. Environmental Protection
Agency, EPA/600/8-86-010, National Technical Information Center, Springfield, VA,
Venkatram, A., D. Strimaitis, D. Cicristofaro, 1984: A semiempirical model to_estimate
dispersion of evelated releases in the stable boundary layer. Atmos. Environ., 18, 923-
928. *' . " • '"• • , ^ _ ..
Venkatram, A. 1988: Dispersion in the stable boundary layer. Chapter 5, In Lectures on Air
Pollution Modeling, A. Venkatram abd J. Wyngaard, Eds., Amer. Meteorol. Soc,, Boston,
MA., 1988,
9-30
-------
EPA/600/R-99/030
Weil, J.C., 1988: Dispersion in the convective boundary layer. Chapter 4, In Lectures on Air
Pollution Modeling, A. Venkatram and J. Wyngaard, Eds., Amer. Meteorol. Soc., Boston,
MA., 1988.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
9-31
-------
EPA/600/R-99/030
Locations of MEPSE Sources
y=grid_mepse_36kmjul7Joapi
1JDOO 78
-"OjOOO
HUM
(WE
by
MCNC
July 7,1395 0:00:00
Min=OJOOO at (1,1), Mw=3DOO at (48,4S)
Figure 9-1 Location of MEPSEs depicted in individual grid cells.
9-32
-------
EPA/600/R-99/030
Suit ate
30
Bscat
/'•..Ozone ppb
/'••' Stages
10x10
11:31
11:38 1224
12:31 15:57
16:06
Time
'igure 9-2. Chemical stages of a plume are depicted by aircraft plume data of SO2, ozone,
sulfate, and aerosol scattering coefficient (Bscat) from crosswind traverses through a large
VTEPSE plume. Measurements are from the daytime period of the 23 August 1978 Tennessee
lume Study experiment. (Adapted from Gillani et al, 1981)
9-33
-------
EPA/600/R-99/03G
[ (a) Time-hajjht view"]
ff r i , i ; r : : :
,«C.-..- «• -, *. J. f t* ,"
BlfSSOKS
exxsi)
| (t>) Top viiw)
Plume handover
Figure 9-3. Schematics of the a) time-height view and b) top view of the modeling concept of the
subgrid scale plume in the Models-3 PinG approach.
9-34
-------
EPA/600/R-99/030
1600 -i
1400 -
1200 -
w 1000 H
s
_l
Cu 800 -
0 600
u.
LL
UJ
200 —
0—'
• Zp
*****
***
I 'I
4 6
10 12 14 16 18 20 22 24
RELEASE TIME (GMT)
Figure 9-4. Example of the final plume height after plume rise for hourly releases from a MEPSE
300 m stack height.
9-3.5
-------
EPA/600/R-99/030
2400 -i
2200 —
2000 —
1800 —
1600 —
f? 1400 —
£ 120° H
S
m 1000 _|
800 —
600 —
400 —
200
0—'
hT
•fZ!
I ' I ' I
246
10 12
1*
I 'I ' I
16 IS 20
TIME (LSI)
Figure 9-5. Time variation of the plume bottom (hb) and top (hT) heights and the PBL height
(Zi) for a nocturnal release.
9-36
-------
EPA/600/R-99/030
o
i
UJ
3
_j
CL
1SOOOO
140000
130000-
120000 —
110000 —
100000 —
90000
80000 —
70000 —
60000 —
50000
40000
30000 —
20000 —
10000 —
0 —
Wp
•
/
I
10
I
12
\
14
16
I
18
I
20
TIME (LST)
Figure 9-6. Time variation of plume width (Wp) from a nocturnal plume release. A plume width
of about 30 km is achieved during the morning period about 10 hours after release.
9-37
-------
EPA/600/R-99/030
*
"
"""•"
.^~
""" *
*•
""*" " ""
<%
/~^
r*
..... ».__... j..^-... -_......-«,
c
-^
«U£
on
•«*
-<
Surlaceimissiona DeposJ!k>n
Plume
cress-section
' at time t
1 at(t+dt)
— Entrapment
Figure 9-7. An illustration of the PinG module formulation depicting the relevant processes;
a) time-height view and b) cross-sectional view.
9-38
-------
EPA/600/R-99/030
Mefewofogy
PDM
Chemistry- Transport
I
PinG
[Figure 9-8. Flow diagram of the PDM processor and PinG module with .associated programs in
{the Models-3 CMAQ system. ..
9-39
-------
VO
-k
O
PinG MODEL RESULTS FOR CPP MEPSE
NASHVILLE DOMAIN (DAY=1995188)
70i
.
60:
ft ;
| 50:
Z :
9 40-
5 :
ff^ '
5 30:
1U
O •
iC. 2O *•'
8 I
10:
0-
^_ « • »- ^
^^^-•'"* a"~""^*~-^
_^,— — *^" __^ — "**" ^ ^^ ^^.A-""^" ~~^^*"--^, ""*--— ~^^
*— ^ -— *- X^*^*
'^X* v"^
\ (V A I
\\ \ j
\ r
\ j
& r
V \ f
VV
XxH»
I I I I | I I I i | I I I I | ! I I I | I I I I I I I I | i I I I I I i I
-«J -15 -10 -50 5 10 15 20
DISTANCE ACROSS PLUME (KM)
ooo OZONE H PUUME 14 FHOM CTM PING CS4
tfHISION - HNQ^Ss«i14 ~ RLE-BNA-MOL
tigure 9-9 Plume ozone concentrations in an expanding cross-section in the subgrid scale plume
phase at various limes (o -14:00, * -14:30, 0 - 15:30, A -15:30, • - 17:30). Plume section was
released at 13:00. Symbol at each edge is the grid boundary concentration.
•I
o
1
VD-
-------
EPA/600/R-99/030
Chapter 10
AEROSOLS IN MODELS-3 CMAQ
Francis S. Binkowski*
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
The aerosol module of the CMAQ is designed to be an efficient and economical depiction of
aerosol dynamics in the atmosphere. The approach taken represents the particle size distribution
as the superposition of three lognormal subdistributions, called modes. The processes of
coagulation, particle growth by the addition of new mass, particle formation, etc. are included.
Time stepping is done with analytical solution to the differential equations for the conservation
of number and species mass conservation. The module considers both PM25 and PM,0 and
includes estimates of the primary emissions of elemental and organic carbon, dust and other
species not further specified. Secondary species considered are sulfate, nitrate, ammonium,
water and organic from precursors of anthropogenic and biogenic origin. Extinction of visible
light by aerosols represented by two methods, a parametric approximation to Mie extinction and
an empirical approach based upon field data. The algorithms describing cloud interactions are
also included in this chapter.
'On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Francis S. Binkowski, MD-80, Research Triangle Park, NC 27711. E-mail:
fzb@hpcc.epa.gov
-------
EPA/600/R-99/030
'< . . ' '?
1Q Q AEROSOLS IN MODELS-3 CMAQ
Inclusion of aerosol particles in an air quality model presents several challenges. Among these:
are the differences between the physical characteristics of gases and particles. In treating gases
in an air quality model, the size of the gas molecules is not usually of primary importance. In
contrast, particle size is of primary importance. The interaction between condensing vapors and
the target particle depends in an important way on the particle size in relation to the mean free '
path in the atmosphere. For gases, once the concentration is known, the corresponding number
of molecules is known. This is not the case for particles. Thus, including aerosol particles in an
air quality model means choosing how the total number, total mass, and size distribution of the
particles is represented. Once this choice is made, then important physical and chemical q
processes involving particles must be represented. Particles may be emitted into trie air by J
natural processes such as wind blowing dust from a desert. Human activities may disturb the soil
to allow wind to blow soil particles off the ground. Sea salt particles come into the atmosphere
by wind driven waves on the sea surface. Volcanic activity is another source of particles for both
the troposphere and the stratosphere. Particles can be made in the atmosphere directly from
chemical reaction. The most important example of this is the transformation of sulfur dioxide, a
by-product of fossil fuel combustion, into sulfate particles. Hydroxyl radicals .attack the sulfur
dioxide and make sulfuric acid that then may nucleate in the presence of water vapor and
ammonia to produce new particles. If there are particles already present in the atmosphere, the
new sulfate may condense on the existing particles or nucleate to form new particles depending
upon conditions which are only recently beginning to be understood. Reactions of organic -•
precursors such as natural monoterpenes and anthropogenic organic species with ozone and other
oxidants or radicals make new species that condense on existing particles or make new particles
depending upon conditions. Combustion sources emit particles composed of mixtures of organic
carbon and elemental carbon. The exact mixture of organic and elemental carbon is a strong .-
function of the conditions of combustion. Once these particles are in the air, they may grow by
condensing of species upon them as has already been mentioned. For a large group of particles
made in the air, i.e., secondary particles, growth may be related to relative humidity because of
water condensing on the particles. Another gas-particle interaction is the chemical equilibration
of species within or on the surface of a particle with gases and vapors within the air. Unlike
gases, particles coagulate, e.g., collide and form a particle whose mass and volume are the sums
of the masses and volumes of the colliding particles. Thus, adding particles to an air quality -
model means adding a new set of physical processes.
In designing the aerosol component of CMAQ the following assumptions were made. Any
representation of particles had to be consistent with observations of particles. The representation
had to be mathematically and numerically efficient to minimize computer time. And finally the
representation had to be usable for regional to urban simulations. These assumptions led to a
choice of two methods. The first method would be to model particle behavior in set of bins of
increasing size. This approach is quite popular and is described originally by Gelbard et. al.
(1980) and more recently by Jacobson (1997). The second approach, the one chosen for
implementation in CMAQ, is to follow Whitby (1978) and model the particles as a superposition
of lognormal subdistributions called modes. The sectional method using the discrete size bins;;
requires a large number of bins to capture the size distribution. If one wishes to model several
10-2
-------
EPA/600/R-99/030
chemical components then the number of components is multiplied by the number of bins. This
leads to a very large number of variables that must be added to an air quality model to capture
particle behavior. In the modal approach, using the three modes suggested by Whitby(1978),
only three integral properties of the distribution, the total particle number concentration, the total
surface area concentration, and the total mass concentration of the individual chemical
components in each of the three mode. The current approach differs from that taken by
Binkowski and Shankar (1995) where the sixth moment was chosen as a third in integral
property in place of the second moment. That moment as chosen because of a mathematical
simplification (see Whitby and McMurry, 1997). The mathematical simplifications of the modal
method allow analytical solutions to be used for the aerosol dynamics. The current approach uses
numerical quadratures to calculate all of the coagulation terms. The numerical quadratures were
compared with the analytical expressions exhibited in Whitby et al. (1991) and are accurate to six
decimal places. The choice of using numerical quadratures was made to reduce the memory
requirements associated with a variable geometric standard deviation and because the second
moment unlike the sixth moment does not have an analytical form.
The aerosol component of the CMAQ is derived from the Regional Particulate Model (RPM)
(Binkowski and Shankar, 1995) which in, turn, is based upon the paradigm of the Regional Acid
Deposition Model (RADM), an Eulerian framework model (Chang et al., 1990). The particles
are divided into two groups, which are fine particles and coarse particles. These groups generally
have separate source mechanisms and chemical characteristics. The fine particles result from
combustion processes and chemical production of material that then condenses upon existing
particles or forms new particles by nucleation. The coarse group is composed of material such as
wind-blown dust and marine particles (sea salt). The anthropogenic component of the coarse
particles is most often identified with industrial processes. The common EPA nomenclature used
in air quality refers to PM2 5 (particles with diameters less than 2.5 u,m) and PMIO (particles with
diameters less than 10u,m). Note that PM,0 includes PM25. Thus, in the present context, coarse
particles are those with diameters between 2.5 and 10 um. Then, the mass of the coarse particles
is the difference between the masses in PM10 and PM2 5.
As already noted, the aerosol particle size distribution is modeled using the concepts developed
by Whitby (1978). That is, PM25 is treated by two interacting subdistributions or modes. The
coarse particles form a third mode. Conceptually within the fine group, the smaller (nuclei or
Aitken), /-mode represents fresh particles either from nucleation or from direct emission, while
the larger (accumulation),/-mode represents aged particles. Primary emissions may also be
distributed between these two modes. The two modes interact with each other through
coagulation. Each mode may grow through condensation of gaseous precursors; each mode is
subject to wet and dry deposition. Finally, the smaller mode may grow into the larger mode and
partially merge with it. These processes are described in the following subsections. The
chemical species treated in the aerosol component are fine species sulfates, nitrates, ammonium,
water, anthropogenic and biogenic organic carbon, elemental carbon, and other unspecified
material of anthropogenic origin. The coarse-mode species include sea salt, wind-blown dust,
and other unspecified material of anthropogenic origin. Because atmospheric transparency or
visual range is an important air quality related value, the aerosol component also calculates
estimates of visual range and aerosol extinction coefficient.
10-3
-------
EPA/600/R-99/030
10.1 Aerosol Dynamics
The particle dynamics of this aerosol distribution are described fully in Whitby et al. (1991) and
Whitby and McMurry 1997); therefore, only a brief summary of the method is given here.
(Note: In the following equations repeated subscripts are not summed.)
10.1.1 Modal Definitions
Given a lognormal distribution defined as
N
i/27c m<
exp
-0.5
Incr
(10-1)
where N is the particle number concentration, D the particle diameter, and Dg and crg the
geometric mean diameter and standard deviation of the distribution, respectively. The fclh
moment of the distribution is defined as
Mk = I* zA(ln Dj d(ln £>)
(10-2)
with the result
In
CTS
(10-3)
MQ is the total number, AT, of aerosol particles within the mode suspended in a unit volume of air.
For A = 2, the moment is proportional to the total paniculate surface area within the mode perj
unit volume of air. For A = 3, the moment is proportional to the total particulate volume within
the mode per unit volume of air. The constant of proportionality between M2 and surface area is
7t; the constant of proportionality between A/3 and volume is ?t/6. Note that the geometric Pn is the average density of the nth species. The third moment for the coarse mode is
Sir
10-4
-------
EPA/600/R-99/030
obtained in a similar manner. Given a value of third moment concentration and number
concentration, the geometric mean standard deviation and the geometric mean diameter for each
mode is diagnosed from
i 2 1 (~, r,,^ ~
in CT =-r[21nlMj -3
§3v. *• v - - j - ^
n3_ M3 (10-5b)
JVexp
The prediction equations for number, second moment and species mass are given in Section
10.1.4.
10.1.2 New Particle Production by Nucleation
The CMAQ aerosol component has a choice of two particle production mechanisms, those of
Harrington and Kreidenweis (1998a,b) and Kulmala et al. (1998). Both of these methods predict
the rate of increase of the number of particles, J, (in number per unit volume per unit time) by
the nucleation from sulfuric acid vapor. In order to predict the rate of increase on new mass and
new second moment an assumption about particle size is necessary. Following work by Weber et
al. (1997), it is assumed that the new particles are 3.5 nm in diameter. Weber et al. reported
measurements of the concentration of particles that are in the size range 2.7 to 4.nm. For
simplicity we have chosen 3.5 nm as a representative diameter.
Using either of these methods, the production rate of new particle mass [ jig m"3 s"1 ] is then
d Mass _ 7CD£/3 j
dt ^ 35 (10-6a)
and that for number [ m"3 s"1 ] is
dNum _ j (10-6b)
dt
and that for second moment [ m2 m"3 s-1 ] is
2 r
(I0-6c)
where d^ is the diameter of the 3.5 nm particle and p is the density of the particle (taken as
sulfuric acid) at ambient relative humidity (Nair and Vohra, 1975).
10-5
-------
EPA/600/R-99/030
10.1.3 Primary Emissions . -. •:. :
The EPA emission inventory for PM2 5 and PM,0 does not currently contain information about
neither size distribution nor chemical speciation. In the CMAQ work, the assumption is that the
major part of PM2 5 particulate mass emissions are in the accumulation mode with a small
fraction in the Aitken mode; i.e. a fraction of 0.999 of PM25 is assumed to be in the accumulation
mode and the remaining fraction, 0.001, is assigned to the Aitken mode. Sensitivity studies will
be conducted to evaluate this assumption. In order to estimate the emissions rate for number and
second moment from the mass emissions rate an assumed mass size distribution is required. It is
convenient to express the emission rate for number, Eg, and that for second moment £2 in terms
of a total emissions rate for third moment. This is shown schematically as follows where En is
the mass emissions rate for species n and pn is the density for that species
(10-7a)
(10-7b)
(10-7c)
where the sum is taken over all emitted species.
In Equation 10-7b,c, EQ and £2 schematically represent the emissions rates for the various
modes. In Section 10.1.4, the nomenclature used to represent the emissions rate for number for
each of the three modes will be respectively Enf, En,; and Encor.
J • .!• seas, t •.
We have chosen values of 0.3 urn for the geometric mean diameter for mass, D v, and 2.0 for the
geometric standard deviation, a for the accumulation mode. The corresponding values for the
Aitken mode are 0.03 jam and 1.7, and those for the coarse mode are 6 urn and 2.2.
The current emissions inventory estimates that 90% of PM10 is fugitive dust, and that 70% of this
dust consists of PM2 5 particles. The paradigm adopted for the CMAQ is that fogitive dust is aT"
coarse mode phenomenon with a tail that overlaps the PM2S range. Therefore, 90% of PMIO
emissions are assigned entirely to the coarse mode species ASOIL. Sulfate emissions are treated
differently in CMAQ than in RPM. In RPM sulfate emissions were treated as particles and
distributed between the Aitken and accumulation modes. In CMAQ, the photochemical module
has sulfate emissions incorporated into the chemical solver. Thus, the production rate for sulfuric
10-6
-------
EPA/600/R-99/030
acid will include direct emissions of sulfate. This rate is passed from the photochemical module
to the aerosol module. Assigning fractional amounts of emitted PM25 and PM10 to the specific
species in Table 10-1 is a matter of ongoing discussions with those responsible for preparing the
national emissions inventory.
10.1.4 Numerical Solvers
The numerical solvers for the two fine particle modes in the Models-3 aerosol component have
been modified from those in RPM, which followed from Whitby et al. (1991). The major
difference is that the RPM solvers linearized the quadratic term for intramodal coagulation in the
equation for modal number concentration. The new solvers in CMAQ retain this quadratic term.
The number concentrations for the Aitken and accumulation modes are denoted as JV/ and Nj
respectively. Intramodal coagulation coefficients are functions only of the geometric mean
diameters and geometric standard deviations for each mode and are denoted as FQH and FQ/J.
Similarly, the intermodal coagulation coefficient for coagulation between the Aitken and
accumulation modes is FQU. For simplicity the following coefficients are defined.
For the Aitken mode:
ai = FOU »bi = Nj FOJJ , and
d Num _ ., d Num „,,„,,,
c — —. 1-£0/5 with —;— from (10-ob);
at at
and for the accumulation mode:
a/ = F0M-, and c/ = EQJ
The emissions rates for number concentration are EQJ and EQJ and are set to values determined
for each mode from Equation 10-7b.
We may now write for the particle number concentrations
-„ .A/2 fc A, . -j (10-8a)
2 (10-8b)
.
dt J
=c .-a N.
J J
Equation 10-8a, a Riccati type equation and Equation 10-8b, a logistics type equation, have
different analytical solutions depending upon whether c/ and GJ are zero or nonzero. These
analytic solutions are used in the CMAQ solver with the coefficients being held constant over
one model time step. In discussing the analytical solutions to Equations 10-8a and b, subscripts
will be omitted for simplicity
10-7
-------
EPA/60Q/R-99/03Q
The solution to Equation 10-8a for c/ # 0 is of the form
r,
P exp(£)f)j
where
For c/ = 0, the solution to Equation 10-8a is of the form
bNfyexp (-b
The solution to Equation 10-8b when cj^Qis of the same form as that to Equation 10-8a except
b = 0. The solution when cj = 0, known as Smoluchowski's solution, is:
The equations for the prediction of second moment, M2, in the Aitken and accumulation modes
are both of the form '
with solutions of the form
exp -
In these equations, production of second moment is denoted by P2 and loss by Lj -For the Aitken
mode, the production term includes the rate of second moment increase by new palrticle
formation from Equation 10-6c, condensational growth (Equation 7a of Binkowski and Shankar,
1995) and by primary emissions from Equation 10-7c. The loss term accounts for the loss of
second moment by intramodal coagulation, as well as including the transfer of second moment to
the accumulation mode by intermodal coagulation. For the accumulation mode, the production
term includes the transfer of second moment by intermodal coagulation, condensational growth
(Equation 7b of Binkowski and Shankar, 1995) and the contribution of primary emissions from
Equation 10-7c. The loss term accounts for intramodal coagulation.
10-8
-------
EPA/600/R-99/030
It is important to note that the history variable in CMAQ is the modal surface area, which, as
already noted, is n time the second moment. For convenience, however, within the internal
aerosol subroutines, the second moment is the treated. Before returning to the main CMAQ
routines, the second moment is multiplied by n. That is why species number 23 and 24 in Table
10-1 are identified as modal surface areas. It is also important to note that the surface area
predicted by CMAQ is the surface area for spherical particles and may not represent the true
surface area available in nonspherical particles or in porous particles such as carbon soot,
Empirical correction factors may be needed for use of CMAQ surface area predictions in certain
applications.
The equations for mass concentration of species n may be written as:
(10-9b)
where Pf = tf + E" + /?"O, and I,-
/¥ jbf/TS$
with cpf = — -r — from Equation 10-6a, when n denotes sulfate, and where E" and E" are the
emission rates and R" is the gas-phase production rate for species n. The factors Q/ and Qj,
defined by Equations A17 and A18 of Binkowski and Shankar (1995) represent the fractional
apportionment of condensing species. FUJ is the coagulation coefficient for the third moment.
Note that the loss of mass in Equation 10-9a is a gain of mass in Equation 10-9b. This represents
the transfer of mass by intermodal coagulation. There is no such transfer of number in Equations
10-8a,b because of the convention that when a smaller particle coagulates with a larger particle
there is a loss of number from the population of smaller particles, but no gain of number in the
population of larger particles. There is, however, a transfer of mass. Equations 10-9a and b have
an analytic solution holding the coefficients constant for the time step of the form:
The solution to Equation 10-9b are by an Euler forward step once again holding the production
terms constant over that time step.
The equation for the prediction of coarse mode mass is
The solution is by an Euler forward step. The equation for coarse mode number is similar
because coagulation is ignored for the coarse mode, and is also solved by an Euler forward step.
10-9
-------
EPA/600/R-99/030
10.1.5 Mode Merging by Renaming
' - ^ £• ' ' ••• : . i,; 3,. ;
In Binkowski and Shankar (1995), the Aitken mode diameters grew over the simulation period to
become as large as those in the accumulation mode. While this is probably true in nature, it
violates the modeling paradigm that two modes of distinct size ranges always exist, This
phenomenon can be modeled by mode merging as follows. The Aitken mode approaches the
accumulation mode by small increments over any model time step when particle growth and
nucleation are occurring. Thus, an algorithm is needed that transfers numberand jmass
concentration from the Aitken mode to the accumulation mode when the Aitken mode forcing
exceeds the accumulation mode forcing and the number of particles in the accumulation mode is
no larger than that in the Aitken mode, - - .>
This algorithm is formulated as follows (Binkowski et al., 1996). When Equation 10-10 is
satisfied, the diameter of overlap, « , for the modal number distributions can be calculated
exactly. Given this diameter, the fraction of the total number of Aitken mode particles greater
than this diameter is easily calculated from the complementary error function
F,m,,, = 0.5[l + erfc (x,,,,,,,)] , where (10-10a)
Y =
Hum fTT,
/2m
and dgni is the geometric mean diameter for the Aitken mode number distribution.
The number concentration corresponding to these particles is transferred to the accumulation
mode, a processes denoted here as renaming the particles. A similar process is used to transfer
mass (third moment) concentration and surface area (second moment) concentration from the
Aitken to the accumulation mode using the complementary error function corresponding to the
third moment.
= 0.5[l+ erfc (*,)], . (10-1 Ob)
In (agi)
where xk = xmm, - •
/2
For numerical stability, the transfer of number and mass is limited so that no more than one half
of the Aitken mode mass may be transferred at any given time step.
3 In (a) " ' ;
This is accomplished by requiring that —=— < x,
'mini '
The fraction of the total number and surface area (k= 2) and mass (k=3) remaining in the Aitken
mode is calculated from the error function of the overlap diameters as:
10-10
-------
EPA/600/R-99/030
" (10-1 le)
q>. -
M2i=
-------
EPA/600/R-99/030
in air quality models uses simplified parametric approaches to model the effect of clouds rather
than modeling the clouds directly. This approach was used in RADM and RPM and is applied in
the first version of CMAQ,
The assumptions for aerosol behavior in clouds are:
* The Aitken (/) mode forms interstitial aerosol which is scavenged by the cloud droplets.
All three integral properties of the Aitken mode respond to in-cloud scavenging.
:••-<,•'•• ' J- «J», •' ,), JK
« The accumulation (j ) mode forms cloud condensation nuclei and thus is distributed as
aerosol within the cloud water. Mass and number in this mode may be lost through
precipitation. Mass but not number is increased by in-cloud scavenging of the Aitken
mode.
* All new sulfate mass produced by aqueous production is added to the accumulation ~-
mode, but the number of accumulation mode particles is unchanged as is the geometric
standard deviation,
-------
EPA/600/R-99/030
with solution (10-13)
where ak (k = 0,2,3) is the attachment rate for interstitial aerosol concentration. The attachment
rate is assumed to be held constant over the cloud lifetime icld- Thg initial values y^O) are
determined after cloud mixing (see Equations 1 1-4 and 1 1-5).
The cloud water aerosol concentration is represented by
(10-14)
dt °
where p is the precipitation removal rate, and P is the production of new sulfate mass by aqueous
chemistry. The first Kronecker delta indicates that only mass (k=3) is increased for the
accumulation mode by chemical production and in-cloud scavenging. The second Kronecker
delta indicates that only number (k=0) is removed by the precipitation removal term in this form.
Mass is removed explicitly in the cloud processor.
The attachment rates, otk, using the form recommended by Prappacher and Keltt (1978) and
including an enhancement factor for the settling velocity of the cloud droplets, v£c are given by;
cxk =
+ 0.5 Pek"3),k = 0,2,3;
Where
Nc and dd are the cloud droplet number concentration and geometric mean
diameter respectively.
Pek = -&-& is a Peclet number. ~ » . _,
The polydisperse diffusivity is given by
f-4k ,
1.246Kn.exp P- 0 Inaj
* \ 2 kj
•
and is the same form as that for dry deposition algorithm (10-18)
(see Binkowski and Shankar, 1995, Equation A29).
The precipitation removal rate for number is given by
10-13
-------
EPA/600/R-99/030
r
fso4l -
L 4J mil
f8SO4l .
L 4J wetdep
t-|"SO4l +r8SO4l
L IJjraiv L 4J /J
\
r"7
(10-19) -v ,
where tcW is the cloud lifetime, [ SSO4]We&fep is the change in sulfate concentration due to
precipitation loss, and [ SO4]/n// is the sulfate concentration at the beginning of the cloud
lifetime, [SO4]scay is the amount of sulfate added from in-cloud scavenging of Aitken mode
sulfate; [ 8SO4WO(f is the amount of new sulfate produced by aqueous chemistry.
10.4 Aerosol Chemistry ;
The aerosol chemical species are listed in Table 10-1. The secondary species sulfate is produced
by chemical reaction of hydroxyl radical on sulfur dioxide to produce sulfuric acid that may
condense on existing particles or nucleate to form new particles. Emissions of fresh primary ,-,
sulfate are treated in the gas-phase chemistry component, and this contributes to the total change
in sulfate from the chemistry component. This is a change from RPM where primary sulfate
emissions were treated as a source of new mass and new particle number. Other inorganic
species such as (ammonia and nitric acid) are equilibrated with the aerosols.
An assumption of the model is that organics influence neither the water content nor the ionic
strength of the system; however, this assumption may not be valid for many atmospheric
aerosols. Although much progress has been made (e.g. Saxena et al., 1995; Saxena and
*~* •* ^ • > *-' , » - . .jiffi" „ H'ljiji:^ ...... "n
Hildemann, 1996), sufficient basic data are not yet available to treat the system in a more
complete and correct way. Over continental North America for PM2 5, sea salt and soil particles
are not considered in the equilibria. Thus, for the initial release of CMAQ, only the equilibrium
of the sulfate, nitrate, ammonium and water system is considered. The equilibria and the
associated constants are based upon Kim et al. (1993a) and shown in Table 10-3.
The aerosol water content is computed using the ZSR method (see Kim et al., 1993a) from:
where W is the aerosol liquid water content [kg m"3], Mn is the atmospheric concentration of the
nth species [moles m"3], and mnQ is the molality [moles kg"3], of the nth species at a value of
water activity (fractional relative humidity) of ctw. The values for the molality as a function of
water activity are calculated from laboratory data from Giauque et aL (1960), Tang and
Munkelwitz (1994), and Nair and Vohra (1975). The ZSR method is used in a somewhat
different way than usual. The water content of sulfate aerosols depends strongly upon the ionic
ratio of ammonium to sulfate. This ratio varies from zero for sulfuric acid to 2.0 for ammonium
sulfate with intermediate values of 1 .0 for ammonium bisulfate, and 1.5 for letovicite. The usual
method would span this range with a single expression; however, Spann and Richardson (1985)
have shown that this is not correct. They proposed a modification which resultedjn a correction
term. A very similar result is obtained by using the ZSR method between the ranges of the ionic
10-14
-------
EPA/600/R-99/030
ratio of sulfuric acid to ammonium bisulfate, ammonium bisulfate to letovicite, and letovicite to
ammonium sulfate. The binary activity coefficients are computed using Pitzer's method and the
Bromley method is used for the multicomponent activity coefficients in the aqueous solution (see
Kim et al., 1 993a) for details.
Two regimes of ammonium to sulfate ionic ratio are considered. The ammonia deficient regime
(in which the ionic ratio of ammonium to total sulfate ion is less than two) leads to an acidic
aerosol system with very low concentrations of dissolved nitrate ion which depend very strongly
on ambient relative humidity. The second regime is one in which the ammonium to sulfate ratio
exceeds two, the sulfate is completely neutralized, and there is excess ammonia. If there is nitric
acid vapor in the system, it will dissolve in the aqueous particles along with the excess ammonia
and produce abundant nitrate.
For cases when the relative humidity is so low that the aerosol liquid water content comprises
less than 20 percent of the total aerosol mass, and the ionic ratio of ammonium to sulfate is
greater than two, "dry ammonium nitrate" aerosol is calculated with the following equilibrium
relationship:
NH4N03(s)oNH3(g) + HN03(g) (10-21)
The value of the equilibrium constant is taken from Mozurkewich (1993) as noted in Table 10-2.
Precursors of anthropogenic organic aerosol (such as alkanes, alkenes, and aromatics) react with
hydroxyl radicals, ozone, and nitrate radicals to produce condensable material. Monoterpenes
react in a similar manner to produce biogenic organic aerosol species. The rates of production of
sulfuric acid and the organic species are passed from the photochemical component to the aerosol
component. The formation rates of aerosol mass (in terms of the reaction rates of the precursors)
are taken from Pandis et al. (1992). These factors are given in Table 10-3.
10.5 Visibility
Visibility is usually defined to mean the furthest distance one can see and identify an object in
the atmosphere. For a detailed presentation on the concepts of visibility, see Malm (1979). In a
perfectly clean atmosphere composed only of nonabsorbent gases, the only process restricting
visibility during daylight is the scattering of solar radiation from the molecules of the gases. This
is known as Rayleigh scattering. Scattering is usually represented by a scattering coefficient. If
absorption is also occurring in addition to scattering, an absorption coefficient may also be
defined. The sum of the scattering and absorption coefficients is called the extinction coefficient.
If absorption is not occurring, the extinction coefficient is defined to be equal to the scattering
coefficient. The visibility in an atmosphere in which Rayleigh scattering is the only optical
process active may be taken as a reference. A useful index for quantifying the impairment of
visibility by the presence of atmospheric aerosol particles is the deciview (Pitchford and Malm,
1994). The deciview index, deciV, is given as
10-15
-------
EPA/600/R-99/030
fa \ (10-22)
where the value of 0,01 [km"1] is taken as a standard value for Rayleigh extinction. The aerosol
extinction coefficient, pejc/ [km"1], must be calculated from ambient aerosol characteristics such
as index of refraction, volume concentration and size distribution.
The extinction coefficient at a wavelength of "k for aerosol may be expressed^ ^
-- (10'23) 1
where the particle distribution is given in a lognormal form as
dV =
where a = , a,, = , ^ = -4j—
A. A. 2 In me Mie extinction efficiency factor, is a
function of a and the index of refraction of the particles. Willeke and Brockmann (1977)
showed that the behavior of the extinction coefficient is a smooth function of the geometric mean
diameter for the volume distribution Dm, and the index of refraction. This smooth characteristic
implies that an accurate approximation to the Mie efficiency can be used in its place to reduce a
very computationally intensive task. The method of Evans and Fournier (1990), a highly
accurate approximation, is used to calculate Qext-
Because routine measurements of aerosol species mass concentrations are often available, but
particle size distribution information is not, an additional method of calculating extinction has
also been included. This is an empirical approach known as reconstructed extinction. The
method is explained by Malm et al. (1994). The formula used here is a slight modification of
their Equation 12 (Sisler, 1998).
$ext [ I/km] = 0.003* f(rh)*{ [ammonium sulfate] + [ammonium nitrate] } (10-25)
+ 0.004 * [organic mass}
+ 0.01* [Light Absorbing Carbon] + 0.001* [fine soil]
+ 0.0006* [coarse mass]
In implementing this method, ammonium sulfate and ammonium nitrate were taken as the sum of
ammonium, plus sulfate, plus nitrate. Organic mass was taken as the sum of all organic species.
Light absorbing carbon was taken as elemental carbon. Fine soil was taken as the unspeciated
portion of PM25 emitted species, and the coarse mass term was not implemented in CMAQ at
this time. The reason for not implementing coarse mass was that the uncertainty in the emissions
was deemed to be too large at the present time. The relative humidity correction, ftrh), is
10-16
-------
EPA/600/R-99/030
obtained from a table of corrections with entries at one- percent intervals. The methodology for
the corrections is given in Malm et al. (1994).
10.6 Summary
The CMAQ aerosol component is a major extension of the RPM. Addition of the coarse mode
and primary emissions now allow both PM2 5 and PMIO to be treated. Ongoing work will
improve the representation of the production of secondary organic aerosol (SOA) material by
including a version of the method of Pankow (1994a,b) as discussed by Odum et al. (1996). This
method, based upon laboratory experiments, calculates the yield of SOA as a function of the
amount of organic material already in the particle phase.
Kleeman et al. (1997) have shown that various source types have size and species information
that may be looked upon as a source signature. This assumes the availability of such source
characteristics for the entire modeling domain. As noted in Section 10.1.3, there are ongoing
discussions with those responsible for the national emissions inventory. As more information
becomes available, identification of source signatures may be possible for a larger domain than
the Los Angeles area, and an effort similar to Kleeman et al. (1997), albeit using a modal
approach, might be undertaken. Other planned improvements for primary particles are the
inclusion of marine aerosol as well as a better treatment of fugitive dust.
Future plans also include an intensive effort to evaluate the CMAQ aerosol component using
atmospheric observations from selected field studies in which aerosol particles were observed.
Comparison with routine visual range observations during the field study periods will provide an
additional method of evaluation.
10.7 References
Binkowski F. S., and U. Shankar, The regional particulate model 1. Model description and
preliminary results. J. Geophys. Res., 100, D12, 26191-26209, 1995.
Binkowski F. S., and U. Shankar, Development of an algorithm for the interaction of a
distribution of aerosol particles with cloud water for use in a three-dimensional eulaeian air
quality model, Presentation at the Fourth International Aerosol Conference, Los Angeles, CA,
Aug. 29 - Sept 2,1994.
Binkowski, F. S., S. M. Kreidenweis, D. Y. Harrington, and U. Shankar, Comparison of new
particle formation mechanisms in the regional particulate model, Presentataion at the Fifteenth
Annual Conference of the American Association for Aerosol Research, Orlando Florida,
October 14-18, 1996.
Bower, K. N. and T. W. Choularton, A parameter!sation of the effective radius of ice free clouds
for use in global climate models. Atmos. Res,, 27, 305-339,1992.
Bowman, F. M., C. Pilinis, and J. H. Seinfeld, Ozone and aerosol productivity of reactive
organics, Atmos. Environ., 29, 579-589, 1995.
10-17
-------
EPA/600/R-99/030
Chang, J. S., F. S. Binkowski, N, L. Seaman, D. W. Byun, J. N. MeHenry, P. J. Samson, W. R.
Stockwell, C. J. Walcek, S. Madronieh, P. B. Middleton, J. E. Pleim, and H. L. Landsford, The
regional acid deposition model and engineering model, NAPAP SOS/T Report 4, in National
Acid Precipitation Assessment Program, Acidic Deposition: State of Science and Technology,
Volume I, Washington, D.C., 1990.
Chaumerliac, N., Evaluation des Termes de Captation Dynamique dans un Modele
Tridimensionel a Mesoechelle de Lessivage de L'Atmosphere, These Presentee a L'Universite" de
Clermont II, U.E.R. de Recherche Scientiflqueet Technique, 1984, : _
Evans, T, N. and G. R. Fournier, Simple approximation to extinction efficiency valid over all
size ranges. Appl. Optics, 29,4666-4670, 1990.
Gelbard, F., Y. Tambour, and J. H. Seinfeld, Sectional representations for simulating aerosol
dynamics. Jour.of Colloid and Interface. Sci., 76, 541-556, 1980,
Giauque, W. F., E. W. Hornung, J.E. Kunzler, and T. R. Rubin, The thermodynamics of aqueous
sulfuric acid solutions and hydrates from 15 to 300 K, J. Amer. Chem. Soc., 82, 62-67, 1960.
Harrington, D. Y, and S. M. Kreidenweis, Simulations of sulfate aerosol dynamics: Part I model
description, Atmos. Environ., 32, 1691-1700, 1998a.
Harrington, D. Y. and S. M. Kreidenweis, Simulations of sulfate aerosol dynamics: Part II model
intercornparison, Atmos. Environ., 1701 -1709, 1998b.
Jacobson, M.Z. Development and application of a new air pollution modeling system-II. Aerosol
module structure and design, Atmos. Environ. ,31, 131 -144, 1997. ; • : * ?
Kim,Y. P., J. H. Seinfeld, and P. Saxena, Atmospheric gas-aerosol equilibrium I.
Thermodynamic model, Aerosol Sci. and Techno!., 19, 157-181, 1993a.
Kim,Y. P., J. H. Seinfeld, and P. Saxena, Atmospheric gas-aerosol equilibrium II. Analysis of
common approximations and activity coefficient calculation methods, Aerosol Set. and TechnoL,
19, 182-198, 1993b.
Kleeman, M.J., G.R. Cass and A. Eldering, Modeling the airborne particle complex as a source-
oriented external mixture. J. Geophys. Res., 102,21355-21372, 1997.
Kulmala, M., A. Laaksonen, and Liisa Pirjola, Parameterization for sulfuric acid/water
nucleation rates. J. Geophys. Res., 103, 8301-8307, 1998.
Leaitch, W, R., Observations pertaining to the effect of chemical transformation in cloud
on the anthropogenic aerosol size distribution, Aerosol Sci. and Techno!., Vol. 25, pp 157-173,
1996.
10-18
-------
EPA/600/R-99/030
Malm, W. C., Considerations in the measurements of visibility, J. Air Pollution Control Assoc.,
29, 1042-1052, 1979.
Malm, W. C., J. F. Sisler, D. Huffman, R. A. Eldred, and T. A. Cahill, Spatial and seasonal
trends in particle concentration and optical extinction in the United States, J. Geophys. Res., 99,
-1347-1370, 1994.
McElroy, M. W., R. C. Carr, D. S. Ensor, and G. R. Markowski, Size distribution of fine
particles from coal combustion, Science, 215, 13-19, 1982.
Middleton, P. B. and C. S. Kiang, A kinetic model for the formation and growth of secondary
sulfuric acid particles, J. Aerosol Sci., 9, 359-385, 1978.
Mozurkewich, M. The dissociation constant of ammonium nitrate and its dependence on
temperature, relative humidity, and particle size. Atmos. Environ., 27A, 261-270, 1993.
Nair, P. V. N. and K. G. Vohra, Growth of aqueous sulphuric acid droplets as a function of
relative humidity,/ Aerosol Sci., 6, 265-271, 1975.
Odum, J. R., T. Hoffman, F. Bowman, D. Collins, R.C. Flagan, and J.H Seinfeld, Gas/particle
partitioning and secondary organic aerosol yields. Environ. Sci. Techno!., 30, 2580-2585, 1996.
Pandis, S. N., R. A. Harley, G. R. Cass, and J. H. Seinfeld, Secondary organic aerosol formation
and transport, Atmos. Environ., 26A, 2269-2282, 1992.
Pankow, J. F., An absorption model of gas/particle partitioning of organic compounds in the
atmosphere. Atmos. Environ., 28, 185-188, 1994a.
Pankow, J.F., An absorption model of gas/particle partitioning involved in the formation of
secondary organic aerosol, Atmos. Environ., 28, 189-193, 1994b.
Pitchford, M. L. and W. C. Malm, Development and applications of a standard visual index,
Atmos. Environ., 28, 1049 - 1054, 1994.
Pratsinis, S. E., Simultaneous aerosol nucleation, condensation, and coagulation in aerosol
reactors, J. Colloid Interface Science, 124, 417-427,1988.
Pruppacher, H. R. and J. D. Klett, Microphysics of Clouds and Precipitation, Reidel, Dordrecht,
Holland, 1978.
Saxena, P. and L. Hildemann, Water-soluble organics in atmospheric particles: a critical review
of the literature and application of thermodynamics to identify candidate compounds, J. Atmos.
Chem., 24,57-109, 1996.
10-19
-------
EPA/600/R-99/030
Saxena, P., L. M. Hildemann, P. H. McMurry, and J. H. Seinfeld, Organics alter hygroscopic
behavior of atmospheric particles, J. Geophys. Res., 100, 18755 - 18770, 1995.
Seinfeld, J. H., Atmospheric Chemistry and Physics of Air Pollution, Wiley, New York, 1986.
Shankar, U. and F. S. Binkowski, Sulfate aerosol wet deposition in a three-dimensional Eulerian
air quality modeling framework, Presentation at the Fourth International Aerosol Conference,
Los Angeles, CA, Aug. 29 - Sept. 2, 1994.
Sisler, J. Personal Communication
Slinn, W. G. N., Rate-limiting aspects of in-cloud scavenging, J. Atmos. Sci., 31, 1172-
1173,1974.
.in . „ ii, il"T Jinn ••
Spann, J. F. and C. B. Richardson, Measurement of the water cycle in mixed ammonium acid
sulfate particles, Atmos. Environ., 19, 919-825, 1985.
Tang, I.N. and H.R. Munkelwitz, Water activities, densities, and refractive indices of aqueous
sulfates and sodium nitrate droplets of atmospheric importance, J. Geophys. Res., 99, 18801-
18808,1994.
Van Dingenen, R. and F. Raes, Determination of the condensation accommodation coefficient of
sulfuric acid on water-sulfuric acid aerosol, Aerosol Sci. Technol., 15, 93-106, 1991.
Weber, R.J., J.J. Marti, P.H. McMurry, F.L. Eisele, D.J. Tanner, and A. Jefferson, Measurements
of new particle formation and ultrafine partilce growth rates at a clean continental site. J.
Geophys. Res., 102, 4375-4385, 1998.
Wesely, M. L., D. R. Cook, R. L. Hart, and R. E. Speer, Measurement and parameterization of
particulate sulfur dry deposition over grass. J. Geophys. Res., 90, 2131-2143, 1985.
Wexler, A. S., F. W. Lurmann, and J. H. Seinfeld, Modeling urban and regional aerosols: I.
Model development, Atmos. Envion., 28, 531-546, 1994.
I I •' • • V" •• : . ' ' • = '
Whitby, K. T., The physical characteristics of sulfur aerosols, Atmos. Environ., 12, 135-159,
1978.
Whitby, E. R.and P. H. McMurry, Modal aerosol dynamics modeling, Aerosol Sci. and
Technol., 27, 673-688, 1997.
Whitby, E. R., P. H. McMurry, U. Shankar, and F. S. Binkowski, Modal Aerosol Dynamics
Modeling, Rep. 600/3-91/020, Atmospheric Research and Exposure Assessment Laboratory,
U.S. Environmental Protection Agency, Research Triangle Park, N.C., (NTIS PB91-
161729/AS), 1991.
10-20
-------
EPA/600/R-99/030
Willeke, K. and J. E. Broekmann, Extinction coefficients for multimodal atmospheric particle
size distributions, Atmos, Environ., 11, 995 - 999, 1977.
Youngblood, D.A. and S,M. Kreidenweis, Further development and testing of a bimodal aerosol
dynamics model, Colorado State University, Department of Atmospheric Sciences Report No.
550, 1994.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
10-21
-------
EPA/6QO/R-99/03Q
Table 10-1 Aerosol Species Concentrations
mass [ u.g m"3 ], number [ # m"3 ]
Units:
{al}
fa2}
fa}
i -
{a4}
I35)
i.!
{a6}
(a?}
{38}
{a9}
jii'1' "
{alO}
{all}
?.' '
{a!2}
{a!3}
{a!4}
{315}
{a!6}
{al?}
{a!8}
{a!9}
ASO4J
:~ ASO4I
ANH4J
•"L
ANH4I
ANO3J
• •>,(• f r "> -~*-\L ' t i j •
Aitken mode biogenic secondary biogenic organic mass
Accumulation mode elemental carbon mass
Aitken mode elemental carbon mass
Accumulation mode unspecified anthropogenic mass
Aitken mode unspecified anthropogenic mass
Coarse mode unspecified anthropogenic mass
Coarse mode marine mass
Coarse mode soil-derived mass
NUMATKN Aitken mode number
NUMACC Accumulation mode number
NUMCOR Coarse mode number
SRFATKN Aitken mode surface area
SRFACC Accumulation mode surface area
AH2OJ Accumulation mode water mass
AH2OI Aitken mode water mass
10-22
-------
EPA/600/R-99/030
Table 10-2. Equilibrium Relations and Constants
(Kimetal., 1993a)
Equilibrium Relation
HSO;(ag) o H+(aqf) + S042-(ag)
NH3(g)oNH3(ag)
NH3(ag) + H2O(ag) o NH;(ag) + OH~(aq)
HN03(0)oH*(aq)+NO,-(ag)
NH4N03(s)oNH3(g)-«-HN03(g)
H2O(aq) o H+(aq) + OrT(aq)
Constant
[Hi
SO|-]yH.ysoS-
[HSO;]yHSoi-
[NH3(aq)]yNH3
^NH3
[NH;
.[°H"]YNH*YOH-
[NH3(ag)]YNHjaw
KI
.NOs'JYH^NOf
^HNO3
°NH3°HN03
[H +
.OH~]YH*YOH-
aw
K(298.15)
1.015E-02
57.639
1.805E-05
2.511E06
5.746E-17*
1.010E-14
a
8.85
13.79
-1.50
29.17
-74.38*
-22.52
b
25.14
-5.39
26.92
16.83
6.12*
26.92
Units
mol / kg
mol / kg atm
mol / kg
mol2 / kg2 atm
atm2
mol2 /kg2
o
K)
U)
The constants a and b are used in the following to adjust for ambient temperature
1
298.15 [K]
These values are only used by Kim et al. (1993a,b). The values used in the CMAQ are from Mozurkewich (1993):
K =ex l 18.87 - - 6.025 In
=exp jl 18.87 -
(f)J
where Mozurkevich reports in nanobars squared. This yields a value for the equilibrium constant of 43. 11 [nb2] at 298. 15 K.
-------
EPA/600/R-99/030
Table 10-3. Organic Aerosol Yields in Terms of Amount of Precursor Reacted
(From Pandis et al. (1992) and Bowman et al. (1995))
Gas-Phase Organic Species
C8 and higher aJkanes
Anthropogenic internal alkenes
monoterpenes
toluene
xylene
cresol
Aerosol Yield
[Hg m~3 / ppm(reacted)]
380
247
740
424
342
221
V
L: ,
10-24
-------
EPA/60Q/R-99/03Q
Chapter 11
CLOUD DYNAMICS AND CHEMISTRY
Shawn J. Roselle* and Francis S. Binkowski"
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
ABSTRACT
Chapter 11 describes the cloud module that is currently incorporated into CMAQ. This module
simulates the physical and chemical processes of clouds that are important in air quality
simulations. Clouds affect pollutant concentrations by vertical-convective mixing, scavenging,
aqueous chemistry, and removal by wet deposition. The CMAQ cloud module includes
parameterizations for sub-grid connective precipitating and non-precipitating clouds and grid-
scale resolved clouds. Cloud effects on both gas-phase species and aerosols are simulated by the
cloud module.
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Shawn J. Roselle, MD-80, Research Triangle Park, NC 27711. E-maii:
sjr@hpcc.epa.gov
**On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
11.0 CLOUD DYNAMICS AND CHEMISTRY
11.1 Background
Clouds play an important role in boundary layer meteorology and air quality. Convective clouds
transport pollutants vertically, allowing an exchange of air between the boundary layer and the
free troposphere. Cloud droplets formed by heterogeneous nucleation on aerosols grow into rain
droplets through condensation, collision, and coalescence. Clouds and precipitation scavenge
pollutants from the air. Once inside the cloud or rain water, some compounds dissociate into
ions and/or react with one another through aqueous chemistry (i.e., cloud chemistry is an
important process in the oxidation of sulfur dioxide to sulfate). Another important role for
clouds is the removal of pollutants trapped in rain water and its deposition onto the ground.
Clouds can also affect gas-phase chemistry by attenuating solar radiation below the cloud base
which has a significant impact on the photolysis reactions.
The Models-3/CMAQ cloud model incorporates many of these cloud processes. The model
includes parameterizations for several types of clouds, including sub-grid convective clouds
(precipitating and non-precipitating) and grid-scale resolved clouds. It includes an aqueous
chemistry model for sulfur, and includes a simple mechanism for scavenging.
11.2 Model Description
The cloud model can be divided into two main components, including the sub-grid cloud model
(sttbcld) and the resolved cloud model (rescld). For large horizontal grid resolutions, the grid
size will be larger than the size of a typical convective cloud, requiring a parameterization for
these sub-grid clouds. The sub-grid cloud scheme simulates convective precipitating and non-
precipitating clouds. The second component of the cloud model considers clouds which occupy
the entire grid cell and have been "resolved" by the meteorological model. The rate of change in
pollutant concentrations ( m,) due to cloud processes is given by:
dt
eld
subcld
dt
(11-1)
rescld
The terms on the right-hand side of Equation 11-1 are solved separately at different times. The
influence of sub-grid clouds are instituted once an hour while the resolved clouds impact
concentrations every synchronization timestep. Each subcomponent of the cloud model is _
described in detail below.
11-2
-------
EPA/600/R-99/030
11.2.1 Subgrid Convective Cloud Scheme
— /{mixing, scavfaqchem, wetdep) (11-2)
subcld
The current sub-grid cloud scheme in CMAQ was derived from the diagnostic cloud model in
RADM version 2.6 (Dennis et al., 1993; Walcek and Taylor, 1986; Chang, et al., 1987; Chang, et
al., 1990), Seaman (1998) noted that most convective parameterizations are based on the
assumption that "the area of an updraft is small compared to the area of the grid cell" and most
parameterizations can be used at grid resolutions a small as 12 km (Wang and Seaman, 1997). In
CMAQ, subgrid clouds are considered only for horizontal grid resolutions on the order of 12 km
or more. Seaman (1998) also pointed to a study by Weisman et al. (1997) that showed that
explicit cloud models can resolve convection for resolution finer than 5 km. Within CMAQ, for
resolutions of 4 km or less, vertical convection is assumed to be resolved at the grid level;
therefore, the resolved cloud model will be the only cloud scheme used at small grid scales.
The effects of sub-grid clouds on grid-averaged concentrations are parameterized by modeling
the mixing, scavenging, aqueous chemistry, and wet deposition of a "representative cloud"
within the grid cell. For all sub-grid clouds, a 1-hour lifetime (rc/rf) has been assumed. Sub-grid
clouds can be either precipitating or non-precipitating, and the non-precipitating subgrid clouds
are further categorized as pure fair weather (PFW) clouds and non-precipitating clouds
coexisting (CNP) with precipitating clouds.
The subgrid cloud model determines if precipitating or nonpreeipitating clouds exist over each
grid cell. Precipitating clouds are simulated when the meteorological preprocessor (currently the
Mesoseale Model version 5 or MM5, Grell et al., 1994) indicates precipitation from its
conveetive cloud model. The CMAQ implementation differs from the RADM in that only the
convective precipitation amounts from MM5 are used to drive the subgrid precipitating cloud.
RADM used the total precipitation (convective and nonconvective precipitation) to drive the
subgrid cloud model. In CMAQ, the nonconvective precipitation is used in the resolved cloud
model. Nonpreeipitating clouds are modeled if the moisture and temperature profiles support the
development of a cloud (Dennis et al., 1993). Nonpreeipitating clouds are modeled only when
the relative humidity of the source level is above 70% and the calculated cloud base is below
1500 m for PFW clouds or 3000 m for CNP clouds. For both precipitating and nonpreeipitating
cloud types, the geometry of the cloud (base, top, and spatial extent) are determined next. The
cloud base is calculated by lifting a parcel of air from the cloud source level (the level between
the surface and 650 mb with the highest equivalent potential temperature) to the lifting
condensation level (LCL). The cloud top calculation depends upon the cloud type and
atmospheric stability. For precipitating clouds in unstable conditions, the cloud top is found by
following the moist adiabatic lapse rate from the cloud base up to the level where it becomes
cooler than the surrounding environment. For precipitating clouds under stable conditions, the
cloud top is set as the first layer above the cloud base in which the relative humidity falls below
11-3
-------
EPA/6QQ/R-99/030
65%, but limited to less than the 600 mb height. For nonprecipitating clouds, further restrictions
are placed on the cloud top. The cloud top calculation for nonprecipitating clouds uses the same
relative humidity criterion as the precipitating clouds, but the cloud top is allowed to extend up to
500 mb for CNP clouds and only to 1500 m for PFW clouds. If the atmosphere is unstable, the
nonprecipitating cloud top may be reduced in height if the parcel method calculated a lower
cloud top. The fractional cloud coverage calculations depend on cloud type and have been
described thoroughly by Dennis et al. (1993). For precipitating clouds, the model uses a
parameterization similar to approach of Kuo (1974). The fractional coverage parameterization
for the nonprecipitating clouds is based on relative humidity.
The convective cloud simulated by the sub-grid cloud model is considered to be composed of air
transported vertieally-from below the cloud, entrained from above the cloud (for precipitating
clouds), and entrained from the sides of the cloud. Concentrations of pollutants for each layer of
the cloud are calculated by:
— eld P"/• _ \—down — ~1 , x — up
/«/ c^) = /;«[(i - A/* >»/ + /;,we»«/ c^> J + (i - /;„, >*<• ci 1-3)
where fs,Je is the fraction of entraining air originating from the side of the cloud. For
nonprecipitating clouds, no entrainment of air from above the cloud is allowed and therefore
fSKh=\. The entrainment,^,, is calculated by iteratively solving conservation and
thermodynamic equations (Dennis et aL, 1993; Chang et al, 1990,1987; and Walcek and Taylor,
IT. . fir W ft , . -
1986). The terms w/ and mt represent the above and below cloud concentrations,
respectively. Once the cloud volume has been determined, vertically-averaged cloud
temperature, pressure, liquid water content, total water content, and pollutant concentrations, are
computed with liquid water content (PFC) as the weighting function (gives the most weight to the
layers with the highest liquid water content). Thus, the average pollutant concentrations within
the cloud are calculated by:
J
(11.4)
*fli*n v '
With the averages over the cloud volume, the processes of scavenging, aqueous chemistry, and
wet deposition are considered. The final step in cloud mixing is the reapportioning of mass back
into individuaf layers. This is accomplished using cloud fractional coverage, initial in-cloud
concentrations, final in-cloud concentration, and the initial vertical concentration profile. For
11-4
-------
EPA/600/R-99/030
precipitating clouds, the average pollutant concentration for the grid cell within the cloud layers
is computed by:
—eld
(*>'o)
cfrac + mt (z, t0)[ 1 - c/rac] (11 -5)
where cfrac is the fractional cloud coverage. There are variations on this equation for below
cloud, above cloud, and for nonprecipitating clouds.
11.2.1.1 Scavenging and Wet Deposition
Pollutant scavenging is calculated by two methods, depending upon whether the pollutant
participates in the cloud water chemistry and on the liquid water content. (1) For those pollutants
that are absorbed into the cloud water and participate in the cloud chemistry (and provided that
the liquid water content is > 0.01 g/m3), the amount of scavenging depends on Henry's law
constants, dissociation constants, and cloud water pH. (2) For pollutants which do not participate
in aqueous chemistry (or for all water-soluble pollutants when the liquid water content is below
0.01 g/m3), the model uses the Henry's Law equilibrium equation to calculate ending
concentration and deposition amounts. The rate of change for in-cloud concentrations (m?let) for
each pollutant (/*) following the cloud timescale (Tcld) is given by:
-cU
-1
Tcld
where a, is the scavenging coefficient for the pollutant. For subgrid convective clouds, Tcld is 1
hour and for grid resolved clouds it is equal to the CMAQ's synchronization timestep. For gases,
the scavenging coefficient is given by:
1
washout
i, TWF
1 T
where H, is the Henry's Law coefficient for the pollutant, TWF is the total water fraction given
by:
TWF = —* (11-8)
WrRT.
11-5
-------
EPA/600/R-99/030
where pH20 is the density of water, WT is the mean total water content (kg/m3), R is the
Universal gas constant, and T is the in-cloud air temperature (K). The washout time,
represents the amount of time required to remove all of the water from the cloud volume at the
specified precipitation rate (Pr),
and is given by:
_ c
-washout ~ (11*9)
Here, &ZM is the cloud thickness.
The accumulation mode and coarse mode aerosols are assumed to be completely absorbed by the
cloud and rain water. Therefore, the scavenging coefficients for these two aerosol modes are
simply a function of the washout time:
1
(11-10)
*r*
washout
The Aitken mode aerosols are treated as interstitial aerosol and are slowly absorbed into the
cloud/rain water. This process is discussed in detail in the aerosol chapter (Chapter 10).
The wet deposition algorithms in CMAQ were taken from the RADM (Chang et al., 1987). In
the current implementation, deposition is accumulated over 1-hour increments before being
written to the output file. The wet deposition amount of chemical species / (vwfep/) depends upon
the precipitation rate (Pr) and the cloud water concentration (mfttf):
*cU —dd
vtdep^ I mi Prdt (11-11)
o
Deposition amounts are accumulated for each of the modeled species, but the user specified
which species are written to the output file. This is handled in the Program Control Processor
(see Chapter 1 5).
11.2.1.2 Aqueous Chemistry
The aqueous chemistry model evolved from the original RADM model (Chang et al., 1987; and
Walcek and Taylor, 1986). The model considers the absorption of chemical compounds into the
cloud water; the amount that gas-phase species absorb into the cloud water depends on
thermodynamie equilibrium, while accumulation-mode aerosols are considered to have been the
nucleation particles for cloud droplet formation and are 100% absorbed into the cloud water.
Then the model calculates the dissociation of compounds into ions, oxidation of S(IV) to S(VI),
and wet deposition. The species that participate in the aqueous chemistry are given in Table 11-
11-6
-------
EPA/600/R-99/030
1. This version of the aqueous chemistry model differs from Walcek's scheme in that it tracks
contributions from gases and aerosols separately. It also considers the scavenging of interstitial
aerosols, and it allows for variable-length cloud time scales.
Table 11-1. List of Species Considered in the Aqueous Chemistry Model
Gases
Aerosols
SO2
HNO3
N205
C02
NH3
H202
03
formic acid
methyl hydrogen peroxide
peroxy acetic acid
H2SO4
SO4 (Aitken & accumulation)
NH4+ (Aitken & accumulation)
NO3" (Aitken, accumulation, & coarse)
Organics (Aitken & accumulation)
Primary (Aitken, accumulation, & coarse)
CaCO3
MgC03
NaCl
Fe3*
Mn2+
KC1
Number (Aitken, accumulation, & coarse)
11.2.2 Resolved Cloud Scheme
At any grid resolution, clouds may be resolved by the MM5, which could include stratus,
cumulus, or cirrus type clouds. The resolved clouds have been simulated by the MM5 to cover.
the entire grid cell. No additional cloud dynamics are considered in CMAQ for this cloud type
since any convection and/or mixing would have been resolved and considered in the vertical
wind fields provided by MM5. A resolved cloud horizontally covers the entire grid cell and
—eld • — .'••••-.
vertically extends over the whole depth of the layer, thus m, and nn are equivalent in resolved
clouds. These clouds are activated in MM5 when the humidity is high enough for water vapor to
condense, and then MM5 computes cloud and'rain water amounts according to any of several
microphysical submodels. Using the total of the condensed cloud water and rain water reported
by MM5, the CMAQ resolved cloud model then considers scavenging, aqueous chemistry, and
wet deposition. The average liquid water content We in a model layer (z) for the resolved cloud
is given by:
(11-12)
11-7
-------
EPA/600/R-99/030
where Qc(z) is the cloud water mixing ratio (kg/kg), QR(z) is the rain water mixing ratio (kg/kg),
p(z) is the air density (kg/m3). Precipitation amounts for resolved cloud layers, P/z), are derived
using the MM5 non-convective precipitation amounts (J?J, apportioned into individual model
layers with the vertical profile of condensed liquid water as follows:
Pr(z)=Ra
(11:13)
Once quantities for precipitation rate, liquid water content, etc. have been calculated, then the
scavenging, aqueous chemistry, and wet deposition are solved using the same procedures as in
the subgrid clouds.
dim
dt
dmi
resold
dim
(11-14)
aqchem
Several assumptions have been made in the current implementation of the resolved cloud model.
(l) The lifetime of the resolved cloud computations varies based on the synchronization timestep
of CMAQ. (2) Following the method of operator splitting, the effect of the resolved clouds on
pollutant concentrations occurs at the end of the cloud lifetime, thus no exchange between layers
is permitted during the cloud life-cycle. (3) The pollutants, cloud water, and rain water are
uniformly distributed within the grid cell. Because of the separation of MM5 from CMAQ, we
do not have the information to do precipitation fluxes. Even if a complete cloud precipitation
model was developed within CMAQ, there is no guarantee that it would be consistent with what
was done in MM5.
11.3 Conclusions
One of the concepts for Models-3 was that multiple modules may exist for each physical process
of the air quality model. The implementation described here is the first module available for
modeling cloud physics and chemistry. Other subgrid cloud models (i.e., the Kain-Fritsch (1990,
1993) and Betts-Miller (1986)) are under consideration and may be included as optional modules
for CMAQ. In addition, a more detailed resolved cloud model is under development which will
include a microphysical submodel for following the evolution of the cloud (i.e., cloud droplet
formation, growth of rain droplets, and descent through model layers to the ground). It will also
consider resolved cloud lifetimes which extend beyond the CMAQ synchronization timestep,
thus maintaining the partition between gas and aqueous-phase pollutants during the gas-phase
chemistry calculations. The current implementation of the cloud model in CMAQ will be
evaluated using available datasets and will be used as a reference for evaluating future cloud
modules for CMAQ.
11-8
-------
EPA/600/R-99/030
11.4 References
Belts, A.K, and M.J. Miller, 1986. A new convective adjustment scheme. Part II: Single column
tests using GATE wave, BOMEX, ATEX, and arctic air- mass data sets. Quarterly J. Roy.
Meteor. Soc., 112, 693-709.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madronieh, P. Middleton, W.R. Stockwell, and CJ.
Walcek, 1987. A three-dimensional Eulerian acid deposition model: Physical concepts and
formation. J. Geophys. Res., 92, 14681 -14700.
Chang, J.S., P.B. Middleton, W.R. Stockwell, C.J. Walcek, I.E. Pleim, H.H. Lansford, F.S.
Binkowski, S. Madronich, N.L. Seaman, D.R. Stauffer, D. Byun, J.N. McHenry, P.J. Samson, H.
Hass., 1990. The regional acid deposition model and engineering model, Acidic Deposition:
State of Science and Technology, Report 4, National Acid Precipitation Assessment Program.
Dennis, R.L., J.N. McHenry, W.R. Barchet, F.S. Binkowski, and D.W. Byun, 1993. Correcting
RADM's sulfate underprediction: Discovery and correction of model errors and testing the
corrections through comparisons against field data, Atmos. Environ., 26A(6), 975-997.
Grell, G.A., J. Dudhia, and D.R, Stauffer, 1994, A description of the fifth generation Perm
State/NCAR mesoscale model (MM5). NCAR Tech. Note NCAR/TN-398+STR, 138 pp.
Kain, J.S. and J.M. Fritsch, 1990. A one-dimensional entraining/detraining plume model and its
application in convective parameterization. J. Atmos. Sci, 47,2784-2802.
Kain, J.S. and J.M. Fritsch, 1993. Convective parameterization for mesoscale models: The Kain-
Fritsch scheme. The Representation of Cumulus Convection in Numerical Models, Meteor,
Monogr., 46, Amer. Meteor. Soc., 165-170.
Kuo, H.L., 1974. Further studies of the parameterization of the influence of cumulus convection
on large-scale flow, J. Atmos. Sci., 31, 1232-1240.
Seaman, N.L., 1988. Meteorological Modeling for Air-Quality Assessments: A NARSTO
Review Paper, Submitted to Atmos. Environ.
Walcek, C J. and G.R. Taylor, 1986. A theoretical method for computing vertical distributions
of acidity and sulfate production within cumulus clouds, J. Atmos. Sci 43, 339-355.
Wang, W. and N.L, Seaman, 1997. A comparison study of convective parameterization schemes
in a mesoscale model. Man. Wea. Rev., 125, 252-278.
Weisman, L.M., W.C. Skamarock and J.B. Klemp, 1997. The resolution dependence of
explicitly modeled convective systems, Man. Wea. Rev., 125, 527-548.
11-9
-------
EPA/600/R-99/030
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality {CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
11-10
-------
EPA/600/R-99/030
Chapter 12
METEOROLOGY-CHEMISTRY INTERFACE PROCESSOR (MCIP) FOR
MODELS-3 COMMUNITY MULTISCALE AIR QUALITY (CMAQ) MODELING
SYSTEM
Daewon W. Byun* and Jonathan E. Pleim"
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
Ruen Tai Tang
Dyntel Corporation
Research Triangle Park, NC 27711
AI Bourgeois
Lockheed Martin/U.S. EPA Scientific Visualization Center
Research Triangle Park, NC 27711
ABSTRACT
The Meteorology-Chemistry Interface Processor (MCIP) links meteorological models such as
MM5 with the Chemical Transport Model (CTM) of the Models-3 Community Multiscale Air
Quality (CMAQ) modeling system to provide a complete set of meteorological data needed for air
quality simulations. Because most meteorological models are not built for air quality modeling
purpose, MCIP deals with issues related to data format translation, conversion of units of
parameters, diagnostic estimations of parameters not provided, extraction of data for appropriate
window domains, and reconstruction of meteorological data on different grid and layer structures.
To support the multiscale generalized coordinate implementation of the CMAQ CTM, MCIP
provides appropriate dynamic meteorological parameters to allow mass-consistent air quality
computations. The current implementation of MCIP links MM5 meteorological data to CMAQ
CTM. Because its code has a streamlined modular computational structure, adapting the system to
other inputs only require inclusion of a reader module and coordinate related routines specific for
the meteorological model.
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Daewon W. Byun, MD-80, Research Triangle Park, NC 27711. E-mail:
bdx@hpcc.epa.gov
** On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/60Q/R-99/030
12.0 METEOROLOGY-CHEMISTRY INTERFACE PROCESSOR (MCIP) FOR
ijIQDELS-! COMMUNITY MULTISCALE AIR QUALITY (CMAQ) MODELING
SYSTEM
12.1 Introduction
Models-3 Community Multiscale Air Quality (CMAQ) modeling system is a next generation
modeling system designed to handle research and application issues for multiscale (urban and
regional) and multi-pollutant (oxidants, acid deposition, and particulates) air quality problems. Its
Chemical Transport Model (CTM) uses a generalized coordinate system. To accommodate
meteorological inputs from a variety of meteorological models using different coordinate systems,
CMAQ CTM (CCTM) requires information about the coordinates and grid as well as the
meteorological data. The Meteorology-Chemistry Interface Processor (MCIP) links a
meteorological model with CCTM to provide a complete set of meteorological data needed for air
quality simulation. Because most meteorological models are not built for air quality modeling
purposes, MCIP takes care of many issues related to data format translation, conversion of units of
parameters, diagnostic estimations of parameters not provided, extraction of data for appropriate
window domains, and reconstruction of meteorological data on different horizontal and vertical
grid resolutions through interpolations as needed. Considering these functions, it is not difficult to
see that MCIP is a key processor in the Models-3 CMAQ system.
In the Models-3 CMAQ system, the role of MCIP is further expanded to enforce consistency
among the meteorological variables. The consistency among meteorological parameters and the
way they are utilized in a CTM greatly influence the success of air quality simulations. This issue
becomes a dominant concern for the CCTM, which uses a generalized coordinate system, because
it should be able to deal with data from different meteorological models that may or may not use
fully compressible formulations (or assumptions on the atmospheric dynamics such as hydrostatic
or nonhydrostatic approximation). Chapters 5 and 6 of this science document provide detailed
descriptions on the generalized coordinate system.
CMAQ's MCIP provides similar functions as the meteorological preprocessor for Regional Acid
Deposition Model (RADM) (Chang et al., 1987, 1990). MCIP's code has a streamlined
computational structure, incorporating many of the physical and dynamical algorithms necessary to
prepare meteorological inputs used by CMAQ. Some of the planetary boundary layer (PEL)
parameterizations are extensively updated subroutines of the RADM's meteorological preprocessor
which was described in Byun and Dennis (1995). MCIP is highly modularized to accommodate
data from different meteorological models. This versatility is accomplished by allowing
incorporation of a different set of input modules for a specific meteorological model. At present,
two sets of input modules are available. One links to the output of Pensylvania State
University/National Center for Atmospheric Research (PSU/NCAR) Mesoscale Modeling System
Generation 5 (MM5) with CCTM and the other links to meteorological data already in the Models-3
input/output applications programming interface (I/O API) format. MM5 can be run either in
hydrostatic mode using a time dependent terrain-following hydrostatic pressure coordinate, or in
nonhydrostatic mode using a time independent terrain-following reference pressure coordinate.
For the details on how MM5 simulations are conducted and how the reference state is determined,
refer to Seaman et al. (1995), Seaman and Stauffer (1993), Dudhia (1993), Grell et al. (1994),
Stauffer and Seaman (1993), Tesche and McNally (1993), and Haagenson et al. (1994). To
12-2
-------
EPA/600/R-99/030
characterize past atmospheric conditions properly, MM5 is usually run with Four Dimensional Data
Assimilation (FDDA) for air quality simulations. It incorporates the results of intensive
meteorological observations into the model simulations so that the uncertainties associated with
meteorological input for a CTM are minimized (Stauffer and Seaman, 1993; Seaman et al., 1995).
Different output data generated by the different options, such as dynamics cloud parameterization
and surface-PBL algorithms used for the operation of MM5 operation, can be handled by MCIP
accordingly.
This chapter provides a detailed description of the functions and data flow of MCEP and
formulations used for estimating parameters diagnostically. Although this chapter mainly describes
the current implementation of MCEP written for MM5, it also provides key information necessary
to build different MCEP versions for other meteorological models. It is hoped that developers of
different MCEP versions can concentrate their efforts to read in data files from different
meteorological models with minimal modifications for diagnostic routines and output processes.
Section 12-1 describes basic functions and data dependency of MCIP. Section 12-2 deals with
meteorological data types, coordinates, and grids. Section 12-3 contains descriptions of diagnostic
algorithms used for estimating parameters necessary for air quality simulation and Section 12-4
describes additional parameters needed for the generalized coordinate system. Section 12-5
provides key operational operational details, such as building and executing MCIP including how
to set up grids/domains and environmental variables for different runtime options. Refer to Figure
12-1 for the structure of the contents of the present chapter.
12-3
-------
BPA/6QO/R-99/030
Meteorological
Model Outputs
Environmental
Variable Inputs
Landuse Data
DATA
DEPENDENCIES
MOP Moliles
(Section 12.5,1)
UKtview fSratm 12.!)
BulMil* MCP
(Stolen 12.5.2)
DattTypci
lnJt Grids
fSection 12.21
ExeeuinjjMCIP
(Section 12.5.31
Providing boundary
layer paranrters
Defining Grid&
Domain
(Section 12.5.4)
inconsistert
meteorological
model atinutfe.e..
PBLhelgltl,
deposition velocities,
clout! parameters }
Figure 12-1. Contents and Structure of Chapter 12
12.1.1 MCIP Functions
One of MCIP's functions is to translate meteorological parameters from the output of a mesoscale
model (e.g., MM5) to the Modefs-3 I/O API format (Coats, 1996), which is required for
operations of Models-3 CMAQ processors. Some other necessary parameters not available from
the meteorological model are estimated with appropriate diagnostic algorithms in the program. The
key functions of MCIP are summarized below.
Reading in meteorological model output files:
:'!, .'1'1 . . i-M, , •,•:;:; ' " , •' • • ',' ' ' 'jij"1',
The EPA-enhanced MM5 version (Pleim et al., 1997) generates not only the standard MM5 output
but also several additional files that contain detailed PEL, cloud, and surface parameters. MCJP
reads these files and stores the information in the memory for further processing. Essential header
information is passed to the Models-3 I/O API file header.
Extraction of meteorological data for CTM window domain:
In general, the CCTM uses a smaller computational domain than MM5. Also, MM5 predictions in
the cells, nearthe boundary may not be adequate for use in air quality simulation. Therefore, MCIP
extracts only the portion of the MM5 output data which falls within the CCTM's main domain and
boundary cells.
12-4
-------
EPA/600/R-99/030
Interpolation of coarse meteorological model output for finer grid:
When user requests meteorological data on a finer resolution grid than that simulated in the
meteorological model, MCIP interpolates profile data using simple bilinear interpolation.
Collapsing of meteorological profile data if coarse vertical resolution data is
requested;
MCIP performs a mass-weighted averaging of data in the vertical direction. For example, 30-layer
meteorological data may be lumped into 15 layers, or 6 layers for the CCTM.
Computation or passing through surface and PBL parameters;
Depending on the user options, MCIP either passes through surface and PBL parameters simulated
by the meteorology model directly or diagnoses them using the mean wind, temperature, and
humidity profiles, surface data, and detailed landuse information available.
Diagnosing of cloud parameters;
When important parameters needed for processing cloud effects in the CCTM are not provided by
the meteorological model, MCIP diagnoses cloud information (i.e., cloud top, base, liquid water
content, and coverage) using a simple convective parameterization. The information can be used in
the CCTM to process aqueous-phase chemistry and cloud mixing as well as to modulate photolysis
rates that reflect the effects of cloud.
Computation of species-specific dry deposition velocities;
MCIP computes dry deposition velocities for important gaseous species using either diagnosed
PBL parameters or the surface/PBL information passed through from the meteorological model.
Generation of coordinate dependent meteorological data for the generalized
coordinate CCTM simulation;
Many of the coordinate-related functions traditionally treated in a CTM have been incorporated as a
part of the MCIP functions. This change was necessary to maintain modularity of the CCTM
regardless of the coordinates used and to eliminate many coordinate-dependent processor modules
in the CCTM. By incorporating dynamically consistent interpolation methods and associated
subroutines in the CCTM, the dynamic and thermodynamic consistencies among the
meteorological data can be maintained even after the temporal interpolations.
Output meteorological data in ModeIs-3 I/O API format;
MCIP writes the bulk of its two- and three-dimensional meteorological and geophysical output data
in a transportable binary format using the Models-3 input/output applications program interface
(I/O API) library.
12-5
-------
* 1
EPA/600/R-99/030
12,1.2 MCIP's Data Dependency
«*. % < ,» . . •.
**. . c 1 I ,- J- • . <• •
MCEP processes meteorological model output files in order to provide environmental data needed
for the other computational subsystems in ModeIs-3 CMAQ. Landuse data is also required to
generate additional meteorological information needed for air quality simulations. MCBP utilizes
this and profiles of temperature, moisture, and wind components to estimate parameters for the
turbulence and surface exchange characteristics. When the meteorological model computes all the
necessary* information, it can be passed through the MCDP as well. The inputs for MCIP consist of
operational inputs and meteorological model output files. These inputs are described below.
12.1.2.1 Environmental Variable Inputs
The user can select a computational path among the internal process options and define parameters
in MCEP output files by specifying several UNIX environmental variables. These settings allow
MCDP to be configured to fit a particular meteorological model simulation and to follow process
steps requested by the user. First, it defines the mode of meteorological model run to check if the
meteorology data linked have been generated using options compatible with the MCEP. Then it
assigns filenames of meteorology model output files, defines CTM model window domain offset
coordinates in terms of the meteorological model grid definition, determines simulation time and
duration of the MCIP, and assigns appropriate landuse data file type. Refer to Section 12-5 and
Table 12-10 for the details of the UNIX environmental variables used in MCEP for these settings.
O" - ~- * ' . ~ '-?f p " W ' • " M< , r- 1 -, •- • _ , , f, , • c, • ~ 1
12.1.2.2 Meteorology Model Outputs
L • ;l* ;i ' .: . '. . < ' • < • • \ :
It is anticipated that MCEP will include a unique set of reader modules for a variety of
meteorological models. Most of the idiosyncrasies of meteorological data from a specific
meteorological model should be resolved in this module. They include the number of files,
information on the data such as where they are stored and in what format, etc. In the following,
we provide a description of the output files from MM5 as an example.
Standard MM5 output
The standard MM5 time-stepped grid-domain output contains most of the key meteorological data
written by the MM5 subroutine OUTTAP. Each time-step includes a header record. This header is
read by MCEP at each time-step, but only the header from the first time-step is used and the
subsequent headers are ignored. Data following each header consist of two- and three-dimensional
arrays. The time-step interval for the meteorological data is taken from the header record on the
first time-step.
EPA-added MM5 output
EPA and MCNC added two more output files for additional information to facilitate air quality
modeling. One includes additional two-dimensional boundary layer parameters and flux values
arid the other contains the detailed Kain-Fritsch (Kain and Fritsch, 1993) cloud data file, which
describes locations and cloud lifetimes of convective clouds. Currently, only the first file is used
to process CMAQ dry deposition module options and the second file is not actively used because
12-6
-------
EPA/600/R-99/Q30
the CCTM does not yet support a corresponding aqueous-phase Kain-Fritsch cloud mixing
module.
12.1.2.3 Landuse Data
MCIP requires landuse data that define surface characteristics in order to compute dry deposition
velocities and other PEL parameters. Depending on the PBL and dry deposition modules desired,
the needs for landuse data vary. The RADM dry deposition algorithm (Wesely, 1989) needs 11-
category fractional landuse data. On the other hand, the new CMAQ deposition algorithm requires
cell-averaged parameters defining surface exchange characteristics (i.e., landuse dependent
parameters). The latter algorithm assumes that the landuse dependent subgrid effects are processed
in the meteorology modeling system. To avoid incorrect averaging of the land-surface parameters,
it distinguishes a dominant water cell from other landuse types.
Table 12-1. Relations Among MM5, CMAQ/RADM, and USGS Landuse Categories
MM5
1
2
3
4
5
6
7
8
9
10
11
12
13
None
MM5 Category
Urban Land
Agriculture Land
Range-Grassland
Deciduous Forest
Coniferous Forest
Mixed Forest/Wet
Land
Water
Marsh or Wet Land
Desert
Tundra
Permanent Ice
Tropical Forest
Savannah
MCIP
1
2
3
4
5
6
7
9
11
8
7
3*
3
10
CMAQ/RADM
Category
Urban Land
Agriculture
Range
Deciduous Forest
Coniferous Forest
Mixed Forest and
Wet Land
Water
Nonforest Wet
Land
Rock, Open Shrub
Barren Land
Water
Range*
Range
Mixed
Agriculture/Range
Land
USGS
1
2
3
4
7
8
9
10
12
16
13
14
17
6
15
18
19
20
21
22
23
24
25
26
27
11
5
USGS Category
Urban/built up
Dry cropland & pasture
Irrigated cropland & pasture
Mixed dryland/irrigated pasture
Grassland
Shrubland
Mixed shrubland/grassland
Chaparral
Broadleaf deciduous forest
Deciduous coniferous forest
Evergreen coniferous forest
Sub alpine forest
Evergreen broadleaf
Woodland/cropland mosaic
Mixed forest
Water
Herbaceous
Forested wetlands
Barren or sparsely vegetated
Shrub & brush tundra
Herbaceous tundra
Bare ground tundra
Wet tundra
mixed tundra
Perennial snowfields or glaciers
Savannah
Grassland/cropland mosaic
Mapping 'Tropical Forest" to "Range" is not appropriate in most locations.
12-7
-------
EPA/6QO/R-99/030
Currently, MCIP accepts four different types of landuse data. Two types of landuse input (one
from MM5 directly, or the other from TERRAIN, a preprocessor for MM5 system) contain 13-
category data. The third one is a preprocessed landuse data set in an ASCII file. Recently, we
added a landuse processor (LUPROC) for the USGS vegetation information (described below) to
improve landuse data quality. For the case of MM5, the percentages of MMS's 13 landuse
category are transformed into the CMAQ/RADM's 11-category fractions in the landuse reader
module. The USGS vegetation category is also transformed into the 11-category fractions. Table
12-1 provides the conversion rules for landuse types from MM5 13-category, and USGS 27
(http://edcwww.cr.usgs.gov/landdaac/glcc/ nadocl_l.html) category to CMAQ/RADM's 11-
category. The USGS North America land cover characteristics data base has 1-km nominal spatial
resolution and is based on 1-km AVHRR data spanning April 1992 through March 1993. This
data base has been adapted to the CMAQ's base map projection which uses Lambert conformal
projection with origins at latitude 40° N and longitude 90° W.
Landuse Processor (LUPROC)
*•:;*,* . * . . •»••-.-) • t
CMAQ's Landuse Processor (LUPROC) is a special processor that provides a high-resolution
landuse data base for the system. MCIP requires landuse data that define surface characteristics to
compute dry deposition and other PEL parameters. Depending on the PEL and dry deposition
modules desired, the needs for the landuse data are somewhat different.
LUPROC windows out the landuse data for the user-defined domain and converts percentages of
27 vegetation categories in the database into the fractional landuse data in RADM's 11-category.
The output of the LUPROC should have the same resolution as the CCTM domain and the
LUPROC domain_should include the boundary cells in addition to the CCTM's computational
domain. In the near future, MM5 will be upgraded to allow use of the USGS land cover
characteristic data as an option. When the CMAQ dry deposition algorithm is used and necessary
PEL parameters are provided by the MM5 directly, LUPROC will not be needed,
12.1.3 Computational Structure
; . _. jffi . f
The MCIP data flow diagram, Figure 12-2, shows the key processing sequences. First, MCIP
executes the one-flme processes such as: reading the header, processing other operational
information for the meteorological output, reading appropriate landuse data, and generating time
independent MCIP output files. Then, MCIP loops through the time stepped (hourly or sub-
hourly depending on the meteorology data) input data from a meteorology model performing the
functions described above. The processing sequence of MCIP is summarized below:
11 ,,,,ii; 'I .. : s
• GETMET: reads and extracts meteorology data from standard MM5 output for the CCTM
window domain, converts variables into SI units and process special files (e.g., two-
dimensional surface/PBL data and Kain-Fritsch cloud files);
• PBLPKG/PBLSUB: computes PEL parameters using diagnostic method if desired;
• BCLDPRC_AK: computes diagnostic convective cloud parameters if needed;
» SOLAR: computes solar radiation parameters;
* RADMDRY/M3DDEP: computes dry deposition velocities; and
12-8
-------
EPA/600/R-99/O30
METCRO_OUT & METDOT_OUT: computes additional meteorology data required for the
generalized CTM, interpolates mean profile data into finer grid resolution if needed, and
output Models-3 I/O API meteorology files.
START
get eov. variables |
3ETMET
• read met. data
• reconcile coordinate
• unit adjustment
• horizontal interpolation
• process special files
(e.g., 2D met. & cloud flies)
• compute Jacobian, entropy, density
YES
J_
PBLPKG
compute all PEL parameters
PELS UP
supplement PEL parameters
END
Figure 12-2. Flow Chart of MCIP Showing Key Data Processing Sequence.
12-9
-------
EPA/600/R-99/030
12.2 Data Types, Coordinates, and Grids
MCIP's essential role is to provide consistent meteorological data for the CMAQ modeling system.
Therefore, it is very important to understand the vertical coordinate used in the meteorological
modeling system. For example, MM5 can be used either in hydrostatic mode, in which the
hydrostatic pressure (time dependent) is used to define the terrain-influenced vertical coordinate, or
in nonhydrostatic mode, where a reference hydrostatic pressure (time independent) in a normalized
height form is used as the vertical coordinate. Many parts of current MCIP code deal with these
differences in the vertical coordinate specifically. In the following we provide technical
information related with coordinates and grids used in MCIP.
•* ' - **. ' . - , s „ ',
12.2.1 Meteorological Data Types
Many different combinations of approximations are used for describing the atmosphere in
meteorological models. Therefore, classification of MCIP output parameters based on detailed
classification of vertical coordinate types (such as geometric height, pressure, terrain-following
coordinates, etc.), and their application approximations (such as hydrostatic and nonhydrostatic),
can be exceedingly complex for the CTM with generalized coordinate implementation. Unlike
previous use of detailed classification of meteorological coordinates for determining meteorological
data type, we classify the coordinate types mainly based on the temporal dependency of the
Jacobians (explained later). The benefit of this distinction is obvious. For example, a height
coordinate which is time independent may require only a time constant file for describing the
vertical coordinate while a dynamic pressure coordinate requires several parameters related with the
coordinate description that needs to be stored in a time dependent file.
A similar distinction is made based on the need for describing data in different horizontal positions,
such as flux-point data for horizontal wind components and cross-point for most other scalar
parameters. For MM5, two horizontal wind components are defined at so-called 'dot' points while
all other scalar values are at 'cross' points, following Arakawa-E grid definition. Refer to Figure
12-3 for the definitions of Arakawa C-grid and E-grid (Mesinger and Arakawa, 1976). For the
CCTM, certain flux data are defined on the Arakawa C-grid in which the flux points are not
collocated in Jc'-and x2 -directions. Therefore, MCIP interpolates MMS's dot-point wind
components linearly and multiplies the result with the two-point averaged density to provide the
flux-point momentum component data. Both the flux-point and dot point data require array sizes
larger than the cross-point data by one cell in each horizontal direction.
CMAQ utilizes contravariant wind components, instead of the regular wind components, to advect
tracer species. Figure 12-4 shows that the east-west component of the contravariant wind field, u,
is placed at the xl -direction flux points (marked with the square symbol) and the north-south (x2-
direction) component of the contravariant wind field, v, is placed at the Jc2 -direction flux points
(marked with the triangle symbol). The vertical wind component is defined at the full-layer height.
The Models-3 I/O API requires individual data components to have exactly the same temporal and
spatial dimensions in a file. Because the flux-point data and dot-point data both need additional
column or row positions, we can combine them together in a so-called DOT file, which is larger
than the CRO file by one cell for each horizontal direction. It is important to note that because the
12-10
-------
EPA/600/R-99/030
flux points are different from the dot-points, they must be shifted by a half cell east or north
(depending on whether they are square-point or triangle-point flux values) for graphical
visualization of the flux-point parameters.
(C)
(B)
rp*~rpr~rp!'~i u'v~r" "-p^TT
pu—s—pu—s—pu -s—pu
hv—J—pv——pv—
I I I I I I
u,v
L-e
H
>v^>v+j
h
i ' i ' i ' i 'I i >
—pv-J—pv——pv— u,v-4—ti.v u,v-+-u,v
1 I ' I ' I ' I I I I
•H
pu—Q—pu—s—pu—s—pu
I ' I ' I ' I 'I I I I
I—pv—I—pv—I—pv—I u,v—J—u,v—L—u,v—«—u,v
Figure 12-3. Two Different Grid Point Definitions, Arakawa C- and 5-grids, Used in MCIP.
u and v are horizontal wind components, s represents a scalar quantity, and p is density of air. For
MM5, p" is used instead of p.
k+J/2
1-1/2
Figure 12-4. Computational Grid Points and Corresponding Indices Used in MCIP.
Markers are for dot points (•), cross points (x), x-direction flux points (D), y-direction flux
points (A), and vertical cross point at layer interface (®).
12-11
-------
EPA/600/R-99/030
L t :,' . ' . • ; • • -I
Furthermore, when these two flux components are assigned to form vectors in a visualization
program like PAVE (Thorpe, 1996), the starting points of the vectors in the visualization are
relocated at the dot-points although the two components are actually not collocated there.
Data for the boundary cells are defined with a special I/O API boundary data type. Depending on
the need to describe boundary mass flux accurately, one may want to have a boundary domain with
NTHIK cells (see Figure 12-5). The boundary grid is represented as the external perimeter to the
main grid. This perimeter is NTHIK cells wide (where you may use a negative NTHIK to indicate
an internal perimeter as used by such air quality modeling systems ROM and RADM). The
boundary array is dimensioned in terms of the sizes of the arrays surrounding the main domain.
Current Models-3 CMAQ system uses NTHIK=\ throughout the system components.
Finally, the dimensionality of the parameters (i.e., whether they contain three-dimensional or two-
dimensional information) is used to distinguish the data type of meteorological parameters. In
vertical direction, we define half and full layer positions based on the values of the generalized
vertical coordinate, (jt3). Vertical wind component and flux values are examples of the full-layer
parameters. Although full-layer data require one more data point vertically, often the flux values at
the bottom boundary (i.e., at the lowest full-layer) are zero. Therefore, we do not need to use
additional data types for these, as long as we save the non-zero lowest full-layer data in a
corresponding two-dimensional data type file separately. (Refer to Table 12-2 for dimensions used
for each grid points.) Because a file represents a set of data with the same data type (with only a
few exceptions) in the Models-3 system, locating meteorological parameters from an appropriate
I/O API file is relatively easy. Table 12-3 summarizes the possible data types for the
meteorological parameters. Depending on the choice of coordinates and grids, some of data types
may not be relevant. For example, the current version of MCIP does not use GRID_DOT_3D and
MET_DOT_2D data types. Appendix 12A provides the list of MCEP output parameters in each
data file.
12-12
-------
EPA/600/R-99/030
r(5) y(7)
r(4) y(6)
r(3) y(5)
r(2) y(4)
r(l) y(3)
r(0) y(2)
r(-l) yd)
c(-l) c(0) c(l) c(2) c(3) c(4) c(5) c(6) c(7) c(8) c(9) c(10) c(ll) c(12) =NCOLS+NTHIK
x(I) x(2) x(3) x(4) x(5) x(6) x(7) x(8) x{9) x(10) x(ll) x(I2) x(13) x(14)=NCOLS+2*NTHlK
M
1 NCOLS_EXT=NCOLS+2*NTHIK
Note: Individual boundary component grids are treated as a I/O API 2-D file
S(l:NCOLS+NTHtK, 1-NTHIK:0)
E(NCOLS+1:NCOLS+NTHIK, I:NROWS+NTHIK)
N(1-NTHIK:NCOLS, NROWS+1:NROWS+NTHIK)
W(1-NTHIK:0,1-NTHIK:NROWS)
Figure 12-5. Grid Data Structure for Models-3/CMAQ Showing Main Grid and Boundary
Components. Here NTHIK=2, NCOLS=IQ, NROWS=\0 are used for the illustration.
L. NROWS_EXT=NROWS+2*NTHIK; ^
i
r
NTHIK
61
49
95
93
62
50
96
94
9! ^ 92
?
89 § 90
Q,
1
87 Q 88
85 J 86
83 tf
81
79
77
75
73
84
82
80
78
63
51
«IIJU —
(XOR
13
•
64
52
INC
G,Y<
14
2
65
53
RID
! X
I A
RIG)
15
3
66
idary
54
NCO
16
Bounc
4
67
Com]
55
LS,N
17
ary C
5
68
ioncn
56
SlOW
18
'ompi
6
69
-N
57
5) = <
19
ment-
7
70
58
10,10:
20
s
8
71
59
i
c
i
!
>
1
r
21
9
72
60
*»-
22
10
47
45
43
41
39
48
46
44
42
40
o
37 1 38
35 5 36
33 I 34
*§_....
31 § 32
f
H
29 30
27
25
23
11
28
26
24
12
/
/ /
r(12) y(14)
y(I3)
r(10) y(!2)
r<9)
r(8) y(lO)
r<7) y(9)
r(6) y(8)
12-13
-------
EPA/600/R-99/030
Table 12-2. Dimensions and Indices for Grid Points used in MCEP and CCTM
x1
x3
x3
Dot-Point
(start: end)
dimension
index
(1:NCOLS+1)
NCOLS+1
/±l/2
(1:NROWS+1)
NROWS+1
m±\!2
N/A
Cross-Point
(start: end)
dimension
index
(hNCOLS)
NCOLS
/
(1:NROWS)
NROWS
m
N/A
Jt1-
direction
Flux Point
(1:NCOLS+1)
NCOLS-hl
1±\I2
(1: NROWS)
NROWS
m
N/A
x2 -direction
Flux Point
(1:NCOLS)
NCOLS
t
(1: NROWS +1)
NROWS+1
m±l/2
N/A
Full-Layer
( start: end)
dimension
index
N/A
N/A
(0:NLAYS)
NLAYS+1
fctl/2
Half-Layer
(start: end)
dimension
index
N/A
N/A
(1;NLAYS)
NLAYS
k
Table 12-3. Classification of Meteorological Data Types in the CMAQ System,
Data types GltlDJDOT_3*D and MET_DOT_2D (in gray color) are not used in CMAQ currently.
Fiii
ILf.
Time Independent
GRID_CRO_2D
GRID_DOT_2D
GRID_BDY_2D
GRID_CRO_3D
GR1D_DOT_3D
GRID_BDY_3D
Time Dependent
MET_CRO_2D
MET_DOT_2D
MET_BDY_^D
MET_CRO_3D
MKl_UUi_3U
MET_BDY_3D
12.2.2
Coordinates
12.2.2.1 Horizontal Coordinates and Grid
MCIP can be configured with horizontal coordinates based on conformal map projections, such as
Lambert Conformal, Polar Sterographic, and Mercator. Table 12-4 summarizes the necessary
information for the description of the conformal map projections as defined in Models-3 I/O API.
To generate I/O API files, which require exact definitions of the grid and coordinate system, users
need to provide all the necessary map projection information in addition to vertical coordinate and
layering definitions. The Models-3 CMAQ system uses square cells for the horizontal grid
representation. A modeling domain is defined with the integer multiples of the square cells in E-W
(column-wisejl and N-S (row-wise) directions. Although a rectangular cell shape can be handled
by the 1/6 API, the fractional time splitting approach used for the modularization of CTM
processes requires use of square grid to maintain the accuracy of the finite differencing algorithms
consistently. The effect of different horizontal coordinate is reflected in the values of the map scale
12-14
-------
EPA/600/R-99/OSO
factor. Because of the need to couple the map scale factor (m) with other state variables and wind
components, MCIP provides map scale factor for both the dot- and cross-points.
Table 12-4. Map Scale Factors and Parameters Defining Horizontal Coordinates in Models-3 I/O
API (Coats, 1996)
Coordinate
lat.-long.
Lambert
Mercator
graphic
UTM
ID in
Models-3
I/O API
LATGRD3=1
LAMGRD3=2
MERGRD3=3
STEGRD3=4
UTMGRD3=5
Map Parameters
N/A
Pa = $ < p = 0
two latitudes that
determine the
projection cone.
p = A.0. central
meridian
Pa=£,, /},=A0:
latitude longitude of
coordinate origin
within the tangent
circle.
P: angle between
cylinder axis and the
North polar axis
p — A p — ^ '
latitude and longitude
of the point of
tangency.
P : angle from true
north to jc2-axis
Pa is the UTM zone
Pp,P not used
Map Scale (m)
N/A
m = l
sinOr/2-4)r«a»(>r/4-0/2>T
sin(/r/2— $) [_tan(7r/4 — $, /2)J
t 1 sin(7r/2 — ,^e,)
for the center of
coordinate system.
(Jc',x2)are in meters
f*-l »2 \ /-I ^%
\X cent,X cria ) \^st V0)
for the center of
coordinate system.
(5',*2)are in meters
(.X cem,X cent) are offset
from the UTM coordinate
origin.
tx',i2)are in meters
12.2.2.2 Vertical Coordinates and Grid
The Models-3 CMAQ system allows many popular vertical coordinates used in meteorological
models. We expect that the Models-3 I/O API will be extended to include several other vertical
coordinates. The I/O API follows the definitions of vertical coordinates as used in meteorological
models. However, in CMAQ the vertical coordinates are redefined to increase monotonically with
height. This precaution is needed because the CMAQ code is expected to handle generalized
coordinate system. We found that this constraint in the vertical coordinate is extremely useful for
12-15
-------
EPA/6GG/R-99/03Q
the implementation of CTM in the generalized coordinate system by removing possible sign errors
wherever vertical differentiations are involved. For example, the Jacobian information for a
coordinate can bedirectly used to replace some of the derivatives involved with the generalized
vertical coordinate. This internal change does not require redefinition of the vertical coordinates in
the MCIP output files. The impact of this constraint is limited to an include file that defines the
coordinates and vertical layers. In a typical Models-3 operation, this include file is automatically
generated by the system framework through the use of the coordinate/grid manager. In the
following two typical application examples are presented; one for the time dependent coordinate
and the other for time independent coordinate.
Tjyoe dependent hydrostatic sigma-pressure coordinate
For the MM5 system, this coordinate is used when the hydrostatic option is chosen. In general,
this coordinate can be used not only for the hydrostatic atmosphere simulations but also for
nonhydrostatic cases. For example, when a regional scale hydrostatic meteorological model
provides hydrostatic values, the same coordinate can be used in a nonhydrostatic nested model as
described by Jaung (1992). In MCJDP implementation, we should view this coordinate as an
example of processing meteorological data with a dynamic vertical coordinate definition rather than
an example of a hydrostatic coordinate.
The CCTM utilizes a generalized coordinate system that allows construction of the vertical layering
consistent with the meteorological coordinate used. However, to maintain the vertical coordinate
definition monotonically increasing with height, the coordinate definition for the CCTM has been
modified as:
Jc'=!Sl-;a-=J^ = ^- (12-1)
where /?(*', Jc2,*3,?) is the hydrostatic pressure, pT is the pressure at the model top which is held
constant, /^(je1, Jc2,/) is the surface pressure, and p" = ps — pT. Because the pressure used in the
definition of the vertical coordinate is in hydrostatic balance, we have:
f,t : ap _
L ' & .
\ "'
The Jacobian is then defined as:
1 dp
dz
p* dz
PS
(12-3)
and it is time dependent because the surface pressure and air density are time dependent.
Tiine independent reference hydrostatic sigma-pressure coordinate
In MM5 this coordinate is specifically used for the simulation of the nonhydrostatic atmosphere.
The layer structure is entirely different from the case when hydrostatic option is used. The
" * ; 12-16
-------
EPA/600/R-99/030
coordinates defined with this option are time independent and have very similar characteristic to
those of the normalized geometric height coordinates. The only difference is that the vertical layer
thickness is defined with a nonlinear function of the geometric height. Here, we should view this
coordinate as an example of processing meteorological data with time independent vertical
coordinate.
The terrain-following reference hydrostatic pressure (op ) coordinate, again in a monotonically
increasing form, is given as:
\r£L (12-4)
" " Pos-Pr Po,
where p0(xl,x2,x3) is the hydrostatic pressure of the reference atmosphere, pT is the pressure at
the model top, pm(xl,x2) is the reference surface pressure which is determined by the topographic
height zf, and /?* = pw - pT.
Because the reference atmosphere is in hydrostatic balance, we have:
%=-«
dz
(12-5)
The Jacobian is then defined as:
dz
Po
_ Pa
(12-6)
and it is time independent because it is a function of the time-independent surface pressure and the
density of the reference atmosphere. Here, the vertical coordinate z represents a height above the
lowest point in a modeling domain, or the mean sea level (MSL) height if there is no place lower
than the MSL. Usually, this type of vertical coordinate accompanies with a simple description of
reference temperature profile of the base state. For example, in MM5 the base temperature profile
is defined with a simple expression:
P,(z)
(12-7)
where the reference values are chosen such that Tos is a sea level reference temperature in K, p^
1000 mb = 10 Pascal, and A, which is set to be 50 K, is a measure of atmospheric lapse rate
represented in temperature difference for the e-folding depth. With Equations 12-5 and 12-7 and
the Equation of State, we could relate the reference pressure in terms of height z in a differential
equation:
12-17
-------
EPA/600/R-99/030
g
dz R lTos-Aln(pao)]+Aln(pe(z))
which can be readily solved for p0(z) by the separation-of- variables technique. The solution is a
quadratic equation in terms of In pa. For typical tropospheric conditions, the base pressure can be
related with the geometric height by taking the positive square root term:
(12-9)
where b = Tos-A(lnPoo), and c = z-[^(lnPoof + b(lnPoo)}.
K Z
Equation 12-9 shows that/?0(^) is a monotonic, but nonlinear, function of z.
12.2.3 Modification of Grid Structure
In CMAQ, the horizontal and vertical coordinate information (such as the map projection
parameters) for CTM simulation domains are required to be exactly the same as (or a derivative of)
the master coordinates used in the meteorological model simulation. However, the horizontal grid
structures of the CTM domains can be redefined depending on the need for air quality simulations.
Here, the concept of grid family is introduced. It is as a set of grid domain specifications with the
following properties:
1. the same coordinate origin and map projection;
2. a window domain for the parent domain; or
3. a nest (multi-stage) domain from a window or the parent domain.
The Models-3 I/O API file header helps to describe the horizontal grid and coordinates
unambiguously to position the background map correctly and to provide relations among the
members (multi-stage nesting domains) of the grid family. Only with a limited manner, the user is
allowed to modify the grids in the vertical directions because of the concern that ad hoc
interpolation may destroy the integrity of the meteorological data.
12.2.3.1 Windowing
The windowing functions in MCIP extract MM5 output for a CCTM window domain. As a rule of
thumb, a CCTM uses a smaller computational domain than the domain used by meteorological
models, because predictions in the cells near the boundaries may not be suitable for use in air
quality simulation. Therefore, MCIP extracts only the portion of the MM5 output data which falls
within the CTM's main domain and boundary cells. The CTM domain should be located at least
four or five cells inside the MM5 domain to minimize the boundary effects. MCIP can, however,
generate output for the CMAQ domain as large as 2 cells smaller than the meteorology domain,
because it generates files for the boundary cells with NTHIK=\, (Note in MM5 terminology, it
12-18
-------
EPA/600/R-99/030
may seem like the CMAQ domain is smaller by three because a DOT point concept is used in MM5
system for the grid definition.)
Limitation on the number of horizontal cells (to 120 cells) of the meteorology grid is coming from
the MCIP's parameters defined in MCIPPARM.EXT:
PARAMETER ( M&XI = 120, ! MET domain size in N-S dir.
& MAXJ = 120 ) ! MET domain size in E-W dir.
When the meteorology data has larger dimensions, MAXI and MAXJ should be modified
accordingly. Also, there could be some limitations with reader routines specific for meteorology
models. Because these reader modules are essentially foreign codes, there is no easy way to
generalize the input process of the meteorological data. Refer to Figure 12-6, which shows the
relation among the meteorology domain, the window domain for meteorological data extraction,
and the CTM main and boundary domains.
12-19
-------
\o
I i
to
1=1
, . . . .
Relations among MM grid, wrtented-CTM grid, and CTM grid for NDX=3
WM
YOR1G.MM
M
:.
|
x !
-.1-
i
1 !
; x i
— j.. .| . ^~
.....
i
X |
i
X
* -: :
! i
-x ;
X
. i .... ..
X
f ,
1
t X
'. x
X
LJO
S,
x
x
x
X
"""
X
i !
s/
,
x
J=l
V
i
1
a"
' r
—
-
—
i WIIIK^I
Xj x
x j «
X
X
X
p
^
X
X
i X
x
X
X ' X
X
X
X
X
x
X
X
X
X
X
Js
(XORIG.YORIG)*
X | X
X
(XORIOJC.YORIOJC)
H) X
'
'
ORIGJvt
X
M
x
X
1
X
X
X
X
x I - -
X
X
X
«
J&
S
X
X
x
*$
X
i
X X i X
x
•*-[
X
)XJiCM-*
•
->| wc(*-
X
X
x : -
M X
X
X
X *
X s X
CENT,VCBN1>
x ;x *
,
._._
,
i
r
x
X
X
»
0,0)
X
X|x
X | X
I
X'x
X " X
X ' X
x: «
„ i „
. QJT^ nrvM.
X »
EXTENI
x i
' • i" • !
' I
: i
i j
x |x
X ; X
X
X
|
i
x JX| «
X j X
X | X
x X
I
« X
ED CTM
WINI
x
X
X
DO!
iOW
1 x
i
i
x !x
X \ X
X i X
X ' X
X X
» X
« • X
IAIN — 1
DOMA11
;x
—
—
-
s
1
~"
!
X
,
,
X
!
j
x i
1
x:
Xi
-MM DO
—
—
MAI
I-M
X
X i
X
X ]
X \
V
.5
is — : —
AXJJtfM
(XORJO.MM, VOfUG.MM)
Figure 6, Relations among the meteorology domain, window domain for data extraction,
and CTM's main and boundary domains.
-------
EPA/600/R-99/030
12.2.3.2 Horizontal Interpolation
A horizontal interpolation function is provided to help users generate higher resolution data than the
input meteorological data. This function is used when higher resolution emissions data is available
but the detailed variations in meteorological fields can be neglected. This option is useful only
when the surface and PEL parameters are diagnosed in MCEP. If interpolation is desired, NDX
(defined in MCBPPARM.EXT) must be modified before the MCDP is compiled so that data arrays
are dimensioned properly. When interpolation is desired, MCIP copies contiguous grid cell values
from the coarse grid to the fine grid and then performs a two-dimensional bilinear interpolation on
the three-dimensional data. The interpolated temperature profile is updated using fine resolution
landuse data to reflect land-sea boundaries in the profile appropriately. The interpolated
temperature, moisture, and wind components profiles, together with the detailed landuse data, are
used to estimate surface and PEL parameters for the finer resolution. The user should be reminded
that this procedure generates higher resolution meteorological data without enhanced physics, thus
the newly generated data may have consistency problems. Therefore, such interpolation should be
used sparingly for cases such as testing higher resolution emissions data with a finer resolution
CTM. The procedure is never meant to replace or minimize the need for higher resolution
meteorological model runs. It should be noted that the interpolated temperature and moisture
profiles result in different estimations of cloud parameters (such as cloud bottom and top heights,
fractions, liquid water contents) as determined by the diagnostic Anthes-Kuo cloud routine.
12.2.3.3 Vertical Layer Collapsing
A vertical collapsing function is supplied to generate a smaller data set for testing CTM in a smaller
computer system. If desired, MOP collapses MM5 profile data for the coarse vertical resolution
data as defined by the user. MOP performs a mass-weighted averaging of data in the vertical
direction. For example, 30-layer MM5 data may be averaged into 15 or 6 CTM layers. During the
collapsing procedure, the layer description is modified accordingly. The resulting profile may have
consistency problems. This option is usually used to generate meteorology data for a system test-
run for code debugging and development purposes. It is also appropriate to study the effects of the
vertical resolution in air quality simulation such as presented in Byun and Dennis (1995). Refer to
Table 12-5 as an example for the layer collapsing.
Defining the vertical layering structure for the CMAQ system requires consideration of several
factors. Depending on the layer definitions used, the model results will be affected considerably.
The implications of the layer definition are pervading across the entire system. For example, to
determine mass exchange between the boundary layer and free troposphere, a good resolution near
the boundary layer top is preferable. Also, different cloud parameterizations may perform
differently, depending on the layering structure. Aerodynamic resistance, which influences dry
deposition velocities, is a function of layer thickness and the boundary layer stability. For
emissions processing, the layer thickness affects the plume rises from major stacks. Also, the
vertical extent of the area emission effects is limited by the thickness of the lowest model layer for
the CCTM. Although 6-layer vertical grid definition is provided with the tutorial simulation
examples, we do not recommend it to be used for regulatory applications because of the difficulties
in simulating certain processes, such as dry deposition under stable atmospheric conditions.
12-21
-------
BPA/600/R-99/030
Current limitation on the number of vertical layers to 30 comes from one of the MCIP's parameters
defined in MOPPARM.EXT:
PARAMETER { MSXK
= 30 )
! MET number of layers
When a meteorological data has more number of layers, the parameter MAXK should be increased
accordingly. Collapsing is done automatically when the COORD.EXT file (see Models-3 CMAQ
User's Guide for the details) for the output grid has smaller number of layers than the input grid.
Table 12-5. An example of layer collapsing from 15 to 6 <7-layers. Full and half
-------
EPA/600/R-99/030
12.3 Estimation of Physical Parameters
MCIP's essential role is to provide consistent meteorological data for the CMAQ modeling system.
However, the meteorological models used for air quality study may not provide important
boundary layer parameters at all, or may predict those at very coarse temporal resolution, or may
only compute a subset of the needed parameters. In such a case, it becomes necessary to estimate
remaining meteorological parameters using certain diagnostic methods. MOP allows either the
direct pass through of the PBL parameters provided by MM5, or they can be computed from the
mean profiles of temperature, humidity and momentum together with the surface landuse data. In
the following, we explain the diagnostic methods used in MCDP, Note that the approaches
introduced here may not be consistent with MM5 directly and therefore may produce somewhat
different spatial distribution patterns for certain parameters. Basically, the diagnostic routines treat
meteorological model outputs as the pseudo radiosonde observations.
When desired, MCDP estimates key parameters for cloud distributions based on Anthes-Kuo
parameterization. They include precipitation rate, cloud fraction, and cloud base and top heights.
MCIP also provides estimated dry deposition velocities for various chemical species in the RADM
and carbon bond 4 (CB-4) mechanisms,
12.3.1 PBL Parameters
Depending on the user option, MCDP either passes through MM5 predicted surface and PBL
parameters or estimates them us the MM5 profile data and detailed landuse information. The
algorithms used for the diagnostic computation of PBL parameters are provided below.
12.3.1.1 Surface Flux Related Parameters
We utilize a diagnostic method based on similarity theory to estimate the turbulence flux related
parameters. When the meteorological model uses very high vertical resolution and the thickness of
lowest model layer is less than 20 m, we can utilize the surface similarity theory to determine
turbulence parameters for both stable and unstable atmosphere.
For the computation of the surface layer parameters, MCD? utilizes analytical solutions suggested
by Byun (1990) to minimize the needs for numerical iterations in solving the flux-profile relations.
The method has been reviewed and successfully applied in several surface layer studies (e.g.,
Hess, 1992; Lo, 1993 and 1995). Also, a weather research model, the Advanced Regional
Prediction System (ARPS) (Xue et al., 1995) utilizes it in the description of the surface layer. The
algorithm is summarized below.
The nondimensional surface layer profile functions for momentum (m) and potential temperature
(0ft) are defined as:
kzdU z
— = m(—) (12-10a)
M, at L
12-23
-------
EPA/600/R-99/030
=« (12-10b)
where U and 0 are the horizontal wind speed and potential temperature in the boundary layer,
respectively, o*_5s the friction velocity, and 0. is the temperature scale representing the surface heat
flux (eovarianee of potential temperature and wind fluctuations) divided by the friction velocity. L
is the Monin-Obukhov length defined as:
(12-11)
The similarity functional forms proposed by Businger et al. (1971) are used in MCEP as follows:
For moderately stable conditions ( 1> z/L > 0 ) we have:
; $m = l + Pm- (12-12a)
^=Pr0(l + A), (12-12b)
where Pr0 is the Prandtl. number for neutral stability and J3m and fth are the coefficients determined
thgough field experiments. Refer to Table 12-6 for the values of these coefficients used in MCIP.
For unstable conditions (z/L < 0 ), we have:
^=(1-7 £/•« ' (i2-13a)
r &=(l-yAr" (12-13b)
Li
where jm and jh are coefficients of the profile functions. In addition, we added a function for the
very stable condition ( z/L "2.1 ) to extend the applicability of the surface layer similarity following
Holtslagetal.(1990):
fc^CA+) (12-14b)
; ^, ;*, Li
To estimate surface turbulence fluxes with the layered data from a meteorological model, we utilize
the integrated flux-profile relationships: For ( z/L <1 ) we have profile functions represented as a
modified logarithmic functions, when integrated from the height of the roughness length (z0),
12-24
-------
T-
k
L
EPA/600/R-99/030
(12-15a)
(12-15b)
u Z0 L L
and for strongly stable conditions, (z/L >1), we use direct integration of the profile functions:
ft \n — + -
Pmm ^
za L
(12-16a)
(12-16b)
The iff functions are given by (e.g., Paulson 1970), for moderately stable conditions (1£ z/L > 0)
\tr f—\ = -B - £
¥*m^ , I Pm j
and for unstable conditions:
(12-17a)
• (12-17b)
-2 tan"
'(*.)
(12-18a)
'£1
k£/
(12-18b)
where x = (l-rM)"4, ^0 =(l-y« )"4, and
One important fact to note here is that although the similarity functions for stable conditions,
Equations 12-12a-b and 12-14a-b, are continuous at z/L =1, their integrated profiles are not
continuous.
The flux profile relations described above are used to relate the Monin-Obukhov stability
parameters with the readily computable bulk Richardson number Rib:
12-25
-------
• II, ill!1
EPA/600/R-99/030
,. _g(z-zn)A0
(12-19)
For moderately stable conditions (/> z/L > 0), we have:
(12-20)
and for strongly stable conditions, (z/L >1), we can derive a similar solution following the same
procedure as described in Byun (1990) to give:
c=-
7 7
<• I 1 *u
J^K[ i f M
PV-/JV [2 f* +U
1/9
(12-21)
Given a positive bulk Richardson number, one should use both Equations 12-20 and 12-21 to
compute £and choose the value in the correct range between the moderately stable and strongly
stable conditions.
For unstable conditions, we use an efficient analytical approximation by Byun (1990), for
(12-22)
where sb —
^-zn
/?4
PrJ
(12-23)
9 y y
•^ L' m '
54
12-26
-------
EPA/600/R-99/030
IP
, „ |; and
The above equations are used to estimate friction velocity, «., by estimating the bulk Richardson
number using the layered meteorological data:
M = _ _ (12-24)
*
Heat flux can be found by using the temperature scale estimated with:
H = -pCpu.6. (12-25)
where
There have been several suggestions for the coefficients of the surface layer profile functions. For
example, Businger et al. (1971) suggested:
An=4.7, A, =6.35, y. = 15.0, 7A=9.0, Pr0=0.74
with the von Karman constant value of 0.35. On the other hand, Hogstrom (1988) reanalyzed the
experimental data used for the determination of the similarity functions and suggested use of:
&, =6.0, ft =8.21, ym = 19.3, 7» = H.6, Pr0=0.95
with the von Karman constant value of 0.40. In the current version of MCIP, the latter set of
coefficient values are used (refer to Table 12-6).
Moisture flux is found by using similar method as the potential temperature:
(12-27)
where mixing ratio scale, q., is determined with an equation similar to Equation 12-26 for given
Aq = q} — q,. (q, and qs are the mixing ratios at the lowest model layer and at the surface,
respectively).
12-27
-------
EPA/600/R-99/030
ill' ".• n.:'fii ,i: » '" . '„ „ • . -,..., :»". ...••. i ' ' : . I:
I:!1'.,,. • "• 4* .:,,.,: : ••• • •• .IF, ni, . ••' ; ••
12.3.1.2 Estimation of Surface Fluxes Using PBL Profile Functions
Occasionally meteorological models use the lowest model layer thicker than 40 m or so. Under
this condition, it becomes difficult to believe that the lowest layer always belongs to the surface
layer all the time, especially for very stable conditions. To avoid this limitation, MCIP estimates
surface heat, momentum and moisture flux transfer parameters from MM5 surface wind and
temperature by using the boundary layer similarity profiles proposed by Byun (1991). In this
section, we describe the methods used in MCIP for estimating surface fluxes when the thickness of
the lowest model layer is too thick to rely on the surface layer similarity. The intention is to
provide a diagnostic method that can estimate the surface fluxes even when the lowest model layer
extends above the surface layer. It is difficult to expect that a layer with 40 m thick, for example,
continuously belongs to the surface layer though out a day.
Table 12-6. Parametric Constants Used to Describe the PBL in MCIP
Parameter
Symbol
Value
von Karman constant
Coefficient in stable profile function for momentum
Coefficient in stable profile function for heat (scalar)
Coefficient in unstable profile function for momentum
Coefficient in unstable profile function for heat (scalar)
Prandtl number for neutral stability
Critical Richardson Number
Maximum bulk Richardson number
Minimum bulk Richardson number
Minimum magnitude of Monin-Obukhov length
Scale height ratio for neutral stability
Zilitinkevich CH
Neutral value of similarity function for wind component parallel to surface stress
Neutral value of similarity function for wind component normal to surface stress
Reference height
K
ft.
A
7m
Ricr
max(/?/B)
min( RiB)
abs(L)
A(0)
B(0)
0.4
6.0
8.21
19.30
11.60
0.95
0.25
0.70
-4.75
4.0
0.07
0.80
1.70
4.50
10m
The wind and temperature predicted by a meteorological grid model represent layer averaged
values. In order to simplify computation of the surface fluxes, we apply the assumption that the
predicted wind for the lowest model layer has the same direction as the surface stress (i.e.,
um + vm ** Um' where «m and vm are latitudinal and longitudinal components of the first layer mean
wind on the map and Um is the layer mean wind speed in the direction of surface stress). Applying
the PBL momentum profile functions of Byun (1991) and integrating them vertically from z0
(roughness length) to the top of the lowest CTM layer (zF1 < h), one can obtain wind and potential
temperature profiles in the form:
12-28
-------
EPA/600/R-99/030
Um=jPu(H,rlg,-nF), (12-28)
&m-&n=~-P0(^r\0,i}F) (12-29)
where 77 = —, i]F = ——, and fj, = —. For a detailed description of the notation used, refer to
h h L
Byun (1991) and Byun and Dennis (1995). Initially, the atmospheric stability (ji) is approximated
by the analytical solutions of flux-profile relationships described earlier. Then, we compute fi, u*t
and 9* using Equations 12-28 and 12-29, with the Newton-Raphson iteration. Equation 12-29 is
then used to estimate temperatures at heights 1.5 m and 10 m in MCEP.
12.3.1.3 Utilization of Sub-grid Scale Landuse Information
Accurate description of atmospheric turbulence is one of the important elements in modeling the
deposition of pollutants. For Eulerian air quality models, grid-average surface roughness based on
sub-grid scale landuse information has played an important role in the characterization of the
surface condition, which in turn determines intensity of turbulence in the atmosphere. To represent
the atmospheric deposition process utilizing the available sub-grid landuse information, Walcek et
al. (1986) introduced a method to estimate friction velocity for each landuse patch, which is used in
the calculation of the subgrid-scale aerodynamic resistance. Also, several estimation methods have
been proposed for the effective roughness length for use in meteorological grid models. Compared
with the latter methods that estimate the representative grid values for the given sub-grid
information, Walcek et al. (1986) emphasized the description of the sub-grid flux estimation by
introducing a somewhat ad hoc, but useful assumption. They assumed that the quantity t/w. is
constant both for the cell averaged parameters and for the individual landuse patches, i.e.:
Uu. = Ujii. = constant, (12-30)
where the non-subscripted variables refer to grid-averaged values while the ./-subscripted values
refer to the corresponding quantities over individual landuse types. Equation 12-30 is an intuitive
expression of the often observed condition that where wind speed is high, the turbulence is low,
and vice versa, under similar pressure gradient forcing. Strict validity of Equation 12-30 could be
controversial, however, this approach is more realistic than those that assume constant wind or
constant friction velocity. In the dimensional analysis point of view, the condition is a statement
about the conservation of kinematic energy in the presence of surface friction. A derivation leading
to Equation 12-30 is provided below using a combination of the surface layer similarity and the
mixing length theories.
The momentum flux is related with mean wind gradient as:
(12-31)
az
12-29
-------
EPA/600/R-99/030
I! ' ' 4: . , • ••••': _ t
'i "' Jiiir ! in! - •• . . • ;ii ' ' " ;. '"'"
and the eddy diffusivity for momentum is given by:
t ':""' ,«JI ' ' ' .• i, . '• -E '.
no ,_
(12-32)
i • !"* ";,; , ' . ....... • f ., ' ' ' : , i ..... •'•
IL , .A ..... , ••;••! . : ...... •. *•< ..... " ' ...... ,. , ' ' - " '- ..... '
When we apply assumptions such as steady state flow and horizontal homogeneity of the each
landuse patch, the momentum conservation equation can be simplified to give:
-v,, (,2-33)
az az p
Coriolis forcing is neglected in deriving Equation 12-33 because we are dealing with only the sub-
grid scale variation of momentum field.
The right hand side of Equation 12-33 represents the pressure gradient forcing imposed over the
grid and is therefore not dependent on sub-grid representation. Within the surface layer, which is
treated as a constant flux layer, Equation 12-33 is further simplified to give:
_ _ = V n = const. (12-34)
az p
Combining Equations 12-31, 12-32 and 12-34, one obtains
•-const. (12-35)
kz
For the atmosphere at neutral stability, 0m = 1, and the same expression should be applicable for
each landuse patch. Then, Equation 12-35 becomes identical to Equation 12-30 at a given
reference height z - zr.
In addition to Equation 12-30, Walcek et al. (1986) assumed that the surface roughness length can
be averaged as
(12-36)
where f}'s are fractions of different landuse types in a grid cell. Walcek et al. (1986) stated that
Equation 12-36 conforms to a logarithmic wind profile. In reality, however, Equation 12-36 can
be derived from a simple geometric averaging and it does not produce a logarithmic wind profile.
On the other hand, Mason (1987) suggest an averaging method based on the logarithmic wind
profile by introducing a blending height (lb) concept:
-i
(12-37)
12-30
-------
EPA/600/R-99/030
Although this approach is a practical averaging method of the roughness length, this by itself
cannot be used to estimate landuse-dependent friction velocities. Therefore, here we compare
several other roughness length averaging schemes that satisfy both Equation 12-30 and the
logarithmic wind profile function for the neutral condition:
(12-38a)
and
(12-38b)
where zr is the reference height. The objective is to compute landuse specific friction velocities
from the cell average wind and friction velocity values while providing a consistent averaging
scheme for the roughness length. The simplest method for estimating the grid average values from
the sub-grid scale information is using the linear summation with the fractional weight:
(12-39)
where Xj is a physical quantity and j^fj = 1. Because UjU. = constant, one can rewrite Equation
12-30 with Equation 12-39 as follows:
(12-40)
If the surface wind is a quantity that follows Equation 12-39, a relation between the grid average
and sub-grid landuse dependent roughness lengths can be found as:
(12-41)
In order for the expression to be useful, M,;. in Equation 12-41 should be eliminated. From
Equations 12-38a-b and 12-40, a simple relation for «,;. is found in terms of known quantities:
1/2
(12-42)
Substituting Equation 12-42 into Equation 12-41, we obtain a simple relation for the effective
roughness length that satisfies both the logarithmic wind profile and the linearly-additive wind
speed assumptions under the approximation, Equation 12-30:
12-31
-------
EPA/600/R-99/030
(12-43)
The reference height zr needs to be within the regime where the log-linear wind profile can be
satisfied. This means that it should be far away from z0, but still within the surface layer. In the
current MCIP, we are using zr=10m. The new method conserves wind speed and £/,-«. under the
linear summation. .(Equation 12-39). With this assumption the turbulence momentum flux ( «? )
can be summed linearly when it is scaled with the factor [ln(zr /z0)/ln(zr /Z0y)l. Depending on the
degree of inhomogeneity, this factor can be substantially different from unity, making less
consistent with the expectation that fluxes from different patches can be summed up. One may
expect that Equation 12-43 is sensitive to the choice of reference height zr', however, the average
roughness length is not strongly sensitive to the reasonable value of zr between 1 to 10 m.
An alternative approach to the linearly-additive wind speed assumption is to assume that the sub-
grid scale momentum flux can be summed linearly:
(12-44)
Using Equation 12-39 and the logarithmic wind profiles, one can find:
• •••••..- 1 . •
fl245)
In order that Equation 12-45 to be useful, (t/./t/) should be expressed in terms of z „, zai, and zr.
Ml, -S.II '.," . ' ...... .. ........ ,, '
This is accomplished by dividing Equation 12-38a with Equation 12-38b and substituting (M. /«.;)
with Equation 1 2-30 to get:
y» I 1 y / \ ^ '
This leads to an averaging method for the sub-grid scale roughness length:
-l"
(12-47)
Another popular assumption is that at the reference height (or blending height) the drag coefficients
representing the individual landuse patches in a cell can be summed linearly, i.e.:
.ff - ..... , J
(12-48)
12-32
-------
EPA;600/R-99/030
Unlike the two other assumptions introduced, this assumption does not depend on Equation 12-30.
The resulting equation for average roughness length is then identical to Equation 12-37.
Z0 = zr exp
v"2'
j[\n(zr/zoj)]
(12-37')
Although the assumption in Equation 12-48 has been used often in the literature, it is not intuitive
that the sub-grid drag coefficients can be added linearly. The drag coefficient simply quantifies the
turbulence exchange characteristic of a landuse patch. By using this assumption together with the
Equation 12-30, one can readily show that:
4
(12-49)
It is difficult to expect that u? is a physical quantity that can be added linearly. In MCIP, we have
implemented both Equations 12-43 and 12-47 as a user option. Results of the comparison of the
two recommended methods should be available as the model evaluation project progresses.
12.3.1.4 Boundary Layer Heights
Boundary layer height is a key parameter that determines the domain of atmospheric turbulence in
which pollutants disperse. It is used as a fundamental scaling parameter for the similarity theory in
the description of atmospheric diffusion characteristics. Estimating PBL height has been one of the
key functions of meteorological pre-processors for air quality models. Below, we summarize PBL
height estimation algorithms used in MCIP.
Unstable conditions
MCIP estimates the PBL height using the vertical profiles of potential temperature and the bulk
Richardson number with an algorithm similar to the one reported in Holtslag et al. (1995). The
bulk Richardson number of each model layer with respect to the surface is given as:
* Z"(0"To)
00(UkH)2
(*/,)*= g""To (12-50)
B) * k2
where subscript H represents values at the layer middle (i.e., half sigma level).
First, the index of the PBL top (kPBI) is determined at the layer when RiB first becomes larger than
max( RiB). Then the proration factor of the bulk Richardson number relative to max( RiB) is
computed with:
( ma\(RiB)-(RiB)k"L~t\
f«.="H1' (*/,)*«-(*//)*«-r
12-33
-------
EPA/600/R-99/030
wnere max(/JIB)=0.7 (refer to Table 12-6). Depending on the value of fKt, the index kPBL and
fg,t are modified as follows:
for fm> 0.5, f'Ria = fMi -0.5; UPBL = kPBL, and
I •" * : "'M, •*
for /«, < °'5> /», = /«. + °-5 ' kPBL = kPBL ~ 1'
Once the fraction and index for the PEL top are determined, we estimate the initial PEL height
with:
The above proration procedure ensures gradual increase of PEL height. Without the procedure,
the resolution of PEL height is limited by the layer thickness of the model. In the RADM
preprocessor, the PEL height is determined simply at the layer where the potential temperature first
becomes warmer than the surface temperature for convective conditions. Compared with the
RADM method, the present method takes into account effects of the wind shear as well.
i i, *" •/ * , '«*» ' is "I
Stable conditions
For stable conditions, the PEL height is determined by the maximum of the PEL height computed
with above method and the stable boundary layer height given by the Zilitinkevich's (1989)
formula:
h -1
hosr-"2
^PSIT
(12-53)
Limiting PEL heights
Unlike the PEL height estimation algorithms based on temporal integration of surface heat flux
(e.g., Carson, 1973; Betts, 1973; Driedonks, 1981), the above diagnostic algorithm could predict
temporally disconnected PEL heights when the hourly meteorological data change abruptly. To
minimize this effect, following limits on the PEL height are imposed.
1, Compare with PEL height for neutral conditions, and take maximum except for the tropical
areas:
I itiPS_L=m2x{hPBL, A0% (12-54)
v • 15 - ,1" j
2. Compare with the urban boundary layer height, which is approximated with:
"UBL ~ 0 ~ /aria* J^PBLmin ~*~ Jurban"UBLmin ( * 2-55)
12-34
-------
EPA;600/R-99/030
where furban is the fraction of urban landuse in a cell. In MCIP, the minimum PEL height
for urban area and other landuse types are set to be hUBLmin=3QQ m and hPBLltlin=50 m.
Then, take a maximum of the two to reflect the effect of urban landuse:
H PBL= ma\{hPBL, hUBL} (12-56)
This step is introduced to apply the urban landuse classification, which is consistent with
the one used for the emission processing, for the determination of the PEL height.
3. Limit the PEL height with a maximum value (hPBLiruu=3QQQ m) in case the temperature
profile does not have a capping inversion:
HPBL=mn{hPBL, hPB[MUU}. (12-57)
12.3.1.5 Aerodynamic Resistances
Aerodynamic resistance describes the ability of the atmospheric turbulence to transport pollutant to
the surface for the deposition. In this regard, it may well be described with the probability
concept. If 100 particles, say, are at height 20m, how many of them can reach the surface to be
available for deposition during a given time interval? Or, what is the transport rate of particles in
the air to the surface? The ratio represents the maximum potential deposition rate of particles
subjected to atmospheric turbulence. However, even those particles which arrive at the surface
may not all be deposited because of other resistances, which usually are parameterized using
characteristics of surface and gaseous elements.
There have been many efforts to compare different formulations for dry deposition velocities.
While there are a lot of uncertainties in describing atmospheric processes, the different
aerodynamic resistance formulas in the literature are mainly originated from the differences in the
applications and approximations of the same PEL theories and formulas. Among the components
involved in atmospheric resistance computation, formulations for aerodynamic resistance have the
least controversy. To compute the aerodynamic resistance, the parameterization of the eddy
diffusivity for the PEL should be known. The eddy-diffusivity formulations used in the derivation
of aerodynamic resistance are discussed below.
Eddy diffusivity formulations
The similarity theory suggests that eddy diffusivity in the surface layer for heat flux is given by:
K, = kUlZ '• (12-58)
The profile function 0A is defined in Equations 12-12b and 12-13b. Eddy diffusivity in the PEL
(above the surface layer) (Brost and Wyngaard 1978) is given as:
12-35
-------
EPA/600/R-99/030
\3/2
-— (12-59a)
for stable atmosphere (z/ L > 0), and as:
^ = few.z(l - -r5-) (12-59b)
"•PBL
for unstable atmosphere (z/L < 0), where the convective velocity scale wt is defined as
it/3
(12-60)
Aerodynamic resistance formulations
We often characterize the ability of turbulent atmosphere to carry pollutants to the vegetation or
other surface elements using the aerodynamic resistance concept. Certain meteorological models
may use the resistance representation of surface exchange processes in lieu of the conventional
bulk aerodynamic methods and the aerodynamic resistance is available as a part of meteorological
data with other PBL parameters. Conversely, there are models that do not predict resistances and
related parameters needed. In such cases, we need to estimate it using PBL parameters and
profiles of state variables.
The aerodynamic resistance (Ra) and sub-layer resistance (Rh) are parameterized in terms of
friction velocity and surface roughness (Walcek, 1987; Chang et al., 1987 and 1990; Wesely,
1989; Wesely and Hicks, 1977). Although some consider the estimation of aerodynamic
resistance as an integral part of the dry deposition velocity computation, we treat the aerodynamic
resistance as an independent element that characterizes the effects of atmospheric turbulence. Here,
we provide a set of integrated equations, which allow robust estimation of aerodynamic resistance
compared with the method based on the nonintegrated form, which sometimes provides negative
fluxes for unstable conditions. Refer to Byun and Dennis (1995) for details.
General formulation for the aerodynamic resistance at the deposition height zjep (where
concentration is represented) is divided into two components; the resistance in the surface layer
(whose top is at ZSL, height of the surface layer) and the resistance in the PBL above the surface
layer;
S* ...» , '• SL . . *j
"" "" dz f dz /? dz „ . „
o _ f a* _ f__f£. f
a~iKM'JF^) J
12-36
-------
EPA/600/R-99/030
(a) For stable conditions:
ku.
ku,
7 7 -7 1
/ **SL \ i f3 ^"SL **o
(-1+1-
(12-62a)
(12-62b)
:£fiL. ry, =^^and775L=^
L ^" hn, SL h,
(b) For unstable conditions:
Pr
ku.
In
-In
(12-63a)
(12-63b)
12.3.2
Dry Deposition Velocities
The term dry deposition represents a complex sequence of atmospheric phenomena resulting in the
removal of pollutants from the atmosphere to the surface of the earth. The rate of transfer of
pollutants between the air and exposed surfaces is controlled by a wide range of chemical,
physical, and biological factors, which vary in their relative importance according to the nature and
state of the surface, characteristics of the pollutant, and the strength of turbulence in the
atmosphere. The complexity of the individual processes involved and the variety of possible
interactions among them prohibits simple generalization of the process. Nevertheless, a
"deposition velocity," analogous to a gravitational falling speed is of considerable use. In practice,
the knowledge of dry deposition velocities enables fluxes to be estimated from airborne
concentrations. Two dry deposition estimation methods, that from Wesely (1989) currently as
implemented in CMAQ, and the new Models-3/CMAQ approach are presented.
12.3.2.1
RADM Method
The RADM dry deposition module in MCIP calculates deposition velocities of sixteen chemical
species (Table 12-7). It requires various ancillary 2-D meteorology fields such as the PEL height,
mixing scale velocities, Monin-Obukov length, etc. They are usually estimated from horizontal
wind components, temperature and humidity profiles. The dry deposition flux of each chemical
species from the atmosphere to surface is calculated in the CCTM by multiplying the concentration
in the lowest model layer with the dry deposition velocity. The dry deposition velocity (Vd) is
computed from the resistance-in-series method;
12-37
-------
EPA/6(K¥R-99/030
where Rtt is the aerodynamic resistance, Rh is the quasi-laminar boundary layer resistance, and Rc
is the canopy (surface) resistance. Refer to Figure 12-7. y is usually estimated from a series of
resistances to vertical transfer and surface uptake. Aerodynamic resistance (Ra ) is a function of
turbulent transfer in the atmospheric surface layer and can be estimated in several ways depending
on the instrumentation used or parameterizations provided.
K- ' i«'. sa • ;. ;•• • • ; : \ ' . . i; -. „ - ,: r • ,. *-, |. . ;.•
The canopy resistances (Rc) for SOz and Oj are estimated from available measurements as a
function of season, insolation, surface wetness, and land type (Walcek et al., 1986; Shieh et al.,
1979; Fowler, 1978). The surface resistances for all other gaseous species, due to the lack of
extensive measurements, are qualitatively scaled to the SO2 and <9j surface resistances according to
their reactivity and solubility. The surface resistance for particulate sulfate is parameterized in
terms of stability and friction velocity in the surface layer (Wesely, 1989), based on limited studies
that do not include water surfaces. In CMAQ this method only applies to the treatment of surrogate
'lf'n'/f ',, ,•!' ''' ,"!' ''lifflilffl:1 ' ''"'!'!'!' i r'1 " '"in1! " ..... ,' V ', '!"... • ' n ' « ••, „» ', V w n, IT ,,,« ....... , , ...... , ... , • iio ,i <•»
gas-phase representation of sulfate species. Size dependent deposition velocities for particle
depositions are estimated inside the CCTM's aerosol module. See Chapter 10 of this document
and Binkwoski and Shankar (1995) for details.
The laminar sub-layer resistance (Rh) depends on the landuse specific friction velocity and
molecular characteristics of gases. Using the landuse dependent friction velocity or the cell average
friction velocity, Rb's for heat (Rbk) and trace gases (Rhl) are estimated with the Schmidt number
(Sc):
D - ~Sc2n (for heat) ( 1 2-65a)
ku*
-' 2
Rbx=- — Scx2n (for trace gas species) (12-65b)
where Schmidt number is defined as the kinematic viscosity of air (v = 0. 146 cn^s'1 ) divided by
molecular diffusivity ( i.e., Sc = v/Dg for heat and Sc^ = v/Dgx for trace gases). Here, v is
kinematic viscosity of air; Dg is molecular diffusivity of air (heat); and Dgx is molecular diffusivity
of trace species. For heat, the molecular thermal diffusivity ( £>g) is 0.206 cm2s~1 ; for water
vapor, molecular diffusivity ( D^) is 0.244 cm2s~l ; and, for ozone molecular diffusivity ( D,,03) is
0.159 ewiV"1 (molecular diffusivities of other chemical species are included in the model).
12-38
-------
EPA/600/R-99/030
Ambient Concentration
ft
ft
lower canopy ground, water, cuticle, leaf
rj snow stem tissue
Vegetation
Figure 12-7. Schematic Diagram of Pathway Resistances Used in RADM Dry Deposition Model
Resistances with subscript x are for different chemical species.
Rc represents the canopy resistance defined as:
1 1 ]
R,
(12-66)
where ra = stomatal resistance;
rTO = mesophyl resistance;
rto= resistance of the outer surface of leaves in the upper canopy;
rdc = resistance for the gas transfer affected by buoyant convection in canopy;
rclx = lower canopy resistance (uptake pathways at the leaves, twig, etc.);
rac = resistance that depends on the canopy height; and
r = resistance of soil, leaf litter, and other ground materials.
Stomatal resistance for water is obtained using following equation:
200
400
where
0.1 + 0,
— stomatal resistance for water;
in= minimum stomatal resistance for water, specified in Table 12-7;
(12-67)
12-39
-------
EPA/600/R-99/030
Te = surface air temperature in the canopy in Celsius temperature; and
Gnv =- solar radiation reaching at the canopy in W/m2 unit,
Then stomatal resistance for trace gas species is obtained with:
(12-68)
where Dgw is the molecular diffusivity of water. Ratio of molecular diffusivity of water to that of
each trace species is provided in Table 12-7.
The mesophyl resistance is parameterized as:
V " ?«. "I
v-1
where Ht]e is Henry's gas constant (in mole atm"')for the species, andfnxis the reaction intensity
factor,
e>." - • •• •' i«t \ i , • •;• 5;^ , • • , ,
Both the Henry's gas constants and reaction intensity factors are provided in Table 12-7 as well.
r^ is related with the upper canopy resistance for water ( rlu) which is provided in Table 2 of
Wesely(1989)as:
(12-70)
is parameterized in terms of the available solar radiation and the slope of local terrain:
Kl + 10000)"' (12-71)
where 6 is the slope of local terrain in radians. rckwd rg!a are estimated based on the respective
resistance values for SO7 and O, as follows:
i ........ • ..... • ............. .-./ • • 2 .... •
'( H f T1
Si-+ (12-72a)
'f FT f V1
— *«• -i Jox
where subscript S and O are for SO2 and O5, respectively. All of these values, and ruc (the
resistance that depends on the canopy height) are landuse dependent and listed in Table 2 of
Wesely(1989).
12-40
-------
EPA/600/R-99/030
Table 12-7, Gaseous Species Treated in RADM Dry Deposition Module and Their Properties
Relevant to Estimating Resistance Components as Implemented in MCIP
(Modified from Wesely, 1989.)
Gaseous species
Sulfur dioxide
Sulfate
Nitrogen dioxide
Nitric oxide
Ozone
Nitric acid vapor
Hydrogen peroxide
Acetaldehyde
Formaldehyde
Methyl hydroperoxide
Peroxyacetic acid
Formic acid
Ammonia
Peroxyacetyl nitrate
Nitrous acid
Carbon monoxide
Symbol
SO2
SO4
NO2
NO
03
HNOj
H202
ALD
HCHO
OP
PAA
ORA
NH3
PAN
MONO
CO
DsJDgx
1.9
-
1.6
1.3
1.6
1.9
1.4
1.6
1.3
1.6
2.0
1,6
1.0
2.6
1.6
1.2
/jr.,, (mole atm"')
1.0X10*
-
0.01
0.002
0.01
1.0X10"
1.0X105
15.0
6.0X1 0*
240.0
540.0
4.0X10ft
2.0X10"
3.6
1.0X105
0.001
/«
0.0
-
0.1
0.0
1.0
0.0
1.0
0.0
0.0
0.1
0.1
0.0
0.0
0.1
0.1
0.0
12.3.2.2 ModeIs-3/CMAQ Dry Deposition Model
The Models-3/CMAQ dry deposition (M3DDEP) module estimates dry deposition velocities
according to the same electrical resistance analog represented by Equation 12-64. M3DDEP uses
common components with the new land-surface model that has recently been added to MM5.
Specifically, the aerodynamic resistance and the canopy or bulk stomatal resistance are the same as
those used in the modified MM5 (MM5PX) for computing evapotranspiration. Since the land
surface scheme includes soil moisture and has an indirect nudging scheme for improving soil
moisture estimates, the resulting stomatal resistance estimates should be better than those achieved
with a stand alone dry deposition model. Pleim and Xiu (1995) give a description of an early
prototype of the land surface model which is now coupled to MM5. Pleim et al (1996) and Pleim
et al. (1997) briefly describe the dry deposition model and some studies comparing model results
to field measurements of surface fluxes and PEL heights. The description which follows here is
partially drawn from these sources. .
When using the M3DDEP option in the CMAQ system the aerodynamic resistance, as well as the
bulk stomatal resistance (rslbw; discussed below) is provided to MCIP from the MM5PX. Note that
these parameters are computed as part of the land surface model in the MM5PX and are not
available from the standard MM5. In MM5PX the aerodynamic resistance /?a is computed
assuming similarity with heat flux such that:
12-41
-------
EPA/60Q/R-99/030
Ra=pCp(0g-ei)/H-Rbh (12-73)
where Ry, is the quasi-laminar resistance for heat (defined in Equation 12-65a), 0g and 0, are
potential temperature of the ground surface and the air, respectively, in the lowest model layer, and
H is the sensible heat flux defined in Equation 12-25. If the sensible heat flux is very small (as
during transition periods) the surface layer theory for neutral conditions is used as follows:
R _ ISqJ *L I for \H/(p C\\ < i0-15 [K m s'1] (12-74)
*«. UJ
Where z, is the height of model layer 1 and za is the roughness length. The heat and momentum
fluxes are derived from flux-profile relationships as in the MM5 (see Grell et al., 1995). The quasi-
laminar boundary layer resistance accounts for diffusive transfer across a thin laminar layer
adjacent to surfaces. Because of the no-slip condition, turbulent eddies cannot penetrate to a
surface. Therefore, there exists a thin layer of non-turbulent air where molecular diffusion is the
primary mechanism for transfer. While this concept is not relevant for momentum, it is relevant
for any quantity that directly interacts with the surface such as heat, moisture, and chemical
deposition. Therefore, for these quantities, the addition of a resistance based on molecular
diffusion is necessary. Deposition layer resistance varies by the transported quantity because of
differences in molecular diffusivity, which is defined earlier in Equations 12-65a and b.
L ~\- i : ' . . . . . ..
The total surface resistance to dry deposition (Rs) has several components including bulk stomatal
resistance (r^), dry cuticle resistance (rcw), wet cuticle resistance (rw), ground resistance (rg), and
in-canopy aerodynamic resistance
R. =
& i r A ri fvQ ~ f*\ . fvfw I i
* "r JLinuC I " i — "1
1
r
\ rcut rw rg rle+rl,
(12-75)
Bulk stomatal resistance (r^), vegetation fractional coverage (fv), leaf area index (LAI) and
fractional leaf area wetness (fw) are all provided by MM5PX since the same parameters are used in
the land surface scheme. Figure 12-8 shows a schematic representation of Equation 12-75. Note
that r,^, as output from MM5PX, is already a combination of stomatal resistance on a leaf area
basis ra, mesophyl resistance rm, and LAI as described below. A key component of the land
surface model in the MM5PX is the parameterization of the bulk stomatal resistance that is used to
compute evapotranspiration. The bulk stomatal resistance for water vapor is read in to MCIP from
MM5PX and adjusted for chemical dry deposition in the M3DDEP model by weighting with the
ratio of molecular diffusivities for water vapor and the chemical species:
thw. (12-76)
F • • tir**,
12-42
-------
EPA/600/R-99/030
Ambient Concentration
Leaf
ground, water, ground, water,
snow snow
cuticle, leaf
stem tissue
Figure 12-8. Schematic Diagram of Pathway Resistances Used in Models-3/CMAQ Dry
Deposition Model.
In general, the bulk stomatal resistance is related to leaf based stomatal resistance as:
r* = -Tfe + rj
LAI
(12-77)
where Ps is a shelter factor to account for shading in denser canopies. For water vapor and many
chemical species, such as O3 and SO2» rm is assumed to be zero, however, for many less soluble
12-43
-------
EPA/6GO/R-99/030
species a non-zero value should be used. The shelter factor is given by: Ps = 0.3 LAI +0.7 with a
minimum of Ps = 1.0. Leaf scale stomatal resistance, computed as:
' ' ' '
_ _
depends on four functions of environmental factors which influence stomatal function, and the
minimum stomatal resistance (rstmin) which depends on vegetative species. The minimum stomatal
resistance is a bulk parameter which reflects the maximum conductance of a leaf per unit area under
unstressed conditions (well watered, full sunlight, and optimal temperature and humidity). This
parameter is specified in the model according to vegetation type. The key to the model's ability to
shnujate stomatal conductance in real world conditions is the four environmental stress functions,
FM in Equation 12-78. This kind of stomatal model, with independent empirical stress functions,
is often called a Jarvis-type model after Jarvis (1976). Specifically, the land surface model in
MM5PX is based on Noilhan and Planton (1989) (hereafter referred to as NP89) with many
subsequent modifications. The radiation stress function is:
"'•'••' " ' J ' ' j ' ,
(12-79)
J "•" 'rftnln ' 'jimix
<} jp
with / = 0.55— — where rstmax is maximum stomatal resistance which is an arbitrarily large
, .. ....
number (5000 s/m), RG is solar radiation at the surface and the 0.55 factor is an approximation for
the photosynthetically active portion, and RCL is a limit value of 30 W/m2 for forest and 100 W/m2
for crops according to NP89. The only difference from the F/ in NP89 is that the dependence on
LAI has been removed since the effects of leaf shading within the canopy are now accounted for by
the shelter factor in Equation 12-77. Therefore, the F/ as defined here represents the effects of
sunlight on an individual leaf rather than the integrated effect on a canopy. The functions of root
zone soil moisture and air temperature (Fa and F4) were modified to follow the form of logistic
curves as suggested by Avissar e'faJ. (1985). Logistic curves are "S"-shaped and therefore good
for representing a smooth transition from one state to another. Also, logistic curves can be defined
\yltri. varying degrees of abruptness, from an almost linear transition to an almost threshold
behavior, and can be altered while maintaining differentiability. The function of root zone soil
moisture (w2) is:
2 = 1 /(l
(l + exp[-5.0(wv+&w)j) (12-80)
where the available soil moisture fraction is:
w = w*~w™i>
w. — w .
'Yfc "wll
and the half point of the function (where F2 = 0.5) is:
* . ^ ;:! v * •• • • '• • . .;••••. •• - :
12-44
«$ r y
-------
EPA/600/R-99/030
where wwi, is the wilting point and w/cis the field capacity. All soil moisture values (H>) are in
volumetric fraction. In many previous models, including NP89, Jacquemin and Noilhan (1990),
Wetzel and Chang (1988), Mihailocic et al. (1993), Sellers et al. (1986), and Avissar et al (1985),
the function of air humidity (F3) is expressed in terms of vapor pressure deficit between the inside
of the leaf, assumed to be saturated at leaf temperature, and ambient air humidity (vpd= es(Ts)-ea),
However, recent advances in plant physiology research have led to a new generation of stomatal
models based on leaf photosynthesis (Sellers et al., 1997) in which stomatal conductance (g n =
l/rst) is a linear function of relative humidity at the leaf surface (Collate et al., 1991):
where g'st is the stomatal conductance at RHS = 1 and b is the minimum stomatal conductance at
RHS = 0. Clearly, it makes more sense that stomata react to the humidity at the surface of the leaf
rather than the ambient air humidity at some height above the canopy. Although, a physical
mechanism for this linear relationship to leaf surface RH has not been determined, experimental
data shows it to be a very good fit (Ball et al,, 1987). Since leaf surface relative humidity is neither
an easily measured nor a modeled quantity, it must be computed from other parameters. According
to the electrical analog, the humidity at the leaf surface (
-------
EPA/600/R-99/030
Ambient Humidity
— "ya
Leaf Surface
Humidity
Qleaf
Leaf Surface
= 1/g
st
.-I" -a
Figure 12-9. Derivation of Leaf Surface Relative Humidity
The fourth environmental stress function (F4) is related to ambient temperature. Again, we deviate
from the NP89 formulation which used a quadratic function peaking at the optimal temperature of
298 K. Instead, we followed the method of Avissar et al. (1985) which results in a function with a
plateau over a range of optimal temperatures, the idea being that temperature (7^,) inhibits stomatal
function only at extremes of heat or cold. F4 is defined as:
(12-83)
where «y = -0.41 and bT= 282.05 K for Ta £ 302.15 K and aT =1.18 and bj=314 K for Ta >
302.15 ft. Note that the high side of the function extends into higher temperatures than suggested
IIL ,,i,rf :r T1H, » . r IE 'V "'"• 'Lr ° , » , <~J * . . ., , *-"-*
by Avissar et al. (1985) who used a 67= 307.95. The current function is very similar to the
function used by Rochette et al. (1991) and is close to the high side of the NP89 F4 function.
a* " - * -^ ^ try - •. - mmr ~- - SM > . *• • » . > "• » > • ; • _ j • •. n.: • :
•fcj" ' i. ^ *• • «•*. * r\3im I™ j"t ' (j ^*-^ . - fi «. »*J • „ .., . • :, •,. ,* -• ~. - . '*..-,. .„ «myj" • . «•• M; ' v
Resistances fe affof the non-stomatal dry deposition pathways in the M3DDEP module of MCIP
are similar to the dry deposition model developed for the Acid Deposition and Oxidant Model
12-46
-------
EPA/600/R-99/Q3.0
(ADOM) by Pleim et al. (1984) and later evaluated and modified by Padro et al. (1991). As in -
ADOM, surface resistances to ground and leaf cuticles are scaled relative to the most well measured
chemical species such as SO2 and Or The term used for the scaling factor: relative reactivity has
caused considerable confusion since many users try to find a chemical definition for this parameter.
This factor has no chemical definition but is simply meant to be a relative scaling factor for the
removal rate of different species at the ground or cuticle surfaces. It is assumed that the relative
propensity to deposit to different surfaces is similar, so that the same scaling factor can be used for
ground and cuticles:
(12-84a)
(12-84b)
where rgo and rCMO are the ground and cuticle resistances, respectively, for the reference chemical
species (usually SO2 or O3) and Ag and A are the relative reactivities for the reference species and
the modeled species, respectively.
One improvement from the original ADOM model is the inclusion of an in-canopy aerodynamic
resistance which is added in series to the ground resistance for the vegetated portion of the modeled
area (fv) according to Erissman et al. (1994):
rlc=14LA!hc/u. (12-85)
where hc is the height of the canopy. For wet or partially wet canopies (fw), the surface resistance
to wet cuticle is estimated as:
rc» = rcwaKH/a* (12-86)
where KH is the nondimensional Henry's law constant and a* is an aqueous dissociation factpr.
The empirical factor (rcwo) is set to 2.4xl08 s m"1, which is similar to Wesely (1989). For Oj, rcw
is set to 1250 s m"1 based on field measurements. ThepH of the canopy water can be specified to
compute a* for species such as SO2 and NHj which readily dissociate. The canopy wetness
fraction (fw) is provided by the surface model in MM5PX. For deposition to open water, the
ground resistance is replaced by a water surface resistance according to Slinn et al. (1978):
rgw = KH /(4.8e-4 a* «*) (12-87)
The parameters needed for the non-stomatal part of this model are quite uncertain for many
chemical species. Our latest estimation of these parameters for the species in the RADM2 chemical
mechanism for which dry deposition is considered to be an important process are presented in
Table 12-8. The Henry's law coefficients in Table 12-8 are in non-dimensional form as the ratio of
gas- to aqueous-phase.
Molecular diffusivity and Henry's law have the least amount of uncertainty since these have
definite chemical definitions and can be experimentally determined. Relative reactivity, however,
12-47
-------
EPA/600/R-99/030
P": ! ";" '?' "I - " ' : : : '
is less well defined and is estimated according to experimental and modeling studies in the
published literature . The only organic species for which dry deposition is considered significant
are the aldehydes, peroxides, and acids, all of which are soluble in water. For these organic
lumped species the numbers for reactivity are merely educated guesses.
Table 12-8. Chemical Dependent Parameters for M3DDEP
j < ' ..• , ."..., , ,
bpectes
a
(cm2/s) (dimensionless)
Reactivity
(dimensionless)
Henry's law* ( KH)
(dimensionless)
SO2
N02
Oj
ill"""! '"• IL
1 ' " :
-------
EPA/600/R-99/030
been shown experimentally to have a strong stomatal pathway components. Several studies with
this model have demonstrated its ability to realistically simulate both latent heat flux, with a large
fraction from evapotranspiration, and ozone dry deposition, over corn and soybeans (Pleim et al.,
1996; Pleim et al., 1997). These studies are continuing and are being extended to other chemical
species (SO2), and land-use types (deciduous and coniferous forests).
12.3.3 Cloud Parameters and Solar Radiation
Cloud information is developed and used in many different ways throughout the Models-3 CMAQ
system. For example, MM5 includes parameterizations related with subgrid convective clouds,
grid resolved cloud water and microphysics, and cloud effects on radiation. The CCTM needs
cloud information for photolysis, convective transport, and aqueous chemistry. Therefore, MCIP
has an important role in providing cloud information to CTMs by either propagating information
from upstream processors (MM5) or by parameterization.
MCIP has multiple cloud parameterizations and functions depending on the options selected.
There are currently two main options related to MM5 runs, either the standard MM5 version 2.6
which does not include cloud cover and radiation parameters in its output, and MM5PX which
does. Later versions of MM5 (2.7,2.8, and 2.9) also output these additional parameters.
However, when MM5v2.6 or earlier versions are used, it is necessary to compute these parameters
in MCIP. Incident solar radiation at the surface is an important output from MCIP since it is used
to estimate biogenic emissions in the emissions processor.
Another cloud function in MCIP is to diagnose cloud information (such as cloud top, cloud base,
liquid water content, and cloud coverage) which is passed to the CCTM to adjust actinic fluxes for
computation of photolysis rates. This function is executed regardless of the choice of the MM5
version used. In addition, grid resolved cloud water and rain water are propagated through MCIP
from the MM5 output to CMAQ for use in aqueous chemistry and for photolysis calculations at the
4 km grid resolution.
12.3.3.1 Cloud Coverage
The fractional cloud coverage scheme currently used in MCIP is the same as used in MM5PX and
similar to the scheme used in the standard MM5. Cloud cover fraction (f*) above the boundary
layer is computed at each vertical model level k, according to Geleyn et al. (1982), as:
RH" -
l-RHc
where RHC is the critical relative humidity defined as a function of a as:
RHC=\-2
-------
EPA/6QO/R-99/030
We have modified the Geleyn scheme to avoid the overprediction of clouds in well-mixed
boundary layers. Within the convective boundary layer (CBL) when RH > RHC the cloud cover is
estimate by:
l (12-89)
l-RH
c
where the critical relative humidity (RHC) within the CBL is set to be 0.98. The factor 0.34 is from
the suggestion that convective mixing induced clouds within the CBL should not exceed the
fractional areajpf the updrafts at top of CBL, which large eddy simulations estimate to be about
34% (Schumann, 1989; Wyngaard and Brost, 1984) when inactive clouds are disregarded. The
resulting layered cloud fractions, from Equations 12-88 and 12-89, are used for both functions;
namely, to be used in the surface radiation calculation when these parameters are not read from
MM5, and to derive the cloud parameters needed for the photolysis calculations.
' ' .1 . ;I , •. . < • , '•';"';.:- • - •'
For the surface radiation calculation, the layered cloud cover fractions are aggregated into the same
three broad vertical cloud layers (low, middle, and high) as in MM5, assuming maximum overlap.
Each layer is defined by pressure such that the low layer is between 97 and 80 kPa, the middle
layer is between 80 and 45 kPa and the high layer is above 45 kPa. These three cloud layer
fractions are then used in the radiation calculation described in this section (12.3.3.2).
The photolysis model requires cloud information in a different form. For the sake of consistency,
these parameters are estimated from the same layer resolved fractional coverage described above
(Equations 12-88 and 12-89). The photolysis model assumes a single uniform, vertically mixed
cloud layer. Therefore, the required parameters are; cloud top, cloud base, cloud fractional
coverage, and average liquid water content. Cloud top and base layers ( k and kb ) are
f?» • -.;-, -KtJ •..« . . i ... ,i i - • . ' . ,. ;~ . »«, '„ ,
determined by looking up and down from level of maximum coverage to where the fractional
coverage first becomes 50% of the maximum. The layer average cloud cover is then computed
from the volume averaging of layer cloud fractions between klgp and khase :
Once the base, top, and fraction of the cloud layer are determined, the average liquid water content
of the cloud is computed assuming convective characteristics as described by Walcek and Taylor
(1986), Chang et al. (1987), and Chang et al. (1990). The lifted condensation level is assumed to
be at the cloud base, defined as the bottom boundary of cloud layer £tee , where the air is saturated
at the model's ambient temperature. The in-cloud liquid water profile (qc) is then computed as a
fraction of the adiabatic liquid water profile (qj):
(12-91)
12-50
-------
EPA/60CMR-99/030
where a = 0.7exp[(/? - p/c/)/8J + 0.2 and pld is the pressure (in KPa) at the lifting condensation
level according to Warner (1970), The layer liquid cloud water values are then vertically averaged
in the same way as fractional coverage (Equation 12-90).
12.3.3.2 Computation of Solar Radiation Components
To meet air quality modeling needs, MCIP outputs several parameters related with radiation at the
surface including: incident surface shortwave radiation (/?smrf), absorbed surface shortwave
radiation (G^), net long wave radiation at the ground (G/Hj» total net radiation at the ground (Rna),
and surface albedo (A). These radiation components are used in several CMAQ processors. For
example, photosynthetically active radiation (PAR), which is needed for the biogenic emissions
processing, is estimated with PAR = 0.55Rgm]. Depending on which version of MM5 is used,
surface radiation parameters are either passed through from MM5 or computed in MCIP. The
computation of surface radiation parameters in MCIP is identical to the surface radiation option in
MM5, as described in Grell et al. (1995), except that the dependence on zenith angle is added to the
land use specified albedo (Alu) such that:
A = Alu + 0. l[exp(0.00386Z3'2) -1 .OJ (12-92)
where Z is zenith angle in degrees.
Note that MCIP computations of surface radiation parameters, which essentially replicate MM5,
are a stop gap measure for use with MM5v2.6 which does not output these values. All later
versions of MM5 do output these parameters which will be read and passed through by MCIP as is
now done for the MM5PX option. It will be preferable in future applications of Models-3 to
directly propagate the radiation parameters from MM5 to CMAQ so that the more sophisticated
radiation models now available in MM5 can be used. The currently used surface radiation option
includes effects of clouds, aerosols, and water vapor on radiation but neglects radiation effects on
the atmosphere. Therefore, many future applications may utilize either the Dudhia radiation
scheme (Dudhia, 1989) or the CCM2 scheme (Hack et al., 1993) which include atmospheric
radiation effects. See Chapter 3, section 3.3.4.1, for description of MM5 radiation options.
12-51
-------
EPA/600/R-99/030
12,4 Meteorological Data for CCTM with Generalized Coordinate System
One feature of the CMAQ systemfthat is distinct from other Eulerian air quality modeling systems
is its ability to incorporate meteorological models with various different coordinates and dynamics.
This functionality is achieved by recasting meteorological parameters in terms of the variables used
in the governing set of equations for the fully compressible atmosphere in generalized coordinate
system (Byun, 1999a). Key dynamic and thermodynamic parameters are estimated for the given
coordinates in such a way to ensure consistencies among the meteorological data. The consistency
can be maintained throughout CCTM simulations when appropriate temporal interpolation methods
are used.
12.4.1 Thermodynamic Variables: Pressure, Density and Entropy
To facilitate mass-consistent interpolation among the thermodynamic variables, the governing set
of equations for the fully compressible atmosphere is used in the CMAQ system. The system
includes prognostic equations for entropy and air density as suggested by Ooyama (1990). The
algorithms used for estimating the thermodynamic parameters are presented below. For a detailed
discussion on the governing set of equations and mass-consistent interpolation algorithm, refer to
Byun (1999a and b).
12.4.1.1 Pressure, Density of Air and Density of Water Vapor
Like most of the other meteorological models, MM5 does not use a predictive equation for air
density. Instead, air density is estimated with the equation of state using predicted pressure and
temperature. For a terrain-influenced pressure coordinate, when it is applied for hydrostatic
atmosphere, the hydrostatic pressure (p) can be computed with p" available from MM5.
p = p = Pf + Cf~p (12-93)
The total pressure (p) in the terrain-influenced reference pressure coordinate (applied for a
rionhydrostatic atmosphere in MM5) is the sum of the reference pressure and perturbation pressure:
*_ . :p—p'+"p =O»*4- n +n •'.... . (12-94)
Once/? is known, virtual temperature (Tv) and density of the (moist) air are computed with:
Tsa -I (12.95)
:. ;, mj p
•" • •• •> •
where mw and md are the molecular weights of water vapor and dry air, respectively , and
1 ' J :? - r , " :-vt! ;., ?.
-------
EPA/600/R-99/030
The water vapor partial pressure (/?„) can be found from the vapor mixing ratio (r) supplied by
MM5 using the following equation:
pv=r^-pd. (12-97)
Then, the density of water vapor is simply given as:
p =-^—p. (12-98)
1 + r
12.4.1.2 Entropy
In the CMAQ system, entropy is treated as one of the key thermodynamic parameters. Because its
conservation equation follows similar continuity equation for air density, entropy can be
interpolated using the same interpolation scheme as air density. Then, we can reconstruct
temperature and pressure for the intermediate time steps from the interpolated densities (for air and
water) and entropy. This type of interpolation will maintain mass consistency among the
thermodynamic variables for air quality applications. Using the density of moist air and density of
water vapor obtained above, the entropy for moist air can be computed with:
£ = (PA + Pvcw) ln(^~) + PA, - PA ln(—) - PA ln(—), (12-99)
oo rdoo rv*a
where pd = p — pv, p00=105 Pascal, A00 is the specific entropy of saturated vapor, pd is the
density of reference dry air, andpv»0 is the density of reference state water vapor saturated over
water at reference temperature Teo, 273.15 K. The integral constants are computed following
Ooyama (1990):
(12-100)
d\nE(Tc)
dT
c
(12-101)
rvo rv" OB' RJ
where T=T-273.15.
-^^ , (12-102)
c
For the basic formulation of the saturation vapor pressure approximation, we use the AERK
formula recommended by Alduchov and Eskridge (1996):
12-53
-------
EPA/600/R-99/030
(12-103a)
dlnE(fc) be
.dT
c
(12-103b)
where a-610.94, ^=17.625, e=243,04 and£(Fc) is in Pascal. With this information we can
evaluate the integration constants, and thus the entropy using Equation 12-99.
12.4.2 Vertical Jacobian and Layer Height
In the CCTM, Jacobian is used for the definition of vertical and horizontal coordinates/grid system
at gvery model synchronization time step. The Jacobian characterizes the coordinate transformation
and is treated as one of the fundamental parameters defining grid structure of the CTM. Depending
onTtfie horizontal map projections and vertical coordinates, the physical characteristics of the
Jacobian change.
12.4.2.1 Jacobian for Coordinate Transformation
The vertical Jacobian defines the coordinate transformation rules. For a vertical coordinate x* = %,
where £ is a monotonically increasing function of height, the Jacobian is related with the
gebpotential height,
-------
EPA/600/R-99/030
where K is the number of model layer (i.e., NLAYS in Models-3 I/O API), and Jacobian at full
sigma-level is computed for 1 < k
-------
EPA/600/R-99/030
The Jacobian does not depend on MM5 data except for the topographic height and sigma-layer
definition, "
Total Jacobian
The total Jacobian («Jf = J. /m2) is used in the governing equations for the CCTM. The trace
species concentrations are coupled with the total Jacobian. Because of the need to couple with both
the trace species concentrations (defined at the layer middle) and vertical flux variables (defined at
the layer interface), total Jacobians at the half and full level values are stored in the MET_CRO_3D
file. The total Jacobian at the surface, (Ja j divided by the map scale factor squared, is stored
'•-- -'»-.»..««.,«,.•, , \ p / jr
in the MET_CRO_2D file. For hydrostatic applications of MM5 the Jacobian varies with time and
for nonhydrostatic applications it is constant with time. Although we could have saved some file
space by distinguishing this feature, the Jacobians are stored in MET_CRO_3D and
MET_CRO_2D files to maintain the compatibility of data structure for both time-dependent and
time-independent coordinate systems.
»' =• ;-i" . ,1* '•••»• ',•',- » < j ,
For use in the mass-conserving temporal interpolation, Jacobian weighted densities at the half-
levels defined for i ^ fc <]f as:
W '< ' •• :•(). "i:! . t •
(12-112a)
(12-112b)
are stored in the MET_CRO_3D file. Like density, the Jacobian weighted entropy at the layer
middle is obtained for 1 < k £K with:
* " ***"•«„*•-' .
They are stored in the MET_CRO_3D file as the Jacobian-weighted total air density and water
vapor density, and entropy, respectively,
12.4.2.2 Layer Heights
In CMAQ, layer heights are computed using the basic definition of the geopotential height in terms
;"" of Jacobian instead of relying on the coordinate-specific analytic equations, such as a hypsometric
,-' equation for a hydrostatic coordinate. The height above mean sea level (MSL) is defined in terms
of the vertical Jacobian as:
'dS (12-114)
4i *
». ••,.;.- 'V. \ >
The layer heights at the interface and middle of the layers above the ground level (AGL) are
computed with:
12-56
-------
EPA/600/R-99/030
ct»l/2
= ^'
?J
for l
-------
EPA/600/R-99/030
for 0^ / <>L and 0<; m
-------
EPA/600/R-99/030
This equation predicts time rate change of the hydrostatic pressure component whose vertical
gradient is in balance with gravitational acceleration. In the geometric height coordinate, this
equation does not exist. However, the height coordinate does not require surface pressure
tendency to close the system because the coordinate is time independent. Because the hydrostatic
pressure coordinate £, = \ — o. is a material coordinate, the mass continuity equation can also be
used to estimate contravariant vertical velocity component by integrating the wind divergence term
either from the bottom to a level % or from the top to <*.
(12_122a)
m
(12.122b)
where £r = 1.
For other coordinates that do not have diagnostic relations, the contravariant vertical velocity
component should be estimated using standard coordinate transformation
= £3 = + (_mV • v , +W) (12-123)
dt ^ g \oz)
where &r is the height of the coordinate surface and w is the regular vertical component of wind.
For nonhydrostatic MM5 applications, % = I -
-------
: ...... ,. :!:; 'i - ' , • ; • ,.
EPA/600/R-99/030
pj .
output. It also vanishes at the top of the modeled atmosphere. The quantity ( — £•£) is stored in
..... •"• 1111: ..... . '. .1 ' ......... I",!,!' ........ ill '.'I „,,. ..... ..... . . , .. .11. , .«' ' ..',>. ..... i»v' , •••' :' III ...... ..... !" '. ffl
the MET_CRO_3D file from k=\ to K ( i.e., NLAYS) with the variable name (WHAT_JD).
I ........ ,: ....... » , . ..... in ..... ip» ' 1 ....... ....... '"";. j|1 '•• • •" „. •':.•• • •••• .,::: ..... . .••'.•:
12.4.4 Mass Consistent Temporal Interpolation of Meteorological
Parameters
In the CCTM, the temporal interpolations of meteorological data are needed because the output
frequency of the meteorological data is usually much coarser than the synchronization time step of
the chemistry-transport model. Consistency in meteorological parameters such as, Jacobian,
density of air (total), density of water vapor, entropy (or temperature), and pressure are important
for the science modules in the CCTM.
Here, we recommend interpolation schemes for density and velocity fields that maintain mass
consistency. First, the Jacobian and density at a time ta = (1 - a)tn + atn+l between the two
consecutive output time steps, tn and rn+1 , are expected to be interpolated with:
(/{)«=(1 -a)(/{)B + aC/{),H.i (12-126)
(p/4)a =(l-a)(p/4)n +a(p/pn+1 (12-127)
!.' '•' - " ..... ,11 : "i . ....... . : . , ........ . ": i - • '
where 0 £ a
-------
EPA/600/R-&W030
= 3[(/*7.U - (P/,) J -
dt
The density of water vapor (pv) and entropy should be interpolated with the same method. The
density of dry air from the total air density can be obtained by subtracting pv from p. The
interpolated temperature is computed with:
and entropy can be computed with Equation 12-99, The interpolated pressure is computed with the
ideal gas law,
P = (pdRlt+pvRv)T (12-131)
Next, wind components multiplied with the Jacobian-weighted density are interpolated linearly and
the interpolated wind components are obtained by dividing the results with (pJs)a '.
U« (12-I32a) ,
)Ml (12-132b)
(V )a = *** (12-133a)
(P/,).
)g (12-133b)
tt
When density and wind data are not collocated, the above procedure may incur spatial interpolation
errors.
12.4.5 Optional Conversion of Nonhydrostatic Data to Hydrostatic
Meteorological Data for MM5
There may be a special situation when a hydrostatic CTM run is desired while the MM5 run is
made with the nonhydrostatic option. In such a case one may want to redefine the nonhydrostatic
coordinate into hydrostatic one while keeping the profile data as provided. In spite of many
theoretical problems with this kind of data conversion, MCIP provides an option to convert the
12-61
-------
EPA/600/R-99/03Q
nonhydrostatic coordinate into the hydrostatic pressure coordinate. Use of this option is not
recommended for scientifically rigorous computations. However, this is a useful option for certain
purposes such as code development and testing of a hydrostatic CTM.
Nonhydrostatic and hydrostatic coordinates can be compared in terms of the atmospheric pressure.
Pressure in a hydrostatic MM5 run is defined in Equation 12-93 while the same in a hydrostatic
MM5 is defined in Equation 12-94. Then, the difference in the nonhydrostatic and hydrostatic
pressures is:
p* = p*g —
-------
EPA/600/R-99/030
12.5 Operation of MCIP
Because MCIP is a ModeIs-3 conformant processor, it needs to be compiled and executed using a
Models-3 build command (mSbld) with a configuration file. It requires the grid-domain object
include files to define output domain and resolution. In most cases, the user does not need to
define input data domains and resolutions because reader modules can extract the information from
file headers. MCIP code structure, compilation and execution procedures are discussed below.
Additional operational information for preparing MCIP through Models-3 system framework can
be found in Chapter 7 of the EPA Third-Generation Air Quality Modeling System User Manual.
Also, the User Manual Tutorial for the initial public release of Models-3 provides step-by-step
instructions for running MCIP with a set of sample examples.
12.5.1 MCIP Modules
MCIP code is archived with CVS (Concurrent Version System) (Cederqvist, 1993) in the Models-
3/CMAQ system. Currently, eleven module classes are defined for MCIP. The classification
makes it convenient to customize MCIP code for special situations. For example, when different
meteorological inputs and process options are needed to link with the CCTM, appropriate modules
from different classes can be used. Refer to Table 12-9 for the details of module descriptions and
associated source code.
12.5.2 Building MCIP
The MCIP code conforms to the Models-3 coding standard. It is designed to be compiled using
m3bld with a configuration file. Refer to Appendix 12C for a sample MCIP configuration file.
The user needs to retrieve the source code (main program, associated subroutines for selected
modules, and include files) appropriate to the user's choice of optional modules through the CVS
system. There are four kinds of include files for this processor. Three include file types, Models-
3 I/O API include files, MCIP's global include files, and MCIP module specific include files, are
fixed. The fourth include file type describes dimensional information of the input meteorology data
and, as an exception, must be edited to match the number of vertical layers in the input.
The parameter MAXK in MCEPPARM.EXT may need to be changed large enough to accomodate
the dimensions of the number of layers for the meteorological data. Horizontal domain information
is obtained from the header of MM5 files automatically. MCIP expects that include files for the
CTM grid/domain domain (HGRD.EXT, VGRD.EXT, COORD.EXT), for which the data will be
extracted, are provide by the user. Vertical collapsing is done automatically when the number of
layers in VGRD.EXT is less than the number of layers in input data and the coordinate interface
values match as shown in Table 12-5. When a few lowest input model layers are collapsed into
one for the output, only diagnostic option can be used because the surface flux values and
aerodynamic resistance passed through are no longer valid due to the change in the thickness of the
lowest layer. The M3DDEP option must not be used when collapsing the lowest model layers and
doing so will be violating the parameteric values passed through from the meteorological model.
12-63
-------
BPA/600/R-99/030
Table 12-9. MCIP Module Definitions and Associated Source Code
Class
Module
Description
Source Code
driver
mcip
controls main computational flow
mcip.F, initx.F
input
mmS
reads in MM5 output files
getmet_mm5.F,
readmml.F,
readmm2.F, getgist.F,
MM5INPUT.EXT,
MM5HEADER.EXT
r" " .'
landuse
pbl
drydcp
!ism * - *
cloud
solar
nietSd
PI* • .'""i .
output
util
F
f
i-
. ;• ^ J —
fakemet
rams
mSradm
pblpkg
radmdd
: >- ' MW ' f. W
~ cmaqdd
radmkuo
solar_px
mSsup
.„•"!",•, : «H i.
stnd
util
icl
i i
'!' -t
reads in MCIP output files
generates meteorology fields for idealized flow field
study
reads in RAMS I/O API output files
reads in landuse data in RADM dry deposition category
computes PBL parameters
computes dry deposition velocities using Wesely's
RADM method
computes dry deposition velocities using Pleim's
CMAQ method
computes cloud parameters using convective column
assumption
computes solar radiation using the algorithm
implemented in MM5-PX version
computes supplemental three-dimensional variables
, needed for CMAQ
generates MCIP output for standard variable lists
collection of utility subroutines
MCIP specific include files
: "^ f ^
. . . . ,
getmet_m3.F
getmet_m3fake,F
not available
getluse.F
pblpkg.F, pblpwr.F,
sfcflux.F, slflux.F
radmdry.F,
DDEPSPC.EXT
not available
bcldprc_ak.F
solar.F, transm.F
layht.F, verthat.F,
vertnhy.F, verthyd.F,
jacobi.F, metSdsup.F
comheader.F, gridout.F,
metcro.F, metdot.F
bilin2d.F, bilinSd.F,
cvbdx.F, colIapx.F,
cvmgp.F, cvmgz.F,
ratint.F, errmsg.F,
sanity.F
CONST_mete.EXT,
CONST_pbI.EXT,
FILES3_MCIP.EXT,
GROUTCOM.EXT,
INFILES_MCIP.EXT,
LRADMDAf.EXT,
MCIPCOM.EXT,
MCfPPARMtEXT,
MCOUTCOM.EXT,
MDOUTCOM.EXT
12-64
-------
EPA/600/R-99/030
12.5.3 Executing MCIP
12.5.3.1 Run Script Command File
As with other interfaces, an execution run script is used to define key environmental variables
prescribing run-time characteristics and linkage between logical file names specified in the codes
and actual files for input and output. A sample script for MCEP execution can be retrieved from the
CVS archive and edited for the user's particular application. The user may modify environmental
variables for choosing different processing options and for defining input and output files. Refer
to the APPENDIX D for a sample MCIP run script. For processing MM5 output, MCIP may not
need to be co-resident with MM5 if its output files are in IEEE binary format. MCIP can be ran on
any UNIX computer for processing meteorological data in Models-3 I/O API format.
Input parameters METROW, METCOL, and METLAY are defined through the UNIX
environmental variable list. These are used to check if the input meteorology data have correct
dimensions as expected. Within MCEP, the actual values will be picked up from header
information of MM5 data file. When offset values (10, JO) are inconsistent with the information in
MM5 header and in COORD.EXT, MCIP suggests a new set of (10, JO) or (XORIG, YORIG)
values. Make sure to set NDEP to correspond to the number of lowest input layers being collapsed
into the lowest output model layer, so aerodynamic resistances are computed accordingly.
12.5.3.2 Input Files for MM5 Data Processing
Logical names of input files are described below:
• MMOUT_DOMAIN#: This standard MM5 file contains gridded hourly (or sub-hourly)
two- and three-dimensional meteorological data covering the entire MM5 domain.
« MDL3JDOMAIN#: This is optional EPA added MM5 output that contains file contains
PBL parameters needed when the pass-through option is chosen.
• LAND_CRO_2D_J#KM: This is LUPROC output that contains 11-category fractional
land-use data covering the entire MM5 domain. If USGS North American land/vegetation
characteristic data base used as input to LUPROC is not enough to cover the desired
domain, dominant landuse file LANDUSE_DOMAIN# from TERRAIN, a preprocessor
for MM5 system, can be used.
12.5.3.3 Input Files for I/O API Meteorological Data Processing
MCIP can be used to extract meteorology data for a smaller window domain, to collapse number of
layers, or to process meteorological data already in I/O API format further. It treats gridded input
meteorological data as pseudo profiles for temperature, moisture and wind components. Required
I/O API meteorological inputs for this option are; GRU)_CRO_2D, GRJD_DOT_2D,
MET_CRO_2D, MET_CRO_3D, and MET_DOT_3D files. Because it does not use boundary
files as its input, the output domain can be as large as (NCOLS-2)x(NROWS-2) for windowing or
layer collapsing process. This option can be used to generate interpolated meteorology data for
higher resolution grid than the original input as well.
12-65
-------
EPA/600/R-90/b30
Table 12-10. MCIP Environmental Variables
Environmental
Variables
Description
Note
METLAY
METCOL
METROW
10
JO
LUTYPE
IOLUSE
JOLUSE
LCALCPBL
LSLFLUX
LM3DDEP
number of layers in input meteorology
data
column (east-west) direction cell
dimension in input meteorology data
row (south-north) direction cell dimension
in input meteorology data
location of CTM domain origin offset in
row direction = ROW_OFFSET
location of CTM domain origin offset in
column direction = COL_OFFSET
file type for landuse data
location of CTM domain origin offset in
row direction w.r.t. landuse data origin
location of CTM domain origin offset in
column direction w.r.t. landuse data origin
flag for estimating PEL parameters of not
flag for similarity algorithms
flag for dry deposition algorithm
should be larger than or equal to
CTM NLAYS
should be larger than CTM
iSfCOLS
should be larger than CTM
NROWS
offset should be counted based on
the input grid definition
offset should be counted based on
the input grid definition
1: TERRAIN binary 13 category
2: preprocessed ASCII 1 1
category
3: MM5 dominant landuse
category
4: use landuse in GRIDCRO2D
5: use USGS I/O API landuse
file
offset should be counted based on
the input grid definition. For
LUTYPE 4 & 5, IOLUSE is not
used
offset should be counted based on
the input grid definition. For
LUTYPE 4 & 5, JOLUSE is not
used
TRUE: re-compute PEL
parameters
FALSE: pass through from
MM5
TRUE: use surface layer
similarity
FALSE: use PEL similarity
TRUE: use CMAQ dry
deposition
FALSE: use RADM dry
deposition
12-66
-------
Table 12-10, MOP Environmental Variables (continued)
EPA/600/R-99/030
Environmental Description
Variables
Note
LWIND
flag for wind field correction
0 (default): no correction
+/- 1,2,3 wind field correction
options (TBD)
LCALCCLD
flag for cloud algorithm
TRUE: compute cloud
parameters
FALSE; pass through from
MM5
LHYDOUT
flag for hydrostatic output
TRUE: forces hydrostatic data
output with approximations
FALSE: pass through from
MM5
CRO_FTYPE
DOT_FTYPE
flag for I/O API file type
1(default): time dependent
Q (special): time independent
LS AMITY
flag for checking ranges of output
parameters
TRUE: check
FALSE: do not check
BMAX
Maximum boost rate for urban area
2 (default): 100% increase when
urban and water areas are
coexisting
JUDATE
HSTRT
HTOSKIP
HTORUN
NDEP
GRDNAM
EXECUTIONJD
SCENFILE
lOAPIjCHECK^HEAD
ERS
Julian date for the start time
start hour
number of hours to skip MCIP process
number of hours to run
number of deposition layers
grid/domain name for MCIP output files
user defined execution ID
user defined file path
flag for checking I/O API file headers
(yyyyddd)
-------
EPA/6QG/R-99/030
Table 12-11. MCIP File Linkage
Environmental
Variables
- MM51
MM52
LU13
IB " '.:: '•• *•.»> •«*
LOSE
GRID_CRO 2D
GRIDJOOT 2D
MET_CRO_2D
MET CRO 3D
MET_DOT_3D
GRIDJBDY 2D GI
GRID_BDY 3D Gl
GRID CRO 2D Gl
GRID CRO 3D Gl
GRID DOT 2D Gl
MET CRO 2D Gl
MET CRO 3D Gl
MET_DOT_3D_G1
Description
Input filename for MM5 1 data file
Input filename for MM52 data file
Input filename for TERRAIN binary
landuse
input filename for I/O API landuse file
input filename for GRID CRO 2D
input filename for GRID DOT 2D
input filename for MET_CRO_2D
input filename for MET CRO 3D
input filename for MET_DOT_3D
output filename for GRID BDY 2D
output filename for GRID BDY 3D
output filename for GRID_CRO_2D
output filename for GRID_CRO_.3D
output filename for GRID DOT 2D
output filename for MET CRO 2D
output filename for MET CRO 3D
output filename for MET_DOT_3D
Note
for mm5 input module
for mm5 input module
applicable for LUTYPE = 1
applicable for LUTYPE = 4 or 5
for m3 input module
for m3 input module
for m3 input module
for m3 input module
for m3 input module
12.5.4 Defining Grid and Domain for MCIP
12.5.4.1 MM5 Grid and Domain Definitions
MM5 may output as many as nine grid domains (one coarse grid and up to eight nested grid
domains) per MM5 execution. The MM5 developers took the approach of compiling MM5 with
the appropriate dimensions for each study being performed. On the other hand, MCIP processes
; output meteorological data one grid domain at a time. It needs to be compiled explicitly for each
grid domain. For multiple scale MM5 model runs, usually MM5 users have to define all the multi-
level nest domains together with the coarse "mother" grid. The complication in defining MM5
domains as the Models-3 domain objects is caused by somewhat inauthentic use of the left-hand
coordinate "system "and dot-point grid/domain definitions in MM5. For the details on how to define
IyIM5 domains., users are recommended to read MM5 User's Guide. In the following, we uses a
set of multi-level MM5 domains listed in Table 12-12 as the examples of the discussion.
Table 12-12. An Example Set of Multi-level Nesting MM5 Domains
(Note that (1,1) in MM5 corresponds to (ROW, COL) in CMAQ system.)
MM5
Domain
No. (N)
1
2
m> 4
Resolution
(4t=4x)
108km
36km
mm" t4so i
; ,r®kni " ;
Dimensions
(fX,JX) =
(ROWS+LCOLS+l)
(41,61)
(67,82)
'* , (82,100)
(100^82)
Origin Relative to Base
Grid
(U1,J_11)
(l.D
(9,24)
/ji9.333333,35.333333')_
(21.444444,41.555555)
Origin Relative to
Immediate Parent grid
^OWoffsa + l,COL,,ffx, + l)
(l.D
(9,24)
\ ' " '(32,35)"
(20,57)
12-68
-------
EPA/600M-99/030
a) Definitions for MM5 108 km mother grid (N= 1)
In MM5, the grid is defined with the number of dot points. Therefore, to use MM5's grid
definitions, for Models-3/CMAQ's cross-point grid definitions, we need to use following
NCOLS = JX-1 =61-1 =60 (12-138a)
NROWS = IX-1=41-1 =40. (12-138b)
For this grid, no "parent grid" is defined. Origins should be defined with the formula that is
specific to MMS's method of grid/domain definition. Note that we require the MM5 mother
domain's number of cells in horizontal directions (i.e., NCOLS and NROWS) to be even
numbers, so that the dimensions of the four quadrants relative to the center of the coordinates are
identical (Refer to Figure 12-9). The origin of the coarse MM5 grid is defined with
hT hT
(12-139)
*> one' -s ong * ^ r* f^
2, 2,
Because it is the mother grid of the grid family, offsets should be set to zero:
(COLoffset,ROWoffiel} = (0,0) (12-140)
Currently this domain is defined with resolution of 108km (see Table 12-12).
b) Definitions for MM5 36 km first-level nest grid/domain (N=2)
For this grid, the MM5 108km domain defined above is used as the "parent grid." For the nested
domains, NCOLS and NROWS can be either even or odd numbers:
NCOLS = JX - 1 = 67-1 = 66 (12-141a)
NROWS = IX- 1 =82-1 =81. (12-141b)
Because the left-most and bottom corner of the nest 36km domain is defined with (7,,, 7U) in
MM5, we need to convert this information into (COLoffset,ROWoffset) using following equation:
(COLoffxel,ROWoffie!) = (7U -1,/,, -1) (12-142)
Note that MM5 uses left-hand coordinate system, so that the index positions for (/,_,,7,,) are
reversed with those for the Models-3/CMAQ's (COLoffset,ROWoffset). In the following case, the
general equation for computing the offset numbers from (7U,7U) of the MM5 domains will be
presented. The coordinates of the origin is computed with:
(x v } = (xl"tr""+COL Ax1""''"" vpar"" + ROW Avpare'") (12-1431
V^orij'/orij^ — \xorig ^ '-V-'offsa***' >Sorig T KU "offset**.? ' (14 WJ)
12-69
-------
EPA/60Q/R-99/030
c) Definitions for MM5 12 km and 4 km nest grid/domains (N=3,4)
For the third, 12km domaini(N=3),we have:
Pt ' NCOLS = JX -{= 100-l"= 99 ' .-.,'-,.
NROWS = DC-1=82-1 =81
(12-144a)
(12-144b)
and MM5 36km domain is used as the parent grid. Also, for the fourth 4 km domain (N=4), we
have:
NCOjLS = JX - 1= 82-1 =81
NROWS = DC - 1 = 100-1 = 99
(12-145a)
(12-145b)
and MM5 12km domain is used as the parent grid. Usually, the offset values will be known to the
Models-3/CMAQ users.
I' •• ; •?:• ' I '• , , <•.•;. •!! -,. : :. , ;• , • ' •
MM5 Mother Grid/Domain and Coordinate Center
T
oo
&
O
NCOLS
Figure 12-10. A Schematic of the MM5 Mother Grid Definitions.
Because MMS's mother grid is defined from the coordinate center and grows outward, the
numbers of columns and rows for each quadrants relative to the coordinate center should be the
same, i.e., NCOLS and NROWS for the MM5 mother grid should be even numbers. In case the
offset injormationjs not .known a priori, one needs to compute the offset values using the header
12-70
-------
EPA/600/R-99/030
information in the MMS output files. To use the information correctly, we need to know how the
MM5 nest domain's left-most and bottom corner (/^p/u) is defined. In MMS, all the nest
domains are defined relative to the mother grid, and (/,,, J,,) provides the offset information of
the nest domains in terms of the mother grid's resolution. The formula linking a nest MMS's
domain (/,,,7U) and Models-3/CMAQ's offset values (COLeffset,ROWoffsa) are given for /V>2as:
ij
>+2
(COLoffset)t
f^(Axpa"M I Ax)
I "
^ h+2
?2(Ayparer" I Ay)k~
(12-146)
where (Axparen! I Ax) = 3 for the current MMS application example. Because our objective is to
help users to find out (COLoffia,ROWoff:ta) for the nest grid, we need to rewrite Equation 12-146
for (COLoffiet, ROWoffal). Assuming we know all the coarser grid/domains offset values,
(COLoffset, ROWoffat) for the current nesting level N > 2 can be found as:
N-l
= """
(COLoffsts)N = (Ax
] (Jlfl)N - 1 -
f^(Axpanttl IAx)k~2
(ROWoffill)N=(Aypa"'"IAy)
N-2
:-2
(I2-I47a)
(I2-I47b)
Although Equations 12-147 a and b look somewhat complicated, one can readily compute offset
numbers for Models-3 CMAQ's grid/domain definitions for the example set of domains defined in
Table 12-11 as follows:
)^ = [(24
1)] = [23,8]
(12-148a)
[(COLoffietUROWoffitl)}^ = ^{(35l- - 1) - 3° * 23],3[(19| - 1) - 3° * 8]]
= [34,31]
(12-148b)
= [56,19]
(12-148c)
Most of the time, the above MMS domain definitions are provided by the MMS modelers and
Models-3/CMAQ users do not have to re-compute them.
12.5.4.2 CMAQ Grid Definitions
The horizontal grid structure used for the MMS serves as the parent grid for the CCTM domains.
To define a CCTM grid, number of cells to be excluded from the parent domain must be
12-71
-------
EPA/600/R-99/030
determined. It is recommended that the CCTM grid be smaller than its parent MM5 grid by at least
4 grid cells (preferably by 6 grids) to avoid the possible numerical reflection problems at
boundaries of MM5 domain. Once the offset values are known, (x^y^) of the nest domain can
be computed using Equation 12-143. Refer to Appendix 12B for other examples of other CCTM
nest domain definitions.
12.5.5 Extension of MCIP for Other Meteorological Models
The current release version of MCEP can only process MM5 output in binary format or previous
MCIP output in I/O API format. To realize the one-atmosphere concept for meteorological and air
quality modeling, CMAQ utilizes generalized coordinates. One advantage of the generalized
coordinate system is that a single CTM can adapt to any of the coordinate systems and dynamics
commonly used in meteorological modeling. Because most meteorological models are not
originally designed for air quality studies, they lack characteristics that are required for air quality
modeling. Consistencies in dynamic descriptions of atmosphere, physical parameterizations, and
numerical algorithms, where applicable, in meteorological and chemistry-transport models are
critical in determining the quality of air pollutant simulations. This issue becomes more critical
with high resolution air quality study where nonhydrostatic meteorological models must be used.
MCIP is the key processor allowing the consistent linkage between meteorological models and
CMAQ. Currently, different meteorological models are used by different atmospheric modeling
groups forming their own respective user communities. It is because these meteorological models
are applicable for a limited range of spatial and temporal scales. To expand the user base of the
l|[pdcls-3 CMAQ system and to promote the one-atmosphere community modeling paradigm, it is
essential to continuously develop MCIP modules for several popular mesoscale meteorological
models such as RAMS, ARPS, HOTMAC, and others.
To build a versior^of MQQP for processing a set of meteorological model output, several issues
involved with different dynamics and coordinates must be considered. They are; (1) compatibility
of governing set of equations and state variables used, (2) scale limitations in subgrid scale
parameterizations such as cloud, turbulence, and surface exchange processes, and (3) consistency
in numerical algorithms and discretization methods. Before designing MCIP modules for other
meteorological model, developers should identify the structure, format, and frequency of data as
well as their impact on the processing structure of MCIP. MCIP expects time dependent
meteorological data in a structured grid system as the input. As this moment, there is no provision
for handling irregularly spaced data structure either from observations or from a meteorological
modeling system with unstructured adaptive (fixed or dynamic) grid structure, such as the one
used in OMEGA system (Bacon et al., 1996). If there is need for including these data, object
analysis tools must be used to prepare data in a structured grid system.
The essential information related to the coordinates and dynamic assumptions used in
meteorological models is captured in MCIP. Dynamic and thermodynamic state variables are recast
into the fully compressible system as described in Chapter 5 of this volume.
ft is important to know the processing sequence and associated data structures used in the current
MCEP code. Major processing steps in MCIP can be summarized as follows: (1) Grid, coordinate,
and information on atmospheric dynamics are used to compute Jacobian and layer heights; (2) State
12-72
-------
EPA/60Q/R-99/030
variables in meteorological models are recast into density and entropy; (3) Contravariant wind
components are computed; (4) necessary two-dimensional parameters are diagnosed; and (5) Some
of two- and three-dimensional parameters are passed through. Figure 12-11 shows corresponding
data structures used at different phases. Input phase uses arrays with dimensions covering the full
meteorology domain ('F-arrays'), processing phase uses arrays for extended domain covering both
the CMAQ and boundary domains ('X-arrays'). During the output phase, the information in X-
arrays are separated into the CMAQ and boundary domains for respective data types described in
Table 12-3.
Input phase Processing Output phase
NROWS
Met. Domain Extended CMAQ Domain CMAQ Domain Boundary
'F'-arrays 'X'-arrays Dot & Cross Domain
Figure 12-11. Data Structures Corresponding to Input, Processing, and Output Phases of MCIP
Processing
12.6 Concluding Remarks
MCIP is a key processor linking meteorological models to the CMAQ modeling system. Its major
roles are:
1. To read meteorological data from a meteorological model and converts them in Models-3
I/O API format,
2. To provide all the necessary meteorological parameters for air quality simulations. When
necessary, PBL parameters and other derived quantities are computed using gridded
meteorology data and high-resolution fractional land use information.
3. To support multiscale generalized coordinate implementation of the CCTM.
Current implementation of MCEP is mainly for linking MM5 output to CCTM. Reader modules to
Regional Atmospheric Modeling System (RAMS) (Pielke et al., 1992), will be added. Users who
wish to link other meteorological models may need to modify and introduce appropriate input
modules.
12-73
-------
EPA/60Q/R-99/03G
12.7 References
Alduchov, O, A. and R. E. Eskridge, 1996: Improved Magnus form approximation of saturation
pressure. /. AppL Meteorol., 35, 601-609.
Anthes, R. A., E. Y. Hsie and Y.H Kuo, 1987: Description of the Penn State/NCAR Mesoscale
Model vejrfion 4 (MM4). NCAR Technical Note, NCAR/TN-282+STR, National Center for
Atmospheric Research, Boulder, CO. 66pp.
Avissar, R., P. Avissar, Y. Hahrer, and B. A. Bravdo, 1985: A model to simulate response of
plant stomata to environmental conditions. Ag. and Forest Meteorol., 34,21-29.
Bacon, D. P.,'Z. Boybeyi, P.R. Boris, T. J. Dunn, M. Hall, R. A. Sarma, S. Young, and J.
Zack, 1996: An overview of the operational Multi-sclae Environmental model with Grid
Adaptivity. Proc. of the 15th Conf. on Wea. Anal, and Forecasting, Norfolk, VA, August 19-
23, 1996, American Meteorological Soc., Boston, MA.
Ball, J. T., I. E,, Woodrow, and J. A. Berry, 1987: A model predicting stomatal conductance and
its contribution to the control of photosythesis under different environmental conditions. In
Progress in photosynthesis Research, Vol. IV (ed. J. Biggins), pp. 221-234. Martinus Nijhof
Publishers, Dordrecht.
Belts, A. K., 1973: Non-precipitating cumulus convection and its parameterization. Quart. J, Roy.
Meteorol, Soc. 99, 178-196.
Binkowski, F.S. and U. Shankar, 1995: The Regional Paniculate Model. 1. Model description
and preliminary results. J, Geophys. Res., 100 D12, 26191-26209.
Blackadar, A. K., 1978: Modeling pollutant transfer during daytime convection, Preprints, Fourth
Symposium on Atmospheric Turbulence, Diffusion, and Air Quality, Reno, Am. Meteor.
Soc., 443-447.
Brost R. A. and J. C. Wyngaard, 1978: A model study of the stably stratified planetary boundary
layer. /. Atmos. Sci., 35, 1427-1440.
Businger J. A., J. C. Wyngaard, Y. Izumi, and E. F. Bradley, 1971: Flux profile relationships in
the atmospheric surface layer, J. Atmos. Sci., 28, 181-189.
Byun, D. W., 1990: On the analytical solutions of flux-profile relationships for the atmospheric
surface layer, /. AppL meteorol., 29, 652-657.
Byun, D. W., 1991: Determination of similarity functions of the resistance laws for the planetary
boundary layer using surface-layer similarity functions, Boundary Layer Meteorol., 57, 17-
48.
Byun, D.W., 1999a: Dynamically consistent formulations in meteorological and air quality models
for multiscale atmospheric applications: I. Governing equations in generalized coordinate
4 System. /. Atmos. Sci, (in print)
12-74
-------
EPA/600/R-99/030
Byun, D.W., 1999b: Dynamically consistent formulations in meteorological and air quality models
for multiscale atmospheric applications: DL Mass conservation issues. /. Atmos. Sci., (in
print)
Byun, D. W., and R.L. Dennis, 1995: Design artifacts inEulerian air quality models: Evaluation
of the effects of layer thickness and vertical profile correction on surface ozone concentrations.
Atmospheric Environment, 29, No. 1, 105-126.
Carson, D. J., 1973: The development of a dry inversion-capped convectively unstable boundary
layer. Quart. J. Roy. Meteorol. Soc. 99, 450-467.
Cederqvist, P., 1993: Version Management with CVS, Signum Support AB, Linkoping, Sweden.
168pp.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madronich, P. Middleton, W.R. Stockwell, and C.J.
Walcek, 1987: A three-dimensional Eulerian acid deposition model: Physical concepts and
formulation, /. of Geophys. Res., 92, 14,681-700.
Chang, J.S., P.B. Middleton, W.R. Stockwell, C.J. Walcek, I.E. Pleim, H.H. Lansford, S.
Madronich, N.L. Seaman, D.R. Stauffer, F.S. Binkwski, D. Byun, J.N. McHenry, H. Mass,
and P.J. Samson, 1990: The Regional Acid Deposition Model and Engineering Model.
NAPAP report 4.
Coats, C. J., cited 1996: The EDSS/Models-3 I/O Applications Programming Interface. MCNC
Environmental Programs, Research Triangle Park, NC. [Available on-line from
http://www.iceis.mcnc.Org/EDSS/ioapi/H.AA.html.]
Collatz, G. J., C. Grivet, J. T. Ball, and J, A. Berry, 1991: Physiological and environmental
regulation of stomatal conductance, photosynthesis and transpiration: a model that includes a
laminar boundary layer. Ag. and Forest Meteorol., 54, 107-136.
Driedonks, A. G. M., 1981: Dynamics of the well-mixed atmospheric boundary layer. De Bilt
KNMI Sci. Rep. WR 81-2, 189pp.
Dudhia, J., 1989: Numerical study of convection observed during the Winter Monsoon
Experiment using a mesoscale two-dimensional model. J. Atmos. ScL, 46, 3077-3107.
Dudhia, J., 1993: A nonhydrostatic version of the Penn State-NCAR mesoscale model: Validation
tests and simulation of an Atlantic cyclone and cold front, Man. Wea. Rev,, 121, 1493-1513.
Erisman, J. W., A. van Pul, and P. Wyers, 1994: Parameterization of surface resistance for the
quantification of atmospheric deposition of acidifying pollutants and ozone. Atmos. Environ.
28, 2595-2607.
Fowler, D., 1978: Dry deposition of S(>2 on agricultural crops, Atmos. Environ., 12, 369-373.
Fritsch, J.M., and C.F. Chappell, 1980: Numerical prediction of convectively driven mesoscale
pressure systems, 1, Convective parameterization, J. Atmos. Sci., 37, 1722-1733.
12-75
-------
L . • , -. i
EPA/600/R-99/030
(' " ..' Ill , ! ,
Geleyn, J.-F,, A, Hense, and H.-J. Preuss, 1982: A comparison of model-generated radiation
fields with satellite measurements. Beitr. Phys. Atmos., 55, 253-286.
Grell, G.A., J. Dudhia, and D.R. Stauffer, 1994: A Description of the Fifth-Generation PENN
STATE/NCAR Mesoscale Model (MM5), NCAR Technical Note, NCAR/TN-398+5T/?,
f National Center for Atmospheric Research, Boulder, Colorado, 138pp
Haagenson, P. L,, J. Dudhia, D. R. Stauffer, and G. A. Grell, 1994: The Penn State-NCAR
mespscale model (MM5) source code document, NCAR/TN-392+STR, National Center for
Atmospheric Research, Boulder, Colorado.
Hack, f. J., BL A. Boville, B. R Briegleb, J. T. Kiehl, P. J. Rasch.'and D. L. Williamson, 1993:
Description of the NCAR community climate model (CCM2). NCAR Tech. Note, NCAR/TN-
382+STR, 120 pp. [Available from the National Center for Atmospheric Research, P.O. Box
3000, Boulder, CO 80307.]
Hess, G. D., 1992: Observations and scaling of the atmospheric boundary layer. Aus. Meteorol.
T Mag. 41, 79-99.
jF'.iii! . ' ' ''.I, ! i '•',',
Ho'gstrom, U., 1988: Non-dimensional wind and temperature profiles in the atmospheric surface
layers: a re-evaluation. Boundary-Layer Meteor., 42,55-78.
Holtslag, A. A., E. I. F. de Bruin, and H.-L. Pan, 1990: A high resolution air mass
transformation model for short-range weather forecasting. Man. Weather Rev., 118, 1561-
1575.
Holtslag, A.A.M., E.V. Meijgaard, and W.C. DeRooy, 1995: A comparison of boundary layer
diffusion schemes in unstable conditions over land, Boundary-Layer Meteor., 76, 69-95.
? ' •;;:,',' i • :.„; iy *,;;..* / ?\ »„ * , i
Jarvis, P. G., 1976: The interpretation of the variation of leaf water potential and stomatal
conductance found in canopies in the field. Phil. Trans. Royal Meteorol. Soc. of London,
2 Series B, 273, 593-610.
Jacquemin B. and J. Noilhan, 1990: Sensitivity study and validation of a land surface
- parameterization using the HAPEX-MOBILHY data set. Bound-Layer Meteorol., 52, 93-
134.
Jaung, H.-M. H., 1992: A spectral fully compressible nonhydrostatic mesoscale model in
hydrostatic sigma coordinates: Formulations and preliminary results. Meteor. Atmos. Phys.,
"" 50, 75-88.
Kain, J. S., and J. M. Fritsch, 1993: Convective paramterization for mesoscale models: The Kain-
Fritsch scheme. Jthe representation of cumulus in mesoscale models., K. A. Emanuel and D.J.
Raymond, Eds., Amer. Meteor. Soc., 246pp.
Lo, A. K., 1993: The direct calculation of fluxes and profiles in the marine surface layer using
measuremerits from a single atmospheric level. J. Appl. Meteorol., 32, 1890-1990.
12-76
-------
EPA/600/R-99/030
Lo, A. K., 1995: Determination of zero-plane displacement and roughness length of a forest
canopy using profiles of limited height. Boundary-Layer MeteoroL, 75, 381-402.
Madronich, S., 1987: Intercomparison of NO2 photodissociation and UV radiometer
measurements, Atmos.Environ., 21, 569-578.
Mason, P.J., 1987: The formation of areally-averaged roughness lengths. Q.J.R. Meteorol. Soc,,
106, 351-366.
Mesinger, F. and A. Arakawa, 1976: Numerical methods used in atmospheric models. In: Global
Atmospheric Research Programme (GARP), Joint Organization Committee, GARP Publication
Series, No. 17, Aug., 1976: [Available from WMO Secretariat, Geneva].
Mihailovic, D. T., R. A. Pielke, B. Rajkovic, L, J. Tsengdar, and M. Jeftic, 1993: A resistance
representation of schemes for evaporation from bare and partly plant-covered surfaces for use
in atmospheric models. J, Appl. Meteorol., 32, 1038-1054.
Noilhan, J. and S. Planton, 1989; A simple parameterization of land surface processes for
meteorological models. Man. Wea. Rev., 117, 536-549.
Ooyama, K. V., 1990: A thermodynamic foundation for modeling the moist atmosphere. J.
Atmos. ScL, 47, 2580-2593.
Padro, J., G. den Hartog, and H. H. Neumann, 1991: An investigation of the ADOM dry
deposition module using summertime O3 measurements above a deciduous forest, Atmos.
Environ., 25A, 1689-1704.
Paulson, C. A,, 1970: The mathematical representation of wind speed and temperature profile in
the unstable atmospheric surface layer. J. Appl. Meteor., 9, 857-861.
Pielke, R. A., W. R. Cotton, R. L. Walko, C. J. Tremback, M. E. Nicholls, M. D. Moran, D. A.
Wesley., T. J. Lee, and J. H. Copeland, 1992: A comprehensive meteorological modeling
system-RAMS. Meteor. Atmos. Phys.,49, 69-91.
Pleim, J., A. Venkatram, and R. Yamartino, 1984: ADOM/TADAP Model Development
Program, Volume 4, The Dry Deposition Module. ERT P-B980-520, Prepared for OME, AES
of Canada and the Umweltbundesamt, Germany.
Pleim, J. E. and A. Xiu, 1995: Development and testing of a surface flux and planetary boundary
layer model for application in mesoscale models, /. Appl. Meteor., 34, 16-32.
Pleim, J. E., A. Xiu, P. L. Finkelstein, and J. F. Clarke, 1997: Evaluation of a coupled land-
surface and dry deposition model through comparison to field measurements of surface heat,
moisture, and ozone fluxes, Proceedings of the 12th Symposium on Boundary Layers and
Turbulence, July 28-August 1, 1997, Vancouver, BC.
Pleim, J. E., J. F. Clarke, P. L. Finkelstein, E. J. Cooter, T. G. Ellestad, A. Xiu, and W. M.
Angevine, 1996: Comparison of measured and modeled surface fluxes of heat, moisture and
12-77
-------
EPA/6WR-99/03Q
fcc , - . \ r i is
chemical dry deposition. In: Air Pollution Modeling and its Application XI, Ed: Gryning and
- Schiermeier, Plenum Press, New York.
•
Rochette, P., Desjardins, R. L., Dwyer, L. M., Stewart, D. W., Pattey, E., and Dube, P. A.,
1991: Estimation of maize canopy conductance by scaling up leaf stomatal conductance. Ag.
and Forest MeteoroL, 54, 241-262.
• Schumann, U., 1989: Large-eddy simulation of turbulent diffusion with chemical reactions in the
If convective boundary layer. Attnos. Environ., 23, 1713-1727.
1'' '"';' -'.IT ';;!- i"": * - '• • • ! ------ ^ • -.'
Seaman, N. L., D R. Stauffer, and A. M. Lario-Gibbs, 1995: A multiscale four-dimensional data
assimilation system applied in the San Joaquin Valley during SARMAP. Part I: Modeling
design and basic performance characteristics, J. AppL Meteor., 34, 1739-1761.
Seaman, N. L., and D. R. Stauffer, 1995: Simulation of the thermal structure of the San Joaquin
Valley during SJVAQS/AUSPEX/SARMAP using MM5 with four-dimensional data
- • assimilation, paper presented at the AWMA conference, Regional Photochemical Measurement
^ and Modeling "Studies, November 8-12, San Diego, Calif., 1993.
Sellers, P. J., Y. Mintz, Y. C. Sud, and A. Dalcher, 1986: A simple biosphere model (SiB) for
use within general circulation models. J. Atmos. ScL, 43, 505-531.
Sellers, P. J., R. E. Dickinson, D. A. Randall, A. K. Betts, F. G. Hall, J. A. Berry, G. J.
Collatz, A. S. Denning, H. A. Mooney, C. A. Nobre, N. Sato, C. B. Field, and A,
Henderson-Sellers, 1997: Modeling the exchanges of energy, water, and carbon between
continents and the atmosphere. Science, 275, 502-509.
m , , j i ' « .
Sheih, C.M., M.L. Wesely, and B. B. Hicks, 1979: Estimated dry deposition velocities of sulfur
over the eastern United States and surrounding regions, Atmos. Environ., 13, 1361-1368.
Sirois, A., andL.A. Barrie, 1988: An estimate of the importance of dry deposition as a pathway of
acidic substances from the atmosphere to the biosphere in eastern Canada, Tellus, 40B, 59-80.
Slinn, W. G. N., L. Hasse, B. B. Hicks, A. W. Hogan, D. Lai, P. S. Liss, K. O. Munnice, G.
A. Sehmel, and O. Vittori, 1978: Some aspects of the transfer of atmospheric trace constituents
1 past the air-sea interface. Atmos. Environ. 12, 2055-2087.
i.: - tm.-ri- , ...-,.* •.. . j- , , . , -, i • „ • r • < . • .
Stauffer, D. R., and N. L. Seaman, 1993: Evaluation of multiscale four-dimensional data
assimilation techniques developed and tested in MM5 during SJVAQS/AUSPEX/ SARMAP,
paper presented at the AWMA conference, Regional Photochemical Measurement and
Modeling Studies, November 8-12, San Diego, Calif., 1993.
Tesche, T. W., and D. McNally, 1993: Operational evaluation of the SARMAP meteorological
model (SMM1) for episode 1: 3-6 August, 1990, Alpine Geophysics, Crested Butte, CO.
Thorpe, S., cited 1996: PAVE User Guide. MCNC Environmental Programs, Research Triangle
- Park, NC. [Available on-line from http://www.iceis.mcnc.org/
EDSS/pave_doc/SCCS/s.Pave.html.]
12-78
-------
EPA/fOO/R-99/030
Voldner, E.G., L.A. Barrie, and A. Sirois, 1986: A literature survey of dry deposition of oxides of
sulfur and nitrogen with emphasis on long-range transport modelling in North America,
Atmos.Environ., 20, 2101-2123.
Walcek, C.J., 1987: A theoretical estimate of O3 and H2O2 dry deposition over the northeast
United States, Atmos.Environ., 21, 2649-2659.
Walcek, C.J., R. A. Brost, J. S. Chang, and M. L. Wesely, 1986: SO2, sulfate and HNO3
deposition velocities computed using regional landuse and meteorological data,
Atmos.Environ.,20, 949-964.
Walcek, C.J., and G.R. Taylor, 1986: A theoretical method for computing vertical distributions of
acidity and sulfate production within cumulus clouds, J.Atmos.ScL, 43, 339-355.
Warner, J., 1970: On steady-state one-dimensional models of cumulus convection, J. Atmos. ScL,
27, 1035-1040.
Wesely, M. L., 1989: Parameterization of surface resistances to gaseous dry deposition in
regional-scale numerical models, Atmos.Environ., 23, 1293-1304.
Wesely, M. L., and B, B. Hicks, 1977: Some factors that affect the deposition rates of sulfur
dioxide and similar gases on vegetation, /. Air Pollut. Control. Assoc., 27, 1110-1116.
Wesely, M. L., D. R. Cook, R. L. Hart, and R. E. Speer, 1985: Measurements and
parameterization of particulate sulfur dry deposition over grass, J. Geophys. Res,, 90, 2131-
2143.
Wetzel, P. J., and J.-T. Chang, 1988: Evapotranspiration from nonuniform surfaces: a first
approach for short-term numerical weather prediction. Mon. Wea. Rev., 116, 600-621.
Wojcik, G.S., and J.S. Chang, 1996: An Eulerian modeling study of anthropogenic sulfur
budgets and lifetimes for Eastern, North America, submitted to J.Atmos.Chem.
Wyngaard, J. and Brost, J.C. and R. A. Brost, 1984: Top-down and bottom-up diffusion of a
scalar in the convective boundary layer. J. Atmos. ScL, 41, 102-112.
Xue, M., K. Droegemeier, V. Vong, A. Shapiro, and K. Brewster, 1995: ARPS Version 4.0
User's Guide. Center for the Analysis and Prediction of Storms, Univ. of Oklahoma, 380 pp.
[Available from Center for the Analysis and Prediction of Storms, Univ. of Oklahoma,
Norman, OK 73019].
Zilitinkevich S. S., 1989: Velocity profiles, the resistance law and the dissipation rate of mean
flow kinetic energy in a neutrally and stably stratified planetary boundary layer. Bound, Layer
Met., 46, 367-387.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
__
-------
EPA/600/R-99/030
Appendix 12A MCIP Output Data
IVlCIP writes the bulk of its two- and three-dimensional meteorology and geophysical output data
in a transportable binary format using a tailored Models-3 input/output applications program
interface (I/O API) library. Depending on whether the meteorological vertical coordinate is time
dependent or not, the temporal characteristics of certain variables that belong to GRID data types
and MET data types may differ. For example, when the meteorological coordinate is ap, which is
time dependent, Jacobian is also time dependent. On the other hand for apg coordinate, Jacobian
is time independent. However we are using a consistent list of meteorological variables for both
hydrostatic and nonhydrostatic coordinates for as the standard output. Tables 12A-1 through 12A-
6 provide lists of variables in each data type.
I •.•••>.., . ..... . •: ..--..; '.; .
Table 12A1. Variables in GRIDCRO2D
Variable Name
ixs
JjON
WSFX2
HT
ZZE8Q
PRSFCO
!>!
-------
EPA/6OO/R-99/V3O
Table 12A3. Variables in GRIDCRO3D
Variable Name
DSNSO
ENTRPO
JACOBOF
JACOBOM
TEMPO
PRESO
X3HTOF
X3HTOM
Unit
KG/M**3
J/K/M**3
M
M
K
Pascal
M
M
Description
density of reference atmosphere
entropy density of reference atmosphere
total Jacobian at layer face
total Jacobian at layer middle
temperature of reference atmosphere
pressure of reference atmosphere
height of layer face (top) of reference
atmosphere
height of layer middle of reference
atmosphere
Table 12A4. Variables in METCRO3D
Variable Name
JACOBF
JACOBM
DENSA_J
DENSW_J
ENTRP_J
WHAT_JD
OS
QR
QV
TA
PRES
DENS
WWIND
ZH
ZF
JDRATE
Table 12A5. Variables in
Variable Name
UWIND
WIND
UHA.T_JD
VHAT_JD
Unit
M
M
KG/M**2
KG/M**2
J/K/M**2
KG/ (M*S)
KG/KG
KG/KG
KG/KG
K
Pascal
KG/M**3
M/S
M
M
KG/M**2/S
METDOT3D
Unit
M/S
M/S
KG/ (M*S)
KG/ (M*S)
Description
total Jacobian at layer face
total Jacobian at layer middle
Jacobian weighted total air density
Jacobian weighted density of vapor
Jacobian weighted entropy of moist air
J & Density weighted vertical contra-W
cloud water mixing ratio
rain water mixing ratio
water vapor mixing ratio
air temperature
pressure
total density of air
true W component of wind
mid-layer height above ground
full-layer height above ground
time rate change of Jaeob*Density
Description
U-comp. of true wind at dot point
V-conp. of true wind at dot point
(contra_U*Jaeobian*Density) at square
point
(contra_y*Jacobian*Density) at triangle
point
12-81
-------
EPA/600/R-99/030
Table 12A6. Variables in METCRO2D
Variable Name
Unit
Description
PRSFC
JACOBS
EENSAS
WSTAR
RIB
PBL
ZHOF
MOCX
HEX
QPX
MQLX
RMKNI
RBNDXI
RSTOMI
TEMPO
TEMP10
TEMP1PS
SUHF2
ALBEDO
FSOIL
GDW
6SH
RGRND
RNET
RN
HC
CFRACH
CFRACM
CFRACIi
CFRAC
CUOT
WEAR
. VD_KXXX
Pascal
M
KG/M**3
M/S
NODIM
H
M
1/M
WATTS/M**2
WATTS/M**2
1/M
M/S
M/S
M/S
K
K
K
EMPTY
HODIM
NA1TS/M**2
WATTS/M**2
WATTS/M**2
WATTS/M**2
«ATTS/M**2
CM
CM
Fraction
Fraction
Fraction
Fraction
M
M
G/M**3
M/S
surface pressure
total Jacobian at surface
air density at surface
convective velocity scale
bulk Richardson number
PBL height
surface roughness length
inverse of Monin-Obukhov length
sensible heat flux
latent heat flux
inverse of Monin-Obukhov length
inverse of aerodynamic resistance
inverse of laminar bad layer resistance
bulk stomatal resistance for water
skin temperature at ground
air tenqperature at 10 m
air tenqperature at 1.5 m
surface parameter 1
surface albedo
heat flux in soil layers
longwave radiation at ground
solar radiation absorbed at ground
solar radiation reaching surface
net radiation .
accumulated nonconvective hourly precip.
accumulated convective hourly precip.
fraction of high cloud
fraction of middle cloud
fraction of low cloud
total cloud fraction
cloud top layer height
cloud bottom layer height
liquid water content of cloud
dry deposition velocity for species XXXX
Note: List of deposition velocities
VD_SO2
VB_SULF
VDJH02
VDJ5JO
VTLP3
VELB2O2
VD^RLD
VD_OP
VH.ORA
VELPAN
VDJfWNO
VD_CO
"M/S"
"M/S"
"M/S"
"M/S"
"M/S"
"M/S"
"M/S"
"M/S"
•M/S"
"M/S"
"M/S"
"M/S"
"M/S"
"M/S"
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
deposition
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
velocity
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
for species
SO2
SULF
NO2
NO
03
HSO3
H2O2
ALD
HCHO
OP
ORA
NH3
PAN
HONO
CO
12-82
-------
EPA/600/R-99/030
Appendix 12B Examples of Nest Domain Definitions for CMAQ system
12B.I Grid Domain Definitions for the Models-3 CMAQ Tutorial Study
Table 12B-1. Origins for the TUTORIAL MM5 domain, MCEP landuse domain, and CTM
domains. Note that (x,y) corresponds to (COL.ROW) in CMAQ. MCEP landuse domain is
smaller by one cell around than the original MM5 domain. Also, the CTM domain removes three
cells around from the original MM5 domain. Origin is measured from the center of Base 108 km
domain located at 90W and 40N,
Resolution Origin for original MM5 Origin for MCEP Landuse Origin for CTM Domain
(Ax=Ay) Domain Domain (•*<»«'3en*)
OWJarfe) (•***•?«*)
108km
36km
12km
04km
(-3996xl03, -2808xl03)
(432xI03, -432x1 03)
(972x1 03,-36xl03)
(1200xl03, 120xl03)
N/A
(468x1 03,-396xl03)
(984x1 03,-24xl03)
(1204xl03, 124xl03)
N/A
(540xl03,-324xl03)
(1008xl03,0xl03)
(1212xl03, 132xl03)
Table 12B-2. Offset Definitions for the Tutorial MM5 Nest Domains
MM5 Resolution NCOLS NROWS Col. Offset
Domain (km) from Parent
D1
D2
D3
D4
108
36
12
4
74
36
39
36
52
36
36
36
Domain, D1
0
41
47
48.666666
Row Offset Col.
from Parent for
Domain, D1
0
22
26.666666
28.111111
Offset Row Offset
nest for nest
0
41
15
19
0
22
1 1
13
Table 12B-3. Dimensions of Set of MM5, MCEP, and CTM Domains for Each Grid Resolution
for the Tutorial Study
Resolution MM5
108
36
12
4
NCOLS MM5 NROWS MCIP NCOLS MCIP NROWS CTM NCOLS CTM NROWS
74
36
39
36
52 N/A
36
36
36
N/A
34
37
34
N/A
34
34
34
N/A
30
33
30
30
30
30
12-83
-------
EPA/600/R-99/030
12B.2Grid Domain Definitions for the ModeIs-3 CMAQ Demonstration Study
Table 12B-4. Origins for the CMAQ demonstration MM5 domain, MCIP landuse domain, and
CTM domains. Note that (x,y) corresponds to (COL.ROW) in CMAQ. MCIP landuse domain is
smaller by one cell around than the original MM5 domain. Also, the CTM domain removes three
cells around from the original MM5 domain. Origin is measured from the center of Base 108 km
domain located at 90W and 40N.
Resolution Origin for original MM5 Origin for MCIP Landuse Origin for CTM Domain
Domain Domain Conn's'Xm*)
(jc • y • ) (x • v • )
108km
36km
I 12km" '
» 04km"
(-3996xl03,-2808xl03)
(-1080xl03,-1728xl03)
(486xl03,-216xl03)
"(I140xl03,24xi63)
N/A
(-1044xl03,-1692xl03)
(480x1 03, -204xl03)
(1144xl03,28xl03)
N/A
(-864xl03,-1512xl03)
(540xl03,-144xl03)
(1164xl03,48xl03)
Table 12B-5. Offset Definitions for the CMAQ Demonstration MM5 Nest Domains
MM5
Domain
—
K
•i "'
- ,
D1
D2
D3
,D4
Resolution NCOIS NROWS
(km)
108
36
12
4
74
87
102
87
52
81
84
105
Col. Offset Row Offset Coi. Offset Row Offset
from Parent from Parent for nest for nest
Domain, D1
0
27
42.333333
48.555556
Domain, D1
0
22
25
27.222222
0
27
43
56
0
10
42
20
Table B6. Dimensions of Set of MM5, MCEP, and CTM Domains for Each Grid Resolution for
the CMAQ Demonstration Study '
Resolution MM5NCOLS MM5 NROWS MCIP NCOLS MCIP NROWS CTM NCOLS CTMNROWS
: 108
36
12
- , - 4
74
87
102
87
52 N/A
81
84
105
N/A
85
100
85
N/A
79
82
103
N/A
75
90
75
69
72
93
12-84
-------
EPA/600/R-99/030
Appendix 12C Sample MCIP Configuration File
// RCS file, release, date & time of last delta, author, state, land locker]"
// $Header: /project/work/rep/MCIP/src/sunOS5/sun_jn3_12.cfg,v 1.1.1.1 1997/10/13 17:38:23 yoj
Exp $
// what(l) key, module and SID; SCCS file; date and time of last delta:
// @(#)sun_m3_12.cfg 1.1 /project/mod3/MCIP/doc/bldrun/sunOS5/SCCS/s.sun_m3_12.cfg 14 Jun
1997 15:41:54
// This is a configuration file for
model mcip_m3_12;
// f77_flags "-e -N1200 -fast -04";
£77_flags "-e -g -C";
// link_flags "-fast -O4";
link_flags "-e -g -C";
libraries "-L${M3TOOLS}/IOAPI/src_lib/SunOS5 -Im3io \
-L${M3TOOLS}/netCDF/SunOS5 -Inetcdf";
// global { verbose | parse_only | conpile_all | one_step | clean_up } ...
global verbose;
// shared include files 1111111111111111111II111111111111111111111111111
II Models-3 I/O API permanent includes
include SUBST_IOPASMS SM3TOOLS/IOAPI/includes/PARMS3.EOT;
include SUBST_IOEDESC $M3TOOLS/IOAPI/includes/n3ESC3.EKT;
include SOBST_IODKL SMSTOOLS/IOAPI/includes/IODECLS.EXT;
// Models-3 global constants for this processor
include S0BST_CONST SMSMODEL/ICL/src/fixed/const/CONST.EOT;
// Models-3 grid definitions
include SUBST_HGRD_ID $M3MODEL/ICL/src/grid/D_12_15/H3RD_21X21.E5Cr;
include SUBST_VGKD_ID $M3MODEL/ICL/src/grid/D_12_15/V3RD.12Cr;
include S0BST_COORD_ID $M3MODEL/ICL/src/grid/D_12_15/COORD_21X21.EXT;
// Kodels-3 I/O API files for this processor
include SUBST_INFILES $«3MODEL/MCIP/src/icl/icl/INFIIiESJMCIP.E5Cr;
include SUBST_FILES $M3MDDEL/MCIP/src/icl/icl/FILES3JMCIP.E5CT;
include SDBST_«ET_CONST $M3MODEL/MCIP/src/icl/icl/OONST_mete.EOT;
include SUBST_PBL_CONST $M3MODEL/MCIP/src/icl/icl/CONST_pbl.EXT;
12-85
-------
EPA/600/R-99/030
In:. • • ' ..»< .'^mj . *-m
/I HCIP parameters
include SUBSTJffiSRM $M3MODEL/MCIP/src/icl/icl/MCIPPARM.EXT;
include SUB_ST_MCMMN $M3MODEL/MCIP/src/icl/icl/MCIPCOM.EX?T;
include SUBST.JGQOT $M3MODEL/MCIP/src/icl/icl/GROUTCCM,EXT;
include SOBSTJCXXTT $M3MODEL/MCIP/src/icl/icl/MCOUTCCM,EXT;
include SOBST_JCOOT $M3MODEL/MCIP/src/icl/icl/MDOUTCOM.EXT;
include SUBSTJjRADK $M3MODEL/MCIP/src/icl/icl/IjyU]KDAT.EXT;
uiiiiiuiiiiniiiiiiuiuiiiiiiiiii/iiiiiiuiiiuiiiiiiiiiiiiiiiiiniii
jii ' , . , ,0 (,i § , . ;.
// the rocip driver class
module rocip development;
// the landuse class
module mSradm development;
// the input class
"... module m3 development;
// the met3d class
module m3aup development;
// the pbl class
- module pblpkg development;
11', .,-, •.
// the drydep class
n module ractadd development;
// the cloud class
module radmkuo development;
// the util class
module util development;
// the output class
module stnd development;
12-86
-------
BPA/600/R-99/030
Appendix 12D Sample MCIP Run Script
#! /bin/csh -f
# RCS file, release, date & time of last delta, author, state, [and locker]
# SHeader: /project/work/rep/MCIP/src/s»«OS5/sunjn3_12.q,v 1.1.1.1 1997/10/13 17:38:23 yoj
Exp $
# what(l) key, module and SID; SCCS file; date and time of last delta:
# @(i)sun_m3_12.q 1.2 /project/mod3/MCIP/doc/bldrun/sunOS5/SCCS/s.sun_?n3_12.q 19 Jun 1997
08:26:34
# for Sun Spare 20, UltraSparc 2
# method: sun_m3_12.q >&! sun_m3_12.log &
f QSUB -r sun_jn3_12
t QSUB -eo -o /work/you/your_directory/sun_jn3,log
# QSUB -1M 8Mw
# QSUB -IT 1800
date; set timestamp; cat sunjB3_12.cfgj echo " "i set echo
setenv M3HOME /home/models3
setenv M3DATA $M3HQME/datasets
t set executable
*
set EXEC = mcip_jn3_12
tt __ _ __ _ ____ ri_i_ ___ __„ ».«._
f set base directory
#set BASE = /work/you/your_directory
set BASE = $cwd
« _____ «^™ _____ _ __ j.^.^™^ _____
# set working directories
set MCIPDIR = $BASE
set OUTDIR = $cwd
# set input data directories
set DATADIR = $M3HOME/datasets/nostudies
12-87
-------
EPA/600/R-99/030
I' •: ..ft
f (METROW,METaDL,METLAY) is (ROWS, COLS, LAYS) for MM5 data
m MM
setenv METRCW 25
setenv METCOL 25
satenv METLAY 15
§.«.»»._»....-._««.. ».__.._«..__« . —_. —,——. — «.
t (10, JO) : The left bottom loc. of CTM domain in MET terms
i In CTM terms, 10 = ROWMOFFSET, JO = COL_OFFSET
ft~ — — _ — ..««._—»«___««_« «—— — «.•—«—•._._ _..««_ .__.....__»..__.__«._._.._«_...._»
setenv 10 1
setenv JO 1
LUTYPE: Filetype for Landuse data:
1: BINARY 13 category directly from TERRAIN
2: Preprocessed ASCII 11 category
3: Use MM5 internal dominant landuse category
4: Use M3 landuse fractions available in GRIDCRO2D
setenv LUTYEE 4
*„_..____.._..._...„__ .._«.«,.. _„.- . _, .— .« ^ —
I (IOLUSE, JOLUSE) s The left bottom loc, of CTM domain in
I [ lAND USE DATA GRID terms
»_ _«™_^«™_« ..*._ _» , — , —. »
setenv IOLUSE 1
setenv JOLUSE 1
^ • *
§«._»«.»«—..__.»_»._ MM» — — —... :«.~~
I If LCALCPBL is: TRUE, Recompute PBL values
• FALSE, Read PEL values from MK52
f «..,-.. _^..«.«» » « «. «« «™ . «^ —,
a«t«nv Ix^iLCPBL T
§.*....,««.«_.._..»«.** — ™^».—.-..w*™—.-. ^— «^™ , .~ —— ~— «-««. «— , ,
f If LSLFLUX is: TRUE, Use only surface layer similarity
I FALSE, Use PEL similarity
§.«*..-_,«_»._...__.«.« ._»«..«_ »— «. «. — —
setenv LSLFLUX F
»->M_ «._»._ _.„,-_ ». — „ ._.^w^, — „. .«. _—««.—««.
f If LM3ODEP is: TRUE, Use Models-3 D. Dep routtine
I FALSE, Use RKEM. D. Dep. routine
«w.^,_ «,««_«_«._«.-.__™^M«™__».« _..._„ ,-„_—», .. , HMW__ « __«,
uetenv LM3DESP F
i If LCALCCLD is: TRUE: Recompute CLOUD values
12-88
-------
EPA/600/R-9ft/030
# FALSE: Read CLOUD values from MM52
ji__ _ _ ^ _ _^ _
setenv LCALCCLD T
u ___ _»_.. ______ _ «_.._«._...._....__.*_.._____..__...._______..______..___.._
# If LSANITY is: TRUE: check range of output parameters
# FALSE: do not check range of paramters
# -----------------------------------------------------------
setenv LSANITY T
# JUDATE : INITIAL JULIAN STUDY DATE (YYDDD)
# HSTRT : STARTING HOUR OF JUDATE FOR STUDY
# HTORUN : NUMBER OF HOURS TO USE FROM MM5
# HTOSKIP: NUMBER OF HOURS TO SKIP FIRST
#
# 88218 00 0 2 ! John's data
# 88216 12 0 49 ! Dave's data
# 88214 12 10 108 ! Daewon's data
setenv JUDATE 88209
setenv HSTRT 00
setenv HTOSKIP 00
setenv HTORUN 25
#setenv HTORUN 108
#
# input file linkage
# MM51: Filename for MM51 datafile
# MM52: Filename for MM52 datafile (NA if none)
# LU13 : Filename for LU13 datafile (binary from TERRAIN)
4* _____________ __ — _.. _____ — _.__— ._ — -._________.-_______.._ — ___ _ __
#setenv MM51 $INDIR/MM51_es80
#setenv MM52 $INDIR/MM52_es80
#setenv LU13 SINDIR/LU80
tsetenv LUSE $DATADIR/LU11_80
#setenv LUSE /work/bdx/metpp/newmcip/input/LUll_80
# number of iterations for vertical wind correction
# (Use 0, +/- 1,2,3,..)
JA_>_ ____ __„__________.._„..._____.._..„._..__..__.,_.._..___ ___ ___
setenv IWIND 0
# number of deposition layers (default = 1)
# (for layer collapsing, it could be 2, or 3)
#___! -----------------------------------------------------
setenv NDEP 1
12-89
-------
EPA/600/R-99/030
ft «*•.«.*.».«.***.«« — «. — — — __ «__«_.-- M « w. «-*»«._««_.__.
I GROWM: Grid Name (User Defined)
GROK&M
f EXBCOTIONUID {User Defined)
ftw«WM«»»«.~Wm^l.._*l»4MIM.MW^
BCtcnv EXECUMOH.1D MCIPJETA
i SCMFILE: file path (User Defined)
1——W— »««_«— _«—^^™__™__——^^w«»,*™__M__«
ls«tenv SCENFXLE
I If IQAPI_CHECKJffiADERS is: TRUE, Check the headers
I FALSE, Do NOT check headers
£.«.*»**.-.*.-....- ______ ___«., ___ * ____ «_*.».-._.—._ ___ MMM,MU.»_. __ .._*.. __ *._..-
SftteiW IOAP3L_CHEt3OEADERS F
tsctenv L06FIL5 mcip.log
I Remova any previously generated output files
f •»«« »«,™— ——™— ——»—««».————-.—— ™—— «.«.-.—————— —™™—™«,-.— ——•«
f /bin/rm GRID_* MET_*
f/bin/rm . . /output /GR!D_* . . /output/MET_*
f Set up input files
' - L. . %
i. , :f ,t
Stttenv GRID_eHO_2I> $DATAOia." /GRIDCRO2D_TOT12'
satenv GRXD_POT_2D $DATADIR'/GRIEOOT2D_TUT121
setenv MET_CRO_2D $DATA0IR' /METCRO2DJTOT12 '
setenv MET_CKb_3D $DATADIR'/METCR03D_TUT12'
setenv MET_POT_3D $nR3
-------
EPA/60Q/R-99/030
setenv HET_CRO_2D_G1 $OUTDIR'/METCRO2D_tutl2'
setenv MET_CRO_3D_G1 $OOTDIR'/METCRO3D_tutl2'
setenv MET_DOT_3D_G1 $OOTDIR'/METDOT3D_tutl2'
« .
# Execute MCIP
# ™™_™«____™_™^__^™_« „.„_.»_.„._,,______._._„..„_.„—.____..
#cdbx "$EXEC -g -x"; exit
i ja
i/opt/SUNKspro/bin/aebugger «$MCIPDIR/$EXEC'; exit
$MCIPDIR/$EXEC
I ja -cshlt
exitO
ji ™mm™_____«._»™__™__ ™™™™—-»»,~«.______™_
# Move output to proper directory
u __,™___™.™_____™._~.~m.— ™™__-,™™_m™._____™™™___™_
#/bin/inv GRID* . ./outputl/
#/bin/iw MET* ../outputl/
12-91
-------
EPA/600/R-99/030
Chapter 13
THE INITIAL CONCENTRATION AND BOUNDARY CONDITION PROCESSORS
Gerald L. Gipson*
Human Exposure and Atmospheric Sciences Division
National Exposure Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711, USA
ABSTRACT
i
This chapter describes the two processors in the Community Multiscale Air Quality (CMAQ)
modeling system that generate the initial concentrations and boundary conditions that are required
by the CMAQ Chemical Transport Model (CCTM). The major emphasis is on the functionality
incorporated in those processors - the initial condition (ICON) and boundary condition (BCON)
processors. The chapter describes how each processor can be used to generate initial
concentrations and boundary conditions from one of two sources - from predefined default
vertical profiles or from other CMAQ simulation results when model nesting is being performed.
A description of generating initial and boundary concentrations for special tracer species is also
included. The procedures that are used in both processors for interpolating in both the horizontal
and vertical directions are described. A description of methods for converting initial and
boundary concentrations from one chemical mechanism form to another is also presented. Finally,
the manner in which each processor is applied within the Models-3/CMAQ framework is
addressed.
Corresponding author address: Gerald L, Gipson, MD-80, Research Triangle Park, NC 27711. E-mail:
ggb@hpcc.epa,gov
-------
EPA/600/R-99/030
13.0 THE INITIAL CONCENTRATION AND BOUNDARY CONDITION
PROCESSORS
13.1 Introduction
The solution of partial and ordinary differential equations that arise in air quality models requires
both initial concentrations (ICs) and boundary conditions (BCs). This chapter describes the
Community Multiscale Air Quality (CMAQ) modeling system processors that generate these
quantities. In the discussions that follow, they will be referred to as the ICON and BCON
processors, respectively. The ICs and BCs generated by these two processors are used by the
CMAQ Chemical Transport Model (hereafter referred to as the CCTM). For ICs, the CCTM
requires that the concentrations of all model species within each grid cell in the modeling domain
be specified for the start of the simulation. Concentrations of all model species at the boundaries
of the modeling domain throughout the simulation are required as BCs to the CCTM. The
emphasis in this chapter is on how the ICs and BCs are generated rather than on how they are
used in the CCTM. The reader is referred to Chapter 5 for a discussion of the latter.
' * ' : "- ' ' ' ' • • • ' •
ICON andjBCON processors have been designed to process data as automatically as
po'ssible. Users should be aware of certain operational assumptions that are employed by the two
processors and the CCTM to insure that their applications are performed in the intended manner,
however. The discussion below begins with a brief overview of the processors and their
relationship to the CCTM. Several topics that address the various aspects of the IC/BC
processing are then presented, including the input sources, spatial interpolation, species
processing, and mechanism conversions. The last section provides a brief overview of how the
ICON and BCON processors are applied.
13.2 Overview of the ICON and BCON Processors
i .• €>j f • , • r ., . ,
The ICON and BCON processors generate ICs and BCs for individual model species, which
include gas-phase mechanism species, aerosols, nonreactive species and tracer species. The
techniques for selecting and processing species data will be discussed in more detail below. It is
important to emphasize at the outset, however, that it is not necessary for the ICON and BCON
processors to generate ICs and BCs for every species that is being modeled. The CCTM will
attempt to extract ICs and BCs for each species being modeled from the input files. If a species is
not found on the file, the CCTM will automatically set its ICs and BCs to a minimum threshold
limit (i.e., a nominal zero). Any data on an 1C or BC file for a species that is not being modeled
will simply be ignored by the CCTM. Therefore, the user is encouraged to check the ICON,
BCON, and CCTM output logs to insure that the ICs and BCs are being set in the intended
manner.
The ICON processor generates species concentrations for every cell in the modeling domain,
whereas the BCON processor generates species concentrations for the cells immediately
surrounding the modeling domain. At present, the thickness of the boundary cells is limited to
13-2
-------
EPAS600/R-99/030
one, which is the thickness required for all the transport algorithms currently included as part of
the CCTM (e.g., see Chapter 7). The reader is referred to Chapters 6 and 7 for a description of
the CMAQ conventions used in establishing modeling domains and their boundary cells.
Both the ICON and BCON processors write the output ICs and BCs to standard Models-3
Input/Output Application Programmer Interface (I/O API) files: 3-dimensional gridded files for
the ICs and 3-dimensional boundary files for the BCs, The opening, formatting, and writing of
these files are handled automatically by the ICON and BCON processors. It should be noted,
however, that the IO/API has a limit of 120 output variables (i.e., number of species) for any one
IO/API file. If the number of output variables exceeds the IO/API maximum, the ICON and
BCON processors will generate up to two additional files to hold the outputs. If more than 360
output variables are needed, however, code modifications will need to be made to the ICON and
BCON processors and the CCTM to accommodate the additional variables.
13.3 Input Sources
The ICON and BCON processors generate 1C and BC files from one of three input sources. The
first is a time invariant set of vertical concentration profiles. The second source of input data
consists of existing Models-3 IO/API 3-dimensional concentration files, normally generated by the
CCTM. The final source of 1C and BC data involves the generation of special tracer species
concentrations used to test numerical transport algorithms. Each of these will be discussed
below. It should be added, however, that the ICON/BCON processors can be bypassed if the
user desires to use a different method. The only requirement is that the 1C and BC files that are
eventually fed to the CCTM must be standard Models-3 IO/API gridded concentration and
boundary concentration files. The reader is referred to IO/API documentation for a description of
the structure and format of those files (EPA, 1998).
13.3.1 Time Invariant Concentration Profiles
This section describes the manner in which the ICON and BCON processors generate ICs and
BCs from predefined, vertical concentration profiles. The CMAQ system contains a set of
predefined profiles that can be used to generate the requisite ICs and BCs. These profiles give
species concentrations as a function of height, and are spatially independent for the ICON
processor and only minimally spatially dependent for the BCON processor. Both the ICON and
BCON profile data are time independent. Since these data are not highly resolved, they are
typically used when no other information about ICs and BCs is available. It should also be noted
that the predefined profile files can be replaced by user developed profile data provided that the
replacement files are in the same format as the predefined files. The discussion below begins with
a description of the predefined profile data, and is followed by a description of how the user can
prepare replacement data.
The default 1C and BC profiles have been developed for the RADM2 chemical mechanism and the
terrain following sigma-p coordinate system (see Chapter 6 for a discussion of CMAQ coordinate
13-3
-------
EPA/600/R-99/Q30
. Bf- ' - ' .'~!l j» • - ' •; .." i-. T
SI m.ff ? , • • ' . »
types). Both the 1C and BC profiles are intended to represent relatively clean air conditions in the
eastern-half of the United States, and have been formulated from available measurements and
results obtained from modeling studies. Thus, they do not represent any specific time period.
Although profiles for other chemical mechanisms and vertical coordinate systems are not currently
available in the CMAQ system, it may still be possible to use the predefined profiles for these
cases. Special procedures are included with the ICON and BCON processors to allow conversion
from one vertical resolution and/or chemical mechanism to another. These procedures will be
described in sections 13.4.2 and 13.6, respectively.
The predefined profile data are stored in ASCII files in the CMAQ system. Table 13-1 contains a
listing of the CMAQ predefined 1C profile file. Note that line numbers have been added to the
listing for reference. The first three lines contain informational text that is not used by the ICON
processor. Line 4 contains the number of vertical layers and the number of species for which
profiles have been developed, followed by the vertical coordinate values of the sigma-p levels.
The, fifth line is blank and is ignored by the ICON processor. Lines 6 through 46 contain vertical
profiles for 43 different RADM2 gas-phase species. The species name is given in the first column,
followed by mixing ratios (in ppm) for each layer. The first entry for a species corresponds to the
lowest layer, whereas the last entry corresponds to the uppermost layer. Lines 47 through 52
cojtfain analogous information for key aerosol species. The units for ASO4I, ASO4J, and ASOIL
arf ug/m3, and the units for NUMATKN, NUMACC, and NUMCOR are number/m3.
*•> w? ~ - ,
Table 13-2 contains a listing of the corresponding predefined BC profile file. The format for this
file is similar to that for the 1C file, except that one set of profiles is input for each of the four
different boundaries of the modeling domain (i.e., north, east, south and west). The BCON
processor uses the appropriate set of profiles to generate the BCs at each edge of the modeling
domain. Again, line five is blank and is ignored by the BCON processor.
As indicated above, the user can replace a CMAQ predefined profile file with one of their own
provided the same file format is used. As noted in the discussion above, none of the first three
text lines nor the fifth line in either file is used. Thus, those lines can contain any information or
be blank, but they must be present in the file. The remaining data lines are read in FORTRAN
list-directed format, so it is not necessary to align the data by columns. The only formatting
requirement is that all inputs be separated by commas or by one or more blanks. The species
names must not exceed 16 characters and must not contain any blanks. Also, it is not necessary
that a six-layer vertical resolution be used, but at least one layer must be present and no more than
30 are allowed. Note that the number of layer heights on line 4 must be one greater than the
actual number of layers used. These layer heights must correspond to the heights of the layer
interfaces (i.e., not the mid-layer heights), and the first entry for any profile must be for the lowest
level (i.e., ground-level). The input mixing ratios, however, are interpreted as layer average
values as opposed to values at the layer interfaces. Finally, note that the BC vertical profile data
for each edge begins with a line that contains the edge name. Although the processors do not
require that these names be specified explicitly, a line corresponding to the edge name line must
13-4
-------
EPA/600/R-99/030
be included and the processors expect the data for each edge to be input in the order shown in
Table 13-2 (i.e., north, east, south, and west).
13,3.2 CCTM Concentration flies
Both the ICON and BCON processors can be configured to generate ICs and BCs from existing
Models-3 IO/API 3-dimensional, gridded, concentration files. This situation typically arises when
performing a nested model simulation and modeling results are available for a coarser grid.
Unlike the predefined profile files described above, the CCTM concentration files are both
temporally and spatially resolved, necessitating some additional processing procedures. (The
subject of spatial interpolation is described in section 13.4.) It should be added, however, that it
is not necessary to use the ICON processor for a model continuation ran (i.e., a ran in which a
model simulation is started at the exact same time that a previous one ended and the same
modeling domain is being used). In such cases the CMAQ IO/API concentration file from the
previous run may be used directly as the input 1C file to the CCTM. However, it will still be
necessary to generate a BC input file.
Since ICs are required only for the start of a model simulation, the ICON processor will generate
ICs for only one time step. BCs are required throughout a simulation, however, so the user must
prescribe the period of record for which to produce outputs. One set of BCs will be generated at
each time step for that period of record. To illustrate, consider the case in which a model
simulation has been performed for a coarse grid for a 24-hour period of record using 1-hour time
steps (i.e., the model concentrations were predicted at hourly intervals). The IO/API gridded
concentration file for the coarse grid will then contain species concentrations for 25 time steps,
one for the start of the simulation (i.e., the ICs for that simulation) and 24 time steps
corresponding to the model predictions at each hour. Now assume that it is desired to generate
ICs and BCs for a nested model simulation for the same time period. The user would select the
first time step for the ICON output, and the ICON processor would generate ICs for that time
step only. For the BCON processor, the period of record for the output BCs would be obtained
by selecting the same starting time and setting the duration for outputs at 24 hours. This selection
would produce 25 time steps on the output file, one for the start and 24 for each of the
subsequent simulation hours. The manner in which the time controls are set is described in the
Models-3 User Manual (EPA, 1998).
13.3.3 Tracer Species
The ICON and BCON processors also contain the capability to generate ICs and BCs for special
tracer species that can be used to investigate the accuracy of the various transport and diffusion
algorithms available in the CCTM system. This capability has been included primarily to facilitate
the evaluation of alternative algorithms, and thus would not likely be used in most air quality
modeling applications.
13-5
-------
EPA/600/R-99/G30
ICs and BCs can be generated for eight different tracers. Each one can be used to examine
different properties of transport algorithms, and thus can be selected individually for output. They
are chosen by including specially defined tracer names in the list of model species. The defined
names and a brief description of each tracer species follow. (Note that the ICs for these tracers
can be viewed with one of the visualization tools to examine the patterns more closely.)
• IC1_BCO: All ICs in the modeling domain are set to 1.0 and all BCs are set to 0.0.
• 1C 1_BC 1: All ICs in the modeling domain are set to 1.0 and all BCs are set to 1.0.
t- J: I • . . •: . , -, _.- ^ ;-_•*-i . .' .' <
• * ICOJBC1: All ICs in the modeling domain are set to 0.0 and all BCs are set to 1.0.
• STREET: Within any horizontal cross-section of cells, 1C cell concentrations are set to
either 1.0 or 0.0 such that the overall domain pattern resembles a street grid. All BCs are
set to zero.
• CHKBRD: Within any horizontal cross-section of cells, 1C cell concentrations are set to
either 1.0 or 0.0 such that the overall domain pattern resembles a checkerboard. All BCs
are set to zero.
• SPOS_A: ICs and BCs are set such that a concentration mound is centered at cell (10,10)
in the modeling domain, and the concentration profile is defined below.
• SPOS_B: ICs and BCs are set such that a concentration mound is centered at cell (10,10)
in the modeling domain, and the concentration profile is defined below.
• SPOS_C: ICs and BCs are set such that a concentration mound is centered at cell (10,10)
in the modeling domain, and the concentration profile is defined below,
*.. °-
(y-yf
The last three tracer species are designed for superposition tests of transport algorithms. The
base shape of the concentration mounds is described by the Witch of Agnesi surface: where a is
the radius of the mound, xr and yr define the position of the peak of the mound, and hs is the
height of the mound. In the ICON and BCON processors, a is set to 3 and (xr, yr) to (10,10).
13-6
-------
EPA/60Q/R-99JQ30
The three tracer signals hA, hB, and hc for SPOS_A, SPOS_B, and SPOS_C are then defined as
follows:
h. - a ( 1 + h ) •»• q_. , (\3-2a)
A "max v *' "mm ' \* -J +-&J
(13-2b)
hC = -fmaxO + A,) + 2 (13-2c)
where fmax and fmjn determine the amplitude and background values of the signals, and both are
set to 50 in the ICON and BCON processor. With these signals, the addition of mounds B and C
will yield mound A. Thus, the degree to which HA - (hB+ hc ) differs from zero in an advection test
provides a measure of the nonlinearity of the advection algorithm.
13.4 Spatial Interpolation
Both the ICON and BCON processors are designed to account for differences in spatial resolution
between the input files and the IC/BC output files on an automatic basis. As described in
Chapter 6, the CMAQ system is designed to handle different types of horizontal and vertical
coordinate systems. All of the data required to completely specify the horizontal and vertical grid
structure are contained in three INCLUDE files that are generated by the ModeIs-3 framework:
COORD.EXT, HGRD.EXT, and VGRD.EXT. When the CCTM is compiled (or built), these
three files are included with the source code, thereby providing the CCTM with all of the
necessary information on the grid structure that is being used. Both the ICON and BCON
processors operate in a similar fashion, and thus both processors must be built with the same three
INCLUDE files that are used to build the CCTM that will eventually use the ICs and BCs. Grid
information on the IC/BC inputs is extracted from the input files themselves, so no additional
information is required for them. Note that either the horizontal and vertical coordinates for the
input file may be of a different type than that for the output file, or they may be of the same type
but have different extents or resolutions.
13.4.1 Horizontal Interpolation
Horizontal interpolation is required when a CCTM concentration file is being used as input to
either ICON or BCON and the horizontal coordinate system of the input file differs from that
needed for the output file. Such cases typically arise when ICs and BCs are being generated for a
nested modeling domain that has a finer resolution than a coarser, outer domain. Note that no
horizontal interpolation is necessary if both the input and output domains have the same structure
or if the time invariant profiles are being used to generate the ICs and BCs.
Disparities in horizontal grid structure are handled by mapping each vertical column of output
cells to a corresponding column of input cells. The vertical concentration profiles corresponding
13-7
-------
EPA/600/R-99/030
to the input column of cells are then used as the vertical profiles for the output column of cells.
Note that some vertical interpolation may also be required as described in the next section. The
horizontal mapping is done on the basis of cell proximity. The ground-level latitudes and
longitudes at the center of each vertical column of cells are first calculated for both the input and
output domains. These computed values are then used to find which input column is closest to
each output column. Since the ICON and BCON can compute latitude and longitude for any of
the available CMAQ map projections, the horizontal grid coordinate of the input and output files
do not have to be the same type.
13.4.2 Vertical Interpolation
In,some applications, the ICON and BCON may be required to generate ICs and BCs for a
vertical grid system that is different from the one used for the input file. The CCTM is capable of
handling several different vertical grid structures (e.g., sigma-p, sigma-z, pressure, etc.), and each
of those grid types can have different vertical resolutions (i.e., different numbers of vertical layers
or different spacings with the same number of layers). If the vertical structure of the input and
output grids is identical (i.e., both are of the same type and have the same vertical resolution),
then the processing of input data to output data is routine because of the one-to-one
correspondence of the vertical layers. If the vertical grid types or resolutions are different,
however, the ICON and BCON will convert from the input to the output vertical structure in
, ;, «;,i[ , I,) I, •• '- , . . . . . . . " , . " .. .
most cases. Conversions from one resolution to another can be performed using either profile data
of CCTM concentration files, but conversions from one grid type to another can only be
performed when using CCTM concentration files. Both the ICON and BCON processors assume
that the vertical grid-type of the profile data is the same as the required output structure, and thus
no conversions can be performed in that case. The two different vertical interpolation cases that
can be performed by the ICON and BCON processors are described next.
Identical Vertical Grid Types. Different Resolutions. When the input and output vertical grid
types are the same but vertical resolutions are different, both ICON and BCON will derive ICs or
BCs by linear interpolation using the layer heights contained in the input files and those contained
in the COORD.EXT INCLUDE file. These heights are in the units of the vertical coordinate
system (e.g., sigma-p, sigma-z, pressure, etc.) rather than height above ground level as measured
in distance units such as feet or meters. The vertical resolution of the input IC/BC data can be
either greater than or less than the output resolution. Linear interpolation is performed using
layer heights that correspond to the middle of the layers. Output concentrations are not always
interpolated for the bottommost and topmost layers, however. Output concentrations for all
layers with mid-layer heights less than the bottommost mid-layer height in the input file will be set
equal to the concentration of that bottommost layer. Similarly, output concentrations for all layers
above the input topmost mid-layer height will be set equal to the concentration of the topmost
layer.
Different Vertical Grid Types. A complete switch of vertical coordinate types can be carried out
when the following two conditions are met: 1) the input IC/BC data are in standard Models-3
>' I I ! . ; " v
' ;"> ' | ." : [• ' : • 13-8
-------
EPA/600/R-99/030
IO/API files and 2) the gridded 3-dimensional meteorological files that contain the mid-layer
heights for both vertical coordinate systems are also available (see Chapter 12 for a description of
these meteorological files). When these conditions are met, the ICON and BCON processors will
use linear interpolation to convert from one coordinate system to another. The interpolation is
performed in exactly the same manner as described in the previous paragraph, except that the mid-
layer heights are taken from the 3-dimensional meteorological files which give the mid-layer
heights in meters above ground level for both the input IC/BC data and the output IO/API files.
Thus, the meteorological data files containing the layer heights in meters above ground level must
be available for both the input and output grids before the ICON and BCON processors can be
used.
13.5 ICON/BCON Species Processing
The selection of species that are to be modeled and thus require ICs and BCs is controlled by the
Program Control Processor (PCP). A detailed description of that processor is contained in
Chapter 15, and will not be repeated here. A brief overview of some of the relationships between
the PCP and the ICON/BCON processors is presented here to facilitate the description of species
processing, however.
The species that are to be modeled in a CCTM simulation are determined when the user invokes
the PCP in the Models-3 framework to select a gas phase chemical mechanism, whether or not
aerosols are to be included, and whether to include other nonreactive and/or tracer species. The
Models-3 framework generates four INCLUDE files that contain the names of all species that are
to be modeled - GC_SPC.EXT, AE^SPC.EXT, NR_SPC.EXT, and TR_SPC.EXT (collectively
referred to below as the SPC INCLUDE files). The Models-3 framework also allows the user to
invoke special treatment of the ICs and BCs for individual species by defining special surrogate
names and providing corresponding scale factors for the calculation of ICs (EPA, 1998). The
Models-3 framework generates four INCLUDE files that contain this information —
GCJCBC.EXT, AEJCBC.EXT, NRJCBC.EXT, and TRJCBC.EXT (collectively referred to
below as the ICBC INCLUDE files). Note that some of these files may not contain any entries
depending on the user's selection, but all files will be generated by the framework. Just as both
the SPC and ICBC INCLUDE files must be included when the CCTM is built, they must also be
included when the ICON and BCON processors are built.
The SPC INCLUDE files contain the names of all species to be modeled and will therefore require
ICs and BCs in the CCTM. The ICON/BCON processors use the information in both the SPC
and the ICBC INCLUDE files to determine which species data should be extracted from the input
file and be subsequently written to the output file. The ICON and BCON processors also include
an option to convert from one chemical mechanism to another. Since it represents a special
option with its own unique rules, it is described separately in section 13.6. The subsequent
discussion focuses on the treatment of individual species for two cases - one with and one without
associated surrogate names.
13-9
-------
EPA/600/R-99/030
Consider first the case in which a surrogate name is not specified. This situation occurs when a
model species (i.e., one whose name is included in one of the SPC INCLUDE files) does not
have a surrogate name assigned to it. In this case, the ICON and BCON processors simply check
to see if the input file contains data for a species with the exact same name as the model species
name. If none is found, ICs and BCs are not generated for that species. If a match is found, the
data for that species are extracted from the input file, processed, and written to the 1C and BC
output files.
The assignment of a surrogate name to a particular species complicates the processing slightly.
Surrogate names in the CMAQ system arise at least in part from the generalized treatment of the
chemical mechanisms and the fact that species names are not "hardwired" in the CMAQ system.
Their primary function is to link a model species with data that are associated with a species of a
different name. Two examples are presented to illustrate potential uses.
IT i1-: 1 iii J -:- .£- v. ,» , • rf;;5" •*/• , :
, The first example illustrates how surrogates can be used to provide simple links from one species
to another. First, suppose a series of CMAQ model simulations had been performed using the
CMAQ system standard RADM2 mechanism in which formaldehyde is named HCHO. Thus, all
of the data for that species will be associated with the name HCHO in the CMAQ system data
files. Now consider that a researcher wants to make use of that data but will be using their own
version of the RADM2 mechanism in which formaldehyde is named FORM. To use the older
modeling data, the researcher could either change the formaldehyde name in the chemical
mechanism to conform to the original RADM2 name or assign the surrogate HCHO for FORM.
In the latter case, both the ICON/BCON processors and the CCTM are designed to extract IC/BC
data using the surrogate name in place the model species name. This will be explained in more
detail further below.
The second example of using a surrogate name illustrates how scale factors can be used with
surrogates. Suppose a researcher is experimenting with a new chemical mechanism which divides
a particular species PAR into two new species, PAR.1 and PAR2. Further, suppose the researcher
wants to use existing CMAQ model simulations with the PAR species to generate ICs and BCs
for both PAR1 and PAR2. The researcher will assume that 25% of the PAR IC/BC
concentrations will be assigned to PAR1 and 75% to PAR2. To effect this in the CMAQ
modeling system, the researcher can assign the species surrogate name PAR to both PAR1 and
PAR2, and assign 0.25 to the ICBC scale factor for PAR1 species and 0.75 to the ICBC scale
factor for PAR2. The ICON/BCON processors and CCTM will use this information to compute
the proper ICs and BCs for the two species PAR1 and PAR2 from the input PAR values. The
procedures followed by both the ICON/BCON and the CCTM are described next.
The introduction of surrogate names for 1C and BC species affects the manner in which both the
ICON/BCON processors and the CCTM operate. Consider first the processing that takes place in
the ICON/BCON processors. Each species in the SPC INCLUDE files is selected for processing
individually. If a surrogate name is assigned to that species, the IC/BC input files are examined to
see if the surrogate name is included with the inputs. If the surrogate name is on the input file, the
... - . 13-10
-------
EPA/600/R-99/030
data corresponding to that name are extracted from the input file, processed and output under the
surrogate name. Further, the ICBC scale factors are not applied in the ICON and BCON. If the
surrogate name is not found on the input data file, the ICON/BCON processors will then check
the input file using the model species name as if no surrogate name had existed. Again, if a match
is found using the model species name, the corresponding data are extracted and output under the
model species name. If the input file does not contain any data under the surrogate name or the
model species name, then no IC/BC data are generated for that particular model species.
Now consider processing in the CCTM when surrogate names are used for ICs/BCs. The CCTM
processes each species in the SPC INCLUDE files individually. If a surrogate name is specified
for the model species and it is found on the IC/BC input files, the CCTM extracts the data for that
name, applies the scale factor to those data, and then assigns the scaled concentrations to the ICs
and BC concentrations for the model species. If the surrogate species name is not found on the
input IC/BC files, the input file is checked using the model species name contained in the SPC
INCLUDE files. If data for that model species are found, then the data are extracted and used
directly as ICs and BCs (i.e., no scale factors are applied). If data for the species are not found on
the input file, the ICs and BCs are set to the lower threshold limits.
For most routine applications, users need not concern themselves with surrogate mapping issues.
The ICON and BCON processors are designed to generate the appropriate ICs and BCs with
minimal user interaction, provided the correct input files and include files are used with the
processors. Both the ICON and the BCON processors generate a species map on the output log
that reports the relationship between the input and output 1C and BC species. The user can check
these logs to insure that the desired species mapping has been performed. Further, the CCTM's
output log lists those species for which the lower threshold limits are used, and these should be
checked to insure that the ICs/BCs for an important species have not been omitted.
13.6 Mechanism Conversions
The vertical profile data and the CCTM concentration file data that were described above
normally correspond to one gas-phase, chemical mechanism. For example, the predefined,
vertical profile data presently in the CMAQ system were developed for the RADM2 chemical
mechanism. The capability to convert from one gas phase chemical mechanism to another has
been included in the ICON and BCON processors, however. It should be emphasized that there is
not always a definite correspondence between chemical mechanism species, and it is usually
necessary to make several assumptions to make such conversions. This is particularly true for
organic species that may be defined differently in two mechanisms. Nevertheless, some simple
mechanism conversion capabilities have been included for those cases in which approximations are
acceptable for converting from one mechanism to another.
A special routine to generate ICs and BCs for the CB4 chemical mechanism from RADM2 gas-
phase mechanism inputs is included as part of the ICON and BCON processor options. At
present, no other "hardwired" conversion routines are available in the system, but it is conceivable
13-11
-------
EPA/6QO/R-99/030
that some could be added in the future as more chemical mechanisms are added to the CMAQ
system. No conversion from CB4 to RADM2 format has been included because the carbon bond
lumping technique makes it difficult to disaggregate mechanism species into their original
chemical compounds. For a description of these two mechanisms, the reader is referred to
Chapter 8. The hardwired RADM2 to CB4 conversion procedures that are used are the same as
for the generalized conversion processor that is described next.
As a convenience for the user, a generalized mechanism conversion processor has also been
included in both the ICON and BCON processors. This conversion procedure requires the user
to generate an ASCII input file that contains mechanism conversion rules. Thus, the user is
responsible for defining how the mechanism conversion is to be performed. An example file
showing the conversion from RADM2 to CB4 is included in Table 13-3. The file contains the
expressions that are used to convert the ICs and BCs from one mechanism form to another. The
ICs and BCs for the CB4 species (i.e., those on the left hand side of the equal sign) are computed
from the RADM2 species concentrations (i.e., those on the right hand side of the equal sign)
according to the equations shown. As is evident, there is a simple one-to-one correspondence for
most species and lumping only occurs for relatively few species. It should also be noted that no
conversions will be performed for those species that are not included in the input file, and thus no
ICs or BCs will be generated for them. As mentioned previously, the user is advised to review the
species map on the ICON/BCON output log to insure that the conversions are being done as
intended. Finally, it should be added that the mechanism conversion option applies only to the gas
phase chemistry species. Aerosol, nonreaetive and tracer species will be processed in the standard
way, regardless of whether a mechanism conversion is being performed.
The format for the mechanism conversion rules is relatively simple. Species names must be 16
characters or less and not contain any blanks. Only one species is permitted on the left-hand side
of the equation, and the species name should correspond to a name in the output mechanism.
The equal sign must be present, and at least one species name must appear to the right of it.
Coefficients for species are allowed on the right-hand side only, and they can be in either integer,
real, or exponential format. They must always be followed by an asterisk to indicate
multiplication. For expressions involving lumping of species, only addition is allowed and it is
indicated by a plus sign. Line-wrap is allowed so that a single lumping rule can span more than
one line. Each conversion rule must conclude with a semicolon, and the last command must be
"END;". Finally, the input files are read in a free form input that generally ignores white spaces
between species names, coefficients, operators and line terminators.
13.7 ICON/BCON Applications
This section provides an overview of how the ICON and BCON processors are applied in
practice. Applications are normally carried out in two steps: compilation (or building) and
execution. As mentioned earlier, when the ICON and BCON executables are built, the grid
INCLUDE files for the modeling domain must be included. In addition, the user must select four
modules that determine the ICON/BCON processing routines to incorporate when the executable
13-12
-------
EPA/600/R-99J030
is built. Two modules must always be present: icon (or bcon in the case of the BCON processor)
and id. These two modules contain processing routines that are common to all applications. One
of the following three modules must be chosen to include the routines to process the specific input
source: I)profile for using a time invariant vertical concentration profile contained in an ASCII
file, 2) mSconc for processing a CMAQ IO/API concentration file, or 3) tracer for generating
special tracer species ICs and BCs. Finally, one of three modules must also be selected to control
the mechanism conversion procedure: 1) noop for no mechanism conversion, 2) radm2_to_cb4
for a CMAQ predefined RADM2 to CB4 conversion, or 3) us,er_defined to process a user
generated ASCII file (e.g., Table 13-3) containing mechanism conversion rules that are used to
generate ICs and BCs for one mechanism using input concentrations for a different mechanism.
The second step involves running the processors with all necessary input files assigned and the
time control parameters set. Both are provided to the processors by means of environment
variables. The input and output files are assigned by means of predefined logical file names. The
ICON processor requires a starting date and time in the Models-3 conventional time stamp
format. The BCON processor requires both a starting date and time and a run duration parameter.
For detailed information on compiling and running these processors, the reader is referred to the
Models-3 User Manual (EPA, 1998).
13.8 References
EPA, 1998. Models-3 Volume 9b: User Manual, Draft EPA report, U. S. Environmental
Protection Agency, Research Triangle Park, N.C.
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
13-13
-------
EPA/600/R-99/030
Table 13-1. CMAQ Predefined Vertical Profiles for Initial Concentrations
l Optional boundary condition: The vertical coordinate of the model to
2 generate these b.c. is the terrain- following sigma coordinate. The number of
3 Sigma layers and defined sigma levels are listed below.
4 6 47 1.00 0.98 0.93 0.84 0.60 0.30 0.00
S
S S02
7 SULF
a NO2
9 NO
10 O3
11 HNO3
12 H2O2
t '" .
13 ALB
14 HCHO
15 OP1
16 OP2
17 PAA
18 ORA1
19 ORA2
20 NH3
21 N2O5
22 NO3
23 PAN
24 HC3
25 HC5
26 HC8
27 ETH
28 CO
29 OL2
30 OLT
31 OLI
32 TOL
33 XYL
34 ACO3
3S TPAN
3S HONO
37 rHNO4
38 'KET
39 GLY
40 MGLY
41 DCB
42 ONIT
43 CSL
44 ISO
4S HO
46 HO2
47 ASO4I
48 ASO4J
49 NUMATKN
0
0
0
0
~6
6
0
0
0
0
0
0
0
0
0
~~o
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
. ,*«0
0
0
0
0
0
0
0
0
4
1
1
.300E-03
.150E-03
.200E-04
.200E-04
.350E-01
.500E-04
.100E-02
.300E-04
.250E-03
.250E-06
.300E-07
.300E-04
.100E-05
.100E-05
.100E-03
.olJOE+oo
.OOOE+00
.200E-04
.400E-04
.400E-04
.200E-04
.OOOE+00
.800E-01
.500E-05
.200E-06
.100E-06
.100E-05
.200E-06
.100E-08
.100S-07
.100E-08
.200E-08
.300E-04"
, f
.250E-06
.250E-06
.2SOE-06
.200E-04
.100E-08
.250E-05
.OOOE+00
.OOOE+OQ
.810E-02
.154E+00
.437S+10
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
7
9
.200E-03
.150E-03
.200E-04
.200E-04
.350E-01
.500E-04
.100E-02
.350E-04
-250E-03
.250E-06
.350E-07
.300E-04
.100E-05
.100E-05
.100E-03
.OOOE+00
.OOOE+00
.200E-04
.400E-04
.400E-04
.200E-04
.OOOE+00
.800E-01
.300E-05
.200E-06
.100E-06
.100E-05
.200E-06
.100E-08
. 100E-07
.100E-08
.200E-08
.350E-04
.250*E-06
.250E-06
.250E-06
.200E-04
.100E-08
.250E-05
.OOOE+00
.OOOE+00
.207E-02
.696E-01
.583E+09
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
3
4
.100E-03
.100E-03
.100E-04
.100E-04
.400E-01
.500E-04
.150E-02
.300E-04
.250E-03
.250E-06
.300E-07
.300E-04
.500E-06
.500E-06
.300E-04
.OOOE+00
.OOOE+00
.100E-04
.320S-04
.320E-04
.160E-04
.OOOE+00
.800E-01
.200E-05
.100E-06
.OOOE+00
.100E-05
.100E-06
. 100E-08
. 100E-07
.100E-08
.200E-08
.300E-04
.250E-06
.250E-06
.250E-06
.160E-04
.100E-08
.250E-05
.OOOE+00
.OOOE+00
.603E-02
.848E-01
.791E+09
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1
3
4
.100E-03
.100E-03
.OOOE+00
.OOOE+00
.500E-01
.500E-04
.100E-02
.20QE-04'
.200E-03
.200E-06
.200E-07
.250E-04
.500E-06
.5QOE-06
.200E-04
.OOOE+00
.OOOE+00
.100E-04
.120E-04
.120E-04
.600E-05
.OOOE+00
.700E-01
.100E-05
.OOOE+00
.OOOE+00
.100E-05
. OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.200E-04
.200S-06
.200E-06
.200E-06
.600S-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
.603E-02
.848E-01
.791E+09
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
3
7
9
.200E-04
.200E-04
.OOOE+00
.OOOE+00
.600E-01
.700E-04
.800E-03
.200E-04
.100E-03
.100E-06
.200E-07
.200E-04
.500E-06
.500E-06
.200E-04
.OOOE+00
.OOOE+00
.100E-04
.400E-05
.400E-05
.200E-05
.OOOE+00
.650E-01
.100E-05
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.20QE-04
.100E-06
.100E-06
.100E-06
.200E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
.207E-03
.696E-02
.583E+08
0.100E-04
0.100E-04
0. OOOE+00
0. OOOE+00
0.700E-01
0.100E-03
0.200E-03
0.100E-04
0.500E-04
0.500E-07
0.100E-07
0.150E-04
0. OOOE+00
0. OOOE+00
0.100E-04
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.500E-01
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.100E-08
0.100E-07
0.100E-08
0.200E-08
0.100E-04
0.500E-07
0.500E-07
0.500E-07
O.OOOS+00
0.100E-08
0.250E-05
0. OOOE+00
0. OOOE+00
1.603E-03
3.848E-02
4.791E+08
13-14
-------
EPA/600/R-99/030
Table 13-1. CMAQ Predefined Vertical Profiles for Initial Concentrations
50 NUMACC 4.1101+08 2.740E+08 1.370E+08 1.370E+08 2.740E+07 1.370E+07
51 ASOIL 1.0001+00 8.000E-01 4.000E-01 4.000E-01 8.000E-02 4.000E-02
52 NUMCOR 1.740E+04 1.392E+04 6.960E+03 6.960E+03 1.392E+03 6.960E+02
13-15
-------
EPA/600/R-99/030
Table 13-2. CMAQ Predefined Vertical Profiles for Boundary Conditions
'4
S., . $3 > *, ;-.,.-.,•• ;i i • h,, *!•:„*. i -
i Optional boundary condition: The vertical coordinate of the model to
2 generate these b.c. is the terrain- following sigma coordinate. The number of
3 sigma layers and defined sigma levels are listed below.
46 47 1.00 0.98 0.93 0.84 0.60 0.30 0.00
5 '
s North
7 l- • • ' " ' '-" ' - "t • • *, .
m\
8 SO2
9 SOLF
10 NO2
11 NO
12 O3
13 HN03
14 H2O2
15 ALD
16 HCHO
17 JOP1
18 l|i!iOP2
19 PAR.
20 ORA1
21 ORA2
22 NH3
23 N2O5
24 _NO3
25 PAN
26 HC3
27 HC5
28 HC8'
29 ~ETH
nm f, • • »
30 CO
31 OL2
32 OLT
33 OLI
34 TOIi
35 XYI,
36 ACO3
37 TPAN
38 HONO
39 HNO4
40 KET
41 6LY
42 MGIiY
43 DCS
44*ONiT
45 "CSL
16 ISO
47 HO
48 HO2
H ;
0
0
0
0
0
~6
0
0
0
_ p
' o
0
0
0
0
0
0
o
0
0
!™b
o
0
0
0
0
0
0
0
0
" 0
0
0
0
•"b
o
"b
"o
0
0
"o
J
vt
. , m
.300E-03
.150E-03
.200E-04
.200E-04
.350E-01
.500E-04"
.100E-02
.300E-04
.250E-03
.250E-06
.300E-07
.300E-04
.100E-05
.100E-05
.100E-03
.OOOE+00
.OOOE+00
.200E-04"
.400E-04
.400E-04
.200E-04
.OOOE+00
. 800E-01
.500E-05
.200E-06
.100E-06
.100E-05
.200E-06
.100E-08
.100E-07
.10 OS- 08
.200E-08
.300E-04
.250E-06
.250E-06
.250E-06
-200E-04
.lOOE-08
.250E-05
.OpOE+00_
.'dboE+bo"
* ' '
T
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
"o
»r -. •
.200E-03
.150E-03
.200E-04
.200E-04
.350E-01
.500E-04
.100E-02
.350E-04
.250E-03
.250E-06
.35bE-07
.300E-04
.100E-05
.100E-05
.100E-03
.OOOE+00
.OOOE+00
j200E-b4
.400E-04
.400E-04
".200E-04
.OOOE+00
".'sOOE-01
.300E-05
.200E-06
.lOOE-06
.100E-05
.200E-06
.100E-08
.100E-07
.100S-08
.200E-08
.350E-04
.250E-06
.250E-06
.250E-06
.200E-04
.iooE-os'
.250E-05
.OOOE+00
.'odoE+bo
'•
. ;
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.100E-03
.100E-03
.100E-04
.100E-04
.400E-01
.500E-04
.150E-02
.300E-04
.250S-03
.250E-06
.300E-07
.300S-04
.500E-06
.500E-06
.300E-04
.OOOE+00
.OOOE+00
.100E-04
.320E-04
.320E-04
.160E-04
.OOOE+00
.800JB-01
.200E-05
.100E-06
.OOOE+00
.100E-05
.100E-06
.100E-08
.100E-07
. 100E-08
.200E-08
.300E-04
.250E-06
.250E-06
.250E-06
. 160E-04
. 100E-08
.250E-05
.OOOE+00
.OOOE+00
.
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
6
.100E-03
.100E-03
.OOOE+00
.OOOE+00
.500E-01
."500E-04
.100E-02
.200E-04
.200E-03
.200E-06
.200E-07
.250E-04
.500E-06
.500E-06
.200E-04
.OOOS+00
.OOOE+00
.100E-04
.120E-04
.120E-04
.600E-05
.OOOE+00
.7bbE-oi"~
.100E-05
.OOOE+00
.OOOE+00
.100E-05
.OOOE+00
.100E-08
. 100E-07
.100E-08
.200E-08
.200E-04
.200E-06
.200E-06
.200E-06
.600E-05
.100S-08
.250E-05
.OOOE+00
.OOOE+00'
' • ' v •"
'U '.
0
0
0
0
0
0
0
0
0
6
0
0
0
0
0
0
0
0
0
0
0
0
b
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.200E-04
.200E-04
.OOOE+00
.OOOE+00
.600E-01
.700E-04
.800E-03
.200E-04
.100E-03
'. 100E-06
.200E-07
.200E-04
.500E-06
.500E-06
.200E-04
.OOOE+00
.OOOE+00
.100E-04
.400E-OS
.400E-05
.200E-05
.OOOE+00
.650E-'Ol"
.100E-05
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.200E-04
.100E-06
.100E-06
.100E-06
.200E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
'
', t
0.100E-04
0.100E-04
0. OOOE+00
0. OOOE+00
0.700E-01
0.100S-03
0.200E-03
0.100E-04
0.500E-04
0.500E-07
0.100E-07
0.1BOE-04
0. OOOE+00
0. OOOE+00
0.100E-04
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.500E-01
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.100E-08
0.100E-07
0.100E-08
0.200E-08
0.100E-04
0.500E-07
0.500E-07
0.500E-07
0. OOOE+00
0.100E-08
0.250E-05
0. OOOE+00
0. OOOE+00
1 . ,
'
*
13-16
-------
EPA/600/R-99/030
Table 13-2. CMAQ Predefined Vertical Profiles for Boundary Conditions
49 ASO4I
50 ASO4J
51 NUMATKN
52 NUMACC
53 ASOIL
54 NUMCOR
55 East
56
57 SO2
58 SULF
59 NO2
60 NO
61 O3
62 HNO3
63 H2O2
64 ALD
65 HCHO
66 DPI
67 OP2
68 PAA
69 ORA1
70 ORA2
71 NH3
72 N2O5
73 NO3
74 PAN
75 HC3
76 HC5
77 HC8
78 ETH
79 CO
80 OL2
81 OLT
82 OLI
83 TOL
84 XYL
85 ACO3
86 TPAN
87 HONO
88 HNO4
89 KET
90 GLY
91 MGLY
92 DCB
93 ONIT
94 CSL
95 ISO
96 HO
97 HO2
2
5
7
2
5
8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.405E-02
.772E-01
.187E+09
.055E+08
.OOOE-01
.750E+03
.OOOE+00
.200E-03
.100E-04
.OOOE+00
.300E-01
.500E-04
.200E-02
.400E-04
.250E-03
.250E-06
.400E-07
.500E-04
.150E-05
.150E-05
.500E-04
.OOOE+00
.OOOE+00
.100E-04
.120E-04
.120E-04
.600E-05
.OOOE+00
.800E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.250E-06
.250E-06
.250E-06
.600E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
2
5
7
2
5
8
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.405E-02
.772E-01
.187E+09
.055E+08
.OOOE-01
.750E+03
.OOOE+00
.200E-03
.100E-04
.OOOE+00
.350E-01
. 500E-04
.200E-02
.400E-04
.250E-03
.250E-06
.400E-07
.500E-04
.500E-06
.500E-06
.500E-04
.OOOE+00
.OOOE+00
.100E-04
.120E-04
.120E-04
.600E-05
.OOOE+00
.800E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.250E-06
.250E-06
.250E-06
.600E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
1
3
4
1
4
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.603E-02
.848E-01
.791E+09
.370E+08
.OOOE-01
.OOOE+03
.OOOE+00
.200E-03
.OOOE+00
.OOOE+00
.400E-01
.500E-04
.200E-02
.400E-04
.250E-03
.250E-06
.400E-07
.500E-04
.500E-06
. 500E-06
. 500E-04
.OOOE+00
.OOOE+00
.100E-04
.120E-04
.120E-04
.600E-05
.OOOE+00
.800E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.250E-06
.250E-06
.250E-06
.600E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
1
3
4
1
4
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.603E-02
.848E-01
.791E+09
.370E+08
.OOOE-01
.OOOE+03
.OOOE+00
.200E-03
. OOOE+00
.OOOE+00
.500E-01
.500E-04
.200E-02
.400E-04
.200E-03
.200E-06
.400E-07
.500E-04
.500E-06
.500E-06
.200E-04
.OOOE+00
.OOOE+00
.100E-04
.800E-05
.800E-05
.400E-05
.OOOE+00
.750E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
. OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.200E-06
.200E-06
.200E-06
.400E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
3
7
9
2
8
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.207E-03
.696E-02
.583E+08
.740E+07
.OOOE-01
.400E+03
.OOOE+00
.200E-04
.OOOE+00
.OOOE+00
.600E-01
.500E-04
.150E-02
.400E-04
.150E-03
.150E-06
.400E-07
.500E-04
.500E-06
.500E-06
.200E-04
.OOOE+00
.OOOE+00
.100E-04
.400E-05
.400E-05
.200E-05
.OOOE+00
.700E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.150E-06
.150E-06
.150E-06
.200E-05
.100E-08
.250E-05
.OOOE+00
.OOOE+00
1.603E-03
3.848E-02
4.791E+08
1.370E+07
3. OOOE-01
5.250E+02
0. OOOE+00
0.100E-04
0. OOOE+00
0. OOOE+00
0.700E-01
0.150E-03
0.150E-02
0.400E-04
0.100E-03
0.100E-06
0.400E-07
0 .500E-04
0.500E-06
0.500E-06
0.200E-04
0. OOOE+00
0. OOOE+00
0.100E-04
0.400E-05
0.400E-05
0 .200E-05
0. OOOE+00
0.650E-01
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.100E-08
0.100E-07
0.100E-08
0.200E-08
0.400E-04
0.100E-06
0.100E-06
0 .100E-06
0.200E-05
0.100E-08
0.250E-05
0. OOOE+00
0. OOOE+00
13-17
-------
EPA/600/R-99/030
Table 13-2.' CMAQ Predefined Vertical Profiles for
r\ . »«
99
100
• 101
102
1 103
104
f< 105
i t' 106
107
108
109
no
, 111
112
... 113
114
*" US
'.". 116
117
118
121
122
123
124
125
126
127
128
,: , 129
130
. 131
,J 132
133
."*- 134
135
136
iM:« 137
:." 139
140
141
142
143
144
145
14 6
|S04I
AS04J
NUMATKN
NUMACC
ASOIL
NUMCOR
South
SO2
SULF
NO2
NO
03
HNO3
H2O2
ALD
HCHO
OP1
OP2
PAA
ORA1
ORA2." '"
NH3 '
12O5
NO3
PAN
HC3
HCS
HC8
ETH
CO
OL2
OLT
OLI
"fpi.
XYL
Acos
TPAN
MONO
HN04
KET
GLY
MGLY
DCS
ONIT
CSL
ISO
HO
HO2
jL
'*"•?'.
9.
2 .
8 1
1.
JO.
0.
0.
0.
0.
0.
o .
0.
0.
0.
0.
0.
0.
raw
0.
J5.
~~o".
0.
0.
0.
0.
0.
0.
J2-
"15.
0.
0.
'"p.
0.
"'""b.
0.
0.
0.
0.
> 3 Is
0.
0.
0.
0.
0.
0.
0.
0.
,a,r • !. .
69~6E-Ol"
583E+09
740E+08
OOOE-01
392E+04
qqoE+oo _
200E-03
IOOE-04
IOOE-04
300E-01
SdOE-04
2bOE-b2
400E-04
250E-03
250E-06
400E-07
100E-03
150E-05"
ISOE-OS'
500E-04 _
OOOE+00 '
OOOE+00
IOOE-04
120E-04
120E-04
600E-05
OOOE+00
700E-01
OOOE+00
OOOE+00
OOOE+00
OOOE+00
OOOE+00
IOOE-08"
100E-07
IOOE-08
200E-08
400E-04
250E-06***
250E-06
250E-06
600E-05
IOOE-08
250E-05
OOOE+00
OOOE+00
3.207E-02
7.696E-01
9.583E+09
2 .740E+08
8. OOOE-01
1.392E+Q4
0. OOOE+00
0.200E-03
0. IOOE-04
0. IOOE-04
0.350E-01
0.500E-04
0.20bE-02
0.400E-04
0.250E-03
0.250E-06
0.400E-07
0.100E-03
0.500E-06
0.500E-06
0.500E-04
0 . OOOE+00
0. OOOE+00
0. IOOE-04
0.12QE-04
0.120E-04
0.600E-05
0. OOOE+00
0.700E-01
O.OOOE+bO
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. IOOE-08
0.100E-07
0. IOOE-08
0.200E-08
0.400E-04
0/250E-06
0.250E-06
0.250E-06
0.600E-05
0. IOOE-08
0.250E-05
0. OOOE+00
0. OOOE+00
3.207E-02
' 7.696E-01
9.583E+09
2.740E+08
8. OOOE-01
1.392E+04
0. OOOE+00
0.200E-03
0. OOOE+00
0. OOOE+00
0.400E-01
0.500E-04
0.200E-02
0.400E-04
0.250E-03
0.250E-06
0.400E-07
0.100E-03
0.500E-06
0.500E-06
0.500E-04
0. OOOE+00
0. OOOE+00
0. IOOE-04
0.120E-04
0.120E-04
0.600E-05
0. OOOE+00
0.700E-01
O.bOOE+00
0. OOOE+00
0. OOOE+00
o.ooos+oo
0. OOOE+00
0. IOOE-08
0.100E-07
0. IOOE-08
0.200E-08
0.400E-04
0.250E-06*
0.250E-06
0.250E-06
0.600E-05
0. IOOE-08
0.250E-05
0. OOOE+00
0. OOOE+00
Boundary Conditions
3.
7".
9.
2.
8.
1.
o".
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
b.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0."
0.
0.
0.
0.
0.
0.
0.
207E-02
696E-01
583E+09
740E+08
booE-oi
392E+04
OOOE+00
100E-03
OOOE+00
OOOE+00
500E-01
500E-04
200E-02
400E-04
200E-03
200E-06
400E-07
100E-03
500E-06
500E-06
300*E-04
"OOOE+00'
OOOE+00
IOOE-04
400E-05
400E-05
200E-05
OOOE+00
700E-01
booE+bo
OOOE+00
OOOE+00
OOOE+00
OOOE+00
IOOE-08
100E-07
IOOE-08
200E-08
400E-04
200E-06
200E-06
200E-06
200E-05
IOOE-08
250E-05
OOOE+00
OOOE+00
" 3
"~"7
9
2
8
1
" 0
0
0
0
0
0
0
0
0
0
0
'0
0
0
o
b
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.207E-03
.696E-02
.583E+08
.740E+07
.OOOE-02
.392E+03
.OOOE+00
.500E-04
.OOOS+00
.OOOE+00
.600E-01
.500E-04
.150E-02
.400E-04
.150E-03
.150E-06
.400E-07
.500E-04
.500E-06
. 500E*-06
.200E-04
. OOOE+00
.OOOE+00
. IOOE-04
.400E-05
.400E-05
.200E-05
.OOOE+00
.650E-01
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.OOOE+00
.IOOE-08
.100E-07
.IOOE-08
.200E-08
.400E-04
. 150E-06
.150E-06
.150E-06
.200E-05
.IOOE-08
.250E-05
.OOOE+00
.OOOE+00
'i.
3.
4 .
1.
4.
6 .
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
0.
603E-03
848E-02
791E+08
370E+07
OOOE-02
960E+02
OOOE+00
200E-04
OOOE+00
OOOE+00
700E-01
1SOE-03
100E-02
500E-05
100E-03
100E-06
500E-08
500E-04
500E-06
500E-06
200E-04
OOOE+00
OOOE+00
IOOE-04
400E-05
400E-05
200E-05
OOOE+00
550E-01
OOOE+00
OOOE+00
OOOE+00
OOOE+00
OOOE+00
IOOE-08
100E-07
IOOE-08
200E-08
500E-05
100E-06
100E-06
100E-06
200E-05
IOOE-08
250E-05
OOOE+00
OOOE+00
13-18
-------
EPA/600/R-99/030
Table 13-2. CMAQ Predefined Vertical Profiles for Boundary Conditions
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
ASO4I
ASO4J
NUMATKN
NUMACC
ASOIL
NUMCOR
West
SO2
SULF
NO2
NO
O3
HNO3
H2O2
KLD
HCHO
OP1
OP2
PAA
ORA1
ORA2
NH3
N2O5
NO3
PAN
HC3
HC5
HC8
ETH
CO
OL2
OLT
OLI
TOL
XYL
ACO3
TPAN
HONO
HNO4
KET
GLY
MGLY
DCS
ONIT
CSL
ISO
HO
HO2
3
7
9
2
8
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.207E-02
.6961-01
.5831+09
.740E+08
.OOOE-01
.4001+04
.300E-03
.300E-03
.1001-03
.500E-04
.3501-01
.SOOE-03
.200E-02
.400E-04
.40QE-03
.40QE-06
.4001-07
-250E-04
.250E-05
.2501-05
.300E-03
.0001+00
.OOOE+00
.100E-03
.800E-04
.8001-04
.4001-04
.OOOE+00
.800E-01
.1001-04
.200E-05
.100E-05
.100E-04
.3001-05
.100E-08
.1001-07"
.1001-08
.200E-08
.400E-04
.400E-06
.400E-06
.4001-06
.4001-04
.100E-08
.250E-05
.OOOE+00
.OOOE+00
3
7
9
2
8
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.207E-02
.6961-01
.583E+09
.740E+08
.0001-01
.400E+04
.200E-03
•200E-03
.1001-03
.5001-04
.400E-01
.500E-03
.200E-02
.4001-04
.400E-03
.400E-06
.400E-07
.250E-04
.250E-05
.250E-05
.300E-03
.OOOE+00
.OOOE+00
.1001-03
.800E-04
,8001-04
.400E-04
.OOOE+00
.800E-01
.100E-04
.100E-05
.200E-06
.500E-05
.200E-05
.100E-08
.100E-07
.100E-08
.200E-08
.400E-04
.4001-06
.4001-06
.4001-06
.400E-04
.1001-08
.2501-05
.0001+00
.0001+00
3.207E-02
7.696E-01
9.583E+09
2.740E+08
8.0001-01
1.400E+04
0.200E-03
0.200E-03
0.100E-03
0.500E-04
0.450E-01
0.500E-03
0.200E-02
0.400E-04
0.4001-03
0.4001-06
0. 4001-07
0.250E-04
0.250E-05
0.250E-05
0.300E-03
0. OOOE+00
0. OOOE+00
0.100E-03
0.8001-04
0.8001-04
0.4001-04
0. OOOE+00
0.800E-01
0.500E-05
0.500E-06
0.1001-06
0.5001-05
0.400E-06
0.100E-08
0.1001-07
0.100E-08
0.2001-08
0.400E-04
0.400E-06
0.400E-06
0.400E-06
0.400E-04
0.100E-08
0.250E-05
0. OOOE+00
0.0001+00
1
3
4
1
4
7
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.6031-02
.848E-01
.791E+09
.370E+08
.0001-01
.OOOE+03
.100E-03
.100E-03
.500E-04
.2001-04
.5001-01
.500E-03
.200E-02
•400E-04
.4001-03
.4001-06
.400E-07
.250E-04
.250E-05
.2501-05
.200E-03
.OOOE+00
.0001+00
.500E-04
.600E-04
.6001-04
.3001-04
.OOOE+00
.800E-01
.5001-05
.3001-06
.0001+00
.300E-05
.400E-06
.100E-08
.100E-07
.100E-08
.2001-08
.4001-04
.400E-06
.400E-06
.400E-06
.300E-04
.100E-08
.250E-05
.OOOE+00
.0001+00
3
7
9
2
8
1
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
.2071-03
.6961-02
.5831+08
.740E+07
.0001-02
.400E+03
.OOOE+00
.OOOE+00
.0001+00
.OOOE+00
.600E-01
.200E-03
.800E-03
.4001-04
.100E-03
.1001-06
.400E-07
.200E-04
.500E-06
.5001-06
.1001-03
.OOOE+00
.OOOE+00
.100E-04
.800E-05
.800E-05
.400E-05
.OOOE+00
.650E-01
.100E-05
.0001+00
.OOOE+00
.OOOE+00
.OOOE+00
.100E-08
.1001-07
.100E-08
.200E-08
.400E-04
.100E-06
.100E-06
.100E-06
.4001-05
.100E-08
.250E-05
.0001+00
.0001+00
1.603E-03
3.8481-02
4.791E+08
1.370E+07
4.000E-02
7.000E+02
0. OOOE+00
0. OOOE+00
0.0001+00
0. OOOE+00
0.7001-01
0. 1001-03
0.200E-03
0.5001-05
0.500E-04
0.5001-07
0.500E-08
0.100E-04
0.5001-OS
0.500E-06
0.500E-04
0. OOOE+00
0. OOOE+00
0.100E-04
0.4001-05
0.4001-05
0.200E-05
0. OOOE+00
0.500E-01
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0. OOOE+00
0.1001-08,
0.100E-07
0.1001-08
0.200E-08
0.500E-05
0.500S-07
0.500E-07
0.5001-07
0.2001-05
0.100E-08
0.250E-05
0. OOOE+00
0. OOOE+00
13-19
-------
EPA/600/R-99/030
V • • W -t
Table 13-2. CMAQ Predefined Vertical Profiles for Boundary Conditions
» if- = . .
196 ASO4I
187 ASO4J
198 NUMATKN
199 NUMACC
200 ASOIL
201 NUMCOR
T}
-------
EPA/600/R-99/030
Table 13-3, Example RADM2 to CB4 Conversion Rules
NO2 =
NO
O
O3
NO3
O1D
OH
HO2
N2O5
HNO3
HONO
PNA
H202
CO
S02
SULF
PAN
FACD
AACD
PACD
UMHP
MGLY
OPEN
CRES
FORM
ALD2
PAR
OLE
TOL
ISOP
ETH
XYL
TERP
END;
1.0
0 .4
7.9
3.9
NO2 ;
NO ;
03 P;
03 ;
NO3;
O1D;
HO ;
HO2 ;
N2O5 ;
HNO3 ;
HONO ;
HN04 ;
H202 ;
CO;
S02;
SULF;
PAN;
ORA1;
ORA2;
PAA;
OP1;
MGLY ;
DCB;
CSL
HCHO + 1.0 * GLY;
* ALD + 2.0 * OLI;
+2.9* HC3
+ 0.8 * OLT
* ETH
* HC8
* KET;
OLT;
TOL;
ISO;
OL2;
XYL;
TERP ;
+ 4.8
+ 0.8
HC5
OLI
4-
13-21
-------
EPA/600/R-99/030
Chapter 14
PHOTOLYSIS RATES FOR CMAQ
Shawn J. Roselle,* Kenneth L. Schere,*" and Jonathan E. Pleim*
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, North Carolina 27711
Adel F. Hanna
MCNC-North Carolina Supercomputing Center
Research Triangle Park, North Carolina 27709
ABSTRACT
The method for calculating photodissociation reaction rates (photolysis rates) for CMAQ is
described in this chapter. The description includes the photolysis rate preprocessor (JPROC) and
CMAQ subroutine PHOT. JPROC produces a clear-sky photolysis rate look-up table. The
look-up table consists of photolysis rates at various altitudes, latitudes, and hour angles. The
look-up table is recalculated for each simulation day and is dependent upon the chemical
mechanism. Within CMAQ, photolysis rates for individual grid cells are interpolated from the
look-up table in subroutine PHOT. PHOT also uses a parameterization to correct the clear-sky
photolysis rates for cloud cover.
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Shawn Roselle, MD-80, Research Triangle Park, NC 27711. E-mail:
sjr@hpcc.epa.gov
On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
140 PHOTOLYSIS RATES FOR CMAQ
14.1 Background
Many chemical reactions in the atmosphere are initiated by the photodissociation of numerous
trace gases. These photodissociative reactions are responsible for most of the smog buildup
detrimental to humans, animals, plant life and materials. In order to accurately model and predict
the effects of air pollution, good photodissociation reaction rate (or photolysis rate) estimates
must be made.
Photodissociation is the conversion of solar radiation into chemical energy to activate and
dissociate chemical species. Examples of species that photodissociate include many important
trace constituents of the troposphere such as NO2, 03, HCHO, CH3CHO, MONO, the NO3
radical, and H2O2 (also see Table 14.1). The simulation accuracy of the entire chemical system is
highly dependent upon the accuracy of photolysis rates, which are the primary sources of radicals
in the troposphere. Photolysis rates (min"1), sometimes called J-values, are computed for
photodissociation reaction (/) by
(14-1)
where, F(A) is the actinic flux (photons cm"2 min"1 nm"1), 0j(A,) the absorption cross section for the
molecule undergoing photodissociation (cm2 molecule"1), <|>j(A,) the quantum yield of the
photolysis reaction (molecules photon"1), and A. the wavelength (nm). Absorption cross sections
and quantum yields are functions of wavelength, and may also be functions of temperature and
pressure; they are unique to species and reactions. Laboratory experiments measuring the
absorption cross sections and quantum yields, have been conducted for many species that
photodissociate in the troposphere. Actinic flux is a radiometric quantity that measures the
spectral radiance integrated over all solid angles per unit area. The spherical receiving surface
distinguishes the actinic flux from the more commonly measured irradiance, which is the radiance
falling on a horizontal surface. Thus, the actinic flux can be called spherical spectral irradiance.
The actinic flux changes with time of day, longitude, latitude, altitude, and season, and is
governed by the astronomical and geometrical relationships between the sun and the earth. It is
greatly affected by the earth's surface albedo as well as by various atmospheric scatterers and
absorbers. Hence, correct model calculation of the temporal and spatial variation of the actinic
flux is critical to obtaining accurate photolysis rates for regional and mesoscale episodic
photochemical modeling.
The current approach taken for setting photolysis rates in the Models-3/CMAQ framework
follows that of the Regional Acid Deposition Model (RADM) (Chang et al., 1987). It includes
two stages of processing: (1) a table of clear-sky photolysis rates is calculated for specified
14-2 '
-------
EPA/600/R-99/030
heights, latitudes, and hours from local noon; and (2) photolysis rates are interpolated from the
table within the CMAQ Chemistry Transport Model (CCTM) based on grid cell location and the
model time, and are corrected for cloud cover. This approach is computationally efficient and has
been shown by Madronich (1987) to give clear-sky photolysis rates within the uncertainty of the
surface-based measurements.
14.2 Preprocessor JPROC: Calculate Clear-sky Photolysis Rate Table
Preprocessor JPROC calculates a table of clear-sky photolysis rates (or J-values) for a specific
date. The table is dimensioned by latitude, altitude, and time. Currently, J-values are calculated
for 6 latitudinal bands (10°N, 20°N, 30°N, 40°N, 50°N, and 60°N), 7 altitudes (0 km, 1 km, 2
km, 3 km, 4 km, 5 km, and 10 km), and ±9 hours from local noon (0 h, 1 h, 2 h, 3 h, 4 h, 5 h, 6 h,
7 h, and 8 h). There is a separate table for each photolysis reaction. In order to compute the
photolysis rates using Equation 14.1, the actinic flux, absorption cross section, and quantum yield
must be determined as a function of wavelength.
The delta-Eddington two-stream radiative transfer model (Joseph et al., 1976; Toon et al., 1989)
is used for computing the actinic flux. The two-stream approximations are limited in application
to cases where the scatter is not highly anisotropic. In computing the actinic flux, a description of
the extraterrestrial radiation, aerosol, ozone absorption, oxygen absorption in the
Schumann-Runge Bands, Rayleigh scattering (WMO, 1985) and surface albedo are provided to
the radiation model.
Extraterrestrial radiation is specified by a user input file. JPROC is flexible enough to use any
extraterrestrial radiation data distribution specified by the user. However, the wavelength
distribution of the extraterrestrial radiation data is important because this is also the distribution
that will be used in the integration of Equation 14.1. Therefore, the user should choose a
wavelength distribution that resolves the features that are important to the photolysis reactions of
interest. A modified WMO extraterrestrial radiation data distribution (Chang, et al., 1990) is used
as input to JPROC, which has a variable wavelength resolution ranging from 1 nm to 10 nm.
The O2 and O3 absorption cross section data are specified by user input files, but it is
recommended that the most recent NASA data (DeMore et al., 1994) be used in the calculations.
Vertical ozone profiles are set by interpolating seasonal profiles from a user input file, and if total
ozone column data are available (such as data measured by the Total Ozone Mapping
Spectrometer (TOMS) instrument aboard the sun-synchronous polar orbiting Nimbus satellite),
then the interpolated vertical profiles are uniformly rescaled so that the profiles integrated total
ozone column value matches the measured total ozone column data. TOMS data are archived
and available at the National Satellite Service Data Center (NSSDC) in the form of digital daily
maps with a resolution of 1 degree latitude by 1.25 degrees longitude. The TOMS data are
averaged over each latitudinal band in JPROC. Nimbus-7 TOMS data are available for years
1978 through 1993; Meteor-3 TOMS data are available for 1991-1994; ADEOS TOMS data are
14-3
-------
EPA/600/R-99/030
available for 1996-1997; and Earth Probe TOMS data are available for 1996-1998. These data
can be downloaded from site http://jwocky.gsfc.nasa.gov.
The albedo data given by Demerjian et al. (1980), which have been used extensively in radiative
transfer models, are given as a function of wavelength and are used in the current version of
JPROC. Currently, a single vertical profile of aerosol attenuation coefficients (Elterman, 1968) is
used in JPROC. An effort is underway to incorporate predicted aerosol parameters from CMAQ
into the photolysis rate calculations.
ti. -; t .-»• • n; at . , ;/, ,,.. m • „ „ - . • „ -.-.- ... - „• >
l- ".,'•" ; ""»_.«» •>• r-'-. T >;f '•- •• • •• •':•• . "• -T:- • >. -- '•'* v . • • t
Several factors in JPROC depend on the vertical profiles of temperature and pressure, including
ozone absorption and the absorption cross sections and quantum yields for individual photolysis
reactions. JPROC interpolates seasonal profiles of temperature and pressure. We use the same
vertical profile data that are used in the RADM (Chang et al., 1987).
iii,,ni. . , ' 1 r ' ,. ,., "' :,!oc(1.6/rcos(e)- 1)] (14-2)
*
where cfrac is the cloud coverage fraction (cloud fraction is interpolated from hourly data for
each grid cell), 8 is the zenith angle, and tr is the cloud transmissivity. Below cloud photolysis
rates will be lower than the clear-sky values due to the reduced transmission of radiation through
the cloud. The cloud transmissivity is calculated by:
14-4
-------
-f)
EPA/600/R-99/030
(14-3)
where/is the scattering phase fiinction asymmetry factor (assumed to be 0.86) and Tdd is the
cloud optical depth. We have replaced the cloud optical depth equation in RADM with one taken
from Stephens (1978). The original RADM equation for Tcld required an estimate of the cloud
droplet radius, which is not readily available. In RADM, the cloud droplet radius was assumed to
be IQum, The empirical formula for Tcld from Stephens (1978);
log(TcW) = 0.2633 1 1 .7095 ln[ log( W)} (14-4)
is only a function of liquid water path (W), where W=LAz (g/m2), L is the liquid water content
(g/m3), and Az is the cloud thickness. The above cloud top factor (F,,) is calculated as:
This equation allows for enhancement of photolysis rates above the cloud due to the reflected
radiation from the cloud. It also includes a reaction dependent coefficient (aj which allows for
further above cloud enhancements (Chang et al., 1987). Within the cloud, the cloud correction
factor is a simple linear interpolation of the below cloud factor at cloud base to the above cloud
factor at cloud top. Once computed, the below, above, and within cloud factor are used to scale
the clear sky photolysis rates to account for the presence of clouds. In the current
implementation, all cloud types (including clouds composed of ice crystals) are treated the same
using the above outlined procedure,
14.4 Summary
The current method for calculating photolysis rates in CMAQ, which was derived from RADM,
uses a preprocessor to compute a look-up table and a subroutine within the chemistry transport
model to interpolate J-values and apply a cloud-cover correction. Other approaches and
enhancements are being developed and tested. One enhancement to be added in the future is the
dynamic link between aerosol predictions and photolysis rate calculations. Another is to replace
the two-stream model with a more comprehensive multi-stream radiative transfer model (Stamnes
et al., 1988). Other absorption cross section and quantum yield data will be added or updated
using the DeMore et al. (1997) revisions.
14-5
-------
EPA/600/R-99/030
14.5 References
Carter, W.P.L., 1990: A detailed mechanism for the gas-phase atmospheric reactions of organic
compounds, Atmos. Environ., 24(A), 481-518.
Chang, J.S., R.A. Brost, I.S.A. Isaksen, S. Madronich, P. Middleton, W.R. Stockwell, and CJ.
Walcek, 1987: A three-dimensional eulerian acid deposition model: physical concepts and
formulation, J. Geophys. Res. 92 (D12): 14681-14700.
I* '•'•"':••'. • j .
Chang, J.S., PJ3. Middleton, W.R. Stockwell, CJ. Walcek, J.E. Pleim, H.H. Lansfbrd, F.S.
Binkowski, S, Madronich, N.L. Seaman, D.R. Stauffer, D. Byun, J.N. McHenry, P.J. Samson, H.
Hass, 1990: The regional acid deposition model and engineering model, Acidic Deposition: State
of Science and Technology, Report 4, National Acid Precipitation Assessment Program.
iL« T; " i , ,. " ;; ..,:
Demerjian, K.L., K.L. Schere, J.T. Peterson, 1980: Theoretical estimates of actinic (spherically
integrated) flux and photolytic rate constants of atmospheric species in the lower troposphere.
Advances in Environmental Science and Technology, Vol. 10, by John Wiley & Sons, Inc.,
369-459. -- - • ,
DeMore, W.B., S.P. Sander, D.M. Golden, R.F. Hampson, M.J. Kurylo, CJ. Howard, A.R.
Ravishankara, C.E. Kolb, and MJ. Molina, 1994: Chemical Kinetics and Photochemical Data for
Use in Stratospheric Modeling: Evaluation Number 11, JPL Pub. 94-26. Pasadena, CA: National
Aeronautics and Space Administration, Jet Propulsion Laboratory.
DeMore, W.B.2 S.P. Sander, D.M. Golden, R.F. Hampson, MJ. Kurylo, CJ. Howard, A.R.
Ravishankara, C.E. Kolb, and MJ. Molina, 1997: Chemical Kinetics and Photochemical Data for
Use in Stratospheric Modeling: Evaluation Number 12, JPL Pub. 97-4. Pasadena, CA: National
Aeronautics and Space Administration, Jet Propulsion Laboratory.
Eljerman, L, 1968: UV, Visible, and IR Attenuation for Altitudes to 50 km; AFCRL-68-0153.
Bedford, MA: Air Force Cambridge Res. Lab.
Gery, M.W., G.Z. Whitten, J.P. Killus, M.C. Dodge, 1989: A photochemical mechanism for
urban and regional scale computer modeling, J. Geophys. Res., 94, 12925-12956.
Joseph, J.H., WJ. Wiscombe, and J.A. Weinman, 1976: The delta-Eddington approximation for
radiative flux transfer, J. Atmos. Sci., 33,2452-2459.
Madronich, S, 1987: Photodissociation in the Atmosphere: 1. Actinic flux and the effects of
ground reflections and clouds, J. Geophys. Res. 92 (D8): 9740-9752.
' i- „ 14'6
-------
EPA/600/R-99/030
Stamnes, K., S. Tsay, W. Wiscombe, and K. Jayaweera, 1988: Numerically stable algorithm for
discrete-ordinate-method radiative transfer in multiple scattering and emitting layered media,
Appl. Opt., 27, 2502-2509.
Stephens, G.L., 1978: Radiation profiles in extended water clouds. II.: Parameterization schemes,
J. Atmos. Sci. 35*2123-2132.
Stockwell, W.R., P. Middleton, and J.S. Chang, 1990: The second generation regional acid
deposition model chemical mechanism for regional air quality modeling, J. Geophys. Res. 95
(D10): 16343-16367.
Toon, O.B., C.P. Mckay, and T.P. Ackerman, 1989: Rapid calculation of radiative heating rates
and photodissociation rates in inhomogeneous multiple scattering atmospheres, J. Geophys. Res.,
94, 16287-16301.
World Meteorological Organization, 1986: "Atmospheric Ozone 1985: Assessment of Our
Understanding of the Processes Controlling its Present Distribution and Change"; WMO Rep. No.
16; Global Ozone Research and Monitoring Project, Geneva, Switzerland.
This chapter is taken from Science Algorithms of the EPA ModeIs-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
14-7
-------
EPA/600/R-99/030
Table 14-1. RADM2 Photolysis Reactions (Adapted from Stockwell et al,, 1990.)
Reaction
Description
O3 + hv - O2 + O'D
O + hv - O + O3P
Np3 + hv -NO +O2
NO3
hv -
HfONO
HNO3
HNO4
H2O2 + hv - OH + OH
HCHO + hv - H + HCO
HCHO + hv - H2 + CO
CH3CHO + hv"(+2O2) - CH3OO + HO2
dH3COCH3 £hv - CH3 + CH3CO
CH3COC2H5 + hv " - ACO3 + ETH
HCOCHO + hv -"HCHO + CO
CO
HCOCHO + hv -
CH3COCHO + hv - ACO3 + HO2
HCOCH=CHCHO + hv - 0.98HO2 +
0.02ACO3
CH3OOH + hv - CH2O + OR + HO2
CH3ONO2 + hv - 0.2ALD + 0.8KET
C,E.O -f hv - products
CO
TCO3
HO + NO2
Ozone Photolysis to O'D
Ozone Photolysis to O3P
Nitrogen Dioxide Photolysis
Nitrate Photolysis to NO
Nitrate Photolysis to NO2
Nitrous Acid Photolysis
Nitric AcicTPhotolysis
Pernitric Acid Photolysis
Hydrogen Peroxide Photolysis
Formaldehyde Photolysis to Radicals
Formaldehyde Photolysis to Molecular
Hydrogen
Acetaldehyde Photolysis
Acetone Photolysis
Methyl Ethyl Ketone Photolysis
Glyoxal Photolysis to Formaldehyde
Glyoxal Photolysis to Molecular
Hydrogen
Methyl Glyoxal Photolysis
Unsaturated Dicarbonyl Photolysis
Methyl Hydrogen Peroxide Photolysis
Organic Nitrate Photolysis
Acrolein Photolysis
14-8
-------
EPA/600/R-99J03O
Chapter 15
PROGRAM CONTROL PROCESSING IN MODELS-3
Jeffrey Young'
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
Based on user choices and input, the Models-3 framework creates objects and include files that
control the complete problem domain under which a CMAQ model is built and executed. The
general concepts underlying the definition of these objects and the creation of the include files
are discussed.
'On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Jeffrey Young, MD-80, Research Triangle Park, NC 27711. E-mail:
yoj@hpcc.epa.gov
-------
EPA/600/R-99/030
I'll . I • .1" ,i; V'f' . '> • • . • F ,
15.0 PROGRAM CONTROL PROCESSING IN MODELS-3
Program control processing (PCP) refers to setting up internal arrays and mappings, global
parameters, and data linkages to establish the complete problem domain in which the programs,
of "models," in the Community Multiscale Air Quality (CMAQ) system are built and executed.
Controlled by user choices, the Models-3 framework carries out this processing automatically
and provides the key to modularity and ensures uniform, consistent internal dimensioning,
looping parameters and solver data in the compiled codes for all the CMAQ system models.
Thus, modularity and consistency in establishing the computational domain and chemical
mechanism for any particular application of the CMAQ system are assured by PCP. Modelers
can easily and relatively safely carry out many different applications at different scales, or with
different chemistry, or different solvers. Model developers can readily study the effects of
different schemes and determine optimal implementations of the science and codes. PCP helps
to establish a true "one-atmosphere" approach to modeling.
PCP involves both the Models-3 framework's graphical user interface and specialized processors
that are launched from the framework. In the Models-3 framework, a user selects the
computational grid and domain characteristics, the vertical layer structure, and the chemical
mechanism to be used in building (compiling and linking) a model (an executable). How these
choices get transformed automatically into compiled code is discussed in the following sections.
This chapter focuses on the CMAQ Chemical Transport Model (CCTM) since the CCTM
requires the most extensive use of PCP. However, some or all of the user choices made to set up
a particular CCTM determine the particular application characteristics for all of the supporting
processors (or "models"). In addition to the CCTM, CMAQ is comprised of eight additional
processors as follows:
• The Models-3 emissions processing and projection system (MEPPS), described in
Chapter 4.
• The emissions-chemistry interface processor (ECIP), described in Chapter 4.
• The plume-in-grid dynamics processor (PDM), described in Chapter 9.
• The meteorology-chemistry input processor (MCIP), described in Chapter 12.
• The land-use processor (LUPROC), described in Chapter 12.
• The initial and boundary conditions processors (ICON and BCON), described in Chapter
13.
• The photolytic J-value processor (JPROC), described in Chapter 14.
• The process analysis processor (PROCAN), described in Chapter 16. PROCAN is
launched by the Models-3 framework and its execution is transparent to the user. The
outputs from PROCAN are used to control CCTM diagnostic outputs (see Section 15.3.1
below).
15-2
-------
EPA/600/R-99/030
Building a particular model in the Models-3/CMAQ system requires the creation'of Fortran
include files that populate the model's codes to set the specific computational grid and domain
characteristics, the vertical layer structure, the chemical mechanism, and (for the CCTM) the
process analysis to be used for the particular model. Once these model configuration objects are
set by the user in the Models-3 Science Manager, the Models-3 framework automatically
generates the required global Fortran include files. The use of global Fortran include files, the
notion of "Global Name Table Data," and the design concepts that govern this implementation
are discussed in Chapter 18, Sections 18.5 and 18.7.
15.1 Domain Configuration
The computational domain configuration data are contained in three include files:
• HGRD.EXT - The horizontal grid dimensions in terms of the number of grid cell columns
and rows and the number of grid cells for which the computational domain boundary is
extended for boundary data. The number of these perimeter boundary cells is also
specified.
• VGRD.EXT - The vertical layer dimensions as the number of vertical layers. This
include file also contains declarations for the layer surface and layer center arrays.
• COORD. EXT - The domain coordinate data:
The map projection type (Lambert, Mercator, Stereographic, UTM, or latitude-
longitude).
The map projection parameters.
The center of the grid's coordinate system with respect to the "mother grid" or
parent grid (usually the main grid from the meteorology pre-processing).
The horizontal grid cell sizes in meters.
The vertical layer type. There are 7 currently defined:
1.
-------
EPA/600/R-99/030
15.2 Input/Output Applications Programming Interface
Input and output data access for the CMAQ system is accomplished mainly through the
Input/Output Applications Programmer's Interface (I/O API) sub-system [1]. The CMAQ system
implementation of the I/O API is described in Chapter 18, Section 18.3. In order to access I/O
API files and use other I/O API functions, it is necessary to specify the required declarations and
parameters. Some or all of the following include files are needed in each subroutine that uses the
I/O API:
F : ; 3 ' : ' .: ' ' - - / .'' " U '. •• •'•?. ^ : - I
• PARMS3.EXT - I/O API dimensioning parameters, file type and file characteristic
parameters, coordinate system and map projection parameters, and vertical layer
parameters.
• FDESC3.EXT - Fortran common blocks that contain a complete I/O API file description.
The data in these common blocks are loaded from an I/O API file when opened and read.
These data must be supplied from the user subroutine when an I/O API file is opened for
writing.
• IODECL3.EXT - Declarations and usage comments for I/O API functions and routines.
• XSTAT3.EXT-Exit codes for the I/O API M3EXIT function. Generally, M3EXIT is
called from any subroutine that produces a fatal error during I/O API access.
15.3 Other CCTM Configuration Control
15.3.1 CCTM Process Analysis
The ModeIs-3/CMAQ system provides a diagnostic tool that allows a user to probe into the way
the science processing is actually being executed in the CCTM. The process analysis tool
optionally provides two types of information, which are termed Integrated Process Rates (IPR)
and Integrated Reaction Rates (IRR). Note that these diagnostic tools apply only for the CCTM
and not for the other models. The reader is referred to Chapter 16 for the details, but briefly the
IPR processing captures the changes to the concentration field for different species, or groupings
of species, for each individual science process modeled. The IRR processing focuses on the gas-
phase chemistry and allows a detailed examination of various characteristics of the chemical
mechanism implemented in the CCTM for the chosen scenario. The data produced by these tools
arc collected in I/O API output files, which can be further processed or examined using the
visualization tools.
The CCTM requires three include files, even if neither IPR nor IRR processing is selected.'
These include files are automatically generated from PROCAN, and the Models-3 framework
then incorporates them in the CCTM build process. The three IPR/IRR include files are:
1 If process analysis is not requested while building a CCTM model executable, the Models-3 framework
automatically supplies the include files.
15-4
-------
EPA/600/R-99/030
• PA_CTL.EXT - The declarations and values of the two flags that determine if IPR or IRR
(or both) processing is to be performed.
« PA_CTM,EXT - Declarations and parameters that define the domain, output file
descriptions, arrays and dimensions, IPR and IRR loop variables, and named common
blocks for computational data.
• PA_DAT.EXT - Data statements containing values for the variables needed to fill in the
common blocks in PA_CTM,EXT.
15.3.2 CCTM Fixed Data
The Models-3/CMAQ system can deal with other types of global include files and reference
them from a user-supplied full path to build a model. The following three categories of global
include files apply to the CCTM. The other processor models (MCIP, ECIP, etc.) also have
include files in this category, but not as extensively.
* FILES_CTM.EXT - The set of logical file names used in the current CCTM
implementation. The include file contains data statements for the file names as variables
used in the codes and values for the variables. The values are character strings that are
UNIX environment variable values set in the scripts launched by the Models-3
framework to run the models.
• CONST.EXT, CONST3_RADM.EXT - These include files contain parameter statements
to define basic and frequently used air quality and meteorological modeling physical and
mathematical constants. Inclusion of these files in the model codes helps to assure
consistency in the science calculations across all the processes and modules.
• BLKPRM_500.EXT - Computational blocking: For computational efficiency and
reduced memory requirements, operations may be performed on groups of grid cells (a
block of cells) at a time. This include file provides convenient definitions to use 500-eell
blocking.
• GRID_DECL.EXT - Declarations for the dimensions and species classes offsets in the
main concentration array. See Chapter 18, Section 18.2.2 for details.
• EMISPRM.chem.EXT, EMISPRM.vdif.EXT - These two include files contain the
declarations that control in which process the emissions sources are injected; either the
vertical diffusion process of the gas-phase chemistry solver process. By using a C pre-
processor (cpp) #ifdef directive, one or the other include file is actually included when the
code is compiled. The cpp directive is set by the user during the model building phase.
The reader is referred to Chapter 18, Section 18.5 for more details.
15-5
-------
EPA/600/R-99/030
a- • > at ' i •
•s -' - . W ' J -' <-.-..
15.4 Generalized Chemistry
15.4,1 Design
In the past, a particular chemical mechanism was generally "hardwired" into a chemical transport
model with mechanism parameters and variables embedded in the solver codes. Implementing
even minor changes incurred a high potential for code and consistency errors. Additionally,
some mechanism changes necessitated extensive coding changes, requiring careful tracing
through all the mechanism dependencies. Worse, not finding these dependencies led to errors
that were not necessarily detected. The Models-3/CMAQ system employs a generalized
chemical.mechanism processor (MP), also called the "mechanism reader," which greatly
simplifies the task of implementing or altering gas-phase chemical mechanisms and provides the
capability of easily and safely using different mechanisms in the CMAQ system.
MP reads an ASCII file that contains a symbolic description of a gas-phase chemical mechanism,
and generates three ASCII Fortran source files. The input ASCII file is called a mechanism
description, or simply, "mechanism."
The ASCII mechanism description file is formatted according to a simple set of rules in a free-
•III." ::. > I 1IIU I lilllli • .1 ' '!"'., . *, .11. •.-,.. , - * .
form format similar to the approaches used by Jeffries et al. [2] and Gery and Grouse [3]. The
types of reactions that are supported are described below in Section 15.4.3, and the specific
formats are described in the Models-3 User Manual, Appendix M [4].
The Fortran files that are generated are incorporated as include files in the CMAQ Fortran codes.
They consist of:
• Fortran named common blocks that declare parameters and explicitly dimensioned arrays
associated with gas chemistry kinetic and photolytic reactions.
• Fortran data statements that fill in the common block arrays with actual values
determined from the input ASCII mechanism file.
The common blocks and the data statements files are global and are used in all the codes that
require gas chemistry information.
Another file that is generated from the mechanism description file by MP, which is transparent to
the user, is a species list found in the mechanism description. This file is used by the framework
to generate all the gas chemistry species global include files that are required in the model codes.
In order to implement the mechanism reader's capabilities, the CMAQ system requires
generalized gas-phase chemistry solvers. At present, there are two such solvers available,
SMVGEAR and QSSA, which are described in detail in Chapter 8, The use of generalized
solvers precludes some code optimizations that can increase performance, but significantly
facilitates implementing new or altered chemical mechanisms.
15-6
-------
EPA/600/R-99/030
15.4.2 Operation
The CMAQ system accounts for chemistry in three forms: gas phase, aerosols (liquid and solid
phase), and the aqueous phase. Therefore, the Models-3 framework controls certain aspects of
how the model species are treated in a model simulation and provides linkages between species
in different phases. Setting up the linkages is accomplished by means of a series of four species
tables in a spread sheet format that are filled out by the user. These tables also allow the user to
control some aspects of the simulation and generated outputs and provide links to data such as
emissions, initial and boundary concentrations, and dry deposition velocities that are generated
by the CMAQ input processors. The mechanism definition and species tables become objects
that the Models-3 framework stores in its database. Therefore if a required chemical mechanism
already exists in the database as a predefined mechanism, the user does not have to enter any
mechanism data into the framework.
Since different gas-phase chemical mechanisms can be used in the CMAQ system and these
mechanisms may employ different species names, it is necessary to supply the linkages among
the gas-phase species, the fixed aerosol species, and the species that participate in aqueous-phase
chemistry. Similarly, it is also necessary to link the gas-phase species to emissions, deposition
velocities, and to aqueous-phase scavenging. In addition, the species tables allow the user to
select which species concentrations are written to the output files and whether or not they are to
be advected or diffused. In the same manner, most of these linkages must be established for
aerosol, non-reactive, and possibly tracer species.
The methodology used to establish links between species names involves the concept of
surrogate names. Surrogate names are used to provide linkages between model species within
the CCTM and to link the model species to those that represent data provided by other CMAQ
processors. For example the emissions-chemistry interface processor, ECIP, that links the
Models-3 emissions processing and projection system (MEPPS) with the CCTM could write a
species named XXX that represents xxx emissions rates. However, the CCTM that uses xxx may
have the corresponding model species named YYY. Using the surrogate name concept,
emissions species XXX gets mapped to model species YYY. The reader is referred to the
Models-3 User Manual [4] for more details on the implementation of the set of predefined
surrogate names used in the current version.
An additional feature that can be applied to any of the surrogate species linkages is the
application of a multiplicative, or scale factor. The user can set the factor to be other than unity
(the default) to easily modify some model input data related to the selected species and process
linkage. For example, this could be used to test the effect of changing the deposition velocity for
a certain species, or to modify initial or boundary conditions data. Also, the factors can be set to
modify certain data between model species groups within the CCTM, such as in the gas-phase to
aerosol linkage.
These linkages are set by the user and are distributed among the four tables that establish the
linkages among gas-phase, aerosols, non-reactive, and (possibly) tracer species. Non-reactive
species are gas-phase, but do not participate in gas-phase reactions. Optional tracer species are
15-7
-------
EPA/600/R-99/030
r., ••» • J •
inert and may have special, user-defined initial and boundary concentrations as well as emissions
sources. The reader is referred to the ModeIs-3 User Manual [4] for more details and a complete
description of the species tables.
The following is a list of the linkages set up in the current implementation of the Models-
3/CMAQ system. The linkages take place with respect to the CCTM or model species name; the
surrogate names and flags point to this name.
r .. IAI -,yr .- .1 , . . • • ' «; _
1^ ModeJUpeeies molecular weight.
2. Emissions species surrogate name.
3. Emissions species scale factor.
4. Initial and boundary concentrations species surrogate name.
5. Initial and boundary concentrations species scale factor.
6. Deposition velocity species surrogate name.
7. Deposition velocity species scale factor.
8. Gas-phase to aerosol linkage species surrogate name.
9. Gas-phase to aqueous-phase linkage species surrogate name.
.» v , ;.*'-,»,•.,.• • - . ** . •- \ •
10. Gas-phase aqueous scavenging linkage species surrogate name.
11. Gas-phase aqueous scavenging linkage species scale factor.
liiii!1 •" "i! HI :i i "is,,. ., • , niSii 1. . • " . , •-1,,, i, . • S'i :u . ,'i'i1
12. Aerosol to aqueous-phase linkage species surrogate name.
13. Aerosol aqueous scavenging linkage species surrogate name.
14. Aerosol aqueous scavenging linkage species scale factor.
15. Non-reactive to aerosol linkage species surrogate name.
16. Non-reactive to aqueous-phase linkage species surrogate name.
17. Non-reactive aqueous scavenging linkage species surrogate name.
18. Non-reactive aqueous scavenging linkage species scale factor.
19. Model species flag for participation in the advection processes.
20. Model species flag for participation in the diffusion processes.
2L Modeljspecies flag for inclusion in the concentration output file.
22. Model species flag for inclusion in the dry deposition output file.
23. Model species flag for inclusion in the wet deposition output file.
24. Tracer species - similar linkages to any of the above.
It should be noted that items 1-7 and 19-23 apply to all of the four groups of species (gas
phase, aqueous phase, solid and liquid aerosols).
15-8
-------
EPA/6QO/R-99/030
15.4.3 Supported Reaction Types
In this section we describe the types of gas-phase chemical reactions supported by the
generalized mechanism processor, MP, in the current version of the Models-3/CMAQ system,
The reader is referred to Chapter 8 for a description of the available mechanisms.
The gas-phase mechanism description file is a list of mechanism reactions. The mechanism
reactions list consists of lines of symbolic descriptions of elementary reactions among modeled
chemical species, followed by a rule-based symbolic description of the reaction rate expression
for that reaction. The reader is referred to Chapter 8 for mechanism description examples.
These mechanism reaction lists are generally "free-form," in that spacing of the symbols on the
page is not important (except, perhaps for readability), and reaction descriptions may span many
lines if necessary. Liberal use of comments is facilitated by allowing for comment lines and
comments embedded within the reaction lines. In addition, each reaction may be labeled for
subsequent reference. The photolysis reactions are given labels for a specific photolytic table
reference. These tables are generated by the J-value processor and are used in the CCTM for the
gas-phase chemistry processing. Currently, there are 10 different rate constant expressions
available for thermal reactions, including Arrhenius-type expressions and the class of so-called
fall-off reactions (Type f, below). New expressions can be added, as necessary,
The type of reaction rate expressions that can be calculated are:
Type 1
Type 2
Type 3
Type 4:
Type 5 :
Type 6 :
Type 7 :
Type 8:
Type 9;
Typef:
k = A
k = A(T/300)B,
k = Ae('OT),
k = A(T/300)Be{-Cflf)
k = k,. = kf /Ae("crT),
k = Akj,,
k = A(1.0 + 0.6P),
= k0+
i
k = k, + k2[M],
k= k = -
where T = temperature in deg K.
where C = Ea/R,
Ea is the activation energy, and
R is the gas constant.
where kf is any previously defined forward reaction.
where k,, refers to the n* reaction.
where P = pressure in atmospheres.
where ko, k2, and k3 are Type 3.
• •
where k, and k2 are Type 3.
Fc
where kg and kr may be Type 1, 2, 3, or 4,
F = 0.6 (usually), and
n = 1.0 (usually).
15-9
-------
. ,_ EPA/600/R-99/030
f '15.4.4 Mechanism Parsing Rules
The mechanism description file that MP reads must follow a set of format rules in order to
describe the reactions listed above. Generally, reserved ASCII characters and keywords on one
line (which may wrap around the page) represent parts of a reaction, including species reactants
and products, stoichiometric coefficients and rate expressions. In addition, labels are supported,
which function to allow one reaction to refer to another and to fix photolytic reaction table names
for photolytic reactions. As mentioned above, the reader is referred to Chapter 8 for mechanism
description examples. Also, Appendix M of the Models-3 User Manual contains a complete
description of these formats.
fin. . . j i «i~ «-, r «i«i -*.••'• .>*•'»•••"'• • •« •• • •>• •-•*•' • -'
; j . , 15.4.5 Chemical Species Include Files
w. 11 t " »» . no*;**! >: 1P>, r» * . '.'-,, ,~, ' •,," ?"••* • , • -"• I • '• i" > •• '
;» 1 4 As described above, MP reads and processes the mechanism description file, and one of the
oCtpufs it generates is a file that contains the list of chemical species found in the mechanism.
The Models-3 framework reads this species file to initialize the gas-phase species table in a
format stoilarjo ^spreadsheet, which the user completes by inserting data such as molecular
""'-" * weights, surrogate names and factors, and processing flags. The initial set-up presented to the
, , _ - user consists of all blank entries in each spreadsheet cell, but with the first column filled in with
; 1 the species found in the MP-generated species file. Thus all the species in the mechanism are
listed in the order in which they were found in the mechanism description file. The top row is
also filled hi with category headers that are related to the linkages described above and that
determine the framework-generated Fortran include files to be used in the codes.
There are three additional tables that are initialized for aerosols, non-reactive and tracer species.
They do not have the first column set up, but the first row is set as in the gas-phase table. It is up
tdthe user to fill in the first column, as with all the other spreadsheet cells. Appendix M of the
Models-3 User Manual describes the details of how to enter data into these Jables.
^7 .. , , • * 7* ^ jgas* * -5 • ^- • . • * -v . » ;*t^ * -^ ' -^ r* m ^ • * , .• * i * . »
_i „:;: Tie following sections describe all the standard include files that are generated from the species
«" IK tables and the behavior of the models when accessing the data declared in the include files.
15.4.5.1 Gas-Phase Reactions
*::>, J"; • •''" : ..... •• •"« vft ••;•• ft. .*':V . .••- ^ . ' ». • . ; ' •-
:-. :",['> » ' '" /•• i.ir 31 ' .'4-. '•. '• . " •>:;.-^r. ; . •- ' " - ,! ^. I'-" ; ' .- ,- . . /
«!-,>, Th? gas-phase reactions data are required in the CCTM by the gas-phase chemistry solvers and
the subroutine PHOT, which calculates the photolysis rate constants. These data are also
-, ,w required in JPROC, the processor model that calculates the table-based photolysis rates (see
Chapter 14). There are two include files associated with this data:
* RXCM.EXT - contains all the declarations, parameter statements and array definitions
associated with the gas-phase chemistry reactions. This file also contains three named
jljl'lj! ^ Kf; ' \ ,P|!|!|!|!|!|!' '„ ;i ..... 'fy \ ..... , ..... ' *"' f ...... ..... ...... , ........ . ....................... ....... . .
common blocks for memory allocation of the data from RXDT.EXT.
,;:,. • •* 1 ..... , •, ::>• ,- ..... it ( ...... *., " ........ I;' W : ' ; ..... \f- ........ /"'.?- ...... ....... :•; ; " • .. .......... *? ; "- ••-*:•
• RXDT.EXT - contains data statements associated with the parameters and arrays declared
hi RXCM.EXT. These data fill the arrays in the named common blocks in RXCM.EXT.
t : , •; . •
~ ..... 15-10
-------
EPA/600/R-99/030
The CCTM driver program must include both of these files in its declaration section. The
subroutines in the gas-phase chemistry solver process module must include only the common
block file, RXCM.EXT, That goes for the subroutine PHOT, as well. PHOT is called from the
gas-phase chemistry "class-driver," the top-level subroutine (see Section 18.2.2 in Chapter 18).
The CCTM driver, by including both files, causes the common blocks to be loaded with the
necessary data used by all the modules that need the chemistry data, including the process
analysis module, PROCAN.
The remaining sections describe the global include files associated with the four species tables.
As mentioned above, these files are automatically generated by the Models-3 framework,
according to the user-specified linkages and flags, and are targeted for inclusion in all model
routines that require the specific chemistry data in them. For additional details the reader is
referred to the Models-3 User Manual [4], Each of the include files contains declarations and
parameter statements for the loop index and array dimensioning for each particular species
group. If a particular species group is not used (e.g., aerosols or tracers), the include file contains
declarations and parameter statements that set the species loop counter to zero and its array
dimensioning parameter to one, thus maintaining complete generality and modularity within the
codes. In addition, by setting the dimensioning value to one (Fortran does not allow zero), the
compiled codes do not waste memory, a typical problem when coding in Fortran with some pre-
determined maximum dimension to hopefully account for all cases.
15.4.5.2 Model Species
The framework generates four model species include files that together contain the names and
molecular weights of all the chemical species available globally to the model.
The model species include files are:
• GC_SPC.EXT - gas-phase model species names and molecular weights.
i
• AE_SPC.EXT - aerosol species names and molecular weights.
• NR_SPC.EXT - non-reactive species names and molecular weights.
• TR_SPC.EXT - tracer species names and molecular weights.
The next three sections describe include files that are associated with data access by the CCTM
from external files. The data are referenced from I/O API files by means of file variable names
that are contained in headers in each of the files.
15.4.5.3 Emissions
The CCTM reads the emissions data from an I/O API file produced by ECIP using the surrogate
name concept discussed above. The data are read by file variable names (surrogates) and stored
in arrays that are linked to the corresponding model species names.
15-11
-------
EPA/60G/R-99/030
The CCTM has the option of emissions sources injected as part of the vertical diffusion process
oras part of the chemistry solver process, where the injected sources are treated as part of the
chemical production terms. To read source emissions data into the CCTM for processing either
in the vertical diffusion or in the gas-phase chemistry modules, the following include files are
required:
• GC_EMIS.EXT - gas-phase emissions surrogate species names with scale factors and
indices that point to the positions of the species in the model species name table (in
OC_SPC.EXT).
• AE_EMIS.EXT - aerosol emissions surrogate species names with scale factors and
» indicesjhaf point to the positions of the species in the model species name table (in
| AEJSPC.EXT).
. , . . .
• _ NRJEMIS.EXT - non-reactive emissions surrogate species names with scale factors and
indices that point to the positions of the species in the model species name table (in
NR_SPC.EXT).
• TR_EMIS,EXT - tracer emissions surrogate species names with scale factors and indices
that point to the positions of the species in the model species name table (in
TR_SPC.EXT).
,,v, y" • v ij / n|i , 'ii| ......
15.4.5.4 Initial and Boundary Conditions
For both the initial and boundary concentrations the CCTM reads I/O API input data by variable
name reference. The processing first checks to see if the model species name is on the file and
reads the data associated with that name. If the model species name is not on the file, it checks
for the surrogate name. If that, too is not available, the input value is set to a minimum value
("model zero"). Additional information on initial and boundary conditions and how they are
implemented in the Models-3/CMAQ system can be found in Chapter 14,
• " GC_ICBC.EXT - gas-phase initial and boundary conditions surrogate species names with
scale factors and indices that point to the positions of the species in the model species
name table (in GCJSPC.EXT),
• _ AE_ICBC.EXT - aerosol initial and boundary conditions surrogate species names with
scale factors and indices that point to the positions of the species in the model species
name table (in AE_SPC.EXT).
• NR_ICBC.EXT - non-reactive initial and boundary conditions surrogate species names
with scale factors and indices that point to the positions of the species in the model
species name table (in NR_SPCEXT).
• TR_ICBC.EXT - tracer initial and boundary conditions surrogate species names with
scale factors and indices that point to the positions of the species in the model species
name table (in TRJSPC.EXT)
' ?•: '• V *%' '• . • ' ' \. "•'•'•• . * If
~ 1*5-19
IM li *-**» " 1, +J I id*
-------
EPA/600/R-99/030
15.4.5.5 Dry Deposition
The following include files are used to read in dry deposition velocities from an I/O API file
produced by the meteorology-chemistry input processor, MCIP. Using the surrogate name
concept discussed above, the data are read by file variable names (surrogates) and stored in
arrays that are linked to the corresponding model species names.
• GCJDEPV.EXT - gas-phase deposition velocity surrogate species names with scale
factors and indices that point to the positions of the species in the model species name
table (in GC_SPC.EXT).
• AE_DEPV.EXT - aerosol deposition velocity surrogate species names with scale factors
and indices that point to the positions of the species in the model species name table (in
AE_SPC.EXT).
• NR_DEPV.EXT - non-reactive deposition velocity surrogate species names with scale
factors and indices that point to the positions of the species in the model species name
table (in NR_SPC.EXT).
• TRJDEPV.EXT - tracer deposition velocity surrogate species names with scale factors
and indices that point to the positions of the species in the model species name table (in
TR_SPC.EXT).
15.4.5.6 Wet Scavenging
In both the scavenging and the cross-phase linkages it is necessary to set up a mapping from the
gas-phase species to the generic species names that are associated with aerosols and aqueous-
phase chemistry and to the names of species that are absorbed by cloud and rain water. Because
of the generality with respect to the gas-phase species names, these generic names are set in the
subroutines associated with aerosols, aqueous chemistry, and removal by in-cloud and
precipitation scavenging. They are linked to the model species names by employing the
surrogate name concept described above.
• GC_SCAV.EXT - gas-phase scavenging surrogate species names with scale factors and
indices that point to the positions of the species in the model species name table (in
GC_SPC.EXT).
• AE_SCAV.EXT - aerosol scavenging surrogate species names with scale factors and
indices that point to the positions of the species in the model species name table (in
AE_SPC.EXT).
• NR_SCAV.EXT - non-reactive scavenging surrogate species names with scale factors
and indices that point to the positions of the species in the model species name table (in
NR_SPC.EXT).
15-13
-------
EPA/600/R-99/030
• TR_SCAV.EXT - tracer scavenging surrogate species names with scale factors and
indices that point to the positions of the species in the model species name table (in
TR_SPC.EXT).
15.4.5.7 Cross-Phase Linkage
• GC_G2AE.EXT - surrogate names for gas-phase model species that participate in aerosol
chemistry with scale factors and indices that point to the positions of the species in the
model species name table (in GC_SPC.EXT).
I ," ' ii i i ' • •». , / ' . •, : • :!', : 'i*:;-': .: .,.-: • .
• GC_G2AQ.EXT - surrogate names for gas-phase model species that participate in
aqueous-phase chemistry with scale factors and indices that point to the positions of the
species in the model species name table (in GC_SPC.EXT).
• AE_A2AQ.EXT - surrogate names for aerosol species that participate in aqueous-phase
chemistry with scale factors and indices that point to the positions of the species in the
model species name table (in AE_SPC.EXT).
• NR_N2AE.EXT - surrogate names for non-reactive species that participate in aerosol
chemistry with scale factors and indices that point to the positions of the species in the
model species name table (in NR_SPC.EXT).
• NR_N2AQ.EXT - surrogate names for non-reactive species that participate in aqueous-
phase chemistry with scale factors and indices that point to the positions of the species in
the model species name table (in NR_SPC.EXT).
• TR_T2AE.EXT - surrogate names for tracer species that participate in aerosol chemistry
with scale factors and indices that point to the positions of the species in the model
species name table (in TR_SPC.EXT).
• TR_T2AQ.EXT - surrogate names for tracer species that participate in aqueous-phase
chemistry with scale factors and indices that point to the positions of the species in the
model species name table (in TR_SPC.EXT).
15.4.5.8 Operational Choices
The Models-3 framework provides the capability of allowing the user to control how some of the
processing is carried out with respect to the chemical species in the CCTM. The following lists
describe the include files that the framework generates for the various processes:
1. Advected species: In order to save memory storage or to minimize computation, the user
may choose not to advect certain modeled species such as some radicals. In general, the
special "counter species" used in a mechanism should not be advected.
• GC_ADV.EXT - names of gas-phase model species that are advected and indices
that point to the positions of the species in the model species name table (in
GC_SPC.EXT).
15-14
-------
EPA/600/R-99/030
• AE_ADV.EXT - names of aerosol species that are advected and indices that point
to the positions of the species in the model species name table (in AE_SPC.EXT).
• NR_ADV.EXT - names of non-reactive species that are advected and indices that
point to the positions of the species in the model species name table (in
NR_SPC.EXT).
• TR_ADV.EXT - names of tracer species that are advected and indices that point
to the positions of the species in the model species name table (in TR_SPC.EXT).
2. Species that undergo diffusive processes: The comment above, for advected species,
applies also to diffused species.
• GC_DIFF.EXT - names of gas-phase model species that are diffused and indices
that point to the positions of the species in the model species name table (in
GC_SPC.EXT).
• AE_DIFF.EXT - names of aerosol species that are diffused and indices that point
to the positions of the species in the model species name table (in AE_SPC.EXT).
• NR_DIFF.EXT - names of non-reactive species that are diffused and indices that
point to the positions of the species in the model species name table (in
NR_SPC.EXT).
• TR_DIFF.EXT - names of tracer species that are diffused and indices that point to
the positions of the species in the model species name table (in TR_SPC.EXT).
3. Species that are saved to the dry deposition output file: Generally, all the dry
deposition species that are modeled would be saved to the output file, but a user might
elect to reduce the number and save only a subset.
• GC_DDEP.EXT - names of gas-phase model species that are written to the dry
deposition output file and indices that point to the positions of the species in the
model species name table (in GC_SPC.EXT).
• AE_DDEP.EXT - names of aerosol species that are written to the dry deposition
output file and indices that point to the positions of the species in the model
species name table (in AE_SPC.EXT).
• NR_DDEP.EXT - names of non-reactive species that are written to the dry
deposition output file and indices that point to the positions of the species in the
model species name table (in NR_SPC.EXT).
• TR_DDEP.EXT - names of tracer species that are written to the dry deposition
output file and indices that point to the positions of the species in the model
species name table (in TR_SPC.EXT).
4. Species that are saved to the wet deposition output file: See the comment in item 3.
• GC_WDEP.EXT - names of gas-phase model species that are written to the wet
deposition output file and indices that point to the positions of the species in the
model species name table (in GC_SPC.EXT).
• AE_WDEP.EXT - names of aerosol species that are written to the wet deposition
output file and indices that point to the positions of the species in the model
species name table (in AE_SPC.EXT).
15-15
-------
EPA/600/R-99/030
• NR_WDEP.EXT - names of non-reactive species that are written to the wet
deposition output file and indices that point to the positions of the species in the
model species name table (in NR_SPC.EXT).
• TR_WDEP.EXT - names of tracer species that are written to the wet deposition
output file and indices that point to the positions of the species in the model
species name table (in TR_SPC.EXT).
!!!!:;,.,l . •» ill inn . . „„ : , •• . '] ' '" . •" ;• ji . , I
5. Species that are saved to the concentration output file: See the comment in item 3.
• GC_CONC.EXT - names of gas-phase model species that are written to the
concentration output file and indices that point to the positions of the species in
the model species name table (in GC_SPC.EXT).
• AE_CONC.EXT - names of aerosol species that are written to the concentration
output file and indices that point to the positions of the species in the model
species name table (in AE_SPC.EXT).
• NR_CONC.EXT - names of non-reactive species that are written to the
'concentration output file and indices that point to the positions of the species in
the model species name table (in NR_SPC.EXT).
• TR_CONC.EXT - names of tracer species that are written to the concentration
output file and indices that point to the positions of the species in the model
species name table (in TR_SPC.EXT).
15.4.5.9 Tracer Species
The use of tracer species is purely user-determined. An application with tracers can provide the
modeler with insights into how the model is simulating various physical processes, like
advection or diffusion. If tracer species are to be modeled, the user must have created the special
table entries appropriate to the application. In addition special data, such as emissions or initial
and boundary tracer concentrations, must be created in the corresponding I/O API files. The
specialized data creation is outside the scope of the Models-3 framework and must be carried out
by the user in conjunction with the application that is being modeled.
15.6 Conclusion
In this chapter we have described how PCP helps to establish modularity and consistency for
particular applications of models in the CMAQ system. The Models-3 framework processing,
through PCP, enables model developers and model users to study the effects of different
implementations of the science and codes or to execute different applications at various scales
with different chemical mechanisms or with different numerical solvers. The design and
implementation of the concepts in the program control processing helps to establish a true "one-
atmosphere" approach to modeling.
15.7 References
[1] "The EDSS/Models-3 I/O API" http://sage.mcnc.Org/products/I/O API/
15-16
-------
EPA/600/R-99/030
[2] Jeffries H. E. (1990), "User's Guide to Photochemical Kinetics Simulation System PC-
PKSS Software Version 3," Chapel Hill, N.C. 27514
[3] Gery M. W. and R. R. Grouse (1990), "User's Guide for Executing OZIPR," EPA/600/8-
90/069, U. S. Environmental protection Agency, Research Triangle Park, NC 27711
[4J EPA Third-Generation Air Quality Modeling System Models-3 Volume 9b, User
Manual, Appendices, June 1998, EPA-600/R-98/069(b)
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
15-17
-------
EPA/600/R-99/030
Chapter 16
PROCESS ANALYSIS
Gerald L. Gipson"
Human Exposure and Atmospheric Sciences Division
National Exposure Research Laboratory
U. S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
The implementation of process analysis techniques in the Models-3 Community Multiscale Air
Quality (CMAQ) modeling system is described in this chapter. These techniques can be used in
Eulerian photochemical models such as the CMAQ Chemical Transport Model (CCTM) to obtain
information that provides insights into how model predictions are obtained. This type of
information is particularly useful when modeling nonlinear systems such as atmospheric
photochemistry. The two techniques available in the CMAQ system - integrated process rate
(IPR) analysis and integrated reaction rate analysis (IRR) — are each described. The manner in
which IPR analysis can be used to determine the relative contributions of individual physical and
chemical processes is presented. Descriptions of how to employ IRR analysis to elucidate
important chemical pathways and to identify key chemical characteristics are included. Finally, the
procedures used to apply each technique in the CMAQ system are also described.
Corresponding author address: Gerald L. Gipson, MD-80, Research Triangle Park, NC 27711. E-mail:
ggb@hpcc.epa.gov
-------
EPA/600/R-99/030
t*' -'' v" v vi ' ' • • . • •- ..-••*. • i f.
??:••. n !' ij5 tJ '. • ' . ' ' ' f 1 " ' , " • * •' "
16.0 PROCESS ANALYSIS
'- ' ., • - *'>. * -.- f" -.r : i i , :.'.'
A major function of air pollution models is to predict the spatial and temporal distributions of
ambient air pollutants and other species. For complex Eulerian grid models, output concentration
fields of these species are determined by solving systems of partial differential equations. These
equations define the time-rate of change in species concentrations due to a series of physical and
' chemical processes (e.g., emissions, chemical reaction, horizontal advection, etc.). Since most
grid models are configured to output only the concentration fields that reflect the cumulative
effect of all processes, information about the impact of individual processes is usually not
available. Grid models can be configured to provide quantitative information on the effects of the
chemical reactions and other atmospheric processes that are being simulated, however (Pleim,
1990; Jeffries and Tonnesen, 1994; Jang et al., 1995a,b). This type of information has been used
to develop various process analyses that provide descriptions of how a model obtained its
predictions. This chapter provides background information on these methods and describes how
process analysis is implemented in the Models-3 Community Multiscale Air Quality (CMAQ)
modeling system.
Although process analysis does not have to be included in a grid model application, it can provide
supplemental information that can be quite useful in assessing a model's performance. Quantifying
the contributions of individual processes to model predictions provides a fundamental explanation
of the reasons for a model's predictions and shows the relative importance of each process. This
information can be useful in identifying potential sources of error in the model formulation or its
inputs. It can also be useful in interpreting model outputs, particularly with respect to
understanding differences in model predictions that occur from a change to the model itself or to
its input. Further, information provided from the chemical process analysis can be used to
determine important characteristics of different chemical mechanisms. This is particularly useful
for investigating mechanistic differences under different chemical regimes (e.g., VOC versus NOX
limiting conditions).
The inclusion of process analysis in a model application is generally carried out in two steps.
First, the model itself is "instrumented" (i.e., additional code or modules are added to the model)
to produce supplemental outputs about the contributions of individual processes and different
chemical reaction pathways to the model predictions. These data are then used with the
concentraJionJfieWsjn postprocessing operations to provide quantitative explanations of the
factors affectmg junodel's predictions. Although several specific postprocessing techniques have
been developed to reveafparticular model features (e.g., Jeffries and Tonnesen, 1994; Jang et al.,
1995a,b), process analysis data can be extracted and analyzed in many different ways. The
implementation of process analysis in the CMAQ system has been structured to facilitate data
extraction for subsequent model analysis. Although the main focus of this chapter is on data
extraction techniques, some example process analyses are also presented to illustrate particular
applications.
16-2
-------
EPA/600/R-99/030
For purposes of discussion, it is convenient to separate process analysis into two parts: integrated
process rate (IPR) analysis and integrated reaction rate (1RR) analysis. The first deals with the
effects of all the physical processes and the net effect of chemistry on model predictions. IRR
analysis deals with the details of the chemical transformations that are described in the model's
chemical mechanism. In general, IPR analyses are generally much easier to apply and understand
than IRR analyses since the latter typically requires a fairly thorough understanding of
atmospheric chemistry. Thus, the discussion below describes each analysis method separately,
starting with IPR analyses. Users not familiar with atmospheric chemistry details may wish to
omit the sections on IRR analysis. It should be added that either analysis method can be applied
independently of the other.
The CMAQ implementation of process analysis includes a flexible user interface that allows the
user to request only those particular outputs that are needed for model analysis. The information
generated for both IPR analysis and IRR analysis is controlled by the Process Analysis Control
Program (PACP). The PACP processes user-specified commands to instrument the CMAQ
Chemical Transport Model (hereafter referred to as the CCTM) to generate the specific outputs
that are selected by the user. As a consequence, the PACP must be invoked before configuring a
CCTM and running a simulation. The details of how the PACP works and the syntax for the
commands are covered in the Models-3 User Manual (EPA, 1998), and they will not be repeated
in their entirety here. Nevertheless, some of the command syntax and some simple examples are
presented in the discussions that follow to illustrate how the PACP is used to set up a particular
process analysis.
16.1 Integrated Process Rate Analysis
The governing equation for Eulerian models is the species continuity equation. Application of
the continuity equation to a group of chemically reactive species results in a system of partial
differential equations (PDEs) that gives the time-rate of change in species concentration as a
function of the rates of change due to various chemical and physical processes that determine the
ambient species concentrations. As noted in the introduction, the concentration fields that are the
numerical solutions to these PDEs reveal only the'net effects of all processes. This section is
concerned with how the contributions of individual processes are determined and used in process
analyses. The first two subsections deal with the calculation and use of IPRs in general. The last
two subsections describe the CMAQ implementation of IPR analysis and the use of the PACP to
set up an IPR analysis.
16.1.1 Computation of Integrated Process Rates
All Eulerian models utilize the technique of operator splitting. As a result, it is relatively easy to
obtain quantitative information about the contribution of individual processes to total
concentrations. In operator splitting, solutions to the system of PDEs are obtained by separating
the continuity equation for each species into several simpler PDEs or ordinary differential
equations (ODEs) that give the impact of only one or two processes. These simpler PDEs or
16-3
-------
EPA/600/R-99/030
*. :»
I- ; >•'*
ODEs^arejhen .solved separately to arrive at the final concentration. To illustrate, consider the
simple case of two-dimensional horizontal advection of a single species in the absence of any
other processes, for which the governing equation can be expressed as follows:
dt
dx
dy
= 0
(16-1)
where c is the species concentration, and u and v are the x- and ^-components of the wind
velocity vector, respectively. With operator splitting, this 2-dimensional equation is split into two
I-dimensional operators, one for each direction:
|c
dt
(16-2a)
and
(16-2b)
. ... , .. . - .
These two equations are then solved sequentially, with the solution to the first being used as the
initial condition for the second. The solution to the second equation then represents the final
solution and gives the net effect of 2-dimensional advection.
The final solution for the example presented above can also be represented as follows:
'<*!•
= c(t) * (Ac), + (Ac)
(16-3)
where c(t+&f) is the final solution, c(t) is the initial condition for the 2-dimensional problem, and
(Ac), and (Ac),, are the changes in concentration produced by each of the 1 -dimensional
operators, (Ac), and (Ac),, give the impact of each operator in moving from the initial to the final
concentration and are equivalent to the results obtained by integrating the process rates
individually. Hence the term integrated process rates is used to describe them. Note that they can
Se computed with little additional work since they are simply equal to the difference between the
final and initial concentrations for each operator.
From the above example, it should be evident that a general mathematical representation of IPRs
for individual processes can be expressed as follows:
(16-4)
where (Ac),, is the change in a species' concentration due to operator n, Ln is the differential
operator associated with a process, and Af is the model synchronization time step (which is
16-4
-------
EPA/600/R-99/030
equivalent to Afsyncin Chapter 6). Refer to Chapter 6 for a discussion of the various time steps
used in CMAQ. The integration in equation (16-4) is performed by the model regardless of
whether it has been instrumented for process analysis. Thus, it is only necessary to save the (Ac)n
to obtain process analysis capabilities. In a few cases, however, models may be structured such
that one operator deals with two or three processes simultaneously. In those cases, the (Ac),,
obtained after the integration would represent the compound effect of all of those processes and it
would not be possible to discern the impacts of the individual processes. Normally, the only way
that information could be obtained would be to integrate the process rates separately. In some
instances, however, it may still be possible to isolate the impacts of the individual processes
without performing additional integrations. For example; the CCTM treats vertical diffusion,
emissions, and dry deposition simultaneously in one operator, but the amount of material
deposited by dry deposition is tabulated and the amount of material that is emitted is known. As a
consequence, mass balance techniques can be used to compute the contribution of each process
subsequent to the simultaneous integration of the process rates without separately integrating
each process rate. Thus, the IPRs that are available for process analysis are to some degree
determined by the underlying structure of the photochemical model that is being used and by the
effort that is invested in separating individual components when processes are coupled in a single
operator.
Analogous to equation (16-3), the concentration at the end of a time step can be expressed as
follows:
c(/+AO = c(t) + (Ac). , (16-5)
«=i
where the model is assumed to have N operators. It should be noted that most IPRs can be either
positive or negative since most processes can cause concentrations to either increase or decrease.
Further, it should also be evident that the IPRs in the above expression are additive. Thus, for
example, the IPRs for horizontal advection and vertical advection could be summed to give one
IPR that represents the net impact of the two advection processes. The differential operators (£„)
themselves are most often nonlinear, however. Because of these nonlinearities, the magnitude of
the IPRs for most processes would change if the order of the model's operators was altered or
even if only one of the operators was changed. Thus, the additive property for IPRs holds only
for a particular application of the model.
16.1.2 Example IPR Analyses
The tabulation and subsequent output of IPRs provide the user with quantitative information on
the effects of individual processes, and these can be examined and depicted in a number of ways.
Figures 16-1 and 16-2 contain two types of displays that were developed to depict process
contribution data graphically (Jeffries, 1996). Figure 16-1 is a time series plot showing both
predicted concentrations and integrated process rates. This type of plot shows the hourly
variations at a cell (or group of cells if the data are aggregated) of a predicted species
16-5
-------
EPA/600/R-99/030
concentration and the change in concentration caused by each process (i.e., the IPRs). Plots of
this type illustrate the variations in process contributions during the simulation period. Figure 16-
2 shows process contributions and total concentrations for different model formulations, in this
case, different grid resolutions. Here the data have been aggregated over several cells and hours,
but could be developed for a single cell or time period just as well. This figure highlights how
cumulative process contributions are altered by the different model formulations. For a more
thorough discussion of how this type of data can be used to assist in the evaluation of a model's
performance, the reader is referred to Jang et al. (1995a and 1995b) and to Pleim (1990).
16.1.3 Implementation of IPR Analysis in the CMAQ System
* ' •' "* T '?' , : • ' , .• • v,
The previous section illustrated that instrumenting a model for process analysis involves providing
the capability to capture IPRs, Since an IPR can be calculated for every combination of process
and species, the amount of output data that can be generated is substantial. As a consequence,
the PACP has been designed to allow substantial flexibility in selecting the particular IPRs for
output. This is accomplished primarily by allowing the user to choose only those particular
species/process combinations that are of interest. Additional control and flexibility are provided
by including the ability to produce lumped IPRs, by allowing special species families to be defined,
and by providing controls to limit the size of the modeling domain for which outputs are
generated. These will be illustrated in the examples presented below.
The physical processes that are simulated in the CCTM and hence are available for IPR analysis
are shown in Table 16-1. As will be illustrated below, these processes are referenced in the PACP
by the codes shown in the first column. The procedure for selecting specific IPRs for output is
species oriented. Hence, a user selects a species and then indicates which processes will be
included in the IPR output. The species are referenced by their model names. In addition, the
user may also define a family of species that is a linear combination of the model species (e.g.,
defining NOx as the sum of NO and NO2) and extract IPRs for the family. This can be useful in
saving disk space occupied by the output IPR files when information about individual members of
a family is not needed.
It should be apparent from Table 16-1 that IPRs for some species will always be zero (e.g., those
species not emitted always have zero IPRs for that process). Thus, the size of the output file can
be minimized by not extracting those IPRs. The PACP also contains an option that allows the
user to limit the amount of output data by extracting IPR outputs for only part of the modeling
domain. Currently, the user is restricted to selecting a single, contiguous block of cells within the
domain for the IPR outputs. The block is defined relative to the modeling domain by selecting a
starting and ending column, row and level. A possible future enhancement would be to provide a
graphical interface to allow the user to select any particular cell or group of cells within the
domain.
The CCTM model has been instrumented to write the IPRs to an output file at the same time as
the output concentration files are written. Thus, the IPR outputs represent the cumulative impact
16-6
-------
EPA/600/R-99/030
of the process integrations over the entire output time interval. Process contributions over shorter
time intervals can be obtained by increasing the frequency of writing outputs of both the
concentration fields and the IPRs, Finally, the IPR output files are standard Models-3 IO/API
gridded files, and can be viewed with the Models-3 visualization tools described in Models-3 User
Manual (EPA, 1998),
16.1.4 Use of the PACP to set up an IPR Analysis
This section illustrates how the PACP can be used to generate the IPR data for an analysis.
Details on formatting inputs and using the PACP are contained in the Models-3 User Manual
(EPA, 1998). This section borrows from that discussion to illustrate how the PACP is used. The
user selects and controls the form of the IPR output data by means of a PACP command file. A
few predefined command files are available to set up an analysis, or users can generate their own
files to customize their analyses. A command file consists of a series of commands and definitions
that contain instructions for generating IPR outputs. The commands are input in a free form
format to facilitate encoding, and they contain special keywords that have specific meaning to the
PACP. The commands related to IPR analysis have been divided into two groups: Global
commands and IPR output commands. A description of the commands within each group will be
presented first, followed by an example illustrating how these commands are used. In the
description that follows, the syntax for each command is given first, with bold type used for
PACP keywords and normal type used for user supplied input. Alternative inputs are separated
by vertical bars and completely optional inputs are enclosed in curly braces.
Global Commands:
OUTPUT_DOMAIN = {LOCOL[n,] + HICOL[n2J + LOROW[n3l + HIROW[n4J
+ LOLEV[n5l + HILEV[n6]};
The OUTPUT_DOMAIN command provides the capability to limit
the IPR output data to only one portion of the modeling domain.
The n{ in brackets are numbers that define the bounds of the output
domain relative to the number of columns, rows, and vertical levels
in the modeling domain. Thus, for example, the value for n, must
be greater than or equal to one and less than or equal to the number
of columns in the domain. If this command is included, at least one
domain specifier must be present, and the end of the domain is used
for any that are missing. If the command is omitted entirely, output
is generated for the entire domain.
DEFINE FAMILY familyname = {c,*}species, {+ {c2*}species2 + ...}•
The DEFINE FAMILY command is used to define a group of
species as members of a family. The user specified "familyname"
16-7
-------
EPA/600/R-99/030
must be unique, and can be referenced in subsequent commands.
|L: ' ' The Cj are numerical coefficients that default to one if not specified;
"specieSj" are the names of individual model species.
ENDPA;
"The ENDPA command signifies the end of the command input in
' "the PACP command file.
>', • • «• J~ !l ' L !•• •- / ••• . , • .: -. • - • -',
IPR Output Command:
IPR_OUTPUT species|familyname|ALL {= pcode, + pcode2 + ...};
The IPRJDUTPUT command defines specific IPR outputs to be
...... , generated during a CMAQ simulation. A species name, family
** "* ' * < keyword ALL must follow the IRRJDUTPUT
keyword. The keyword ALL refers to all model species. IPRs are
- • — generated for the selected species or family, and they are controlled
by the specified values of pcodej, where pcodej corresponds to one
of the process codes listed in Table 16-1. If no process codes are
_ T:-.. "spe^ifjec^ IPRS ^n t,e generated for every process. The output
r * variables that are generated are named either species_peodej or
* ..... *" ' ' " "•• ..... S familyname_pcodej
'-" • -•« >*' ".•"VJ-."' l>, .' . . , .. ... • ,.j- - -MfcV • -. ,. •'.-•!•'-.
A listing of an example PACP command file is contained in Exhibit 16-1. To facilitate the
discussion that follows, the commands have been numbered, although this is not required by the
PACP. (Note that all information enclosed by curly braces in a command file is treated as
comments.) Each numbered line represents a command, and the input for each command is
terminated by a semicolon. This particular set of commands causes the CCTM to generate
several individual IPRs for a special user-defined sub-domain. Each of the commands is described
below.
Command 1 is used to restrict the process analysis output to a subset of the modeling domain. As
noted above, output would be generated for the entire computational domain if the
OUTPUTJDOMAIN were omitted. Since keywords for columns and rows are not present, the
PACP default is to include all rows and columns in the modeling domain. The keywords
"LOLEV" and "HILEV" restrict the output for the vertical level to layers 1 through 2. Thus, the
net effect of this command is to limit the IPR output to all cells within the first two vertical levels
of the modeling domain.
Commands 2 and 3 are used to define families of species. The species names to the right of the
equal sign are model species. The effect of the numerical coefficients in the definition of the VOC
16-8
-------
EPA/600/R-99/030
families is to convert the units of the IPRs for individual organics from ppm to ppmC. As will be
seen, these "defined" family names are referenced in subsequent commands.
Commands 4 through 8 actually cause IPR outputs to be generated. As described above, IPRs
are generated for the species and the processes that are referenced by means of the codes listed in
Table 16-1. Since no process codes are specified in commands 4 and 5, IPRs will be generated
for each of the twelve processes in Table 16-1 for the families NOX and VOC (i.e., a total of 24
IPRs). Command 6 causes three IPRs to be generated for species O3, one each for total
transport, chemistry, and clouds. Command 7 causes an IPR for vertical diffusion to be generated
for every model species. The last command signifies the end of the command inputs.
The CCTM will generate the IPR output data in a form comparable to the output concentration
files. Hence, the data are written to standard Models-3 IO/API gridded data files. The units of
the IPR data are normally the same as those for the species concentration data (i.e., ppm for gas-
phase species and either |ig/m3 or number/m3 for aerosols). As described in the example of
creating the family VOC with units of ppmC, however, different output units can be created using
special family definitions.
The example just described was formulated primarily to illustrate how IPR commands are
structured to collect IPR information during a model simulation. In general, the particular data
that are collected for process analysis would be determined by the needs of the study, and thus it
is difficult to define a "default" process analysis. Nevertheless, a minimal process analysis for
studying ozone formation might involve collecting the process contributions for NOx, VOC, and
for ozone. The example in Exhibit 16-1 could be used as the starting point for such an analysis.
Commands 4 and 5 capture all the IPRs for NOx and VOC. Command 6 could be modified to
capture all IPRs for ozone as well. Since it would not normally be necessary to capture vertical
diffusion IPRs for all species, command 7 could be dropped. Of course, the user would still be
required to define the domain for outputs (command 1) and to define the VOC family for the
mechanism that is being used (command 3). Thus, commands 1 through 5 with command 6
modified to collect all IPRs for ozone would provide some very basic process analysis information
on the formatio of ozone during a simulation.
16.2 Integrated Reaction Rate Analysis
The second major component of process analysis is IRR analysis. It is applied to investigate gas-
phase chemical transformations that are simulated in the model. Its primary use to date has been
to help explain how ambient ozone is formed in the chemical mechanisms that are used in
photochemical models (e.g., Jeffries and Tonnesen, 1994; Tonnesen and Jeffries, 1994). Thus,
the CMAQ implementation of IRR analysis currently addresses only gas-phase reactions.
Nevertheless, the concepts should be adaptable to the modules simulating aerosol formation and
aqueous chemistry as well, and this is an area for future enhancement in the CCTM. The
remainder of this section describes how the IRRs are calculated and generated in the CMAQ
system.
16-9
-------
EPA/600/R-99/Q30
16,2.1 Computation of Integrated Reaction Rates
As described in Chapter 8, the simulation of atmospheric chemistry is a key component of a
photochemical air quality model. The operator splitting techniques that are used in Eulerian
photochemical models typically result in a set of nonlinear, coupled, ordinary differential
equations (ODEs) that describe chemical interactions in the gas-phase. Solutions to these ODEs
are obtained using numerical solvers to compute species concentrations as a function of time.
Again, these computed concentrations show only the net effect of chemical transformations. As
has been done for physical processes, a technique has been developed to provide quantitative
information on individual chemical transformations (Jeffries and Tonnesen, 1994). Since the
technique involves integrating the rates of the individual chemical reaction, the method is termed
integrated reaction rate analysis.
As was noted in section 8.3.1, the mathematical expression for the rate of a chemical reaction
takes one of the forms of equation set 8-3. The reaction rate is used to compute the change in
species concentration that is caused by the reaction. Mathematically, this can be expressed as
follows: :
r
(16-6)
where Mt refers to the integrated reaction rate (IRR) for reaction /, A/ is the model
synchronization time step used by the chemical solver, and r, is the rate of reaction /
corresponding to one of the forms of equation set 8-3. The value of M, represents the total
throughput of the reaction, and can be used with the appropriate stoichiometry to determine the
amount of an individual species that is produced or consumed by the reaction. For example,
assume the IRR for the reaction A + B - 2C is 20 ppb for a given time period. Then, the amount
of A and B consumed in that time period by this reaction is 20 ppb, and correspondingly, the
amount of C produced is 40 ppb. Further, the net change in a species concentration due to all
chemical reactions is equivalent to the sum of all its production terms less the sum of all its loss
terms. As a consequence, the contribution of each reaction to the change in concentration of any
species is directly available from the IRRs. With this information, it is possible to identify the
important chemical pathways that affect species concentrations and thereby unravel the complex
chemical interactions that are being simulated.
As described in Chapter 8-4, the chemistry solvers used in air quality models typically employ
marching methods that compute species concentrations at the end of a time step given the
concentrations at the beginning of the step. Since the solvers adjust the time steps to maintain
stability and accuracy, the reaction rates should not vary too greatly over a given time step. As a
eoTfsequence, it is possible to use a fairly simple numerical integration technique to compute the
IRRs. The technique used in the CCTM is the same the one used by Jeffries and Tonnesen (1994)
~ the trapezoid rule. With this method, the IRRs are computed as follows:
16-10
-------
EPA/600XR-99/030
M,((+&t) = M,(t) + [r/0 * rX/+A/)] . (16-7)
Thus, the IRRs are simply computed from the values of the reaction rates at the beginning and the
end of each chemistry integration time step. As the chemistry solver marches through time, the
variable M, accumulates the IRRs over the simulation period. Although an IRR could be
accumulated for the entire simulation, more information can be gained on how the mechanistic
processes vary with time if the accumulated IRRs are periodically output and the value of M, is
reset to zero. In the CCTM, the IRR output is synchronized with the outputs for the
concentration fields and the integrated process rates. Thus, just like the IPRs, the IRJRs that are
output represent the integral of the reaction rates over the output time interval.
16.2.2 Example IRR Analyses
Most IRR analysis performed to date has been devoted to studying mechanistic processes that
affect tropospheric ozone formation. One method that has been used involves analyzing two
important, interacting cycles: the OH radical cycle and the NOX oxidation cycle. This section
contains a brief discussion of these cycles and other important chemical parameters to illustrate
how integrated reaction rate analysis can be used to understand different chemical pathways. For
more comprehensive discussions, the reader is referred to Jeffries (1995), Jeffries and Tonnesen
(1994), and Tonnesen and Jeffries (1994).
\
Ozone is formed in the atmosphere via the photochemical cycle
NO2 + hv - NO + O(3P) (Rl)
O(3P) + O2 - O3 (R2)
NO + O3 - NO2 . (R3)
If these were the only reactions taking place in the atmosphere, an equilibrium condition would be
established that would determine the ozone concentration. These levels are almost always lower
than what is observed in the atmosphere because of other interactions (Seinfeld, 1998). The
presence of free radicals can alter the ozone production rate through the following reactions that
are competitive with R3 :
HO2 + NO - NO2 + OH (R4)
RO2 + NO - NO2 + RO . (R5)
In these reactions, NO2 is produced in reactions that do not consume ozone, thereby introducing a
process by which ozone can accumulate. Furthermore, these reactions are radical propagation
reactions since a hydroxyl radical (OH) is produced from the hydroperoxy radical (HO2) and an
16-11
-------
EPA/600/R-99/030
alkpxy radical (RO) is produced from an alkylperoxy radical (RO2). These product radicals are
then available to participate in other reactions, some of which lead to the regeneration of the
peroxy radicals. Thus, the production of ozone can be viewed as an autocatalytic process since
O3 can be produced without the loss of its precursor NO2. In the atmosphere, however, ozone
production is limited by termination reactions that remove either radicals or NO2 from the system.
The ozone formation process has been represented by means of a schematic diagram, shown in
Figure 16-3a, that consists of two interacting cycles (Jeffries, 1995). The top portion of the figure
is the OH reaction cycle which consists of the three principal categories of reactions involving
radicals: initiation, propagation, and termination. Radical initiation reactions are almost always
photolytic reactions that generate "new" radicals. Examples include the photolysis of
formaldehyde, hydrogen peroxide and ozone . Radical propagation reactions include reactions
such as R4 and R5 in which NO is converted to NO2 but no radicals are lost. Many reactions in
which organic compounds are oxidized also propagate radicals. The termination reactions
remove radicals through the formation of stable products. The bottom portion of Figure 16-3a
represents the NOX oxidation cycle. As with the radical cycle, processes in this cycle are classified
as either initiation, propagation, or termination. The initiation process for NOX corresponds to
emissions of NOX, and the termination process corresponds to reactions in which NOX is converted
to'stable products. The NOX cycle is connected to the radical cycle by means of the propagation
steps that convert NO to NO2.
One form of an IRR analysis that can be conducted involves using the IRRs to develop
quantitative information about the various initiation, propagation, and termination processes.
Figure 16-3b shows a cycle diagram similar to Figure 16-3a in which the IRRs have been used to
compute the numerical values that are shown. These values include net throughput for several
parameters that serve to further characterize the state of the reacting system. For example, the
fraction of OH that is regenerated by the chemical reactions (0.776 in Figure 16-3b) can be
determined from the amount of OH that reacts (142.7 ppb) and the amount that is re-created
(110.8 ppb). This is directly related to the OH chain length parameter (4.46) which corresponds
to the average number of times each new OH is cycled until it is removed from the system. An
analogous chain length parameter (5.13) can be calculated for the NOX oxidation cycle in the form
of the NO chain length. The longer these chain lengths, the greater the potential for O3 formation
per unit of NOX emissions. Other parameters such as the "NO oxidations per VOC consumed"
[(NO-NO^/VOC =1.71] and "O3 produced per O3P generated by NO2 photolysis" ([O3y[O3P]
= 0.951) further quantify the relative efficiency of ozone production, the former being particularly
useful in providing a measure of the reactivity of the organic compounds.
These types of analyses are particularly useful for comparing model results that are obtained using
different chemical mechanisms or that are obtained at different locations with the same chemical
mechanism. Other types of IRR analyses can provide other information as well. For example,
IRRs have been used to allocate the total production of O3 to the individual VOCs, to examine
individual characteristics of various chemical mechanisms such as yields of radicals from different
organics, and to quantify other entities such as the production and loss of odd-oxygen (Ox), which
16-12
-------
EPA/600/R-99/030
can serve as surrogate for tracking the formation of ozone (Jeffries and Tonnesen, 1994;
Tonnesen and Jeffries, 1994). It can be expected that many other types of IRR analyses will be
developed, especially as newer and possibly more complex chemical mechanisms are developed
and used in models.
16.2.3 Implementation of IRR Analysis the CMAQ system
As described in Chapter 8, the CMAQ system is designed to treat chemical mechanisms in a
generalized manner. Since a specific chemical mechanism is not embedded in the CCTM, a
comparable generalized method is needed to link IRR analysis with the chemical mechanism. The
technique that has been incorporated in the CMAQ system is one that provides the user with the
capability to formulate and then generate particular chemical parameters of interest. Presumably,
these parameters would be chosen to reveal special properties of the mechanism and/or to be used
in generating photochemical cycle diagrams such as Figure 16-3b. To illustrate, a special PACP
command file has been prepared for the RADM2 mechanism. For reference, a listing of the
RADM2 mechanism is contained in Exhibit 16-2, and the reader is referred to the Models-3 User
Manual for a detailed explanation of the mechanism format. For purposes of the discussion that
follows, it is sufficient to note that reaction labels are enclosed in "<" and ">" and precede each
reaction, and that the reactants and products in each reaction are model species. Both the
reaction labels and the species names will be referenced in PACP commands.
Table 16-2 lists the chemical parameters that are produced in this example IRR analysis for the
RADM2 mechanism. Most of these parameters could also be generated for other chemical
mechanisms, although the specific calculations that would have to be performed would necessarily
differ because of differences in the mechanisms. The parameters in Table 16-2 are computed as
simple, linear combinations of the IRRs that are calculated for each chemical reaction. The
domain controls that are specified for the integrated process rate outputs apply to IRR outputs as
well. Thus, both the IPR and IRR data will be generated for the same domain and written to
output files at the same time intervals. Similarly, the IRR output files are standard Models-3
IO/API gridded files, and can be used with the standard Models-3 visualization tools. In addition,
a special visualization tool that can generate "default" cycle diagrams similar to Figure 16-3b is
available and is described in Models-3 User Manual (EPA, 1998). Note, however, that the IPR
and IRR outputs are written to separate files.
One of the advantages of generating IRR data in the form of chemical parameters is that the
output file storage requirements can often be minimized. Note that there are fewer IRR
parameters in Table 16-2 than there are species in most mechanisms. The major disadvantage to
this approach is that new or different parameters cannot be computed without rerunning a model
simulation. As a consequence, the CMAQ implementation of IRR analysis also contains the
option to capture the complete set of IRRs rather than chemical parameters when a model is run.
With this option, one IRR is generated for each chemical reaction, and these IRRs can then be
manipulated in postprocessing routines to form any particular chemical parameter of interest.
Thus, it would be anticipated that the chemical parameters such as those in Table 16-2 would be
16-13
-------
EPA/600/R-99/030
generated in fairly routine model applications, whereas IRRs for each reaction would be generated
for exploratory analyses. The form of the outputs is controlled by the PACP commands that are
described next.
16.2.4 Use of the PACP to set up an IRR Analysis
As with IPR analysis, the IRR outputs that are generated by the CCTM are controlled by
commands and special operators that are processed by the PACP. The global commands that
were described in the IPR section also apply to the IRR output Thus, family names created with
the DEFINE FAMILY may also be used with several of the IRR commands and operators.
Before describing how the commands are used to construct an IRR analysis, a brief description of
each IRR command and operator is first presented. These are divided into three groups: IRR
global definitions, IRR operators and IRR output commands. Again, the same syntax conventions
are used, i.e., PACP keywords and symbols are in bold type, user supplied values are in normal
type, alternative inputs are separated by vertical bars, and completely optional inputs are enclosed
in braces.
IRR Global Definitions:
r -I* i
IRRJTYPE = FULL|PARTIAL[NONE;
The IRRJTYPE command defines the type of IRR analysis. With
the type set to FULL, IRRs for each reaction will be calculated and
written to the IRR output file, and all other IRR commands will be
,. ignored. IRR_TYPE set to PARTIAL indicates that the IRR
commands following this command are to be processed to
i, Jproduced user defined IRR outputs. Type set to NONE causes all
I *bther IRR commands to be ignored and no IRR output to be
generated. If the command is omitted, type PARTIAL is assumed.
* DEFINE CYCLE cyclename = species,;
The DEFINE CYCLE command is used to compute the net of all
i chemical production and loss of a species involved in more than one
cyclical reaction set. Thus, this quantity is computed by summing
the IRRs for all reactions in which a species is consumed, and then
; -: subtracting that sum from the sum of the IRRs for all reactions in
which the species is produced. The "cyclename" is a user defined
name that must be unique, and can be referenced in subsequent
- IRR_OUTPUT commands.
DEFINE RXNSUM sumname = {±}{c,*} { ± {c2*} ± ...};
16-14
-------
EPA/600/R-99/030
The RXSUM command is used to compute a linear combination of
IRRs for individual reactions that can then be referenced in a
subsequent IRR_OUTPUT command; "sumname" is user defined
and must be unique. The linear combination of IRRs is defined
according to the expressions following the equal signs that specify
the reaction IRRs to sum. The "rxlabl," is the reaction label that is
used by the generalized mechanism to identify each reaction and is
enclosed in "<" and ">". The "c," are optional numerical
coefficients that default to one if not specified.
IRR Output Operators:
PROD[species,I {FROM[species2| {ANDJOR [species3J }}
The production operator (PROD) is used to compute the total
production of a species by summing the IRRs of all reactions in
which species, appears as a product. The optional qualifiers
FROM, AND, and OR restrict the sum to include only those
reactions in which species2 and/or species3 are reactants; "species,"
can be any gas-phase mechanism species or a family of gas-phase
species that was defined using the DEFINE FAMILY command as
described in the IPR section; "species2" or species3" may also be the
keyword HV to restrict the selection to photolytic reactions.
NETP[species,] (FROMlspecies2] {AND]OR [species3] }}
The net production operator (NETP) is very similar to the
production operator PROD since it is used to compute the
production of a species. Whereas the PROD operator includes
every reaction in which species occurs as a product, the NETP
operator includes only those reactions in which the net production
of species, is greater than zero. Thus, if species, appears as both a
reactant and a product with equal stoichiometry in a reaction, the
PROD operator will include it but the NETP operator will not.
This operator is useful for getting the net production of a family,
for example, by eliminating those reactions in which the net effect
of the reaction on the family concentration is zero. The qualifiers
FROM, AND and OR restrict the inclusion of reactions to those in
which species2 and/or species3 are reactants.
LOSS [species ,j (AND|OR [species2] }
16-15
-------
EP^/600/R-99/030
The loss operator (LOSS) is used to compute the total loss of a
species by summing the IRRs of all reactions in which species,
appears as a reactant. The optional qualifier AND restricts the sum
to include only those reactions in which both species, and species2
are reactants. Similarly, the OR qualifier includes all reactions in
which either "species," or "species2" appears as a reactant. The
"species," or "species2" can be any gas-phase species in the
mechanism, a family name that includes only gas-phase mechanism
species, or the keyword HV to restrict the selection of reactions to
those that are photolytic.
NETL[species,] {AND|OR [species2] }}
The net loss operator (NETL) is very similar to the loss operator
since it is used to compute the loss of a species. However.it
includes only those reactions in which there is a net loss of
':.''' i i*!* in •*
• species," and/or "species2" . Thus, if species, appears as both a
f eacfant and a product with equal stoichiometry in the reaction, the
NETL operator will not include it in summing the loss of that
species, whereas the LOSS operator will include the IRR for that
reaction. This operator is useful for getting the net loss of a family
of species.
NET[species,]
The net operator (NET) is very similar to the CYCLE definition
since it gives the net of the production and the loss of a species for
all reactions in which "species," appears either as reactant or a
product; "species," may be any gas-phase, mechanism species or
any family consisting wholly of gas-phase mechanism species.
IRR Output Commands:
IRR_OUTPUT irrname = {c,*}opi|cyclname{qual,}|sumname{qual,}|
{± {c2*}op2|cyclname{qual2}|sumname{qual2}|+ ...
The IRR_OUTPUT command defines a specific IRR output to be
generated during a CCTM simulation. It is constructed by
specifying a linear combination of IRR operators, IRR global
definitions, or IRRs for specified reactions. Each individual term in
the combination must include either one of the five IRR operators
just described (i.e., opj), a cycle name, a reaction sum name, or a
reaction label enclosed in "greater than" and "less than" signs. The
16-16
-------
EPA/600/R-99/030
optional qualifiers (qua!,) for cyclename or reaction sum name can
be either POSONLY or NEGONLY. With these qualifiers, the
defined quantity is included as a term only when it is positive or
negative, respectively. If the name is not qualified, the quantity is
included regardless of sign. The numerical coefficients for each
term (Cj) are assumed to be one unless they are explicitly included.
The irmame that is supplied by the user will be assigned as the
variable name in the IO/APIIRR output file.
DESCRIPTION = 'description1;
The description command is provided to allow the user to specify a
long description of the output variable that will be included on the
IO/API IRR output name. If a description is not specified for an
IRR__OUTPUT variable, the irrname (or short name) will be used in
the output file. If the description command is used, it should be
located immediately following the IRR_OUTPUT command to
which it applies.
Before describing how these commands are actually used, some additional comments are
warranted. First, the specification of any particular IRR output might be accomplished in several
different ways. For example, the net production of a species could be obtained using a CYCLE
definition, a RXNSUM definition, a NET operator, or simply specifying the appropriate sum of
IRRs directly in the IRR_OUTPUT command (i.e., via reaction labels). Although the user is free
to choose any particular approach, some computational efficiencies may be achieved by using the
CYCLE and RXNSUM definitions. The CCTM has been constructed to compute these quantities
just once, and then use them whenever they are referenced in an output command. Conversely,
operator quantities are recomputed every time they are referenced. Thus, it is more efficient to
use the RXNSUM and CYCLE commands when they can be referenced several different times in
IRR_OUTPUT commands. Second, the NETP and NETL operators are probably most useful for
computing the production and loss of species families. When these operators are used, a reaction
is not included in the sum if there is no net loss or production of a family member in the reaction.
Thus, all reactions are eliminated from the computations when a member of a family is formed
from another member of the same family and there is no net impact on the family concentration.
Finally, the sign conventions employed in the CMAQ process analysis need to be defined. IRRs
for individual chemical reactions are always positive. Since IRRs can be subtracted when
computing CYCLE, RXNSUM, and NET quantities, the result can be either positive or negative.
The production and loss operators always produce positive values, however, since individual
IRRs are always summed in their computation.
To illustrate how these PACP commands are used to generate IRR output data, two examples are
presented. The first illustrates how to capture IRRs for each reaction. The second demonstrates
16-17
-------
EPA/600/R-99/030
how PACP commands are used to compute the special chemical parameters in Table 16-2 for the
RADM2 chemical mechanism.
Exhibit 16-3 contains the PACP commands for the first example that corresponds to a full IRR
analysis since IRRs will be calculated for each and every reaction. This option is invoked by the
first command that specifies that the IRR analysis type is FULL . Since no OUTPUT_DOMAIN
command is present, the IRR outputs will be generated for every cell in the modeling domain.
Exhibit 16-4 shows the PACP commands for the example IRR analysis that has been set up for
the RADM2 chemical mechanism. In a PACP command file, all lines that start with an
[' exclamation point (!) in the first column are comments. To facilitate the discussion below,
". ;'K comment lines have been used to block the IRR commands into special groups. The first four
;* groups contain global commands or definitions. All blocks after the first four contain the IRR
,; gQgjfj^jjg fl^ generate the particular parameters listed in Table 16-2.
. " 111 L. *.'" *,,JlfV J" Jl" "™" . ' .,""" 'in , • 1 • ' ' IPf • ' \ "*m "UK1 n. " " ' • -• • "f • - . «' ~
,, * i.. . The commands in the first group simply define the type of IRR analysis and the domain for which
*: * .- .the IRR outputs are to be generated. The second group includes family definitions. These
,'„ commands are of the same form as described for IPRs. The remaining two groups of commands
. ;tl ; define chemical cycles and reaction sums that are subsequently referenced in IRR OUTPUT
, ' * TT yR » * ifif-x i "%«„. ,»..»»„,»», *™, * »», -. - . , , . „ „ * ™™. . „. i, »=, „ _.,.,,. „ , . , —:,.
' tf commands. As noted above, the cycle commands give the net production or loss of a species by
all chemical reactions. Several of the reaction sums that are defined here are also cycles in that
they generate the net effect of a few reactions on the production or loss of a few particular
species. Most of the others are used to define special quantities. For example, the defined
RXNSUM newMO2 in Exhibit 16-4 corresponds to the production of new MO2, where new
refers to an initiation reaction for the radical MO2. As is apparent from the IRR_OUTPUT
commands in the subsequent blocks, the cycle and reaction-sum names are referenced fairly
' '" frequently.
All of the remaining blocks of commands in Exhibit ] 6-4 contain the commands for IRR outputs.
Again, one IRR output is generated for each IRR_OUTPUT command, and the outputs that are
produced correspond to the chemical parameters listed in Table 16-2. As indicated above, each
output is generated by the defined linear combination of predefined cycles, reaction sums, special
IRR operators, and/or specified reaction IRRs referenced by reaction label. It should be evident
that these commands are mechanism specific and require analysis of the mechanism itself to
formulate. Thus, this particular PACP command file would not be applicable to any mechanism
other than the RADM2. It should also be apparent that other important chemical parameters
could be formulated and generated in an analogous manner. In fact, this command file can be
used as the starting point for adding to or modifying some of the selected chemical parameters.
- . • - -• • ^= i ..»*.- m-irt * - •, u (*
•'•' • • -ft- "i" • ' > At:,' i • , f • , .(,«••.• » ,t J , , •'.
As with IPR outputs, the CCTM will generate the IRR output data in a form comparable to the
output concentration files. That is, the data are contained in standard Models-3 IO/API gridded
data files. The IRR outputs are linear combinations of individual reaction throughput, and thus
! ill! ' I ' i. r •«,... ' , ,':; ~ ... ... .. OI-'
have the same units as the gas-phase species concentrations (i.e., ppms). However, it should be
- -• •- 16-18
-------
EPA/600/R-99/030
remembered that these are throughputs calculated by integrating reaction rates over the output
time interval, and not simply abundances at a particular time.
16.3 Conclusion
Process Analysis is a diagnostic method for evaluating the inner workings of a model. Although
specific types of techniques have been performed and used in the past, new ways of examining
and analyzing process data are likely to be developed and used in the future. As a consequence,
the emphasis in the CMAQ implementation of process analysis has been placed on providing easy
methods of extracting key process data from the CCTM simulations. The CMAQ implementation
also provides tools to allow users to customize their analyses. These tools are designed to be easy
to use and not require coding changes to the model.
As noted previously, a process analysis is set up by using the PACP before constructing and
running a CCTM simulation. Both IPR and IRR data can be gathered during a simulation, but
each are written to separate output files. To collect both sets of output during a single simulation
requires that the PACP command file contain both the IPR commands and the IRR commands.
Although the IPR and IRR examples have been presented separately, both sets of outputs can be
produced with a single file containing both sets of commands. Recall, however, that the
OUTPUT_DOMAIN applies to both the IPR outputs and the IRR outputs. Thus, IPR and IRR
data cannot be generated for different parts of the domain in the same simulation.
The PACP program performs a substantial amount of error checking. The program will check
for the proper syntax of the input commands and perform some logic checking. For example, it
checks to make sure that all species referenced in IRR commands are gas-phase mechanism
species and that the members of defined families are either all gas-phase species or all aerosol
species. As is apparent from Exhibit 16-4, however, the inputs for a comprehensive IRR analysis
can be fairly extensive. As a consequence, the PACP produces an output report that summarizes
what IRR and IPR outputs are being requested. One of its major functions is to report on the
effects of the special IRR operators that are used in the PACP command file. A user may wish to
review this report before proceeding to run the CCTM to insure that the desired outputs will be
generated. The reader is referred to the Models-3 User Manual (EPA, 1998) for an example
output report.
Finally, the default configuration for the CCTM is to omit process analysis outputs entirely. Thus,
no process analysis will be generated in this configuration. Any process analysis must be set up
in the Science Manager of the Models-3 framework. The reader is referred to the Models-3 User
Manual (EPA, 1998) for details on how this is done.
16-19
-------
EPA/600/R-99/Q30 '' ' ' "
t • • s ,ii ' • •. j...,
16.4 References
Jang, J. C,, H. E. Jeffries, D. Byun, and J. E. Pleim, 1995a. "Sensitivity of Ozone to Model Grid
Resolution -1. Application of High-resolution Regional Acid Deposition Model", Atmospheric
Environment, Volume 29, No. 21, 3085-3100.
Jang, J. C,, H. E, Jeffries, and S. Tonnesen, 1995b. "Sensitivity of Ozone to Model Grid
Resolution - II. Detailed Process Analysis for Ozone Chemistry", Atmospheric Environment,
Volume 29, No. 21,3101-3114..
Jeffries, H. E., 1995. "Photochemical Air Pollution," Chapter 9 in Composition, Chemistry, and
Climate of the Atmosphere, Ed. H. B. Singh, Van Nostand-Reinhold, New York, N. Y.
Jeffries, H.E., 1996. "Ozone Chemistry and Transport", presentation to the FACA subcommittee
for Ozone, Particulate Matter and Regional Haze Implementation, March 21, Alexandria, Va.
Jeffries, H. E. and S. Tonnesen, 1994. "A Comparison of Two Photochemical Reaction
Mechanisms Using Mass Balance and Process Analysis", Atmospheric Environment, Volume 28,
No. 18,2991-3003.
**•• -H '. Hf" fl. *. ' >>• • >• , , .."...' ^- * .; * r>» , „ *
Pleim, J.E., 1990, Development and Application of New Modeling Techniques for Mesoscale
Atmospheric Chemistry, Ph.D. Thesis, State University of new York at Albany, Albany, New
York.
Seinfeld, J. H. and S. N. Pandis, 1998. Atmospheric Chemistry and Physics, From Air Pollution
to Climate Change, John Wiley and Sons, New York, New York.
Tonnesen, S. and H. E. Jeffries, 1994. "Inhibition of Odd Oxygen Production in the Carbon Bond
Four and Generic Reaction Set Mechanisms", Atmospheric Environment, Volume 28, No. 7,
1339-1349.
This chapter is taken from Science Algorithms of the EPA Modets-3 Community
Multiscale Air Qualify (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Ching, 1999.
16-20
-------
EPA/600/R-99/030
Table 16-1.
CCTM Processes and PACP Codes
PACP Code Process Description
XADV Advection in the E-W direction
YADV Advection in the N-S direction
ZADV Vertical advection
ADJC Mass adjustment for advection
HDIF Horizontal diffusion
VDIF Vertical diffusion
EMIS Emissions
DDEP Dry deposition
CHEM Chemistry
AERO Aerosols
CLDS Cloud processes and aqueous chemistry
PING Plume-in-grid
Note: The following process codes can also be used in the PACP.
XY ADV Sum of XADV and YADV
XYZADV Sum of XADV, YADV, and ZADV
TOTADV Sum of XADV, YADV, ZADV, and ADJC
TOTDIF Sum of HDIF and VDIF
TOTTRAN Sum of XADV, YADV, ZADV, ADJC, HDIF and VDIF
16-21
-------
EPA/600/R-99/030
Table 16-2.
Chemical Parameters Used in Default IRR Analysis
{1} Production of Odd Oxygen (Ox)
{2} Loss of Odd Oxygen (Ox)
{3} Production of NOz from NOx
{4} Production of NOx from NOz
{5} Production of new OH from O1D
{6} Production of new OH other than from O1D
{7} Production of new HO2
{8} Total production of HO2
{9} Production of new RO2
{10} Total Production of RO2
{11} Loss of CO and CH4 by reaction with OH
{12} Production of OH from HO2
{13} Production of NO2 from HO2
{14} Production of NO2 from RO2
{15} Production of PAN and TPAN
{16} Net Production of organic nitrates
{17} Loss of VOCs by reaction with OH
{18} Loss of OH by reaction with organics
{19} Net Production or Loss of HNO3
{20} Loss of HCHO by reaction with OH
{21} Loss of isoprene by reaction with OH
{22} Production of new HO2 from HCHO
{23} Production of HO2 from PAN
{24} Production of HO2 from RO2 and NO
{25} Production of HO2 from RO2 reacting with RO2
{26} Production of RO2 from OH
{27} Production of HNO3 from NO2 reacting with OH
{28} Production of new OH from H2O2
{29} Production of new OH from organic peroxides
{30} Production of OH from HONO
{31} Termination of OH
{31} Termination of HO2
{31} Termination of HO2 by reaction with RO2
{34} Termination of RO2
{35} Termination of RO2 by reaction with HO2
{36} Termination of RO2 by reaction with RO2
{37} Loss of OH by reaction with daughter VOCs
16-22
-------
EPA/6QQ('R-99f'G3Q
Exhibit 16-1.
Example PACP Command File for IPR Analysis
{1} OUTPOT_DOMAIN = LOLEV[1] + HILEV[2];
{2} DEFINE FAMILY NOX = NO + NO2;
{3} DEFINE FAMILY VOC = 2,0*ETH + 2.9*HC3 + 4.8*HC5 + 7.9*HC8
+ 2.0*OL2 + 3.8*OLT +
4.8*OLI + 5.0*ISO 4- 7.1*TOL + 8.9*XYL
+ 1.0*HCHO + 2.4*ALD;
{4} IPR_OUTPUT NOX;
{5} IPRJDOTPUT VOC;
{6} IPR_OUTPDT O3 = TOTTRAN + CH1M + CLDS;
{7} IPR_OUTPUT ALL = VDIF;
{8} ENDPA;
16-23
-------
EPA/600/R-99/030
Exhibit,! 6-2._ Listing of the RADM2 Chemical Mechanism
,1=':' fl
J'lH
:|i,
REACTIONS [ctuj
"fpi>
«F'P2>'
< P3>
< P4>
< P5>
< P6>
< P7>
< P8>
< P9>
• -211'*
«P21»
< 1>
f 2>
1_3*
< 4>
< S>
< 6>
< 7>
* 8>
< 9>
< 10>
< 11>
< 12>
< 13 >
< 14 >
< 16>
< 17*
< 18>
' 4»ls*
•>; 20*
i?2:*
< 22>
/"23>
V24*
<°25»
C?6-*
< 27*
28>
< 29* "
< 30»
< 31>
< 32>
< 33>
/'34>
i ?5>
f 36>
<™"37">
NO2
03
03
HOHO
HNO3
HN04
NO3
NO3
H202
HCHO
ALD
DPI
OP2
PAA
KET
QLY
OLY
HOLY
DCS
OMIT
03P +
03P +
01O +
"oib +
OIB +
03 4-
03 +
O3 +
HO2 +
H02 +
HNO4
H02 +•
HO2 4-
H2O2
MO +
NO +
O3 +
HO3 +
HO3 " +
SO3 4-
SO3 4-
N205
N2OS
HO" 4-
HO ' *
HO" +•
HO 4-
HO +
'cb +•
HO
STH 4-
HC3 +
HC5 *
HC8 +•
OL2 +
OLT 4-
OLI 4-
__ ^_
NO2
.S3.
O2~
H2O
NO
HO
HO2
NO
NO2
H02
„„
.
«
>
•
»
M
m
«
m
-
_
»
.
.
m
K
.
m
m
m
.
.
.
~m
m
.
.
•t
m
,
«
»
HO2 4- H20
4- HO
HO
NO +
NO2
NO
" **NO2^"
HO2""
. ,NO2__
4. H2O
5?2
' 'j?°L.
JjtiOlL^
HO2
.sp'.r
i*o
'So
HO
^
""ib
,"Fp
HO
' 'HO
HO
>
O2
IK
•
. '
-
-
»
_
M
M
».
-
«'
>
•
M
.
• 0
-
.
•
_
"
O3P
DID
O3P
HO
HO
HO2
NO
KO2
2.0* HO
CO
HO2
MO2"
HCHO
ALD
MO2
ACO3
0.13*HCHO 4-
0.45*HCHO 4-
ACO3
0.98*HO2 +
0.20*ALD +
O3
NO
O3P
O3P
2.0» HO
N02
HO2
HO
NO2
HNO4
HO2
H202
M H2O2
- HO2
MONO
a 2.0*NO2
N03
2.0*NO2
NO
HNO3
N2O5
NO2
2lO*HNO3
HN03 ^
I NO3
NCil
SUI.F' + "~HO2
» ,HO2
MO2
ETHP
0.83*HC3P 4-
,025»KET
HC5P +
HC8P 4-
"OL2P"
"OLTP,
OLIP +
4- NO
+ NO
4- NO2
4- NO2
+ O3P
+ HO2"
+ " HO2
4- HO2
4- HO2
+ HO
+ ETHP
1.870*CO
l.SSO*CO f 0
4- H02
0.020*ACO3
0.800*KET +
4- HO
„
ft
ft
tt
tt
f|
n
ft
if
#
+ do "" #*
+ co: ' f"
+ HO #
•f HO #
fi
i*
#
.800*HO2 ft
+ CO ff
+ TCO3 tt
H02 4 N02 ft
i
if
ff
#
t
#
jf
*
It
i.
i.
i.
i.
i.
i.
i.
i.
i.
i.
i .
"i."
i.
i.
i.
i.
i.
i.
i.
i.
i.
6.
6.
1.
3.
2.
2.
1.
1.
3.
# 1.8E-31
4- N02
%3 « 2
%3 # 3
if 7
" 4- NO2
4- NO3
ft 2.1E-27
.20E-139-S20
.0
.08E-34«-2820.
t*
,OE-31*-2.6
if
«<
ff
1
if
ff 2.2E-30
3
41
3
1
1
" 2
" 2
*• „
0 /;
0 /<0301D_RADM88>;
0 /;
0 /;
0 /?
0 /;
0 /;
0 /;
0 /;
0 /;
0*^ /;
'o" /;
0 /;
0 / < HOP_RADM8 8 > ; -
0 / < PAA_RADM8 8 > ;
0 /;
0 / < OLY f o rm_RADM8 8 > ;
0 /;
0 /;
0 /;
0 /;
OE-34*-2.3;
SE-12 « -120.0;
8E-11 ^« -110.0;
2S-11 *« -70.0;
20E-10;
OOE-12 «1400.0;
60E-12 * 940.0;
10E-14 ® 500.0;
70E-12 9 -240.0;
A-3.2 44.7E-12*'-1.4;
« -10900.0 *E< 10>;
& 1.90E-338-980.0;
0 & 2.66E-54S1-3180.0;
.300E-12 i» 200.0;
.5E-11*-0.5;
.300E-39 ® -530.0;
.4000E-13 9 2500.0;
.JOOOE-11 » -150^0;
.'500p'E-14 9~1230.0;
."SOOOE-i2;'
4.3 &1.5E-12*-0.5;
ff 1.10E-27 S -11200.0 *E<21>;
# 2.00E-21; '."'__ '_
tf 2.6E-30
A»
3.2 Jr2.4E-ll*-1.3f
%2 #7 . 2E*-15
-------
EPA/600/R-99/030
Exhibit 16-2. Listing of the RADM2 Chemical Mechanism
< 38>
< 39>
< 40>
<40a>
< 42>
< 43»
< 44>
< 45>
< 46>
< 47>
< 48>
< 49>
< 50>
< 51>
< 52>
< 53>
< 54>
< 55>
< 56>
< 57>
< 58>
< 60>
< 62>
< 64 >
< 65>
< 66>
< 67>
< 68>
< 69>
< 70>
< 71>
< 72>
< 73>
< 74>
< 75>
< 76>
< 77>
< 78>
< 79>
TOL 4-
XYL 4-
CSL 4-
CSL 4-
HCHO 4-
ALD +.
KET 4-
GLY 4-
MGLY 4
DCB 4
DPI 4-
OP2 4-
PAA 4-
PAN 4-
ON IT 4-
ISO +
AC03 +
PAN
TCO3 4-
TPAN
MO2 4.
HC3P 4-
HCSP 4-
HC8P 4-
OL2P 4-
OLTP 4-
OLIP 4-
AC03 4-
TC03 4-
TOLP +•
XYLP 4-
ETHP 4-
KETP 4-
OLN 4-
HCHO 4-
ALD 4-
GLY 4-
MGLY +
DCB +
CSL +
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
HO
NO2
NO2
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
NO
HO
NO
NO3
NO3
N03
NO3
NO3
NO3
= 0.75*TOLP *. 0.250»CSL
-f TOLAER
« 0.83*XYLP 4- 0.170*CSLi
4- XYLAER
m 0.10*HO2 4. 0.900*X02
4. CSLAER
CSL
HO2 4- CO
ACO3
KETP
= H02 + 2.000*CO
= ACO3 + CO
= TCO3
= 0.5*MO2 4-
= 0.5«HC3P 4-
ACO3
HCHO 4-
HC3P
= • OLTP
PAN
ACO3
» TPAN
TCO3
= HCHO
= 0.75*ALD
+ 0.964*NO2 4-
= 0.38«ALD 4-
4- 0.92*NO2 *
= 0.35*ALD 4-
4- Q.24»ONIT
= 1.6*HCHO 4
ALD 4-
= HO2 4-
*• 0 . 1*KET
= MO2
« NO2 4-
+ 0.050-ACO3
N02 +
4- 0.16*GLY
= NO2 + HO2
= ALD 4-
= MGLY +
- HCHO 4-
= H02 4-
= ACO3 4-
= HNO3 4-
= HNO3 4-
= HNO3 4-
» HNO3 4-
4- 0.250»HO2
4- 0.170*HO2
4- 0.900*TC03
0.500*HCHO +0.500»HO
0.500*ALD +
0.500'HO
NO3 4- XO2 (X 300 sq)
+ N02
4- NO2
4- NO2
4- HO2
4- 0.25-KET
0.964*HO2
0.69*KET 4-
0.92»H02
1.06*KET 4-
4- 0.76»NO2 4-
HO2 4- NO2 4-
HCHO + HO2
1.45«ALD 4-
4- NO2
4- NO2
0.920*HO2 4-
4- 0.950*CO
H02 4- 0.17
4- 0.70*DCB
4- .45*MGLY
HO2 4- NO2
NO2 4-
ALD 4-
HNO3 4-
HNO3
HO2 4- 2
AC03
TC03
XNO2 4- 0.
4- NO2
4- 0.09*HCHO
0.08'ONIT
0.04*HCHO
0.76-H02
0.20»ALD
4- NO2
0.28*HCHO
0.890*GLY 4-
4- 2.000*XO2
•MGLY
4- .806*DCB
HO2
2.0*NO2
CO
.ooo*co
+ CO
500'CSL
+ 0.500»CSLAER
< 80>
< 81>
< 82>
< 83>
< 84»
< 85>
< 86>
OL2 4-
OLT 4-
OLI 4-
ISO *
OL2 4-
OLT 4-
OLI 4-
NO3
N03
NO3
NO3
O3
O3
03
= OLN
= OLN
#
f
n
#
#
t*
tt
#
#
it
#
tt
tt
tt
t
«
4-
tt
ft
f
II
#
tj
#
2.
1.
4.
0.
9.
6 .
1.
1.
1.
2.
1.
1.
1.
6.
1.
2.
2.
1.
4.
1.
4.
0.
4.
4.
4.
4.
4.
4 .
4.
10E-12 8 -322.0;
89E-11 « -116.0;
OOE-11;
9*K<40>;
OOOOE-12
8700E-12 « -256.0;
2000E-11 9 745.0;
1SOOE-11
7000E-11
8E-11;
OOOOE-11
OOOOE-11
OOOOE-11
1650E-13A
5500E-11
5500E-11
8000E-12
9500E4-16
7000E-12;
9500E4-16
2000E-12
036'ONIT
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2 m 444
® 540.0
® -409.
8 -181.
® 13543
® 13543
® -180.
® -180.
® -180.
« -180.
S -180.
® -180.
® -180.
® -180.
.0;
;
0,'
0;
.0;
.0;
0;
0;
0;
0;
0;
0;
0;
0;
0,110*MGLY
tt
«
M
n
tt
»
tt
(»
M
u
n
n
ft
ft
= OLN + OLIAER 8
« OLN
= HCHO 4- 0.400*ORA1 4- 0.
4- 0.120*HO2
» 0.53*HCHO 4-
+ 0.20*ORA2 4-
= 0.18*HCHO *
0.500'ALD 4-
0.23*H02 +
0.72*ALD 4-
420*CO
0.33*CO 4-
0.22*MO2 4-
0.10'KET 4-
ff
s
0
0
s
0
4.
4 .
4 .
4.
4.
4.
6.
1.
6.
1.
1.
2.
2.
1.
3.
5.
1.
.20
.10
1.
.23
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
2000E-12
OOOE-13
400E-12
OOOE-13
400E-12
400E-12
200E-11;
OOOE-12
OOOE-11
230E-11
810E-13;
200E-14
*ORA1
*HO
3200E-14
«CO 4- 0 .
8 -180,
® -180.
® -180.
* -180.
» -180.
« -180.
® 2058.
® 1900.
« 2058.
® 1900.
® 1900.
® 2923.
8 1895.
® 975.0
® 2633.
® 2105.
06*ORA1
0;
0;
0;
0;
0;
0;
0;
0;
0;
o,-
0;
0;
0;
;
0;
0;
16-25
-------
EPA/600/R-99/030
JP1 . lillllll1
Ill • •- "" 1 1 . .• • . • .
i;, ", , i I , • . .. «•
Exhibit 16-2. Listing of the RADM2 Chemical Mechanism
< 87>
< 88>
< 89>
< 90 >
i 91>
S 92 >
< 93 >
< 94>
< 95>
< 96 >
< 97>
< 98>
•5 99>
<100>
<101>
<102>
<103>
<104>
<105>
<106>
<107>
<108>
<109>
<110>
<111>
I-
lit
y2>
i in
<113>
<114>
<138>
<115>
, ,n
<116>
<118>
f
|:, " '
<120>
JL>
11
<122>
<123>
<124>
ISO +
HO2 +
H02 +
H02 +
HO2 +
H02 +
HO2 +
HO2 +
HO2 +
HO2 4-
HO2 +
H02 +
HO2 +
H02 +
HO2 +
MO2 +
MO2 +
MO2 +
MO2 +
MO2 +
MO2 +
M02 +
MO2 +
MO2 +
MO2 +
1! ' 1
MO2 +
J MI||. „
M02 +
MO2 +
MO2 +
ETHP +
,1 "
HC3P +
HCSP +
HC8P +
,,
OL2P +'
OLTP +
OLIP +
KETP +
ACO3 +
AC03 +
03
MO2 «
ETHP «
HC3P •
HCSP -
HC8P «
OL2P -
OLTP «
OLIP .
KETP =
ACO3 .
TOLP -
XYLP -
T COS ' -
OLN
MO2
ETHP
HC3P
HCSP
HCSP
OL2P
OLTP
OLIP
KETP
ACO3
U 111: ill
TOLP
"'" ''
"XYLP"
TCO3
OLN
AC03
1lr „,
AC03
ACO3
.
ACO3
I f ]jn ': •
m
+
m
+
m
m
+
0.29«ORA2
OLIAER
O.S3*HCHO
0.20«ORA2
DPI
OP2
OP2
OP2
OP2
OP2
OP2
OP2
OP2
PAA
OP2
OP2
OP2
ONIT
1.5*HCHO
0.75*HCHO
0.84«HCHO
1.000«HO2
0.77»HCHO
1.000*HO2
0.80*HCHO
1.000*HO2
1.55«HCHO
1.25*HCHO
0.89*HCHO
0.55*KET
0.75«HCHO
HCHO + 0.
0.50*ORA2
HCHO -1- 0*".
o'.7d*DCB
_" HCHO, + O".
2.000«HO2 ".
.50«HCHO +
O.SO«ORA2
0.475*CO
1.75«HCHO
ALD -1- 0.5
0 . S • ORA2
,77«ALD +
0.50*M02 +
0.41*ALD +
0.50*MO2 +
0.46*ALD -1-
0.50*MO2 +
0.80*HCHO
6".5*MO2 -1-
ALD -f 0.
6^5*1102 : +
"o"ll725l"*ALD
6.5*HO2 +
MGLY -1- 0
O.S*ORA2
2.0*MO2
MO2 + 0.
0.70*DCB
+ 0.26*HO2 +
+ 0.500*ALD
+ 0.23*HO2 +
+ HO2
+ HO2 +
+ 0.770«ALD
+ 0.41«ALD +
+ 0.46«AL0 +
•1- 0.350«ALD
•1- 0.750*ALD
+ 0.725«ALD
•1- 0.750*MGLY
0.14*HO -1- 0.31
ft
+ 0.33*CO -1- 0
• 0.22*MO2 + 0.
8 7
ft 7
ft 7
S 7
S 7
0.75*ALD
-1- 0.260«KET
0.75«KET
1.39*KET
•1- H02
+ H02
+ H02
+ H02
0
ft
ft
8
ft
ft
ft
ft
#
#
7
1
1
4
3
2
1
1
1
1
i '*.
*MO2
7.2900E-15 ® 1136.0;
.20»ORA1
10*HO
.230E-14 a 2013.0;
.700E-14 ® -1300.0;
.700E-14 a -1300.0;
.700E-14 a -1300.0;
.700E-14 ® -1300.0;
.700E-14 a -1300.0;
.700E-14 ® -1300.0;
.700E-14 ® -1300.0;
.700E-14 a -1300.0;
.700E-14 a -1300.0;
.700E-14 a -1300.0;
.700E-14 a -1300.0;
.700E-14 a -1300.0;
.700E-14 9 -1300^0;
.700E-14 a -1300.
.90E-13
.40E-13
.20E-14
.40E-14
.90E-14
.40E-13
.40E-13
.70E-14
.70E-14
a -220.0
a -220.0
a -220.0;
a -220.0;
a -220.0;
0;
;
;
® -220.0;
® -220.0;
a -220.0;
8 -220.0;
S*HO2 + 0.5*MO2
17«MGLY + 0.
+ |2'.0«HO2
45*MGLY -t- 0.
0.445*GLY +
+ 0.025*ACO3
+ XO2
•1- 0.50«H02
IS'GLY
806*DCB
0.055*MGLY
+ 0.460*H02
+ ALD + NO2
ft
ft'
ft
ft
ft
9
1
1
9
1
.60E-13
.70E-14
.70E-14
.60E-13
.70E-14
a -220.0;
a" -220.0';'
® -220.0;
a -220.0;
a -220.0;
*HO2 + 0.5*MO2
0.26* KET +
0.5*ORA2
0.75*KET +
0.5*ORA2
1.39»KET +
0.5*ORA2
+ O.S*ALD +
0. 5*ORA2
5* HCHO '+ 0.5
_" oVs*ORA2
+ 6"I'55*KET +
0.50*M02 t
-5*HO2 -1- 0.
170*MGLY -1- 0
+ HO2
0 . 5*H02
0.5*HO2
0.5«H02
O.S«HO2
•HO2
0.14*HCHO
0.5*ORA2
5*MO2
. 16*GLY
ft
ft
#
ft
ft
ft
ft
ft
ft
ft
3
1
8
7
3
3
4
4
1
4
.40E-13
.OOE-13
.40E-14
.20E-14
.40E-13
.40E-13
. 20E-14
.20E-14
.19E-12
.20E-14
8 -220.0;
a -220.0;
® -220.0;
"a -220.0;
'"a"' -22 0.0;
"a -220.0,v
a -220.0
a -220.0
a -220.0;
a -220.0
16-26
-------
EPAJ6QO/R-99J030
Exhibit 16-2. Listing of the RADM2 Chemical Mechanism
<125> ACO3 + XYLP = MO2 + 0.45»MGLY + 0.806*DCB
+ HO2
<126> AC03 + TCO3 = M02 + 0.92»HO2 + 0.89'SLY
+• 0.11'MGLY +• 0.05«AC03 +• 0,95»CO
<139> AC03 + OLN
<140> OLN
2.0«XO2
HCHO + ALD -I- 0.5-ORA2
+ NO2 +• 0.5* MO2
OLN = 2.0*HCHO + 2.0*RLD + 2.0*HO2
<127>
<128>
<129>
<130>
<132>
<133>
<134>
<135>
<13S>
XO2 4
XO2 *
XO2 4
X02 4
XO2 4
XNO2
XNO2
XNO2
XNO2
XNO2
TERP
TERP
TERP
HO2
MO2
AC03
XO2
NO
NO2
HO2
MO2
AC03
XNO2
HO
NO3
03
OP2
HCHO +
MO2
NO2
OH1T
OP2
HCHO +•
MO2
TERPAER + HO
TERPAER + NO3
TERPAER + O3
HO2
H02
# 4.20E-14
-220.0;
#
t
*
ft
»
ft
8
8
»
f
#
8
tt
ft
ft
ft
i.
4 .
3.
7.
1.
4 .
3 .
4.
4.
7.
1.
4 .
3 .
1.
1.
1.
19E-12
20E-14
60E-16
70E-14
70E-14
20E-14
60E-16
2000E-12
2000E-12
70E-14
70E-14
20E-14
60E-16
0*K<37>;
0*K<82>;
0»K<86>;
8 -220.
9 -220
® -220.
• -1300
• -220.
8 -220
® -220.
0;
0
.
0;
.0;
0
0
;
0
;
9 -180.
® -180.
® -1300
• -220.
8 -220
® -220.
0
;
0
0
;
0;
0
0
;
;
endmech
16-27
-------
EPA/600/R-9W030
Exhibit 16-3". Example PACP Command File for a Full IRR Analysis
XRRJTYKB • F0LI.J
HNDPA;
16-28
-------
EPA/600/R-99/030
Exhibit 1 6-4. Example PACP Commands for a, Partial IRR Analysis
! Example PACP Command File illustrating Partial IRR Analysis
1 »»»«»<
! IRR type and domain commands
IRRTYPE = PARTIAL;
OOTPOT DOMAIN = LOLEV[1] 4- HILSV[2] ;
! Family Definitions
DEFINE
DEFINE
DEFINE
DEFINE
DEFINE
DEFINE
DEFINE
FAMILY
FAMILY
FAMILY
FAMILY
FAMILY
FAMILY
FAMILY
OX
NOZ
NOX
VOCA
RO2
VOC
dauHC
- O3 +
TPAN
= PAN 4-
OLN +
- NO *
= OL2
= MO2
OLTP
TCO3
= {CH4
OLI
GLY
= CSL
PAA
+
4-
+
4-
4-
•f
NO2 4- 2»NO3 + O3P 4-01D +
4- OLN 4- HNO3 4- ONIT;
TPAN 4- MONO 4- HNO4 + NOB
HNO3;
NO2;
OH 4-
ETHP -i
4- OLIP
+ XO2 •(
+ } CO H
ISO 4-
MGLY H
KET 4-
PAN +
OLT 4-
^ HC3P
ISO;
4- HCSP
PAN + HN04 + 3«N205
4- N205 4- ONIT 4-
+ HC8P *
+ TOLP 4- XYLP + ACO3 +
i- XNO2;
h BTK *
TOL +
^ DCB;
GLY +
ONIT;
• HC3 +
HCS +
CSL + XYL 4-
MGLY 4-
DCB +
HCS
HCHO
OP1
OL2P +
KBTS +
4-
4-
4-
OL2 + OLT +
AI43 + KET 4-
OP2 +
! Cycle Definitions
DEFINE CYCLE PANcyc = PAN,'
DEFINE CYCLE TPANcyc =» TPAN,-
DEFINE CYCLE HONOcyc = HONO;
DEFINE CYCLE HN04cyo = HNO4 ;
! Reaction Sum Definitions
DEFINE RXNSOM NO3cyc
DEFINE RXNSOM N205cyc
DEFINE RXNSOM H2O2_OHcyc
DEFINE RXNSOM HNO3_OHoyc
DEFINE RXNSUM OPl_OHcyc
DEFINE RXNSUM OP2_OHcyc
DEFINE RXNSOM PAA_OHcyc
DEFINE RXNSOM HNO4_HO2cyc
DEFINE RXNSOM OP2_HO2cyc
•ss < 2."7> *" *
3S < 2X3* "" "
« < P9> - .
= < P5> - «
. - •
= - •
< 93> - •
< 98> - •
= - •
= < P6> + •
= - •
< 93 > - «
< 98> - «
t P7> -
t 22>;
e 12> -
e 24>|
e 88>;
c 89> -
e 94 > -
e 99> -
e 97>;
e 11> -
e 89> -
e 94> -
e 99> -
< P8> - < 18> - < 19>!
< 13>;
< 90> - < 91> » < 92> -
< 95> - < 96> - < 97> -
<100>{ - <101>);
< 10>;
< 90> - < 91> - < 92> -
< 95> - < 96> - < 97> -
<100> {- <101>} - <127> -
16-29
-------
EPA/600/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
DEFINE RXNSUM HOXcyc - < 7> - < 8>/
DEFINE RXNSUM newMO2 - + 0,22*< 85> + 0.31*< B6> + 0.22*< 87>;
DEFINE RXNSUM newACO3 =* + + 0.02» + < 77»;
DEFINE RXNSDH newETHP - ;
DEFINE RXNSDH newTCO3 t » + < 78>;
DJEFINE RXNSOT) PAN_ACO3cyc - < 54> - < 53>;
DEFINE RXNSOM TPAN_TCO3cyc = < 56> - < 55>;
JDEPINE RXNSCM propRO2_NO = <67> +• 4.05*<68> •»• 1.5*<126>;
I IRR_OOTPtIT 1: OX Production
f MM^^HHMKBt-BrnvMar»«•!•«•"»«•:«'—""•~araia:n—'^a«i;am>5~-~"—'K;»««»^ —"—si«i^^ ———.._—^^^—- ——^—^a.,«™—.
IRR_OOTPUT OXprod =» NETP [0X1 ;
DESCRIPTION »_'OX Production'|
I IRR_OUTPDT 2: OX toss
IRR_OUTPUT OXloss » NETL[OX];
DESCRIPTION « 'OX LOSS';
i IRR_OOTPUT 3: Production of NOZ from NOX
IRR_OUTPDT NOZfromHOX - PANcyC [POSOMTLY] +
TPANoyo [PQSONMf] +
HONOcyc[POSONLY] +
HNO4cyc[POSOKLY] +
f ,, NO3cyc[POSONLY] +
H2O5cyc[POSONLYJ +
< 24> + 0.036*< 58> + 0.08*< 60>
r - . i ,i 0.024*< 62> + <132>;
DESCRIPTION - 'NOZ produced from NOX' ,-
! IRR_OUTPUT 4 : Production of NOX from NOZ
IRR_OUTPOT NOXfromNOZ - PANcyc [NEGONLY1 +
TPRNcyC [NEGONIjYJ +
HONOcyc [NEGONLY] +
HNO4cyc [NEGONtY] +
N03cyetNEGONLY] +
N2O5cyc [NEGONLY] t
< PS> + + < 51>
< 73> + <138> + <139>
2.0*<140>;
DSSCRIPTION m 'NOX produced from NOZ'j
I IRR_OUTPUT 5: Production of new OH from O1D
ZRR OUTPUT OHf romOlD - PROD [HOJ PROM [O1D] ;
16-30
-------
EPA/600/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
DESCRIPTION m 'OH produced from 01D1;
! IRR_OOTP0T 6: Production of new OH other than from OlD
IRRJDUTPUT newOH = 0.1*< 85> + Q.14*< B6> + 0.1*< 87>
2*H2O2_OHcyc [POSQNLY.] +
HNO3_OHcyc [POSONLY] +
HONOcyc[NEGONLY] +
OPl_OHcyc[POSONLY] +
OP2j3Hcyc[ POSONLY] +
PAAJDHcye[POSONLY];
DESCRIPTION a 'new OH';
! IRRJ30TPTJT 7; Production of new HO2
IRRJDUTPUT newHO2 = 2.0* -f- + 0.8* +
+ 0.98* + +
< 74> + < 76> 4- 0.12*< 84> +
0.23*<,85> + 0.2«*< 86> + 0.23*< 87> +
OPljDHcyc [POSONLY] +
OP2_H02cyC [POSONLY] <•
HN04_HO2cycIPOSONLY];
DESCRIPTION = 'new HO21;
! IRRJDUTPUT 8: Total Production of HO2
I ~»—J-.---.;*™™™™ __—__ — — — —.___—_— z— _; — — — — —.-»««.— »»»»•,•,•,___ : — _______._._.«___
IRR_ODTPUT tOtalHO2 = 2.0* + + 0.8*
+ 0,98* -f
< 74> + < 76> •(- 0.12*< 84>
(HO2new) 0.23*< 85> + 0.26*< 86> + 0.23*< 87>
OPl_OHcyC[POSONLY] +
OP2_H02cyc[POSONLY] +
HNO4_H02cyc[POSONLY] +
{HO2propbyOH} PROD[HO2] FROM [HO] AND [VOC] +
{HO2viaRO2_NO} PROD[HO2] FROM [NO] AND [RO2] +
{HO2byRO2_RO2} PROD[HO2] FROM [R02] AND [RO2] +
{otherOH) HOXcyc[POSONtY] ;
DESCRIPTION = 'total HO2';
! IRR_OUTPUT 9: Production of new RO2
1 = = = := — — -.^.—^^-—J-^^—^g-. _^__ — _™.™.__™____
IRR_OUTPUT newRO2 = newM02 +
newACO3 +
newETHP +
newTCOS +
16-31
-------
EPA/600/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
PAN_ACO3eye [POSONLYJ '+
TPA*r_TC03cyc[POSONLY) ;
' ' A
J IE. • R . .'f
DESCRIPTION » 'newRO2'j
-'
I IRR_tJTPtJT 10: Total Production of RO2
IRRJJOTPUT TotalRO2 « newM02 +
newAC03 +
{newRO2} newBTHP +
newTCOS +
PAH_ACO3cyc[POSONI,Yl +
TPAN_TCO3cyc[POSONLY] +
{prcpRO2_OH} PRODCRO2] FROM [HOJ AND [VOCJ +
~ < 30> + 0.5»< 47> + 0.5*<48>
< S0> + < Sl> +
{propR02_HO} PROOIR02] FROM [NOJ ;
OSSCRIPTIOH « 'Total RO2'|
I IRR OUTPUT 11: Loss of CO & CH4 by reaction with OH
TmmmMmmmmm^mmm'i'mmmmmMmmf
IRR_OUTPOT Losa_CO_CH4 » < 30» -f
~ ~ LOSS [CO] ;
OISCRIFTION « 'LOSS Of CO & CH4 ' ;
1 — «»E^^M«aiS«!«Wl«WM«l!tS"™ """"•^»i«C«»
7 lik_dtfrPOT*i2:"Production of OH from HO2
JRRJDUTPUT H02toOH - HOXcyc [NEGOOTiY] +
2.0*H202_OHcyc[POSONLY);
DESCRIPTION - 'H02 to OH1,-
I IRR_OOTPUT 13: Production Of NO2 from HO2
IRR_ODTPUT MO2fJTOmH02 » < 9>;
DESCRIPTION - «HO2 FROM HO2';
1 IRR_OOTPOT 14: Production of NO2 from R02
IRRJ3UTPUT HO2fromRO2 = < 57> + O.S64*< 5S> * 0.92*« 60> +
"" Q.76*< 62» + < 64> + < 65> H-
< 66> •(• < 67> + < 68> +
• • < S9> -f < 70> •»• < 71> +
16-32
-------
EPA/600/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
DESCRIPTION = 'N02 FROM HO2 ' ;
! IRR_OUTPUT 15: Production of PAN and TPAN
IRR_OUTPUT prodPAN_TPAN = PANcyc + TPANcyc;
DESCRIPTION = 'Production of PAN and TPAN';
! IRRJDUTPUT 16: Net Production of organic nitrates
IRRJDUTPUT netONIT = NET[ONIT];
DESCRIPTION = 'Net production of ONIT';
! IRRJDUTPUT 17: Loss of VOCs by reaction with OH
IRRJDUTPUT 10330H_HC = LOSS [VOC] AND [HO] +
< 30> + < 47> + < 48> +
< 49> + < 50> + < 51>;
DESCRIPTION = 'Loss of HC plus OH1;
! IRR_OOTPUT 18: Loss of OH by reaction with inorganics
| ca — __ — _ — __ — -..___. — —
IRR_OOTPUT lossOH_INORG =< 7>+<14>+<15>+
< 24> •*• < 25> + < 26> +
< 27>;
DESCRIPTION = 'Loss of OH with iorganics';
1 IRRJDUTPUT 19: Net production or loss of HNO3
IRR_OUTPUT netHNO3 = NET[HNO31;
DESCRIPTION = 'Net change in HNO3';
! IRR_OUTPUT 20: Loss of HCHO by reaction with OH
IRR_OUTPUT lossHCHO_OH = LOSS [HCHO] AND [HO] ,-
DESCRIPTION = 'Reaction OH HCHO with OH';
I IRR_OUTPUT 21: Loss of isoprene by reaction with OH
16-33
-------
EPA/6QO/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
2RR_OOTPOT los«ISO_OH . LOSS[ISO] AND [HO];
DESCRIPTION . 'Reaction of ISO with OH1;
P'lRR_3uTPuTl"2s'~Produation'of "new HO2 from HCHO
IRR_OUTPUT nawHOZfromHCHO « PROD[HO2] FROM [HCHO] AND [hv] ,-
DESCRIPTION » 'New HO2 from HCHO1;
! IRRjOUTPOT 23: Production of HO2 from PftN
IRR_OUTPUT HO2fromPAN » PAN_RCO3cyc[POSONLY] ;
D1SCRIPTION - «HO2 from PAN1,-
T IRR_o1uTPUT 24 : Production of H02 from R02 and NO
IRR_OUTPlTr H02fromRO2_NO • PROD[HO2] FROM [NO] MID [RO2]
DESCRIPTION . *'HO2 from RO2 and NO';
T IRR_OUTPUT 25: Production of HO2 from R02 and RO2
IRR OUTPUT HO2fromRO2 RO2 « PROD[HO2] FROM [RO2] AND [RO2];
DESCRIPTION . "HO2 from R02 and RO2';
\
. . • :« .'1 • .' ;- .• -
IRR__GUTPUT"26 :* Production of RO2 from OH
IRR_ODTPOT ROafromOH » PROb[RO2] "FROM [HO];
DESCRIPTION i" 'RO2 from OH1;
V IRR_OUTPUT 27: ProductTon of HNO3 from OH '•*•'NO2
3CRR OUTPUT "HNO3fromOH'NO2 = <" 24>";
DESCRIPTION - 'HNO3 from OH + NO2' ;
1 IRRjOUTPUT 28: Production of new OH from H2O2
IRR_OUTPUT newOH_H2O2 - 2*H2O2_OHcyc[POSONLY];
DESCRIPTION - 'new OH from H2O2';
16-34
-------
EPA/600/R-99/030
Exhibit 16-4. Example PACP Commands for a Partial IRR Analysis
! IRR_OUTPUT 29: Production of new OH from organic peroxides
IRRJDUTPUT newOHJDPl = OPIJDHcyc [POSONIiY] +
OP2_OHcyc [POSONLY] +
PAAJDHcyc [POSONLY] ;
DESCRIPTION - 'new OH from DPI OP2 PAA1 ;
! IRRJDUTPUT 30: Production of OH from HONO
IRRJDUTPUT newOHf romHONO = HONOcyc [NEGONLY] ;
DESCRIPTION = 'new OH from HONO';
! IRRJDUTPUT 31: OH Termination
{_______________________________________________
IRRJDUTPUT OHterm = < 25> + <26> -t- <27> + < 50>
HNO3_OHcyc [NEGONLY] -t-
HONOcyc[POSONLY];
DESCRIPTION = 'OH Termination';
! IRRJDUTPUT 32: HO2 Termination
1 ===============================================
IRRJDUTPUT HO2term * < 20> + < 27> + <101> +
2 * H2O2_OHcyc [NEGONLY] +
HN04_H02cyc[NEGONLY];
DESCRIPTION =- 'HO2 Termination1;
! IRRJDUTPUT 33: Termination of HO2 by reaction with RO2
!======================================================«==
IRRJDUTPUT tertnHO2_RO2 = OPIJDHcyc [NEGONLY] +
OP2_HO2cyc(NEGONLY] +
PAA_OHcyc (NEGONLY] ;
DESCRIPTION = 'HO2 term with RO2';
! IRR OUTPUT 34: RO2 Termination
IRRJDUTPUT termRO2 = .036*<58> + .08*<60> + .24*<62> + .03*<68> +
PAN_AC03cyc(NEGONLY] +
TPANcyc(POSONLY];
16-35
-------
BPA/600/R-99/030
Exhibit i6-4. Example PACP Commands for a Partial IRR Analysis
DESCRIPTION • 'R02 Termination';
• — M_ —_ ^_ —„_ —_ ——___ ——__M—___ — „__ ——___ — ___
I IRR_OOTPUT 351 Termination of RO2 by reaction with with H02
I MM — mm MM__ M___——___ — — — »«xmmma~mmm*'~"~'m«a™"™atmK~~**mm
IRRJTOTPOT terwR02_H02 » OPIJDHcye [NEGONLY] +
"" ~ OW2~OHcyc [NESOSLYJ +
PAAJ3Hcyc [NEGONLY] ,-
DSSCRIPTIOH • 'RO2 Termination with H02'j
I IRR_CXn:PUT 361 Termination of R02 by reaction with RO2
• ^M—_——M___ — ___M—mm_ ^mmm mmm mmmm ^—»»^^« _^_^^^_^^^»_^_]la,___a=^,™ —
*IRR_OUTPDT ter«Rb2_R02 « <102> + <103> + <104> + <10S> + <106>
+ <109> + <110> + <111>
<11S> + <11S> + <117> *
.. <119> + <120» + <121> + <122>
! -. *|| is | 1.030 * + 0.500 * <138> -t-
;- ^ '", 0.500 * <139>»-
t ..-, :f 14 :*•"•'•-•' . :*
DSSCRIPTIOK •' 'R02 Termination with RO2';
5ri' " . «elQ7> 4-
*T ' l.'SIS * <114» 4-
!', "• 'fll", , , |l|'l III ' '!""
|l. IER_OOTPCTa 37s LOBS of OH by reaction with daughter VOCs
IRRJXJTPUT dauHC_OH » LOSS [HO] SND [dauHC] ;
DESCRIPTION'•" 'OH + daughter HC';
r: ,; "f ' • •
16-36
-------
EPA/600/R-99/030
O3 Processes
Charlotte, 287IB O'Brien Off
•Pteza-MonitorGaH
Observed
Hourly Average
Mixing Ratio f
Mixing Ratio
Vert. Trans,
(5-10:110.6
ppb)
Chemistry
(8-18:113.6
ppb)
Honz. Trans.
(5-15: 37.6 ppb)
vert. Trans.
(15-19: 10.2 ppb)
Honz. Trans.
(15-19:-37.33
PPP)
Chemistry
(5-9:-57,4 ppb)
150
140
130
120
&100
.2 90
fi 80
I 70
| 60
a. 50
£ 40
°: 30
1 20
a 10
I °
E -10
-20
-30
-40
-50
O3 Processes
- Charlotte, 287FC O'Brien On
-Plaza Monitor-Sell
Vert. Trans.
(5-8: 74.1 ppb)
Observed
Hourly Average
Mixing Ratio 7-
Predlcted
Mixing Ratio
Chemistry
(8-18: 147.9
Horiz. Trans.
(5-14:52.3 ppb)
Vert. Trans.
(15-19: 11.9 ppb)
"ChemisUy
(5-8:-80.4 ppb)
Vert. Trans.
(8-15:-44.1 ppb)
* Horiz. Trans.
(14-19:-75,7 ppb)
6
10 12
Hours
14
16
18
Figure 16-1. Example process analyses showing contributions of individual processes to model
predicted concentrations (Source: Jeffries, 1996).
16-37
-------
EPA/600/R-99/03Q
ISO
20 km New York area
10-16 EOT
Figure 16-2. Example Process Analyses showing model predicted concentrations and process
contributions for different model configurations (Source: Jeffries, 1996),
16-38
-------
Photochemical Air Pollution
photolysis
= number ofOH cycles=Q/«|* I
> number of cycles a E/< * l /
-------
EPA/600/R-99/030
Chapter 17
AN AGGREGATION AND EPISODE SELECTION SCHEME
DESIGNED TO SUPPORT MODELS-3 CMAQ
Richard D. Cohn
Analytical Sciences, Inc.
Durham, NC 27713
Brian K. Eder* and Sharon K. LeDuc"
Atmospheric Modeling Division
National Exposure Research Laboratory
Research Triangle Park, NC 27711
ABSTRACT
In support of studies mandated by the 1990 Clean Air Act Amendments, the Models-3
Community Multiscale Air Quality (CMAQ) model can be used to estimate pollutant
concentrations and deposition associated with specified emission levels. Assessment studies
require CMAQ-based distributional estimates of ozone, acidic deposition, PM25, and visibility on
seasonal and annual time frames. Because it is not financially feasible to execute CMAQ over
such extended time periods, CMAQ must be executed for a finite number of episodes under a
variety of meteorological classes. A statistical procedure called aggregation, must then be applied
to the CMAQ outputs to derive seasonal and annual estimates.
The objective of this research is to develop an aggregation approach and set of episodes that
would support model-based distributional estimates (over the continental domain) of air quality
parameters. The approach utilized cluster analysis and the 700 mb v and v wind field components
over the time period 1984-1992 to define homogeneous meteorological clusters. A total of 20
clusters (five per season) were identified by the technique. A stratified sample of 40 events was
selected from the clusters, using a systematic sampling technique.
This stratified sample is then evaluated through a comparison of aggregated estimates of the mean
extinction coefficients (b^ to the actual mean bext observed at 201 stations nationwide for a nine
year period (1984-1992). The bext, a measure of visibility, was selected for use in the evaluation
for two reasons. First, of all of the air quality parameters simulated by CMAQ, this visibility
parameter provides one of the most spatially and temporally comprehensive data sets available,
and second, the be)rt can serve as a surrogate for PM2.5 for which little data exists. Results from
the evaluation revealed a high level of agreement (r2 = 0.988) indicating that the aggregation and
episode selection scheme was indeed representative.
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
Corresponding author address: Brian Eder, MD-80, Research Triangle Park, NC 27711. E-mail:
eder@hpcc.epa.gov
"On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce.
-------
EPA/600/R-99/030
17.0 AN AGGREGATION AND EPISODE SELECTION SCHEME DESIGNED TO
SUPPORT MODELS-3 CMAQ
17.1 Introduction^
This chapter describes the development of an aggregation and episode selection scheme to
support the derivation of annual and seasonal estimates of air quality parameters in Models-3
CMAQ, Relevant background information regarding this activity, as well as a precise statement
of the present objectives, is provided in this section. Section 17.2 summarizes the important
~ elements of the approach and the overall strategy, including the rationale behind the methods and
the limitations associated with them. Details of the development of an aggregation and episode
selection are provided in sections 17.3, 17.4, and 17.5. Section 17.6 contains an example
application and evaluation of the methodology using one particular air quality parameter (bcx,),
and section 17,7 provides a summary and discussion.
17.1.1 Background
In support of studies mandated by the 1990 Clean Air Act Amendments, the Models-3
Community Multiscale AirjQuality (CMAQ) model is used by EPA Program Offices to estimate
deposition and air concentrations associated with specified levels of emissions. Assessment
studies and effects models require CMAQ-based distributional estimates (such as annual and
seasonal averages) of ozone, acidic deposition, and measures related to visibility. Such estimates
would ideally be obtained by using CMAQ to simulate atmospheric chemical processes associated
with meteorological conditions occurring on a daily basis over several years. However, for
logistical and cost reasons it is not currently feasible to execute CMAQ over an extended time
period such as a fall year. Therefore, in practice CMAQ must be executed for a finite number of
.episodes or "events," which are selected to represent a variety of meteorological classes. A
statistical procedure called aggregation, must then be applied to the outputs from CMAQ to
f •« derive the required annual-and seasonal-average estimates from this finite number of events. The
objective of the research described in this chapter is to develop such an aggregation approach and
evaluate its effectiveness using the bext.
The basic problem of developing representative meteorological categories has been explored by
other researchers for a variety of purposes, including Fernau and Samson (1990a,b); Davis and
Kalkstein (1990); Eder et al. (1994). The approach used here is based on a variation of the
methods previously used by Brook et al. (1995a,b) in selecting a 30-event aggregation set for the
Regional Acid Deposition Model (RADM). The approach of Brook et al. involved four major
components. Cluster analysis of wind fields was used to determine meteorologically
representative categories. The determination of the number of clusters to retain was based upon
within-group variance patterns and prior work by Fernau and Samson (1990a,b). A procedure for
aggregating the episodic results into annual totals and averages involved frequency-weighted
sums and estimated deposition-precipitation relationships. Event selection procedures were
designed to emphasize-categories that accounted for most of the annual wet sulfate deposition,
17-2
-------
EPA/600/R-W/030
while also representing some winter and dry events. A summary of some specific elements of
their approach is provided in Table 17-1.
Table 17-1 Summary of Methodology Used by Brook et al. (1995a; 1995b) for RADM
x •; - -;' 1. Determination of categories :- ; :•*•' :.«•
::« Used Ward's (1963) method of cluster analysis, which minimizes within-cluster sums of squares, jri an
;. agglomeratiye, hierarchical mode for wind flow parameters, Used eastern North American zonal uart.d
"•: rher|:dfona{ v 66-kPa wind components for 0000 UTC with 5" spatial resolution (two..variables at 48 grid:
nodes over three days). * :• • -:'f- ••• • ^:> :• -V. '. 'i:; -!: ^::
• 'Clustered "consecutive" 3-;day periods from 1979,1981, and 1983,!with subsequent classification of;
'.;: 'remaining "running" 3-day periods from;1979-1985 into one,.pf these clusters. ':.
: . : 2. Determination of number of clusters to retain ..... . , ! -- ••••«••
• ! Examined stepwise increases: In within-group variance with decreasing number of clusters, expressed
; 'through F-statistic. .- :•, ' .• '.•'•..': :J :''
• -Retained 19 clusters based on results for wind flow, sulfate! deposition, and prior work by Fernau and
: Samson (1990a;,b). Used quantitative information but somewhat subjective criteria..
'..:• ,. 3. Development of aggregation procedure . , ; :
• Estimated total deposition for the group of sampled categories, from the sampled events {weighted: •;
, sum,- accounting for the number:of,events sampled from each cluster and frequency of occurrence of <.-.
the clusters).' :' . •..;.:-.:;•* • ; 'f -• .• :. ; '''.-.'•<.'.' -
* Scaled up by ratio of total annual.deposition to an estimate of annual deposition that is based on ;
Aggregating the sampled categories. Estimates used either mean deposition from, a sampled category
or deposition estimated from the;deposltion-precipitatloh relationship in^ th& category. ;
i - .4. Selection of "optimum" set of events : ./;..: ;
» ^Subdivided the 19 categories into wet and dry subsets, resulting in 38 new categories.:; . ;
• Primarily represented categories that accounted for most of the annual wet sulfate^.deposition (at least
715% when combined). Selected 19 events from these categories. Also selected li events from winter
(5) and dry (6) events to represent seasonal deposition differences, dry periods, and possible nonlinear
;. - effects; •''.• . "• '' ;••;. .. • .-;:.".'•...',- '.":-•' •:.'. ,J'J.: . '. • : - -.'".' ":-'• :.-;;- ..;'.: .-.••"•"'••'
>:» Number of events selected from each category was based oh proportionality to the;frequency of J .::
: 'occurrence of the category and the percent offtotal'sulfate accounted for by the category. : : >:
;:* ^Examined 20 potential sets (each randomly:generated) meeting all criteria, selected;the one that :;
;; minimised iRMSE for annual sulfate deposition (primarily) and precipitation {secondarily} at 13 sites.:
;; ..Selected sets in stages, first choosing a set .of 19 events from the wet categories and then aisetof.t.i-
' events from winter and dry periods. . • •,.•:... :•-:;'...
17.1.2 Objectives
The present objectives differ somewhat from those which motivated the earlier research. RADM
was primarily designed to address issues involving acidic deposition. CMAQ addresses a more
17-3
-------
EPA/600/R-99/030
diverse collection of air quality parameters with equal importance. In addition, CMAQ will be
applied to a continental domain that is significantly larger than geographic area for which
estimates are supplied by RADM. The extension from the RADM domain to a continental
domain is extremely ambitious as it relates to episode selection and aggregation. Therefore, the
development of an approach that accommodates this larger continental domain is particularly
challenging.
'• The objective of the activity described in this chapter is the development of an aggregation and
episode selection approach that supports model-based annual and seasonal air quality estimates
• that are at least as accurate (with respect to sampling uncertainty) as those achieved by RADM, in
consideration of the more general applicability envisioned for CMAQ both with respect to air
quality parameters and geographic representation. This accuracy must be preserved while
minimizing any additional cost. In essence, "cost" refers to the number of events for which the
model provides estimates, each of which adds cost in the form of both computational processing
time and human labor.
'i 17.2 Summary of the Approach
'' L ' '' '.'IMS1'1' f'f ......... l|l|l' ' ' •' ||m ' ''' | ' Ai.nJ' !, "' 'I, ..... kijji ' ,•> 1^ | . ,. , •„, | '"'' |f ...... i' ,;| "'• | ||, ,
r : The analysis was carried out in phases, with information gathered in each phase contributing to
I the design of the next. For this reason, this chapter is structured to present complete descriptions
: of ^6 methodology and results achieved in each phase, in sequence, in sections 17.3, 17.4, and
17.5, This section is intended as an overview of the process prior to those detailed descriptions.
This overview consists of descriptions of some key elements of the methodology, the rationale,
scope, and limitations associated with the methodology, and the strategy used to move in phases
- - toward the final result.
17.2.1 Basic Elements of the Methodology
Simply stated, the methods that we have employed involve the determination of meteorologically
representative categories, the selection of events from those categories, and the use of evaluative
tools to ensure that the detailed aspects of those activities are defined in such a way as to achieve
optimal results, to the extent that we can measure optimality.
]t RBM r A -i P. *>"',.- •"",->..; :"v- •'.:""' < ^rWi"* ; 'Ififr' ' .'.' '-j,^ 1, '% *T> '' "J^iJi ..... f
r * "A specific goal is the definition of meteorological categories that account for a significant
proportion of the variability exhibited by the air quality characterizations of interest. The basic
approach used in the current analysis for the determination of categories and event selection
fit COimPonents ls re^atec^ to ^at of Brook et al. (1995a,b), but certain fundamental considerations
have been modified to reflect the differences inherent in the present objectives as described in
section 17.1. The common element is the cluster analysis of zonal u and meridional v wind
,, , , components to define meteorological categories.
The definition of meteorological categories is designed to support the selection of events from
those categories in a process known as stratified sampling (Cochran, 1977). Stratified sampling
"" - *«*-
*'
-------
EPA/600/R-99/030
exploits the internal homogeneity of the meteorological categories, or "strata," to achieve more
precise estimates than would be possible using simple random sampling (i.e., randomly selecting
events without regard to meteorological category). Certain variations of stratified sampling are
relevant to this analysis. One relatively inefficient option for invoking stratified sampling would
involve selecting the same number of events from each category/stratum. This is known as "equal
allocation." An alternative is "proportional allocation," which involves selecting numbers of
events in direct proportion to the size of the stratum. Thus, more events are selected from strata
that contain large numbers of events than from smaller strata. This is potentially much more
efficient than equal allocation, in the sense that it leads to much more precise (i.e., lower variance)
estimates. Estimates exhibiting absolute maximum efficiency (i.e., minimum variance) are
obtained by modifying this method slightly so that the number of events selected from each
stratum is in direct proportion to the product of the size of the stratum times the internal
variability (as characterized by the standard deviation of the measurement of interest) within the
stratum. Thus, strata exhibiting significant variability among events are sampled more heavily
than strata in which events are more uniform. This is known as "optimum allocation," which is
identical to proportional allocation when within-stratum variances are equal.
While wind flow parameters were used to define the meteorological categories, other
meteorological parameters were used in subsequent phases of the analysis to refine aspects of the
episode selection methodology, and as evaluative tools to assess the effectiveness of the
approach. These parameters include visibility (as represented by the bext), temperature, and
relative humidity. Their specific roles are discussed in more detail in the following sections.
17.2.2 Rationale, Scope, and Limitations
As stated previously, the approach to selecting an aggregation and episode scheme is based upon
the definition of meteorological categories that account for a significant proportion of the
variability exhibited by the air quality characterizations of interest. Strictly speaking these
characterizations include parameters such as acidic deposition, air concentrations, and measures
of visibility. Therefore, it might be argued that the definition of categories should be formulated
directly using these parameters that are ultimately of interest. However, it is equally important
that the model simulate the particular transport mechanisms involved in the associated
atmospheric processes, and in particular that source-attribution analyses be facilitated. This
requires that categories be defined with an emphasis on wind flow parameters. Indeed,
characterizations of basic wind field patterns in essence describe frontal passages, along with all of
the meteorological properties typically associated with them.
In view of the importance placed on the accurate simulation of transport mechanisms, a complete
evaluation of the episode selection and aggregation methodology would require an assessment of
the accuracy with which transport is represented. However, this accuracy cannot be measured, as
there is no technique available to support a direct, quantifiable assessment of the representation of
atmospheric transport. In addition, with the exception of bext, there are little air quality data
available with the required spatial and temporal resolution and range to support a direct
17-5
-------
1: 1 £• Pf - AW 3? ' . ; >. - %.v , - : ? > ^ |f x :..-;-; |- . j, .ft f
EPA/600/R-99/030 " .-..-.• •• - .
evaluation of the methods described in this chapter with regard to the outcome parameters that
will be of primary interest in CMAQ.
For these reasons two meteorological parameters (in addition to b^,), which are known to be
related to many of the air quality parameters of interest, and for which appropriately resolved data
are available, were used as evaluative measures in this analysis. Specifically, temperature and
relative humidity were used; however, primary emphasis will be placed upon the b^ which
provides a surrogate measure for fine particles. It must be recognized that this constitutes a
secondary evaluative tool, in the sense that the effectiveness of the approach cannot be directly
measured as it relates to atmospheric transport or to specific air quality parameters, both of which
are primary outcomes. For this reason, the methods were not developed and refined solely
towardtthe goadkif optimizing performance associated with the estimation of visibility. Instead,
this performance was evaluated in combination with other considerations that were believed to be
important but for which performance may not be readily quantifiable.
* "' J8F *"" ' " - japtBf ' t 1 •*, *» - ' *<> 1, < - - . * - >. „• • - , % f • - v -. f , . ^ "
17.2.3 Strategy
The basic strategy used for the selection of events to support aggregation-based estimation is
described in the steps outlined below. The term "cluster" describes a collection of events that are
, defined to be meteorologically similar based upon cluster analysis results. The term "stratum"
', , , m » f • .!•*••.• m », - meteorological parameters described above (visibility, temperature, relative humidity).
^ The relative efficiency of each stratification scheme is defined as the ratio of the variance
associated with simple random sampling to the variance associated with stratified sampling
using that scheme. A large relative efficiency is indicative of a high degree of precision
(lower variance) associated with the estimate of interest. As discussed previously, since
these evaluative meteorological parameters do not afford a complete, direct assessment of
f v : tr -j- i. TtiL* trx t * v"- • ra - . vt -k jr. *< , > » ••.-,:. t, if ; < a. . . . ,
1 ""• i » f r t -4 M "i*i n ••'«!> • " "'• i ' " ' . . " •""» '• , -' ' ' -'- » ""
"•" .»'» . • . . • - 17.5
-------
EPA/600/R-99/030
the validity of process with regard to the parameters to which it is ultimately to be applied,
meteorological considerations were taken into account in combination with the relative
efficiencies in order to select a general stratification scheme. The selected scheme
(seasonal stratification) was considered for more detailed refinement in step 3.
3. We determined an appropriate number of clusters to retain in combination with an
acceptable number of events that would be necessary to achieve the objectives. The
determination was based on estimated standard deviations associated with several
alternative formulations, as well as other considerations. The standard deviations relate
specifically to estimates of the annual means and 90th percentiles of the evaluative
meteorological parameters. The "other considerations" included the objective of matching
or exceeding the performance of the current aggregation methodology used for RADM, as
well as a goal of avoiding unsampled clusters. Unsampled clusters (the inclusion of
clusters for which no events would be selected) were considered to be undesirable because
there would be no information available in the aggregation process to account for events
contained in such a cluster. In concept, our view was that any unsampled cluster should
be combined with another cluster to which it is most similar. In practice, we achieved this
by constraining ourselves to collections of clusters that were adequate to support the
sampling of at least one event from each cluster. Twenty clusters were ultimately
retained, consisting of five clusters defined in each of the four seasons, and the number of
events necessary to achieve the objectives was determined to be 40.
4. A stratified sample of 40 events was randomly selected from the 20 clusters defined in step
3. Proportional allocation was used in determining the number of events to be selected
from each cluster (stratum). Optimum allocation was considered but was not used for
several reasons, including: (1) it requires the quantification of the variance of a primary
outcome parameter, whereas only secondary evaluative outcomes (visibility, temperature,
relative humidity) were available as discussed previously, and (2) the variance of any
outcome parameter varies geographically, so that optimum allocation would likely result in
differing numbers of events depending upon the geographic location, whereas proportional
allocation would not. It was verified that, at least based upon the evaluative outcome
parameters that are available, in most geographic locations the distribution of events that
arises from proportional allocation does not differ substantially from that arising from
optimum allocation.
Details concerning the implementation of this approach are discussed in the following sections.
Sections 17.3 and 17.4 correspond to steps 1 and 2 in the strategy outlined above, and section
17.5 includes a discussion of steps 3 and 4.
17.3 Cluster Analysis of Wind Fields
17-7
-------
*te EPA/60Q/R-9*9/bfo
'.~ The cluster analysis of wind components is described in this section. This includes a description
of the wind field data, the basic analysis technique, variations of the basic technique, and methods
of presentation. Some graphical results of exploratory analyses are also included.
17.3.1 Description of Wind Data
To accommodate the continental domain and to achieve adequate spatial resolution, the cluster
analysis involves data at 336 grid nodes with 2.5° spatial resolution, as obtained from the
_ NCTP/HCAR40-year reanalysis project (Kalnay et aL, 1996). In this analysis, 700 mb wind
r» components for 1800 UTC have been used, in consideration of the mountainous western regions
;*tf in the dqmjun^^qrners of the grid were cut back to guard against excessive influence from
,i|, ocean-based meteorology. Graphical illustrations of this domain are referenced later in this
;JL ' section.
-R £• ... .. , .'i« it '..-•.• • '• , ,'- . . .X • '. i". • •'- ' ' s '.-<,.
17.3.2 Basic Cluster Analysis Technique
Cluster analysis, in the present formulation, involves the classification of a set of observations into
categories that are internally homogeneous with respect to defined multivariate relationships in the
data. In this case "multivariate" refers to the multiple variables used to characterize wind fields,
consisting of the u and v components applicable to each location within the previously described
domain and extending over an event that includes multiple days.
A 12-year period (1984-1995) was considered in the exploratory cluster analyses, later refined to
a 9-year period (1984-1992) for the final clustering upon which episode selection was based.
Since CMAQ is actually run for a 5-day period for each event (the first two days establish initial
conditions, and model predictions from days 3-5 are saved as a "3-day event"), 5-day periods
were clustered rather than 3-day periods. To make the analysis computationally feasible, the first,
third, and fifth days of each 5-day event were considered. Based upon the performance noted by
Fernau and Samson (1990a,b), Ward's method of cluster analysis was used (Ward, 1963),
- minimizing within-cluster sums of squares, in an agglomerative (i.e., moving from many clusters
toward fewer clusters), hierarchical (i.e., once clusters are joined they cannot be separated)
process. Thus, if a single observation (event) is considered to consist of 2,016 elements (the 2 u
w andv components * 336 grid nodes x 3 days considered per event), then the objective of the
,§ cluster analysis is to divide these observations into clusters (categories) for which the within-
" * cji^er^sum ofsquares (sum of squared differences between the elements of individual
;» observations qrjneans) is minimized.
'I* It •»' . ,,-* ,«iik ' _' ', .Or. . fS *t •• X, f ••• • . : - I-'''--, 1V - . ....... . i^, <- . '. 1 .'.
In the exploratory analyses, clusters were initially defined based upon "consecutive" rather than
"running," or overlapping 5-day periods from 1987-1992. Then, each remaining event ("running"
5-day periods from 1984 through 1995) was classified into the cluster that minimizes the sum
(over the 336 grid nodes and three days) of the squared deviations of each u and v component
from the cluster mean u and v. In the final cluster analysis, using 1984-1992 data, consecutive 5-
day periods from 1984 through 1992 were clustered, and remaining events were classified into
b - -a? ..j •. - . • 17-8
-------
EPA/600/R-9P/030
those clusters according to the same criteria described above. Cluster analyses were carried out
using SAS (SAS Institute Inc., 1989); however, due to the extreme computational burden of these
analyses it was necessary to calculate the distance matrix externally from the clustering procedure
itself.
17.3.3 Illustration of Cluster Analysis Results
Representative results from the selected exploratory analyses, as well as all results from the final
cluster formulation, are illustrated graphically. Specifically, cluster definitions are illustrated using
maps of cluster average wind fields. In some cases, these are supplemented with specialized maps
illustrating the intracluster variability in the wind fields. Star chart histograms are also used to
illustrate the frequency of occurrence of events from each cluster, for each month of the year.
To provide proper perspective for a review of this analysis, it is useful to consider preliminary
results for a set of 30 clusters initially defined using annual data from 1984-1995. The most
prevalent cluster (labeled Cluster 1) accounted for 12.19% of all 5-day events during this time
period, with nearly all of those events occurring during the summer months. Map-based graphs of
mean wind vectors for this cluster are contained in Figure 17-1 (a-c). Three maps are included,
representing mean vectors for days 1,3, and 5 of the 5-day event. These maps illustrate largely
stagnant conditions associated with an anticyclonic pattern centered over the southeastern U.S.; a
zonal flow over southern Canada and a trough over the west coast of the U.S.
As a second example, Figure 17-2(a-c) illustrates mean wind vectors for a somewhat less
prevalent cluster that accounted for 3.65% of all 5-day events during this time period (ranking
ninth in overall prevalence and labeled as Cluster 9). Most of these events occurred between the
months of October and April, characterized by northwesterly winds associated with a large-scale
trough moving through the central and eastern portion of the. domain.
While these maps are effective in illustrating average behavior associated with each cluster, they
do not give any indication of the variability inherent in the clusters. Figures 17-4 (a-c) and 17-5
(a-c) contain additional maps that address this issue. The first map (Figure 17-3a) illustrates the
wind field for the third day of an individual event (January 23-27, 1989) that was assigned to
Cluster 9. Comparison of this map to the mean wind field for day 3 of Cluster 9 (Figure 17-2b)
reveals fairly close resemblance between the two. By contrast, the maps in Figures 17-3b and 17-
4c depict the third day from two other events belonging to Cluster 9. These patterns do not
resemble the mean wind field as closely, and are indicative of the variability among individual
events that were assigned to this cluster.
Clearly it would be useful to simultaneously visualize the variability exhibited by all of the wind
fields assigned to each day of Cluster 9, and Figure 17-4 (a-c) represents an attempt to do this.
These maps contain the mean wind vectors for each day of Cluster 9 on a thinned-out grid that
only includes alternating grid nodes. In addition, the maps contain groups of small dots, each of
which depicts the location of the wind vector arrowhead for an individual event assigned to this
17-9
-------
ix- • \\
•-» s m.
EPA/600/R-99/030
cluster. The groups of dots collectively illustrate the distribution of arrowheads for all events
belonging to Cluster 9.
- «h>-.-. .-i.' rum- am r - u • »-'-»~i i ,-«„: -,-**•. , ,..,>... • „. .;.<, aii » . , -, , -r t , ,:l . _ -,,,
The dots surrounding each mean wind vector arrowhead appear in a somewhat circular pattern,
and the extent of spread exhibited by the dots illustrates the degree of variability among wind
vectors assigned to the cluster. Similar patterns characterize the variability associated with other
clusters (not shown). Clearly there is substantial variability associated with the wind vectors
assigned to individual clusters. This variability emphasizes the ambitious nature of the endeavor.
In essence, the goal is to categorize many years worth of meteorological patterns into a finite
number of classes. Furthermore, each meteorological pattern does not simply describe a single
location at a given point in time; it is required to simultaneously represent a broad spatial domain
over a significant temporal period. Indeed, it should not be surprising that a substantial amount of
variability is associated with the result.
Figure 17-5 (a-e) contains star chart histograms of the 30 clusters defined using annual data.
Each chart illustrates the frequency of 5-day events belonging to a given cluster, and the clusters
themselves are ordered according to overall frequency of occurrence. As shown on the charts,
events from Cluster 1 accounted for 12.19% of all 5-day events between 1984 and 1995, those
from Cluster 2 accounted for 12.10% of all events, etc. The numbers arranged radially on each
chart depict the number of events belonging to the cluster from each month of the year. The
length of the line pointing to each month is proportional to this frequency of occurrence, and the
ends of the lines are connected to facilitate the visualization of patterns.
Several observations may be made based upon these charts:
• Although defined using annual data, the cluster frequencies reveal definite seasonal
tendencies. That is, clusters do not occur randomly throughout the year, but rather exhibit
a tendency to occur more frequently within specific seasons. Thus, the annual clustering
procedure successfully identifies and discriminates wind field patterns that are associated
with seasonally distinct meteorological classes.
I.' J^t W-~.»Vk -iS." V ' fT«:* *..'!' 'I'1-' ' ^f "'••iff.f J-TUf ?" •»>_..! t jr J. , ,HI-_ f , r
' yf|?k '.Sisters contajning summer events tend to be quite distinct from those containing
winter events (and vice versa), many clusters contain events from a combined
* If^K, ~ " •• i. »SL^... ,~™™ , , „._- - . ,' . ~ ™ .
-+ *%ansitional" season that includes both spring and fall months.
........ Illlllllllllllllll ............. Ill • -: ....... , ....... ....... ............. • " ........... ' . '.I ' ,,n ..................
The two most prevalent clusters heavily emphasize the summer months; each of these
clusters includes more than twice as many events as any other cluster, with the vast
' majonty of summer events contained m them. ~~
• " "' '« Ul .. .r . . . . . . . . ,
The disproportionate representation of summer events by two of the 30 clusters is not surprising,
since the wind fields are expected to be less variable in the summer. However, seasonal
differences in meteorology and atmospheric chemistry are important in explaining the variability
exhibited by the air quality parameters of interest. The addition of more warm season clusters
17-10
-------
EPA/600/R-99/030
could provide improved resolution in this regard. This is the primary motivation for conducting
analyses using seasonally distinct clustering,
17.4 Evaluation of Alternative Aggregation Approaches
The previous illustration clarifies the motivation for investigating seasonally distinct clustering.
Similarly, regionally distinct clustering offers an alternative that might provide improved
resolution of wind field groupings on smaller spatial scales. An added dimension to the problem is
that the number of clusters to be retained clearly arfects the degree of resolution. To gain an
understanding of the importance of these considerations as they relate to estimation of the
meteorological parameters used as evaluative tools in this analysis, these and other alternatives
were explored in several combinations. These explorations, designed to expose patterns and
trends from which to make an informed selection of a general approach, are described in this
section.
17.4.1 Description of Alternative Approaches
Several variations of the u and v wind vector clustering were investigated. As discussed
previously, these were selected to investigate patterns and properties, and not necessarily to be
considered as final candidates for cluster definitions. They are as follows:
• Annually defined strata with variations in the number of strata: 5, 10, 20, 30, 60, and 90
clusters;
• Seasonally defined strata using warm and cold seasons: totals of 10, 20, 30, and 60
clusters equally divided between the warm season April-September period and the cold
season October-March period;
» Seasonally defined strata approximately equally divided among summer (June, July,
August), winter (December, January, February), and transitional (spring and autumn
combined) seasons: totals of 20, 30, and 60 clusters;
• Seasonally defined strata equally divided between summer and winter seasons, and with
approximately twice as many strata defined for the transitional season: totals of 20 and 30
clusters;
• Seasonally defined strata approximately equally divided among summer, winter, spring
(March, April, May), and autumn (September, October, November) seasons: totals of 20
and 30 clusters; and
• Regionally defined strata: 30 clusters identified in each of four separate cluster analyses
for the northeast, southeast, northwest, and southwest subsets of the domain
17-11
-------
. ...fi EPM600/R-99/P3Q,
- •'• it' My 1 -, L' * >.!«• . * , ,' • » >- , - T. CJ-T-»--Xk« ,.•-.-•.. , »
: We also investigated the alternatives of strata defined to simply mirror the four seasons
*3?', ~"% ' (disregarding all clusters), strata defined by clustering of 3-day events rather than 5-day events,
\^\f." and strata defined by clustering the meteorological parameters used to evaluate the approach.
17.4.2 Description of Meteorological Data
^ .. F°£the alternative^tra^fication|chemess preliminary testing was performed by examining the
J| J: 15% tugrtainty associated with tfie use of cluster-based stratified sampling to estimate the annual
mean~bf daily noontime levels of visibility, temperature, and relative humidity, with primary
emphasis placed upon visibility as discussed previously. Visibility (units of km"1) was specifically
expressed as the light extinction (b^, less observations with precipitation, and less observations
with relative humidity greater than 90%. The light-extinction coefficient is often used to
characterize, visibility, although in general, it has limited ability to predict human visibility. The
visual range vr (ion) can be estimated from the bexl by using the Koschmieder equation:
it '"in 'mi HP'
Temperature is in units of degrees Celsius, and relative humidity in percent. These parameters
were taken from 201 locations in the continental U.S. for which coverage was at least 99%, as
illustrated in Figure 17-6. This was specifically defined as sites exhibiting at least 99%
1' .^i :j!j!j!j!jij,j!j!jiji|i|ii ,,, 'lilji,;('i'1'1'1' 'ijljiji1:1, ,;,|,|,|,|,| ' I,) *** * «* *"J
completeness for light extinction coefficient; allowances were made for missing observations that
were associated with precipitation so as not to bias the inclusion of sites toward drier climates.
17.4.3 Analysis Methods
The average relative efficiency associated with the estimation of mean annual visibility,
temperature, and relative humidity, using each alternative stratification scheme, was used for
comparison of the schemes. Specifically, at each location and for each scheme, the variance of an
aggregation-based estimate (Cochran, 1977) of the annual mean was determined in three ways,
assuming that the estimated mean was calculated using (1) a stratified sample with equal
allocation across strata, (2) a stratified sample with proportional allocation across strata, and (3)
simple random sampling, with a total sample size consisting of the same numbers of events in each
case. Also at each location, the ratio of the simple random sampling variance to the variance
"associated with each of the two_stratified sampling designs was calculated and expressed as the
relative efficiency of each of those designs. Finally, those relative efficiencies were averaged
across sites to provide an indication of the overall efficiency of each scheme.
• *.
This is best explained using a specific example. First, suppose that the mean annual temperature
at a given location is estimated as the average of the daily temperatures from 30 randomly
sampled 3-day events from the period 1984-1992, completely disregarding any information
•f. related to clusters. Suppose that the standard deviation associated with that estimate is 1.5°C, so
' 1 '' IP: H* S ' '- '* ' **"" ^ 41 r IF *li: ' •.' ' iv .' 4;-: • ; i': : f ,
! '" ~' -'•?••' - •• > - i7_12
-------
EPA/600/R-99/030
that a 95% confidence interval would yield the estimated mean ±2.94°C (=1.96x 1.5). Second,
suppose that the mean is instead estimated as a weighted average from 30 3-day events, using two
events per cluster (i.e., stratified sampling with equal allocation), and using the frequency of
occurrence of each cluster as the weight applied to the temperature from the corresponding event.
Suppose that the standard deviation associated with that estimate is 1.0°C, compared to 1.5°C
from simple random sampling. Last, suppose that 30 events are selected using proportional
allocation (i.e., the number of events selected from a cluster is proportional to the number of
events belonging to the cluster), and that the standard deviation of the resulting estimated annual
mean temperature is 0.8°C. These standard deviations (1.5, 1.0, and 0.8) translate to variances of
2.25, 1.0, and 0.64 for simple random sampling, stratified sampling with equal allocation, and
stratified sampling with proportional allocation, respectively. Thus, for this hypothetical location,
the relative efficiency of stratified sampling with equal allocation is 2.25/1.0=2.25, and the relative
efficiency of stratified sampling with proportional allocation is 2.25/0.64=3.52. Proportional
allocation is more efficient than equal allocation in the sense that it leads to lower variances and
therefore tighter confidence intervals bounding the estimated mean.
17.4.4 Results
Tables 17-2 (a-b) and 17-3 present mean relative efficiencies associated with annual means of the
daily noontime temperature, relative humidity, and extinction coefficient, as estimated using
aggregation approaches based upon the various schemes described above. In each table, results
are presented to illustrate the relative efficiency of estimation methods using equal allocation
(equal numbers of events selected to represent each cluster) and proportional allocation (numbers
of events selected in direct proportion to the total number of events categorized into the given
cluster). Relative efficiencies reported in this section are valid for any number of events that
might be selected, as relative efficiency is invariant to sample size under equal or proportional
allocation
In the case of proportional allocation, the relative efficiency actually refers to the minimum
variance that might theoretically be achieved if proportional allocation were carried out precisely.
In practice this might only be possible if a very large number of events were to be sampled, since
the appropriate proportions might otherwise dictate sampling of fractional numbers of events from
some clusters. Therefore, the relative efficiency reported for proportional allocation may be
thought of as an upper limit to the relative efficiency that might actually be achieved in practice.
In all likelihood, this upper limit cannot be attained, but it should be possible to achieve a relative
efficiency occurring somewhere within the range defined by this upper limit and the relative
efficiency associated with equal allocation of events among clusters.
The first six rows in Table 17-2a illustrate relative efficiencies associated with various numbers of
annually defined strata (i.e., clusters emerging from cluster analyses of daily wind field data from
1984-1995 without regard to season). Several observations may be made by inspecting the first
six rows in Table 17-2a:
17-13
-------
.„«,
BPA/600/R-99/030
in 1 ii I
]!» I,
The relative efficiency associated with the estimation of mean temperature is consistently
highest, followed by that of relative humidity. This implies that the meteorological
clusters are most usefiil in distinguishing between events with regard to temperature, and
least useful in distinguishing extinction coefficient.
niiit
1 Hi "i :
In the case of temperature, the variability associated with stratified sampling using wind
field-based clusters is consistently less than the variability associated with simple random
sampling (i.e., relative efficiencies are uniformly greater than 1.0). Thus, in each scheme
the clusters contribute important information that explains variation in temperature.
'2,"\'~ ;--; * "*™t.T-^:"fr-*i«'**:.- ' « 1»*. '"» -'-'Si'K'jm.r **,a T -A' '-. • •
For the estimation of mean relative humidity and mean extinction coefficient, the use of
stratified sampling based upon wind field clusters is consistently more efficient than simple
random sampling if proportional allocation is approximately satisfied. In most cases, the
use of equal allocation actually results in less efficient estimation than simple random
sampling; this reflects a wide range of stratum sizes that would be inappropriately
represented using equal allocation.
• As would be expected, the efficiency associated with proportional allocation increases as
the number of strata_inereases, incorporating more refined representations of the
meteorological classes within the strata.
• • •.• <:*: • •: :•:', , f,. § ..s. ,>; ,», . _ i :,
• The efficiency associated with equal allocation decreases as the number of strata increases.
The stratum sizes become more divergent when more strata are defined, so that the
inefficiency of equal allocation is magnified.
Based upon the well-known properties of stratified sampling discussed above, our objective was
"loHesign and implement a scheme based upon approximate proportional allocation. The numbers
for^equal allocation in these tables merely served to provide a lower bound on the efficiency that
could be realized, since it was known that the precise degree of efficiency reported for
proportional allocation might not be achievable in practice due to the requirement of sampling
integer-valued numbers of events.
As discussed in section 17.3.3 above in regard to the analysis illustrated in the "30 Strata" row of
the table, summer events are disproportionately represented by two of the 30 clusters. To
evaluate the effect of improving the resolution of summer meteorological patterns, as well as the
effects of imposing constraints that would alter the resolution of families of clusters under a
variety of scenarios, various implementations of seasonally distinct clustering were investigated.
Thelast four rows of Table17-2ajUustrate results associated with stratum definitions based upon
a simple warm/cold seasonal dichotomy, in which separate cluster analyses were conducted to
force equal numbers of strata for each season. The following observations may be made based
upon this portion of the table:
17-14
-------
EPA/600/R-99/03G
• With the exception of extinction coefficient estimation under proportional allocation,
stratification using seasonally defined clusters consistently yields improved efficiency over
stratification using the same number of annually defined clusters. Thus, although the
annually defined clusters do adhere to a seasonal pattern, the improved resolution afforded
by the forced inclusion of more warm weather clusters (and reduction of cold weather
clusters) is particularly effective in explaining variation in temperature.
• For seasonally defined clusters, the relative gains in efficiency associated with using equal
allocation are large as compared to the potential gains associated with using proportional
allocation. Thus, in departing from proportional allocation (which is not precisely
achievable in practice as discussed above), seasonally defined clusters are likely to afford
improved efficiency over annually defined clusters.
Results associated with further seasonal stratification schemes are illustrated in Table 17-2b, The
first two rows correspond to approximately equal numbers of clusters divided among summer,
winter, and transitional (spring and autumn combined) seasons, the next two rows correspond to
equal numbers of clusters divided between summer and winter with approximately twice as many
transitional season clusters, and the fifth and sixth rows correspond to approximately equal
numbers of clusters divided among four seasons. In each case, the exact distributions are
constrained to result in total numbers of strata that are directly comparable to the numbers of
strata investigated in other seasonal and annual analyses.
Under proportional allocation, stratification schemes based on three or four seasons offer
significantly improved efficiency in the estimation of mean temperature compared to two-season
and annual stratification schemes with comparable total numbers of strata. They also demonstrate
slight but uniform improvement in the estimation of extinction coefficient, and mixed results in the
estimation of relative humidity.
17-15
-------
EPA/600/R-99/030
Table 17-2a. Mean relative efficiency1 associated with estimation of the annual (1984-1995)
mean of the indicated parameter, using various stratified2 sampling3 approaches relative to simple
random sampling. Continental analysis.
Method
5 Strata
10 Strata
20 Strata
30 Strata
60 Strata
90 Strata
10 Strata Defined Seas-
onally (5 Warm, 5 Cold)
20 Strata Defined Seas-
onally (10 Warm, 10
Cold)
30 Strata Defined Seas-
onally (15 Warm, 15
Cold)
60 Strata Defined Seas-
onally (30 Warm, 30
Cold)
Temperature
Equal
Allocation
Across
Strata
2.34
2.29
2.11
2.08
1.81
1.43
2.62
2.91
3.14
2.74
Propor-
tional
Allocation
2.41
2.50
2.83
2.86
3.05
3.10
2.94
3.54
3.60
3.89
Relative Humidity
Equal
Allocation
Across
Strata
1.14
1.09
0.88
0.86
0.76
0.62
1.11
1.11
1.17
0.98
Propor-
tional
Allocation
1.18
1.22
1.26
1.28
1.32
1.33
1.26
1.35
1.37
1.39
RH-Adjusted Extinction
Coefficient
Equal
Allocation
Across
Strata
1.01
0.90
0.69
0.67
0.59
0.53
0.93
0.89
0.91
0.77
Propor-
tional
Allocation
1.12
1.14
1.16
1.17
1.19
1.20
1.13
1.16
1.17
1.20
-If
•14
i:1 Relatlve^ffjciency is defined as the ratio of the variance associated with simple random sampling to
the variance associated with stratified sampling. The table entries are means of station-specific
efficiency ratios, averaged across stations within the continental domain.
2 Unless otherwise noted, stratum definitions are based on annual clustering of 5-day events from a
temporal subsample of the wind field data; the remainder of the sample is then classified into those
strata.
8 Reflects sampling of 3-day events from 1984-1995.
r, is
f.'t ?
: : f, • 1
17-16
•V •«.-; •. i*
.*.'•
-------
EPM600/R-99M30
Table 17-2b. Mean relative efficiency1 associated with estimation of the annual (1984-1995)
mean of the indicated parameter, using various stratified sampling2 approaches relative to simple
random sampling. Continental analysis..
Method
20 Strata Defined Seas-
onally (6 Summer, 6
Winter, 7 Transitional)
30 Strata Defined Seas-
onally (10 Summer, 10
Winter, 10 Transitional)
20 Strata Defined Seas-
onally (5 Summer, 5
Winter, 10 Transitional)
30 Strata Defined Seas-
onally (8 Summer, 8
Winter, 14 Transitional)
20 Strata Defined Seas-
onally (5 Summer, 5
Winter, 5 Spring, 5 Fall)
30 Strata Defined Seas-
onally (8 Summer, 8
Winter, 7 Spring, 7 Fall)
4 Strata Defined as
Seasons (Dec-Feb, etc.)
30 Strata Defined by
Clustering 3-Day Events
30 Strata; Seasonality
Removed from Met.
Parameters
30 Strata Defined by
Clustering Lt. Ext. Coeff.
Temperature
Equal
Allocation
Across
Strata
2.66
2.60
3.31
2.91
3.23
2.82
2.93
2.24
0.92
0.51
Propor-
tional
Allocation
3.91
4.12
4.06
4.17
3.86
3.88
2.93
2.78
1.28
1.89
Relative Humidity
Equal
Allocation
Across
Strata
0.95
0.87
1.06
0.96
1.11
1.03
1.21
0.97
0.74
0.30
Propor-
tional
Allocation
1.31
1.35
1.33
1.36
1.36
1.38
1.21
1.29
1.15
1.11
^H-Adjusted Extinctior
Coefficient
Equal
Allocation
Across
Strata
0.90
0.78
0.89
0.80
0.92
0.84
1.12
0.76
0.63
0.57
Propor-
tional
Allocation
1.18
1.20
1.19
1.20
1.18
1.19
1.12
1.16
1.09
1.45
1 Relative efficiency is defined as the ratio of the variance associated with simple random sampling to
the variance associated with stratified sampling. The table entries are means of station-specific
efficiency ratios, averaged across stations within the continental domain.
2 Reflects sampling of 3-day events from 1984-1995.
17-17
-------
EPA/600/R-99/030
The final four rows of Table 17-2b address issues that are useful in that they provide additional
perspective for this analysis. They are discussed in sequence below:
» In view of the improvement associated with seasonal stratification schemes relative to
annual stratification, a natural question to ask is whether a seasonal scheme with one
stratum per season (i.e., no cluster analysis) is sufficient to support comparable efficiency.
" Based upon the results in Table 17-2b, this method offers overall improvement relative to
an annual scheme with an approximately equal number of strata (i.e., five strata as shown
in the first row of Table 17-2a). As might be expected, it is significantly less efficient than
the use of 20 strata equally divided among four seasons, but not dramatically so,
« Another natural question is whether the clustering of five-day events (using wind data
from the first, third, and fifth days) has a noticeable impact on this analysis relative to the
I;: clustering of three-day events that previous investigators have pursued. Based upon
~' "^ results in Table IT^b^ thejiefinition of 30 strata using annual clustering of three-day
evehfspfoauces similar results to those associated with 30 strata using five-day event
clustering (fourth row of 17-2a) with respect to the evaluative parameters. Note that this
does not necessarily address differences with respect to the characterization of transport.
3LA ,*>*"•£,-%
A consistent result in these analyses is that the relative efficiency associated with the
j|:**"* estimation of mean temperature is much greater than that associated with the estimation of
?i mean relative humidity, which in turn is slightly greater than the relative efficiency
associated with extinction coefficient. Since the preliminary results demonstrate
associations between the clusters and the seasons (even for annually defined clusters), one
i_ jnight hypothesize mat this characteristic is related to the more pronounced seasonal
trends associated with temperature and the less pronounced trends associated with
r T extinction coefficient. Table 17-2b illustrates relative efficiencies associated with the three
_ "'meteorological parameters following the removal of seasonal trends from each. (Trends
were removed by analyzing deviations of each parameter from a sinusoidal curve fitted to
the raw data at each location.) Indeed, the relative efficiencies for the three parameters
r: '• - =: : are much more comparable in this context, and much more similar to the results for
- extinction coefficient in other analyses. This analysis lends support to the use of extinction
***"' '*' * 5* •'• coefficient as the primary evaluative outcome, because it reflects the ability of each
pr I if jp, ..,_,
5 5"-i = scheme to characterize short-term meteorological patterns apart from long-term seasonal
trends.
* A final investigation in Table 17-2b also relates to the utility of extinction coefficient as the
primary evaluative outcome. Under proportional allocation, the relative efficiencies
ip,!'3,||j9ciated_with it are not dramatically greater than 1.0, indicating that stratified sampling
™" basetfupon wind field clustering produces consistent but not dramatic gains in efficiency
(relative to simple random sampling) in the estimation of mean extinction coefficient. To
put this observation in its proper perspective, it is useful to consider the maximum relative
~ efficiency that might be achieved from any stratified analysis. In particular, a stratification
-------
EPA/600/R-99/030
of events was performed based upon extinction coefficient itself (rather than wind fields).
Under these optimal conditions, the relative efficiency associated with a 30-stratum
scheme was still only 1,45, compared to a range of 1.17-1,20 for other 30-stratum
schemes based upon wind field clusters. Considered in this context, relative efficiencies
encountered in these tables are encouraging. (This clustering of extinction coefficient
provides some useful perspective regarding this evaluative outcome parameter, but is not
pursued as a recommended methodology for estimation based upon the rationale outlined
in section 2,2.)
To evaluate the utility of regionally distinct stratification schemes, the continental domain was
divided into four quadrants and cluster analyses were performed on wind fields within each
region, resulting in four sets of 30 annually defined strata. For each region, relative efficiencies
were summarized as in the previously described analyses. Since the mean relative efficiencies are
constrained to sites within each region, results in Table 17-2a are inappropriate for comparison to
these regional mean relative efficiencies. Therefore, Table 17-3 also includes mean relative
efficiencies only for the sites within each region, from comparable clustering of continental data.
These are displayed in combination with the results for strata defined using regional data.
For example, using 30 strata under proportional allocation, clustering of wind fields in the
northeast quadrant of the domain results in a mean relative efficiency (averaged over sites in that
quadrant) of 1.65 associated with the estimation of mean temperature. For 30 strata, from
clusters defined over the entire continental domain, the mean relative efficiency, over those same
northeast sites is 3.14. The results in Table 17-3 collectively demonstrate that, under proportional
allocation, regional stratification produces either no gains or only slight gains in efficiency in both
the southeast and southwest regions for any of the evaluative parameters. In the northeast and
northwest regions, this technique is markedly less efficient than continental clustering with regard
to temperature, and somewhat less efficient with regard to relative humidity. There is only a
negligible effect on extinction coefficient.
This result has significant importance, because it demonstrates that in the northern half of the
domain, clustering of continental wind field data is actually more effective than clustering regional
data in explaining variation in some meteorological parameters on a regional scale. Thus, the
wind field patterns associated with different levels of temperature (and, less markedly, relative
humidity) in a northern region are more distinctly identified on a continental scale than on a
regional scale. Furthermore, wind field patterns associated with different levels of extinction
coefficient (a primary evaluative parameter due to its close association with fine particles) in a
given region are no more distinctly identified on a regional scale than on a continental scale.
17-19
-------
EPA/600/R-99/030
Table 17-3. Mean relative efficiency1 associated with estimation of the annual (1984-1995) mean
of the indicated parameter, using various stratified sampling2 approaches relative to simple
Imndom sampling. Regional analysis.
Region/Method
Northeast; 30 Strata Def.
Using Regional Data
Northeast; 30 Strata Def.
Using Continental Data
Southeast; 30 Strata Def.
Using Regional Data
Southeast; 30 Strata Def.
Using Continental Data
Southwest; 30 Strata Def.
Using Regional Data
Southwest; 30 Strata Def.
Using Continental Data
Northwest; 30 Strata Def.
Using Regional Data
Northwest; 30 Strata Def.
Using Continental Data
Temperature
Equal
Allocation
Across
Strata
1.41
2.27
2.35
2.02
1.95
1.88
1.17
1.96
Propor-
tional
Allocation
1.65
3.14
2.72
2.62
2.42
2.43
1.66
2.82
Relative Humidity
Equal
Allocation
Across
Strata
0.98
0.77
1.04
0.83
0.94
0.89
0.92
1.01
Propor-
tional
Allocation
1.15
1.20
1.32
1.24
1.26
1.22
1.27
1.47
RH-Adjusted ixtinctior
Coefficient
Equal
Allocation
Across
Strata
0.95
0.61
0.77
0.61
0.81
0.84
0.82
0.71
Propor-
tional
Allocation
1.27
1.28
1.22
1.20
1.01
1.04
1.08
1.03
1 Relative efficiency is defined as the ratio of the variance associated with simple random sampling to
the variance associated with stratified sampling. The table entries are means of station-specific
efficiency ratios, averaged across stations within the continental domain.
2 Reflects sampling of 3-day events from 1984-1995.
Injview ojFjhe_current_objective of matching or exceeding the performance of the current RADM
stratification scheme over the continental domain, it is appropriate to compare the relative
efficiency of the stratification scheme used in the aggregation of RADM output compared to the
relative efficiency of the schemes defined above. Table 17-4 addresses this specific issue. In
addition to the results associated with equal and proportional allocation, this table illustrates the
mean relative efficiencies associated with the actual allocation of the 30 RADM events, which lies
somewhere between those extremes. (Proportional allocation was not a specific goal of the
methods developed for RADM, although it is not inconsistent with the less formally stated goal of
selecting more events from the clusters that accounted for most of the acidic deposition.)
17-20
-------
EPAJ600/R-99/030
Table 17-4. Mean relative efficiency1 associated with estimation of the annual (1982-1985) mean
of the indicated parameter, using various stratified sampling2 approaches relative to simple
random sampling. RADM comparative analysis.
Method
19 Existing RADM Strata
Defined from Clustering
1979-83 3-Day Events
38 Existing RADM Strata
Defined After Separating
Wet & Dry Categories
4 Strata Defined as
Seasons (Dec-Feb, etc.)
Temperature
Equal
Alloc,
0.98
0.83
3.06
Propor-
tional
Alloc.
1.58
1.76
3.05
Actual
Alloc, of
RADM
Events,
1.45
1.12
2.77
Relative Humidity
Equal
Alloc.
0.76
0.59
1.13
Propor-
tional
Alioc.
1.18
1.19
1.13
Actual
Alloe. of
RADM
Events
1.07
0.88
1.02
^H-Adjusted Extinction
Coefficient
Equal
Alloc,
0.67
0.49
1.17
Propor-
tional
Alloc.
1.18
1.17
1.18
Actual
Alloc. of
RA0M
Events
1.11
0.92
1.11
1 The table entries are means of station-specific efficiency ratios, averaged across stations within the
RADM geographic domain.
2 Reflects sampling of 3-day events from 1982-1985.
In the RADM development, 19 strata were defined based upon clustering of 3-day events
occurring between 1979 and 1983. These were then subdivided into wet and dry categories,
ultimately resulting in twice as many (38) strata that are actually employed in aggregation-based
estimation. (Events from the 1982-1985 time period were classified into these strata, with 30
events selected for use in RADM.) The temporal and spatial domains applicable to the RADM
development differ from those of the current analysis; therefore, any comparisons should be
considered in this context.
The RADM domain approximately corresponds to the combined northeast and southeast regions
displayed in Table 17-3. A comparison of Table 17-4 to Table 17-3 results under proportional
allocation suggests superior performance associated with the strata identified using clustering of
continental wind field data, with respect to the outcome measures used here. When the actual
allocation of RADM events is considered, the superiority of proportional allocation associated
with the continental analysis is further elevated.
17.5 Refinement of the Sampling Approach
Based, in part, on the results discussed in section 17.4, a stratified sampling scheme involving
seasonal clustering based upon four distinct seasons was selected for further consideration and
refinement. The general superiority of three- and four-season stratification schemes was
discussed previously in relation to results depicted in Table 17-2 (a-b), and a four-season scheme
was selected following additional considerations regarding differences in emission patterns
between spring and autumn that would not be apparent using our evaluative parameters alone. |
17-21
-------
EPA/600/R-99/030
Another benefit of using this type of scheme is that it naturally lends itself to the development of
se^ojtial^estimate^based upon four-season partitions. The derivation of such estimates from a
two- or three-season scheme would not be as well defined, and the estimates themselves would
likely be less precise.
Having selected this general approach, refinements were needed to determine an appropriate
number of strata, and to arrive at an adequate number of events for sampling. These refinements,
and a description of the sample of events that ultimately was selected, are discussed in this
section. This aspect of the analysis was limited to a refined time frame consisting of the nine-year
' jyg^oj ||om i 984-i 992, which was targeted to ultimately represent baseline meteorology for use
in modeling.
17.5.1 Determination of Appropriate Numbers of Strata and Events
The analysis proceeded under the assumption that, in order to satisfy the goal of matching or
exceeding the performance of the RADM 30-event stratification scheme, a minimum of 30 events
/would be required in the framework of the continental domain. The precision associated with the
estimation of annual means of the evaluative parameters was investigated for a range of 30 to 60
events, and for 16, 20, 24, and 28 seasonally defined strata (4, 5, 6, and 7 strata per season,
respectively).
These numbers of strata were chosen as candidates because 19 wind field-based clusters were
defined in the RADM scheme, which offers a certain degree of resolution with regard to the
characterization of transport (which is not specifically addressed by the evaluative parameters
investigated here). The range of 16 to 28 strata was selected to provide comparable resolution of
wind fields and associated transport, noting that higher numbers of strata result in greater
variation in the sizes of those strata, and this would force a more pronounced deviation from the
goal of proportional allocation. As discussed previously, a primary goal was to ensure that every
stratum is sampled, i.e., that there are no clusters which go unrepresented in the final set of
events. "'"" """""" ~ " " " .-.--.-...__ • »_-,-.... . .
Note that we did not adhere to traditional rules of thumb regarding the determination of
appropriate numbers of clusters to retain. These rules are based upon an assumption that some
finite number of clusters is appropriate to represent the variability inherent in these patterns, and
that additional clusters beyond that point add relatively little information. In reality, clusters
defined during this process represent a continuum, and traditional F-test statistics illustrate this
continuum quite smoothly. There is no magic number of clusters after which the relative
importance of additional clusters drops noticeably. Indeed, as cluster analysis is used here merely
for the definition of strata and not as an end in itself, there is no compelling reason to be restricted
by existing conventions regarding determinations of an optimal number of clusters.
Standard deviations associated with estimation under all of the combinations described above are
illustrated in Table 17-5. Furthermore, the process was repeated after seasonal adjustment of
17-22
-------
EPA/600/R-99/030
each outcome parameter (seasonally was removed by analyzing deviations from a fitted
sinusoidal curve), to ensure that long-term seasonal trends do not unduly influence any pattern.
The standard deviations in Table 17-5 provide an indication of the degree of precision associated
with estimation, and are influenced by the observed within-stratum variability of the parameter.
One would expect the inclusion of more strata to facilitate greater precision, unless this increased
precision is offset by incurred deviations from proportional allocation.
Table 17-5. Standard deviation1 associated with estimation of the annual mean of the indicated
parameter, using stratified sampling2 with 16, 20, 24, or 28 strata and 30, 35, 40, 45, 50, 55, or 60
events. Continental analysis.
No. of
Events
30
35
40
45
50
55
60
30
35
40
45
50
55
60
Temperature, deg. C
16
Strata
20
Strata
24
Strata
28
Strata
Relative Humidity, %
16
Strata
20
Strata
24
Strata
28
Strata
RH-Adjusted Extinction
Coefficient, km"4 (xio3)
16
Strata
20
Strata
24
Strata
28
Strata
NOT SEASONALLY ADJUSTED
0.97
0.89
0,83
0.79
0.74
0.71
0,67
0.98
0.89
0.83
0.77
0.74
0.70
0.67
0.95
0.87
0.81
0.76
0.72
0.69
0.66
0.95
0.87
0.81
0.75
0.72
0.68
0.65
2.38
2.19
2.04
1.93
1.82
1.74
1.66
2.40
2.20
2.05
1.92
1.83
1.74
1.66
2.38
2.20
2.06
1.93
1.82
1.73
1.65
2.40
2.21
2.05
1.93
1.82
1,73
1.65
6.69
6.03
5,71
5.38
5.09
4.78
4.62
6.73
6.08
5.63
5.34
5.06
4.79
4.62
6.73
6.10
5.70
5.33
5.07
4.83
4.60
6,77
6.23
5.71
5.36
5.08
4.83
4.60
SEASONALLY ADJUSTED
0.75
0.69
0.64
0.61
0.57
0.54
0,52
0.76
0.69
0.64
0.60
0.57
0.54
0.52
0.75
0.69
0.64
0.60
0.57
0.54
0.51
0.75
0.69
0.64
0.60
0.57
0.54
0.51
2.32
2.14
1.99
1.89
1.78
1.70
1.62
2.34
2.15
2.01
1.88
1.79
1,70
1.62
2.33
2.16
2.02
1.89
1.78
1.70
1.62
2.35
2.16
2.01
1.88
1.78
1.69
1.62
6.67
6.01
5.69
5.36
5.07
4.77
4.60
6.70
6.05
5.61
5.32
5.04
4.78
4.60
6.70
6.08
5.68
5.31
5.05
4.81
4.59
6.75
6.21
5.69
5.34
5.06
4.81
4.59
1 The table entries reflect the standard deviation associated with the average station-specific variances
(averaged across stations within the continental domain).
2 Results reflect sampling of 3-day events from 1984-1992.
For temperature and relative humidity, the table suggests that for any given number of events, any
effect associated with different numbers of strata is negligible. For extinction coefficient, standard
17-23
-------
EPA/600/R-99/030
deviations for 30 and 35 event schemes increase slightly with increasing numbers of strata. There
"is a. minimum standard deviation associated with 20 strata for the 40-event scheme, and only
negligible effects across strata for greater numbers of events.
The objective of this determination was to develop a sufficiently large set of strata to provide
some resolution with regard to the characterization of transport, yet be sufficiently concise to
avoid any reduction in precision that would result from unsampled strata or from the inability to
approximately satisfy proportional allocation. In consideration of these criteria and the above
results, 20 strata were determined to constitute an appropriately sized set.
Figure 17-7 (a-c) provides perspective regarding the relative precision provided by sample sizes of
25 or more eventsassociated with estimation of annual means ojfthe evaluative parameters, based
upon the use of 20 seasonally defined clusters as strata. The standard deviation displayed in each
u i iji, graph is actually associated with the average of the variances across sites. These graphs illustrate
,ti« ~ ", the advantage of stratified sampling with proportional allocation relative to simple random
sampling. They also illustrate the relative gains in precision (expressed as reduction in standard
deviation) that are realized as the number of events is increased. The star symbols plotted on the
graphs indicate the actual standard deviation that would be realized for selected sample sizes
under a 20-stratum scheme, with events distributed in accordance with proportional allocation to
the extent possible. The selected sample sizes were chosen based upon practical limitations
invplving the number of events that might realistically be implemented. The stars do not fall
7 sfrictly on the proportional allocation curve because of limitations associated with the sampling of
integer-valued numbers of events.
,, __ These graphs indicate that, for temperature, it might be practical to achieve a standard deviation
in the range of 0.6—1.0°C, and that a much larger sample size would be needed to reduce the
standard deviation below 0.5°C. Similarly, standard deviations associated with estimates of mean
relative humidity can possibly be achieved in the range of 1.5-2.5%, and realistic standard
deviations associated with extinction coefficient might be in the range of 0.0045-0.0070 km"1.
Geographic variability with respect to many of these results is described later in this section.
Although the estimation of mean levels of parameters is likely to be a primary point of emphasis
for many model-based results, the accurate estimation of extremes is also of significant
importance. This issue was specifically addressed by investigating the precision associated with
the estimation of the 90th percentiles of the evaluative parameters.
In_contrast tojthe standardjdeviations associated with estimation of the mean, there is no closed-
.'? if $* * - form solution to determine the yariabflity associated with the estimation of 90th percentiles.
'% Thgrefore, a Monte Carlo-type resampling approach was utilized to estimate these standard
' deviations. This specifically involved randomly selecting 200 artificial samples of actual data,
each consisting of the required number of events, from the 20 seasonally defined strata. From
each sample and at each site location, the 90th percentile of the parameter was estimated. The
variance of the resulting collection of 200 estimates was averaged across sites, and the associated
. .. 17-24
-------
EPA/600/R-9fl/030
standard deviation served as an estimate of the precision associated with the particular sample
size. This exercise was repeated for sets of 30, 40, 50, and 60 events.
The graphs in Figure 17-8(a-c) illustrate the resulting standard deviations, along with the standard
deviations associated with estimation of the mean using the indicated number of events. As would
be expected, the standard deviations associated with estimation of the 90th percentile are higher
than those associated with estimation of the mean of each parameter. This difference is most
pronounced for extinction coefficient, with standard deviations for 90th percentile estimation
being approximately three times those for mean estimation. For relative humidity, they are
approximately twice as large. The difference is least pronounced for temperature, where the
increase is less then 20%. In each case, there is only slight, gradual improvement in the precision
of 90th percentile estimation for sample sizes of greater than 40 events.
The next step is to arrive at an appropriate number of events to be distributed among these 20
strata. Table 17-6 displays the standard deviation associated with the estimation of the annual
mean of each evaluative parameter (both raw and seasonally adjusted) that would result from
samples consisting of 30, 40, 50, and 60 events. For comparison purposes, these standard
deviations represent average variances restricted to the RADM geographic domain. The table
also displays the standard deviation associated with the aggregation of 30 3-day events in a
sample stratified using the original RADM clusters. From this, it is clear that any number of
events would be sufficient to provide improved resolution relative to the RADM scheme, in the
context of the evaluative parameters reviewed here.
Although these results suggest that a 30-event sample would be sufficient to meet the objective of
matching or exceeding the performance of the RADM approach with regard to estimation
precision for these outcome parameters, other results displayed in this section demonstrate clear
improvement in the precision associated with estimation of both means and extremes by moving
from a 30-event sample to a 40-event sample. In addition, a 40-event set is needed to ensure
equal precision to the RADM approach with regard to the estimation of wet deposition amounts.
The RADM set of 30 events included 20 events from categories that were identified as "wet", i.e.,
for which average wet SO42" deposition exceeded the median for each cluster. In other words, 20
events were selected from the "wettest" 50% of all events.
This oversampling of wet events was originally pursued to ensure the adequate representation of
those events that contributed most significantly to wet deposition, because the accurate
characterization of wet deposition was the primary purpose of RADM at that time. However,
concerns have since arisen that the disproportionate representation of these events may have
introduced an overall bias with regard to the ambient concentrations of pollutants that are
influenced by cloud cover and precipitation. In view of these concerns, and since wet deposition
is not the primary focus for CMAQ, the oversampling of wet events was deemed inappropriate for
present purposes. In the absence of this oversampling, it is still necessary to include a sufficient
number of events to ensure that wet deposition is characterized as accurately here as in the
RADM approach. This would require approximately 20 wet events, and the same median-based
17-25
-------
EPA/600%99/030
definition of wet versus dry events implies that approximately 20 dry events should also be
included. Therefore, a total of 40 events was deemed necessary to satisfy all of the objectives.
Table 17-6. Standard deviation1 associated with estimation of the annual mean of the indicated
parameter, using stratified sampling2 with 20 strata and 30, 40, 50, or 60 events. RADM
comparative analysis.
No.
of
Eve
nts
30
40
50
60
30
40
50
60
Temperature, deg. C
20 Strata
Existing
RADM
Sample
(30 Events)
Relative Humidity, %
20 Strata
Existing
RADM
Sample
(30 Events)
RH-Adjusted Extinction
Coefficient, km"1 (XIQS)
20 Strata
Existing
RADM
Sample
(30 Events)
NOT SEASONALLY ADJUSTED
0.96
0.82
0.73
0.66
1.78
2.34
2.01
1.79
1.63
2.58
6.65
5.60
5.01
4.59
8.28
SEASONALLY ADJUSTED
0.74
0.63
0.56
0.51
0.82
2.32
1.99
1.77
1.61
2.41
6.64
5.60
5.00
4.58
8.23
1 The table entries reflect the standard deviation associated with the average station-specific variances
(averaged across stations within the RADM geographic domain).
2 "Existing RADM Sample" results reflect sampling of 3-day events from 1982-1985. Other results
reflect sampling of 3-day events from 1984-1992.
Recalling that for 40 events the precision associated with the estimation of mean annual extinction
coefficient (the primary evaluative parameter) was optimized using 20 seasonally defined strata
(Table 17-5), a final plan was adopted for sampling 40 events from 20 strata (5 strata per season)
using approximately proportional allocation.
Figure 17-9(a-d) displays star chart histograms of the 20 clusters defined as strata in this
- arrangement. Each chart illustrates the frequency of occurrence of 5-day events belonging to a
given cluster, and the clusters themselves are ordered according to overall frequency of
occurrence. The numbers arranged radially on each chart depict the number of events belonging
to the cluster from each month of the year. Map-based graphs of mean wind vectors for day 3 of
each cluster are contained in Figures 17-10 through 17-29.
17-26
-------
EPA/600/R-99/03Q
In order to examine the impact of this scheme geographically (in the context of the precision
associated with the estimation of mean levels of the evaluative parameters), we examined
graphically (not shown), the standard deviation at each site location alongside the actual mean
with which that standard deviation is associated. Similarly, analogous information associated with
the estimation of 90th percentiles were examined. This examination confirmed the relative
geographic uniformity with respect to the standard deviations, which serves as support for the use
of "average" standard deviations in drawing conclusions throughout this section.
17.5.2 Selection of Stratified Sample of Events
A stratified sample of events was randomly selected from the 20 seasonally defined strata for the
period 1984-1992. The sample was selected without replacement to ensure that no single day
was selected into more than one five-day event, i.e., that there was no overlap between selected
events. Systematic sampling (Cochran, 1977) was used within each stratum for which more than
one event was to be selected. Specifically, all events assigned to the stratum were ordered
chronologically, an event was selected near the beginning of that ordering, and subsequent events
were selected to be evenly spaced throughout the remainder of the ordering. If k events were to
be sampled from a cluster containing n events, to illustrate the simple case in which nlk is integer
valued, the first event would be randomly selected from any of the chronologically first nlk events,
and every (n/K)th subsequent event would be selected. The purpose of this approach was to
ensure appropriate representation over the entire nine-year period.
Table 17-7 displays the total number of events belonging to each stratum, the number of events
sampled, and the dates of the sampled events. These dates are the middle dates of the three-day
events for which the model ultimately is to be run (i.e., the last three days of the sampled five-day
event). This sample of 40 events includes representation from every month of the year, and from
every year during the period 1984-1992. Table 17-8 illustrates this representation by displaying
the number of events selected from each month and from each year.
17-27
-------
EPA/6QO/R-99/030
Table 17-7. Stratum sizes, number of sampled events per stratum, and dates of events in sample.
Dates shown are for middle day of 3-day event.
Stratum
1
2
3
4
5
6
7
8
0
10
11
12
13
14
15
18
17
18
19
20
Season
Spring
Summer
Autumn
Winter
Spring
Winter
Spring
Summer
Summer
Winter
Autumn
Autumn
Winter
Autumn
Summer
Autumn
Winter
Summer
Spring
Spring
Total # of
Days in
Stratum,
1084-1992
292
267
238
210
200
188
185
171
171
170
168
150
139
135
129
128
102
90
89
62
Number
of
Sampled
Events
3
3
3
3
2
2
2
2
2
2
2
2
2
2
2
2
1
1
1
1
Event Dates
12 March 1985, 08 May 1987, 27 March 1990
17 July 1985, 20 August 1987, 10 August 1990
08 September 1986, 12 October 1988, 08 October 1991
04 January 1986, 15 December 1988, 02 December 1992
07 May 1984, 06 March 1990
03 January 1987, 07 January 1992
01 April 1986, 26 March 1991
05 August 1986, 29 June 1992
07 August 1984, 12 July 1989
18 January 1984, 25 January 1989
18 October 1985, 12 September 1991
17 November 1987, 14 September 1992
19 February 1985, 27 January 1990
17 October 1988, 24 November 1991
03 July 1987, 09 July 1992
25 November 1985, 07 November 1990
18 December 1989
22 July 1989
09 May 1990
30 April 1991
Hi ...... i ........ ^'iii' ir
»*•»
. ,J
-n
I I
r ra •'• m * /I •!
J* »'~ l?t J W »
-------
EPA/600/R-99/030
Table 17-8. Number of sampled events representing each month of the year and each year from
the period 1984-1992.
Month
January
February
March
April
May
June
July
August
September
October
November
December
Number of Events in
Sample
6
1
4
2
3
1
5
4
3
4
4
3
: -Year ; .. '">•
1984
1985
1986
1987
1988
1989
1990
1991
1992
Number of Events in
Sample
3
5
4
5
3
4
6
5
5
17.6 Application and Evaluation
In this section, examples of the aggregation calculation for annual mean concentrations and total
wet depositions, applicable to this sample of events, are provided. Following this example is a
description of an evaluation exercise in which the aggregation calculation was carried out for light
extinction coefficient, and the aggregated estimates were compared to the actual values based on
data from all of the days in the period.
17.6.1 Application of the Aggregation Procedure
Aggregation calculations will be applied to model-based depositions and concentrations obtained
for each sampled event, to achieve unbiased estimates for annual and seasonal means and other
summary statistics within each grid cell. Since the goal of sampling from every defined stratum is
achieved in this approach, these calculations are simplified in comparison to earlier aggregation
methods (NAPAP, 1991). In essence, these aggregation calculations merely produce weighted
means, totals, or other summary measures, from the sample of events.
To illustrate the aggregation approach, consider the estimation of a mean annual air concentration
using model output for the 40 events selected above. These events represent 20 strata; denote
these using the subscript /, /=1,2, . . ., 20. Let/ denote the frequency of occurrence associated
17-29
-------
EPA/600/R-99/030
with stratum /, i.e., the total number of 3-day events belonging to the stratum during the period
1984-1992. For an individual grid cell, also let
represent the mean model-based concentration associated with all sampled events from stratum /.
Thus, for strata with a single sampled event, it is just the event mean concentration in the grid cell.
For strata with two or three sampled events, it is the mean concentration for all of those events.
Then the estimated annual mean air concentration is given by
20" _
Mean Air Concentration =
MODEL,
(17-2)
1=1
Estimates for most other parameters (e.g., dry depositions) and other summary statistics are
calculated using similar methods. The calculation for wet deposition is different, primarily
because the weighting is partially dictated by precipitation. Let
.............. • ..... MODEL,'
MODEL,'
_ represent the mean 3-day modeled deposition for sampled events in stratum /, the mean 3-day
modeled precipitation for sampled events in stratum i, and the total measured precipitation
accounted for by all events belonging to stratum /, respectively. Then the estimated total annual
wet deposition is given by
20
Total Wet Deposition ='£
iiii .......
. ........ iir IB
M°DEL'
MODEL
1
3x9
(17-3)
Thjsjexpression can be thought of as a weighted sum in which the model-estimated wet
concentration for a stratum is weighted by the total measured precipitation associated with the
stratum. The final component of this expression is included to reflect the fact that each day is
counted three times in the calculated sum (due to the use of 3 -day events) and that the strata are
defined over a nine-year period.
17-30
-------
EPA/600/R-99/030
17.6.2 Evaluation
In order to determine the effectiveness of the aggregation technique and subsequent episode
selection, comparisons were made between the observed mean b^s for the period 1984-1992 and
the aggregated estimates of that mean using the stratified sample of events described in 17.5,2 and
listed in Table 17-7. This preliminary evaluation, which includes simple regression analysis, is
similar to that performed on RADM (Eder and LeDuc, 1996; Eder et al., 1996).
Results comparing the observed and aggregated mean bex,s (based on Equation 17-2) are
promising as seen in the scatterplot provide in Figure 17-30. The correlation between the 201
observed and aggregated mean bex( was very high correlation (r2 = 0.988). Estimates of the
regression coefficients between the observed and aggregated mean bex!s reveal an intercept value
of-0.0012 that is not significantly different from zero (a = 0.05).- The slope (1.018), however; is
significantly different from 1.0, (a - 0.05) indicating a slight tendency for these particular
episodes to provide an over-estimate of the expected mean bext. This slight, "apparent" bias is,
however, well within the expected variability associated with the particular set of episodes in this
stratified sample. To wit, selection of a different random set of episodes would just as likely
result in a slight under-estimate of the expected mean bexl.
Perhaps a better way to illustrate the effectiveness of this technique can be shown through an
examination of the percent deviation in aggregate estimates of the mean b^, (where the deviations
are relative to the observed mean (aggregated bex[ - observed bcxt / bcxt observed). These percent
deviations, which were calculated over the period 1984 -1992, are presented in Figure 17-31, For
the most part, the deviations are within ± 10%, indicating excellent agreement between the actual
mean bcxt and the aggregated estimates of the mean bext.
The slight over prediction tendency mentioned above appears to be somewhat spatially biased as
also seen in Figure 17-31. As seen in the top of the figure, areas of generally positive deviations
(aggregation approach yields a higher bexu hence lower visibilities than observed) appear to
concentrate from east Texas into the southeastern states and again in the upper midwest between
Minnesota and the Dakotas. The states of California and Idaho also exhibit positive deviations.
Negative deviations, presented in the bottom of the Figure 17-31, tend to predominate from the
northeast states into the Great Lake States and southwestward toward the states of Kansas, New
Mexico and Arizona. This spatial dependence of the estimates is, once again, well within the
expected variability. Selection of a different random set of episodes would likely result in a
different pattern of positive and negative deviations, as there is a natural tendency for sites at
close proximity to behave in a similar fashion.
The scatter plot in Figure 17-30 also reveals an increase in variance about the regression line
starting at an observed mean bcxt of 0.085. Unlike the positive bias, discussed above, this increase
in variance does not appear to be spatially biased, but rather exhibits a random distribution across
the domain. This is represented in Figure 17-31 by the scattering of the larger biases (i.e. biases >
5.0%) evenly across the domain.
17-31
-------
EPA/600/R-99/030
17.7 Summary and Discussion
The objective of this research was to develop a new aggregation approach and set of events to
support model-based distributional estimates of air quality parameters (acidic deposition, air
concentrations, and measures related to visibility) over the continental domain. The basic
approach is to define meteorological categories that account for a significant proportion of the
variability exhibited by these air quality parameters, as well as the particular transport mechanisms
involved, so that source-attribution analyses are facilitated. This requires that categories be
defined with an emphasis on wind flow parameters. To this end, the cluster analysis of zonal «
and meridional v wind field components has been used to determine meteorologically
representative categories.
The research described in this chapter was carried out in three phases:
• Phase ] : Various stratification schemes were evaluated and compared on different
temporal and geographic scales to support the selection of a preferred general
methodology. The selected methodology involved clustering of wind field data over the
continental domain within each of the four seasons, and defining strata to be equivalent to
. the resulting clusters. This methodology demonstrated superior relative efficiency
compared to methods defined on an annual time frame or on a regional scale for estimates
B: r involving the evaluative meteorological parameters, and is designed to support seasonal
estimation with both simplicity and precision.
"' ' '"' '''*'"' !'!|l|"!l!'''!l' ' ' ' l"'1'1 ' ' ' " ' | ' " ' ' 'I'!'1' ""'" '' ' 1,1, I"'1'"'!1' in " i, " |l,i i n1 ' '• ''l;:
» ^ Phase 2: Determinations were made regarding appropriate numbers of clusters and events
- 1 to support sampling using the general methodology selected in Phase 1 . The resulting
«* scheme involved a total of 20 clusters (5 per season), and 40 events, defined over the time
, -r li-iii ,-*«,. .if IP. '«•« .^ ..«. ,. . .. _..>. r „ ..... '.*. . ..
period 1984-1992. This scheme affords superior precision to previous approaches for
estimates involving the evaluative meteorological parameters, supports approximately
equivalent representation of wet events to those approaches without oversampling, and
provides adequate resolution of wind field patterns that characterize transport.
• Phase 3: A stratified sample of events was selected under approximate proportional
allocation, using systematic sampling within strata, in accordance with the scheme
determined in Phase 2, This sample was successfully evaluated through a comparison of
aggregated estimates of the mean b^ to the actual mean b^ revealing a high level of
agreement, although there was a slight tendency of the aggregation and episode selection
technique to over-estimate the expected mean b^.
The goal of this research was to categorize many years worth of meteorological patterns into a
few classes. This represents a very ambitious goal, and it should not be surprising that there is
substantial variability associated with the wind vectors assigned to individual clusters.
Nevertheless, the results described above suggest that the approach achieves a reasonable
characterization of frontal passage scenarios and leads to clusters that explain variation in the
17-32
-------
EPA/600/R-99/030
evaluative meteorological parameters used in this analysis (temperature, relative humidity, and
visibility), and therefore can be used to achieve improved estimates of these parameters relative to
estimates obtained from simple random sampling. Moreover, the constrained definition of distinct
seasonally-based clusters brings further improvement to the ability of the clusters to explain the
variation in these parameters, and therefore leads to more precise estimates associated with them.
The evaluative parameters were selected for their known associations with many air quality
parameters of interest, thus suggesting that the clusters should also be effective in defining strata
from which events can be selected to estimate those air quality parameters.
17.8 References
Brook, J.R., PJ. Samson, and S. Sillman, 1995a. Aggregation of selected three-day periods to
estimate annual and seasonal wet deposition totals for sulfate, nitrate, and acidity. Part I: A
synoptic and chemical climatology for eastern North America. Journal of Applied Meteorology,
34, 297-325.
Brook, J.R., P.J. Samson, and S. Sillman, 1995b. Aggregation of selected three-day periods to
estimate annual and seasonal wet deposition totals for sulfate, nitrate, and acidity. Part II:
Selection of events, deposition totals, and source-receptor relationships. Journal of Applied
Meteorology, 34, 326-339.
Cochran, W.G., 1977. Sampling Techniques. Wiley & Sons, New York.
Davis, R.E. and L.S. Kalkstein, 1990. Development of an automated spatial synoptic
climatological classification. International Journal of Climatology, 10, 769-794.
Eder, B. K., J. M. Davis and P. Bloomfield, 1994. An automated classification scheme designed
to better elucidate the dependence of ozone on meteorology. Journal of Applied Meteorology,
33, 1182-1199.
Eder, B. K., S.K. LeDuc and F. Vestal, 1996. Aggregation of selected RADM simulations to
estimate annual ambient air concentrations of fine particulate matter. Ninth Joint Conference on
Applications of Air pollution Meteorology with the A&WMA, Jan.28-Feb. 2, Atlanta, GA.
Eder, B. K. and S.K. LeDuc, 1996. Can selected RADM simulations be aggregated to estimate
annual concentrations of fine particulate matter? Proceedings of the International Specialty
Conference on the Measurement of Toxic and Related Air Pollutants, May 7-9, Research
Triangle Park, NC.
Fernau, M.E. and PJ. Samson, 1990a. Use of cluster analysis to define periods of similar
meteorology and precipitation chemistry in eastern North America. Part I: Transport patterns.
Journal of Applied Meteorology, 29, 735-750.
17-33
-------
EPA/600%99/030
Femau, ME. and P.J. Samson, 1990b, Use of cluster analysis to define periods of similar
meteorology and precipitation chemistry in eastern North America. Part II: Precipitation patterns
and pollutant deposition. Journal of Applied Meteorology, 29, 751 -761.
Kalnay, E., M. Kanamitsu, R. Kistler, W. Collins, D. Deaven, L, Gandin, M. Iredell, S. Saha, G.
White, J. Woollen, Y. Zhu, M. Chelliah, W. Ebisuzaki, W, Higgins, J. Janowiak, K. Mo, C.
Ropelewski, J. Wang, A. Leetmaa, R. Reynolds, R. Jenne and D. Joseph, 1996. The
NCEP/NCAR 40-year reanalysis project. Bull. Anter. Meteor. Soc., 77, 437-471.
NAJPAP, 199L National Acid Precipitation Assessment Program 1990 Integrated Assessment
Report, National Acid Precipitation Assessment Program, 722 Jackson Place NW, Washington,
D.C.
SAS Institute Inc., 1989. SAS/STAT^ User's Guide, Version 6, Fourth Edition, Volume 1. SAS
Institute Inc., Gary, NC.
Ward, J.H., 1963. Hierarchical grouping to optimize an objective function. Journal of the
American Statistical Association 58, 236-244.
Tiiis chapter is taken from Science Algorithms of the EPA Modeh-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S.
Clung, 1999.
17-34
-------
EPA/600/R-99/030
Fig. 17.1 (a) Mean wind vectors for day 1 of annually defined cluster
1 (of 30).
17-35
-------
II1 111 '' Mill ><
;: *
I I! Ill,
EPA/600/R-99/030
Fig. 17.2 (c) Mean wind vectors for day 5 of annually defined cluster
9 (of 30).
Fig. 17.3 (a) Actual wind vectors for 25 January 1986, day 3 of cluster
9 (of 30).
17-38
-------
EPA/600/R-99/030
Fig, 17,3 (b) Actual wind vectors for 13 January 1988, day 3 of cluster
9 (of 30).
Fig. 17.3 (c) Actual wind vectors for 8 January 1990, day 3 of cluster 9
(of 30).
17-39
-------
EPA/600/R-99/030
Fig. 17.4 (a) Mean and distribution of wind vectors for day 1 of
cluster 9 (of 30).
• "=- x '
1 i'
,f
Fig. 17.4 (b) Mean and distribution of wind vectors for day 3 of
Ijclujter 9 (of 30).
,: ,: :i ' Hi
17-40
I
-------
EPA/600/R-99/030
Fig, 17.4 (c) Mean and distribution of wind vectors for day 5 of cluster 9 (of 30).
17-41
-------
EPA/600/R-99/030
•».» Frequencies of 30 Clusters (Annually Defined)
Clusfar No. 1: 12.19%
Cluster No, 2; 12.10%
' ClusUr No. 3: 5.53%
ClustirNo. 5: 4.96%
Cluster No. 4: 5.37%
Fig. 17.5 (a) Monthly frequencies of annually derived clusters for clusters
17-42
i n
-------
EPA/600/R-99/030
Frequencies of 30 Clusters (Annually Defined)
Cluster No. 7: 4.29%
Clus+nr No. 8: 4,09%
Cluster No. 9: 3.65%
Jan
0»c
17
Clustar No. 10: 3.61%
Cluster No. 11: 3.47%
Cluster No. 12: 3.40%
Fig. 17.5 (b) Monthly frequencies of annually derived clusters for clusters 7-12.
17-43
-------
EPA/600/R-99/030
Frequencies of 30 Clusters (Annually Defined)
ipt
Cluster No. 13: 2.79%
Cluster No. 14: 2.79%
Cluster No. 15: 2.67%
ClusJer No. 17: 2.17%
Cluster No. 16: 2.35%
JO" _ .
S.p
Cluster No. 18:
,• * ' ^
2.08%
Fig. 17.5 ^ Monthly frequencies of annually derived clusters for clusters 13-18.
17-44
-------
EPA/600/R-9&/030
Frequencies of 30 Clusters (Annually Defined)
Cluster No. 19: 1.85%
80%
Cluster No. 21: 1.76%
Cluster No. 22: 1.67%
Cluster No. 24: 1.55%
Fig. 17.5 (d) Monthly frequencies of annually derived clusters for clusters 19-24.
17-45
-------
BPA/60G/R-99/030
Frequencies of 30 Clusters (Annually Defined)
Cluster No. 25: 1.48%
Jan
Cluster No. 26: 1.44%
Clusior No. 27: 1.30%
Jan
Cluster No. 28: 1.21%
-~ Cluster No. 2i: 1.14%
Cluster No. 30: 1.00%
Fig. 17.5 (e) Monthly frequencies of annually derived clusters for clusters 25-30.
17-46
-------
EPA/60Q/R-99/030
Ftg. 17,6 Site locations for 201 meteorological parameters used in the evaluation.
17-47
-------
EPA/600/R-99/030
Standard Deviation Versus Sample Size: Temperature
Average (Across Stations) Standard Deviation Associated with
Estimation of 1984-1992 Mean from Aggregation of 3-Day Episodes
o
g 2.0
2. 1-5
2
TJ
1.0-
0.5-
.3 0.0 q
50 100 150 200 250
Number of Episodes in Sample
300
Samplng
Slirpl» Random (Sampling
Stralltl>d: ProBortlonalAltoc.
* * * Aehl»vabl» wf 20 Buata
Fig. 17.7 (a) Standard deviation of estimated mean temperature
versus sample size.
Standard Deviation Versus Sample Size: Relative Humidity
Average (Across Stations) Standard Deviation Associated with
Estimation of 1884-1992 Mean from Aggregation of 3-Day Episodes
I":.
4.0 -
^ 3.5 -
~ 3.0-
> 2.5 -
° 2.0-
TJ
« 1.5-
TJ
g 1.0-
4-t
M 0.5-
0.0 -
50 100 .150 200 250
Number of Episodes in Sample
300
Sampling
Slwpl* Random &&ng
Btttt»l»d: ProBOMkiMlAloc.
* * * AcMtvabl* *I20 SUatA
il'Ji ':
Fig. 17.7 (b) Standard deviation of estimated mean relative humidity
versus sample size.
v- : :,;::": •- • : , : 17-48
-------
EPA/600/R-99/030
Standard Deration Vortus Sample Sat: RH-Adj Extinction Coett
Av»rag» (Across Stations) Standard Deviation Associated with
Estimation of 1864-1602 Mean from Aagragtlion of 3-Day Episodes
0.010 -
"I o.ooa -
^ 0.006 -
TJ 0.004 -
a
1 0.002 -
CO
0.000 -
SO 100 150 200 250
Number of Episodes in Sample
300
Sampling
GtntMtd: PriportfonilAlDt.
Achbtfabt* wT3D Strata
Fig. 17.7 (c) Standard deviation of estimated mean extinction
coefficient (adjusted for relative humidity) versus sample size.
Standard Deviation Versus Sample Size: Temperature
Avaraga (Across Stations) Standard Deviation Associated with Estimation of
1934-1962 Milan & 9Mn Pcfl from Aggregation of 3-Day Episodes In 30 S3r*ta
o
ai
43
I
a
•a
1.2-
1.0-
0.8-
0.6-
0.4-
0.2-
0.0 -
20 30 40 50 60
Number of Episodes in Sample
70
Estimation of..
Fig. 17,8 (a) Standard deviation of estimated mean and 90th
percentile temperature versus sample size.
17-49
-------
EPA/600/R-99/030
Standard Deviation Versus Sample Size: Relative Humidity
Avenue (Acros* Station*) Standard Davlatton Auodat«d with Estimation of
19B4-1882 M««n & 90th Pefl from Aoflr«g»aon of 3-Day EpteoctM In 20 Strata
6 H
.4 -
3 -
-------
EPA/6OO/R-99/O30
Frequencies of 20 Clusters (5 Winter, 5 Spring, 5 Summer, 5 Auiumn)
Cluster No. 1: 8.89% (SPRING)
Cluster No. 3: 7.25% (AUTUMN)
Cluster No. 5: 6.09% (SPRING)
Cluster No. 2: 8.13% (SUMMER)
Cluster No. 4: 5.39% (WINTER)
Cluster No, 6: 5.72% (WINTER)
Fig. 17.9 (a) Monthly frequencies of seasonally derived clusters for clusters 1-6.
17-51
-------
EPA/600/R-99/030
Frequencies of 20 Clusters (5 Winter, 5 Spring, 5 Summer, 5 Auiumn)
Cluster No. 7: 5.63% (SPRING)
o.c
o
Mov
0 ,
Mar
38
Q,
'
*
uar
14
Ocl
0
Aug , . Jun
o' J-' o
CIu*»«r No. 9: 5.2IX (SUMMER)
Clusitr No. II: 5.12% (AUTUMN)
Cluster No. 8: 5.21% (SUMMER)
Cluster No. 10: 5.18% (WINTER)
Cluster No. 12: 4.57% (AUTUMN)
Fig. 17.9 (b) Monthly frequencies of seasonally derived clusters for clusters 7-12.
17-52
-------
EPA/600/R-99/030
Frequencies of 20 Clusters (5 Winter, 5 Spring, 5 Summer, 5 Autumn)
Clusier No. 13: 4.23% (WINTER)
Cluster No. 15: 3,93% (SUMMER)
Cluster No. 17: 3,1 1% (WINTER)
Cluster No. 14: 4.1 1% (AUTUMN)
Cluster No. 16: 3.90% (AUTUMN)
Cluster No. 18: 2.74% (SUMMER)
Fig. 17.9 (c) Monthly frequencies of seasonally derived clusters for clusters 13-18.
17-53
-------
EPA/600/R-99/030
frequencies of 20 Clusters (5 Winter, 5 Spring, 5 Summer, 5 Autumn)
Cluster No. 19: 2.71% (SPRING)
Cluster No. 20: 1.89% (SPRING)
Fig. 17.9 (d) Monthly frequencies of seasonally derived clusters for clusters 19-20.
••i* mmmm •* >, i, , ~ i ' i - • am „ ' - •>? v , . K . S. iBl. -J l ' . v if ) ' *«
Fig. 17.10 Mean wind vectors for day 3 of seasonally (Spring) defined
cluster 1 (of 20).
17-54
-------
EPA/600/R-99/030
Fig. 17.11 Mean wind vectors for day 3 of seasonally (Summer)
defined cluster 2 (of 20).
Fig. 17.12 Mean wind vectors for day 3 of seasonally (Autumn) defined
clusters (of20).
17-55
-------
EPA/600/R-99/030
Fig. 17.13 Mean wind vectors for day 3 of seasonally (Winter) defined
.cluster 4 (of 20).
•m -9
Fig. 17,14 Mean wind vectors for day 3 of seasonally (Spring) defined
cluster 5 (of 20),
17-56
-------
EPA/600/R-99/030
Fig. 17.15 Mean wind vectors for day 3 of seasonally (Winter) defined
cluster 6 (of 20).
Fig. 17.16 Mean wind vectors for day 3 of seasonally (Spring) defined
cluster 7 (of 20).
17-57
-------
EPA/600/R-99/030
Fig, 17.17 Mean wind vectors for day 3 of seasonally (Summer)
...defined cluster 8 (of20).
* Fig. 17.18 Mean wind vectors for day 3 of seasonally (Summer)
* defined cluster^ (of 20).
17-58
-------
EPA/600/R-99/030
Fig, 17.19 Mean wind vectors for day 3 of seasonally (Winter)
defined cluster 10 (of 20).
Fig. 17.20 Mean wind vectors for day 3 of seasonally (Autumn)
defined cluster 11 (of 20).
17-59
-------
EPA/600/R-99/030
; ;:r ;» ; , tU yV'iit • Lit 11 ' !
,. i: ' - i. ill I! "":: ii „ ,1 : iiiiiii i ill ,$•• '• ic" >:;T i. JM ,F,-
...11; f
3 irfilll
Fig. 17.21 Mean wind vectors for day 3 of seasonally (Autumn)
defined cluster 12 (of 20).
Fig. 17.22 Mean wind vectors for day 3 of seasonally (Winter)
defined cluster 13 (of 20).
17-60
' "" - I )
-------
EPA/600/R-99/03G
Fig. 17.23 Mean wind vectors for day 3 of seasonally (Autumn) defined
cluster 14 (of 20).
Fig. 17.24 Mean wind vectors for day 3 of seasonally (Summer)
defined cluster 15 (of 20).
17-61
-------
EPA/600/R-99/030
Eifc-
M? ! •,,*
it i \ , a .
Fig. 17.25 Mean wind vectors for day 3 of seasonally (Autumn)
defined cluster 16 (of 20).
Fig, 17.26 Mean wind vectors for day 3 of seasonally (Winter) defined
cluster 17 (of 20).
' " ': ' " ": : 17^62
«k
-------
EPA/600/R-99/030
Fig. 17.27 Mean wind vectors for day 3 of seasonally (Summer) defined
cluster 18 (of 20).
Fig, 17.28 Mean wind vectors for day 3 of seasonally (Spring) defined
cluster 19 (of 20).
17-63
-------
EPA/600/R-99/030
Fig. 17.29 Mean wind vectors for day 3 of seasonally (Spring)
defined cluster 20 (of 20).
Extinction Coefficient (km -1
c «•*
*JZ A -\C
&
"ra
0.1
S
i
* „.
-1
*
«**
t*
t»T
^"
w**
r
t •
t
*-»
^»-
0,05 0.1 0.15
Observed Mean
0.2
Fig. 17.30 Scatterplot of the observed mean b^, (km"1) versus the
aggregated estimate of the mean bm (km"1) for the period 1984-
1992.
17-64
-------
EPA/600/R-99/030
4MB
Fig. 17.31 Spatial variation of the bias of the aggregated estimates of the mean bext
(km-1) for the period 1984-1992. (Deviations (%) are relative to the observed mean:
aggregate-observed/observed). Top figure indicates sites with positive bias, bottom
figure sites with negative bias.
17-65
-------
EPA/600/R-99/030
17-66
-------
EPA/600/R-99/030
Chapter 18
INTEGRATION OF SCIENCE CODES INTO MODELS-3
Jeffrey Young
Atmospheric Modeling Division
National Exposure Research Laboratory
U.S. Environmental Protection Agency
Research Triangle Park, NC 27711
ABSTRACT
A complete group of CMAQ science codes has been integrated into the Models-3 system
following a set of rules to ensure compatibility and compliance with design principles that enable
modularity and flexibility and that allow easy modification and replacement of science process
components. This chapter describes the concept of classes and modules, the Models-3
input/output application programming interface, science code configuration management, and
model building and execution concepts
*On assignment from the National Oceanic and Atmospheric Administration, U.S. Department of Commerce,
Corresponding author address: Jeffrey Young, MD-80, Research Triangle Park, NC 27711. E-mail:
yoj@hpcc.epa.gov
-------
EPA/600/R-99/030
18.0 INTEGRATION OF SCIENCE CODES INTO MODELS-3
18.1 Introduction
Integration of the Community Multiscale Air Quality (CMAQ) science into the Models-3 design
paradigm takes place at the level of transforming the science, described by systems of partial
differential equations, into a discretized and parameterized numerical model. The integral parts
that arc involved in the process are the coding language (Fortran 77, currently), an operational
design, a set of coding rules, computer platforms, model data access methods and storage
management, and a public code repository. Models-3 is a user interface system designed to
facilitate the development and use of simulation models by scientists, model developers and
I,,,,!!, i' "|, jijiiiiiiri mi'" f it„ iiii'i iiiiiiiiiiiinii ,,*I ,,,, j 7 r
regulatory users. The Models-3 framework provides a high-performance computational structure
for a community modeling system that currently contains distinct emissions, meteorological, and
£, , f. chemical transport air quality models. In this chapter we define a model as a single, complete
executable built from a set of compiled subroutines that define either a core environmental
simulation model or a processing system that provides data to a simulation model. The
framework also links the models to data management tools and analysis/visualization software.
A complete set of CMAQ science codes has been integrated into the Models-3 system. The
development and integration of these codes has followed a set of principles relating to design and
implementation concepts that are discussed in this chapter.
We have promulgated a small set of software requirements and integration rules related to these
concepts so that science codes can be developed that conform to Models-3 requirements. With
tills approach, the general user community may readily develop and execute models that use the
rich and growing base of science codes and data that reside in Models-3. A user may contribute
•*, m to that base and allow others to integrate his developments by applying these integration rules.
- With access to all the released, or publicly available source files, a developer can supplement his
own codes and readily develop and test model versions within a highly modularized building
environment. On the other hand, people whose primary interest is studying the effects of control
strategies or regulatory applications can easily build and execute standard model versions for
different emissions scenarios and modeling domains.
The Models-3 and CMAQ systems have been developed to foster community access, and to
facilitate the improvement of the CMAQ system over time by the easy inclusion of new
scientific models and modules developed by researchers in the larger scientific community.
The CMAQ system contains a number of models needed to cany out an air quality model
— . . application. Defining a "model" to be an executable built for a specific application, the codes
Qjjj / j . thajjnjikejip thejnodel are relatively general. To build a specific model, the user selects a set of
fundamental criteria that determine the application, such as the modeling domain, the chemical
mechanism, and physical process solvers. The reader is referred to Chapter 15 for details.
Defining a generic model as the set of codes from which a particular application can be built, the
following generic CMAQ models are currently available in the system:
• ICON provides the required initial conditions for a model simulation as concentrations of
specified individual chemical species for the complete modeling domain.
18-2
:.M M >\-.\.f'.f^ k. - -';?«,. ' i,. •:*. V-V _ '•&' ' .
-------
EPA/600/R-99/030
• BCON provides the needed simulation boundary conditions as concentrations of
individual chemical species for the grid cells surrounding the modeling domain.
• Emissions-Chemistry Interface Processor (ECIP) transforms emission files produced
by the Models-3 Emissions Processing and Projection System (MEPPS) into an hourly,
3-D emissions data file for the CMAQ Chemical Transport Model (CCTM). For a
detailed discussion of the MEPPS operations, refer to Chapter 6 of the Models-3 User
Manual [6],
• The Meteorology-Chemistry Interface Processor (MCIP) interprets the output from a
meteorological model, such as the Perm State/NCAR fifth-generation Mesoscale Model
(MM5) [8], and prepares the data for use in the CCTM.
• The Landuse Processor (LUPROC) provides a high-resolution landuse database for the
CMAQ system. For example, MCIP uses the output from LUPROC to obtain surface
characteristics for computing dry deposition and other planetary boundary layer
parameters.
• The photolysis model (JPROC) computes photolytic rate constants (j-values) for the
gas-phase chemistry used in the CCTM.
• The Plume Dynamics Model (PDM) generates plume dimensions and positions along
with other related data for use in applying the plume-in-grid module in a CCTM
simulation.
• The CMAQ Chemical Transport Model (CCTM) simulates the chemical and physical
processes affecting tropospheric pollutants and estimates pollutant concentrations (e.g.,
ozone, particulate matter (PM2.5 and PMio), and carbon monoxide) and acid deposition.
The reader is referred to the Models-3 User Manual [6] for more details and descriptions of the
usage of these models.
Each generic model consists of a complete set of codes that can be compiled and linked into
different model executables by means of specific user-selected options with regard to groups of
code called modules, the horizontal grid and vertical layer domain, the chemical mechanism, the
computer platform, and compiling options.
One of the key issues in modeling is the manner in which developers and users must deal with
reading, writing and using model data. Models-3 provides a user-friendly Input/Output
Applications Programming Interface (I/O API) library [4] that enables a universal approach to
managing data across subroutines, models, platforms, and networks.
The organization of the remainder of this chapter is as follows: We discuss the concepts that lead
to and the implementation of classes and modules in the next section (18.2). We describe I/O
API usage and function in Section 18.3. Code management, an important consideration for
complex models, is discussed in Section 18.4. In Section 18.5 we discuss how a model is
organized and constructed through Models-3. Issues in regards to executing a particular model
are discussed in Section 18.6. Using the Models-3 framework to construct and execute a model
18-3
-------
EPAv'600/R-99/030
is summarized in Section 18.7. Finally in Section 18.8, we discuss concepts and issues that
address Models-3 compliant coding practice and how to ensure code development that is
conformant to the Models-3 system.
18.2 Classes and Modules
18.2.1 Operational Design
TTie design concept of code classes is used to facilitate the plug&play capability in the Models-
3/CMAQ system. Science process modules are grouped into classes that are primarily based on
the^meHSpHtting paradigm used in a CMAQ Chemical Transport Model (CCTM). Generally
each class is associated with a particular science process. A module consists of a complete set of
subroutines capable of modifying the concentration field related to the science process associated
with the module's class. The CCTM is designed so that each module computes the changes in
the concentration field (CGRID) specific to that particular science process. A science module
operates on trie entire three-dimensional gridded concentration field for a period of time called
the synchronization time step interval. The module performs whatever looping over that grid
and over whatever internal time steps that are necessary to complete its function.
Table 18-1 lists the classes and associated processes and modules currently available for the
CCTM: In budding a particular CCTM, one module is selected from each class.
The other CMAQ models (i.e., those other than the CCTM) do not generally follow the time-
splitting science process paradigm. However, for convenience and consistency in model
building, and for code organization, we have extended the concept of classes and modules to
apply to the other CMAQ models. Class and module organizations similar to those in Table 18-1
exist for the other models but are less extensive.
:"-: - r, , . • <« - .;, it . -.r , ..ii'i., 1 :.. i,;; :.
18-4
-------
EPA/600/R-99/030
Table 18-1. The Classes, Processes, and Modules Available in CCTM
CLASS
(PROCESS)
MODULES
driver (Control Model Execution)
init (Initialize Model)
hadv (Horizontal Advection)
vadv (Vertical Advection)
hdiff (Horizontal Diffusion)
vdiff (Vertical Diffusion)
adjcon (Advected Mass Adjustment)
phot (Photolytic Rate Constants)
chem (Gas Chemistry Solver)
aero (Aerosol Solver)
aero_depv (Aerosol Dry Deposition)
cloud (Cloud and Aqueous Chemistry)
ping (Plume-in-Qrid)
procan (Process Analysis)
couple (Couple for Transport)
util (Utility Processing)
ctm
init
hbot (Bott's scheme)
hppm (Piecewise Parabolic Method)
hadv_noop (no operation)
vbot (Bott's scheme)
vppm (Piecewise Parabolic Method)
vadv_noop (no operation)
unif (Uniform eddy diffusion)
hdiff_noop (no operation)
eddy (eddy diffusion)
vdiff_noop (no operation)
denrate
adjcon_noop (no operation)
phot
phot_noop (no operation)
smvgear (SMVGEAR solver)
qssa (Vectorized QSSA solver)
chem_noop (no operation)
aero
aero_noop (no operation)
aero_depv
aero_depv_noop (no operation)
cloud_radm (RADM cloud scheme)
cloud_noop (no operation)
ping_smvgear (uses SMVGEAR solver)
ping_qssa (uses QSSA solver)
ping_noop (no operation)
pa
gencoor
util
18.2.2 CGRID
To facilitate the science processing for the different chemistries involved in CMAQ, the
concentration field array CGRID, is partitioned into four species classes: gas chemistry, aerosols,
non-reactive, and tracer species. This order is mandatory and maintained throughout the
processing. A subroutine, CGRID_MAP provides the means to correctly index into the CGRJD
array for each science process that deals with individual classes of species. The tracer species
group is user-determined and may be empty in CGRID. Aqueous chemistry is dealt with solely
in the cloud processing, and currently aqueous species are not transported outside of the cloud
processing, therefore do not appear explicitly in CGRID. A subroutine call to CGRID_MAP
provides pointers to the locations of the species classes within CGRID. A process that deals only
with one of the classes (e.g. aerosols) can thus determine which part of CGRJD it needs to
access.
18-5
-------
EPA/600/R-99/030
.18.2.3,. Class Driver
Generally, each module within a class requires a "class-driver," which is the top level subroutine
within the module - it is at the top of the call chain for that module. Exceptions for this paradigm
exist for the util, procan, phot, and aero_depv classes. The class-driver presents a fixed name
and calling interface to the driver class subroutine SCIPROC that calls all the science process
modules. The calling interface consists of a subroutine argument list containing CGRID, the
current scenario data and time, and a time step vector. Each class also contains a "no-operation"
module, if appropriate. The no-op module has the fixed name and call interface for that class,
but performs no calculations. When called, it merely returns control to SCIPROC.
The following algorithm illustrates the processing and call sequence used in a CMAQ model for
fractional time steps processed non-symmetrically with respect to the chemistry processes:
Load CGRID from initial data; set initial Date/Time
Get the TimeStep vector: TimeStep[1] = output time interval
TimeStep[2] = synchronization time interval
_ " TimeStep[3] «= horizontal advection time interval
Foreach output_time_jstep
Call Couple { CGRID, Date, Time, TimeStep )*
Foreach sync_step [in output time interval]
Call Horizontal_Advection { CGRID, Date,
Call Vertical_Advection ( CGRID, Date,
Call Mass_Adjustment ( CGRID, Date,
Call Horizontal_diffusion ( CGRID, Date,
Call De-Couple ( CGRID, Date,
Call Vertical_diffusion ( CGRID, Date,
Call Plume_in_Grid ( CGRID, Date,
Call Gas_Chemistry ( CGRID, Date,
Call Aerosol { CGRID, Date,
Call Cloud { CGRID, Date,
m •: ,*• Advance Date/Time by synchronization time interval
Call Couple ( CGRID, Date, Time, TimeStep )*
End Foreach sync_step
W'. • - -t - .-.nt -•* . - .;-.-, * ...«~>«-jr '•<..• ^ •.;•*» > £• i . i . 9' 11 -.
_ .h Call De-Couple ( CGRID, Date, Time, TimeStep )*
Write Concentration File
-« • End Foreach output_time_atep
* Couple O3HID concentrations with, or de-couple CGRID from the air density x the
Jacobian of the computational grid. See Chapter 6 for a complete discussion on the
concept.
Not shown in the algorithm are processes that deal with source emissions (in either
Vertical_diffusion or Gas_ehemistry), dry deposition (in Vertical_diffusion), and wet deposition
(in Cloud). Also not shown are calls to process analysis routines after each science process,
which compute the integrated process rates for that particular process. Such calls are also
embedded in science process modules that affect CGRID, such as emissions injection in either
the vertical diffusion process or gas chemistry. The Science Process Code Template in Section
18.8.3 below illustrates how these calls can be integrated within a module.
Time,
Time,
Time,
Time,
Time,
Time,
Time,
Time,
Time,
Time,
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
TimeStep )
18-6
-------
EPA/600/R-99/03Q
In order to maintain the Models-3/CMAQ modularity and data-independence, the science
process class-drivers must adhere to coding standards with regard to the calling interface, file
data I/O, and standardized, uniform domain and mechanism data within global include files. See
Section 18.5, below, and Chapter 15 for a more complete discussion on the use of global include
files. These coding standards have not been propagated to the level below the class-driver. At
the sub-module level there are no strict I/O or coding standards, however we offer suggestions to
facilitate the potential incorporation of a module into the Models-3 system in Section 18.8 below,
18.2.4 Synchronization Time Step
There are many different time scales that are important in modeling. The CCTM deals with four
levels of time stepping:
1. Output time step - the time interval for which output data is written to disk.
2. Synchronization time step - the time interval for which the science processes,
represented by science modules, are considered to run independently of the other
processes. This is the time interval during which no interaction with the other time-split
science processes needs to take place.
3. Advection time step - the time interval over which horizontal advection occurs. It may
be less than or the same as the synchronization time interval.
4. Local or internal time looping - a possibly variable time interval that subdivides the
synchronization time step and is dependent on the user's implementation of a particular
algorithm.
There may also be some science processes that don't fit into this type of time scale hierarchy.
The sub-grid cloud time step in CCTM for example, has a fixed lifetime that may span the
synchronization time interval.
In the CCTM the synchronization time step is essentially the time interval over which the
chemistry processes are considered to be time-split and independent of the other processes.
The proper relationship between the first three time steps listed above is determined by the
"ADVSTEP" algorithm. Based on a user-supplied output time step interval, and optionally an
upper and lower limit for the synchronization time step interval,1 the algorithm:
• Ensures the synchronization time interval evenly divides the output time interval.
» Ensures the synchronization interval to not be greater than the upper limit.
• Determines a horizontal advection time interval that ensures all horizontal advection
calculations satisfy a Courant condition with respect to the horizontal winds for each
output time step. See Chapter 7 for details about the Courant condition requirement.
llf these limits are not specified, the algorithm uses defaults of 900 and 300 seconds,
respectively.
18-7
-------
EPA/600/R-99/Q30
• Attempts to establish a horizontal advection time interval as close to the synchronization
\ m ,'• jn{ervgji as possible, but evenly divides the synchronization interval.
• In case the Courant condition restriction forces an advection time interval to be less than
the lower synchronization interval limit, sets advection time intervals to be as close to the
lower limit as possible but still evenly divide2 the lower limit.
« If none of these criteria can be satisfied, reports the issue and aborts the execution.
18.3 Input/Output Applications Programming Interface
A model needsjiccess to external data. Data access, including output to external files, is done at
the module level. The Models-3 I/O API [4] provides a standardized interface to external data
that enables a user to follow the object oriented design concept of encapsulation. For example,
all basic and some derived meteorological data are calculated outside the CCTM (see MCIP
documentation in Chapter 12, e.g.) and available to a module through calls to the Models-3 I/O
APJ library [4]. Thus the user can avoid the re-calculation of various meteorological variables
within subroutjnesjising different algorithms and parameterizations to re-compute essentially the
same quantities, which may give rise to errors and modeling inconsistencies. It is expected that
the modules treat these I/O API data completely independently, not knowing or caring if files
have been opened or read by any other module. In particular, any subroutine within a module
can open an I/O API file. It is recommended that at least one routine within a module, preferably
the class-driver, should open any I/O API files that are required in the module.
In general, the I/O API provides the Models-3 user with a library containing both Fortran and C
routines, which manage all the necessary file manipulations for data storage and access. The
main requirement is that the user follows the I/O API conventions for data structures, naming
conventions, and representations of scenario date and time. The I/O API routines manage access
to the files in such a manner that data can be read or written flexibly, virtually in any order, and
freeing the user from being concerned with low level file manipulation details, such as the order
or format of file variables.
In addition, the I/O API routines can be used for both file storage and cross-media model
coupling using Parallel Virtual Machine (PVM) mailboxes.
For file storage and access, the I/O API, which is the standard data access library for Models-3,
is built on top of NCAR's netCDF files [5]. NetCDF (network Common Data Form) is a library
that provides an interface for array-oriented data access. The netCDF library also defines a
machine-independent format for representing scientific data. Together, the interface, library, and
format support the creation, access, and sharing of scientific data. The netCDF software was
developed at the Unidata Program Center in Boulder, Colorado. The freely available source can
be obtained by anonymous FTP [5],
The I/O API provides a variety of data structure types for organizing the data, and a set of access
routines that offer selective direct access to the data in terms meaningful to the user.
Strictly speaking, this means divides without remainder, not by a factor of two.
18-8
-------
EPA/6DO/R-9W030
Since they are netCDF files, I/O API files share the following characteristics:
1. They are machine-independent and network-transparent. Files created on a Cray
Supercomputer can be read on a desktop workstation (or vice versa) either via NFS
mounting or FTP, with no data format translation necessary;
2. They are self-describing. That is, they contain headers that provide a complete set of
information necessary to use and interpret the data they contain;
3. They are direct access. A small subset of a large dataset may be accessed efficiently
without first reading through all the preceding data. This feature is mandatory in
visualization requiring rapid access of large datasets;
4. They are appendable. Data can be appended to a netCDF dataset along one dimension
without copying the dataset or redefining its structure. For example, the I/O API can add
more time step data to a previously created file; and
5. They are sharable. One writer and multiple readers may simultaneously access the same
netCDF file. This means, for example, that visualization tools can access the file and
display data while the file is being created by the model.
The I/O API has been designed to support a variety of data types used in the environmental
sciences, among them:
• Gridded data (e.g., concentration or meteorological fields);
• Grid-boundary data (for model boundary conditions);
• Scattered data (e.g., meteorology observations or source level emissions); and
• Sparse matrix data (a specialized data type used in an emissions model).
I/O API files support three different time step structures:
1. Time-stepped with regular time steps (e.g. hourly model output concentration fields or
twice-daily upper air meteorology observation profiles);
2. Time-independent (e.g. terrain height); and
3. Restart, which always maintains the last two time step records of output from a running
process as "even" and "odd." The "odd" is available for restart in case of a crash while
the "even" step is being written, and vice versa. Restart data does not consume an
inordinate amount of disk space with only two time steps worth of data at all times.
The I/O API provides automated built-in mechanisms to support production and application
requirements for dataset histories and audit trails:
• Identifiers of the program execution that produced the file.
18-9
-------
- EPAA300/R-99/030
* • Description of the study scenario in which the file was generated.
The I/O API also contains an extensive set of utility routines for manipulating dates and times,
performing coordinate conversions, storing and recalling grid definitions, sparse matrix
arithmetic, etc. There are a variety of related programs that perform various analysis or data-
manipulation tasks, including statistical analysis, file comparison, and data extraction. Section
18.8.3 below illustrates the use of some of these utilities in coding practice, and the web site [5]
provides background materials, a user manual, and a tutorial.
18.4 Code Configuration Management
» - mr* . • im, •**• TW m. -t LJ .-*- • • ' - - j. • "- •. ' • • • > .!: * n . •-•;-•> i :• -i '
18.4.1 The Need
~ ,, Faced with a large and growing community that uses and develops a wide variety of programs,
i,, ti, modules and codes, it is imperative to systematically manage the cross-community access to this
software. Typically, successful management of software involves:
• A repository - a place where all of the public code resides;
., « The concept of archived code - codes that have been deposited into the repository in such
m '£ manneTthat anyone can extract the exact code at a later time. This involves some kind
of transformation program to maintain master copies of the codes with embedded change
tables;
• The concept of revision control - archiving codes results in modifying the tags or unique
revision identifiers in the change tables in the master copies in order to recover the exact
code at a later date; and
• The concept of released code - codes that have reached some state of maturity and have
i , v been designated with some kind of "released" status. They can be used with reasonable
^ k>* expectation of reliability.
The paradigm used employs the following scenario.
1. A user modifies or develops code. The code may be one subroutine or many, possibly
comprising whole science models. The code may originate from "scratch," or be
„ _ extractedjrgm |he repository and modified.
2. After testing or reaching a point of being satisfied with his/her results, he/she decides to
save it in the repository so that others can have access to it.
3. Some archived codes may still be in an experimental, or development, state while others
may be reasonably stable and more completely tested. The latter may be designated as
"released." There is no enforceable means to control access based on an experimental or
released state. The community will have and should have access indiscriminately, well
aware that using development state code is risky.
'.(„•• ""%*.* '*?""; jf*""* * L . X ,' *5 ?• . l , I ' t "^S,^ "M *J- '
f i * * M r? It. -J'J: 'K''.*•.' i. i lit, II f.i.
18-10
-------
EPA/600/R-99/030
4. As the user continues to work with the codes, he/she may make enhancements or
discover and fix errors. The upgrades are then installed in the repository, which
automatically assigns unique revision identifiers.
5. The repository is located where it is conveniently accessible to all pertinent users, and it
is maintained by an administrator who sets and enforces general access rules.
18.4.2 The Tool
There are many configuration management tools both free and commercially available. We
chose The Concurrent Versions System (CVS) [1] mainly because of its versatility. CVS
controls the concurrent editing of sources by several users working on releases built from a
hierarchical set of directories. CVS uses the Revision Control System (RCS) [2] as the base
system. Other reasons that CVS was an attractive choice include:
• It works on virtually all UNIX platforms and many PCs;
• It is publicly available and free;
• CVS is a state-of-the-art system, constantly being improved; and
• MM5 codes are managed by CVS, and MM5 is a primary meteorology model for the
CMAQ system.
From the UNIX man pages (online manual):
CVS is a front end to the Revision Control System (RCS) that extends the notion
of revision control from a collection of files in a single directory to a hierarchical
collection of directories consisting of revision controlled files. These directories
and files can be combined together to form a software release. CVS provides the
functions necessary to manage these software releases and to control the
concurrent editing of source files among multiple software developers [emphasis
added].
The Revision Control System (RCS) manages multiple revisions of files. RCS
automates the storing, retrieval, logging, identification, and merging of revisions.
Another widely used source code management system is the Source Code Control System
(SCCS), which is usually distributed with UNIX systems. The main note is that RCS and SCCS
act on files only, while CVS operates on projects. Working with entire projects works better and
easier with CVS. Multi-developer project development works better with CVS as well. CVS
supports a client/server mode of operation that can be very useful. CVS also allows
customization by adding hooks so that local scripts or programs can be called when executing
various CVS commands. This can be useful to force naming conventions of tags, or reference to
bug-tracking software, etc. Thus CVS adds power and features that are attractive for the
ModelsS-3/CMAQ system.
18-11
-------
EPA/600/R-99/530
18.4.3 The Repository
The repository structure, that is, the UNIX directory hierarchy, follows the class/module
organization discussed above in Section 18.2. The repository is actually divided into many
repositories, one for each generic model. This division makes it easier to maintain the
class/module organization that is important for the model building operation described below in
Section 18.5.
CVS allows for the use of a "modules" file3, which enables a user to easily check out or extract a
complete QJ^L^Q module. For example, a module might be checked out by a user to make code
modifications, and complete modules are checked out during the model building operation.
.„, •ill iiiii: ii: ;,. in ,' ,innnnniiii 'iiiiiiiiinniii. iniiip jinn i ,«« - mi ::i i< . j ;„„.- .jinr1 :. in ii'fj n f jiinn nnwinni /w ? .'iiiiiiii: "f " ,"• i nil I11: if y,. % ""., i. i
The following shows a symbolic CVS UNIX directory tree that represents the current structure
for the CCTM:
II III III fe! :"; ' :•;;; :., III. :'f • ' :l!-,ir '.•.>••• V..,'» f; 1, ,,i,,,r f:1',, \, ;'; '- "• , I | "Jil
3The terminology is unfortunate. The CVS modules file has no intrinsic relationship with the
CMAQ classes/module design implementation.
18-12
-------
EPA/600/R-99/030
...CCTM -+-> CVSROOT --- +-> CVS administrative files
\-> src
+-> drivec — •*•
> ctm
-> RCS files
+-> hadv --- +•
| -f-
| +
+ -> vadv
> hadv_noop • --- +•-> RCS files
> hbot -------- + -> RCS files
> hppm -------- + -> RCS files
> vadv_noop --- +-> RCS files
> vbot -------- + -> RCS files
> ypprn -------- + -> RCS files
> vdiff_noop -- + -> RCS files
> eddy -------- + -> RCS files
> chem_noop --- + -> RCS files
I + --- > smvgear ----- + -> RCS files
I + --- > qssa -------- +-> RCS files
I
+ -> phot --- +• --- > phot_noop --- + -> RCS files
I + --- > phot -------- + -> RCS files
+-> vdiff — +
| +
I
+-> chem --- +
+•-> aero --- + ---
|
I
+-> couple -+ --- > gencoor
1
+-> cloud --+
> aero_noop --- 1-> RCS files
> aero --- • ----- +-> RCS files
+-> RCS files
> cloud_noop — +-> RCS files
> cloudradm — +-> RCS files
+-> RCS files
+-> procan -+ --- > pa
!
+-> hdiff --+ --- > hdiff_noop — +-> RCS files
I + --- > unif -------- + -> RCS files
1
+-> init --- + --- > init -------- •*•-> RCS files
I
+ -> util --- + --- > util -------- + -> RCS files
I
+-> aero_depv +•-> aero_depv_noop +-> RCS files
| +-> aero_depv -+-> RCS files
I
+-> adjcon -+— — > adjcon_noop --»•-> RCS files
I \ --- > denrate ----- +-> RCS files
I
\-> ping --- + --- > ping_noop --- +-> RCS files
+ --- > ping_qssa --- •»•-> RCS files
\ --- > ping_sravgear +-> RCS files
The symbolic tree is shown relative to the subdirectory in the repository named for the CCTM
model. Similar trees exist for each of the generic models. The RCS files are the revision control
history files that contain the change tables to reconstruct the actual source code according to a
specific revision identifier. Also note that the tree closely follows the organization of classes and
modules for the CCTM described in Table 18-1, and contains alternate modules within the
classes. In particular, most classes contain a "no-operation" (_noop) module that allows a user to
essentially turn off that particular science process modeling. This is useful, for example in
debugging, where rapid turn-around is important, and a computationally demanding module that
is not needed can be bypassed.
18-13
-------
EPA/600/R-99/030
18.5 How a Model is Constructed
18.5.1 Object Oriented Concepts
To make the Models-3/CMAQ system robust and flexible, object oriented concepts were
incorporated into the design of the CMAQ system. Incorporating these ideas into the design
helps avoid introducing errors when code modifications are necessary. Additionally, the system
is capable of easy and efficient modification, allowing the user to quickly make models for
different applications.
The implementation language for CMAQ is Fortran 77, which imposes limits on how far one can
go in terms of object oriented design. In particular, since Fortran is a static language, objects
cannot be instantiated dynamically; they must be declared explicitly in the source code to be
created at compile time. However, to encourage a user community that will be contributing code
for future enhancements, every attempt has been made to adhere to the Fortran 77 standard. In
the future, the use of other implementation languages such as Fortran 90 will be considered.
18.5.2 Global Name Table Data
In order to implement modularity and data-independence, we have employed design ideas that
draw heavily from the object-oriented concept of inheritance and code re-use. The data
structures in the codes that deal with the domain sizes, chemical mechanism, I/O API, logical file
names, general constants, and CGRJD pointers are determined by Fortran declarations in data
and parameter statements that are created through the Models-3 system. These data structures
pertain to a particular application (domain, mechanism, etc.) and are meant to apply globally, not
only to one particular CCTM through all its subroutines, but also to all the models that supply
data to the CCTM for that application. These data structures are contained in Fortran INCLUDE
files, which are essentially header files, included in the declaration sections near the top of the
Fortran code source files. The inclusion of these source files is made automatic by using a
generic string that represents the include file and which is parsed and expanded to the actual
include file during a pre-processing stage in the compilation. The Fortran global include files
contain name tables that define:
1. The computational grid domain;
2. The chemical mechanism;
3. The I/O API interface, including logical file names;
4. The global modeling constants; and
5, Other constants or parameters that apply across the model.
In order to effect the implementation of the include files into the code, a special compiling
system, riiSbld, has been developed [3], which reads a configuration file that, based on the
application, completely determines the model executable to be built. The ASCII configuration
file can be generated either by the Models-3 system or by the user following a few, simple
syntactical rules illustrated below. For additional details, the reader is referred to Chapter 15.
18-14
-------
EPA/600/R-99/030
In addition to the global include files, the configuration file contains module commands that tell
mBbld to extract the codes for that module from the model code repository for compilation,
18.5.3 Build Template
The following exhibit is an example of a configuration file that m3bld would use to build a .
model executable: (The numerals at the left-hand margin are labels for the legend and are not
part of the configuration file. The "//" represents a non-parsed, comment line.)
Example Configuration File
U)
(2)
(2)
(2)
(3)
model TUTQ
cpp_flags
£ 77__flags
link flags
libraries
_r4y;
" -Demis vdif
" -e -fast -O4
" -e -fast -O4
« .
/
-xtarget=ultra2 -xcache=16/32/l : 1024/64/1" ;
-xtarget=ultra2" ;
" -i,/home/models3/tools/IOAPI/release/m3io/lib/SunOS5 -ImSio
-L/home/models3/tools/netCDF/SunOS5 -Inetcdf" ;
(4)
(5)
(5)
(5)
(5)
(5)
(5)
(5}
(5)
(5)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6}
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
(6)
global verbose;
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
include
SUBST_BLKPRM
SUBST_CONST
SUBST_FILES_ID
SUBST_EMPR_VD
SUBST_EMPR_CH
SUBST_IOPARMS
SUBST_IOFDESC
SUBST_IODECL
SUBST_XSTAT
SUBST_COORD_ID
SUBST_HGRD_ID
SUBST_VGRD_ID
SUBST_RXCMMN
SUBST_RXDATA
SUBST_GC_SPC
SUBST_GC_EMIS
SUBST_GC_ICBC
SUBST_GC_DIFF
SUBST_GC_DDEP
SUBST_GC_DEPV
SOBST_GC_ADV
SUBST_GC_COHC
SUBST_GC_G2AE
SUBST_GC_G2AQ
SUBST_GC_SCAV
SUBST_GC_WDEP
SUBST_AE_SPC
SUBST_AE_EMIS
SUBST_AE_ICBC
SUBST_AE_DIFP
SUBST_AE_DDEP
/work/rep/include/release/BLKPRM_500 . EXT;
/work/rep/include/release/CONST3__RADM.EXT;
/work/rep/include/release/FILBS_CTM.EXT;
/work/rep/include/release/EMISPRM. vdif . EXT;
/work/rep/include/release/EMISPRM.ehem.EXT;
/work/rep/include/release/PARMS3 .EXT;
/work/rep/include/release/FDESC3 .EXT;
/work/rep/include/release/IODECL3 .EXT;
/work/rep/include/release/XSTAT3 .EXT;
/work/yo j /tgt/BLD/COORD . EXT; f
/work/yoj /tgt/BLD/HGRD. EXT;
/work/yoj /tgt/BLD/VGRD . EXT;
/work/yoj /tgt /BLD/RXCM . EXT ;
/work/yoj /tgt/BLD/RXDT.EXT;
/work/yoj /tgt/BLD/GC_SPC. EXT;
/work/yo j/tgt/BLD/GC_EMIS .EXT, •
/work/yoj /tgt/BLD/GC_ICBC . EXT ;
/work/yo j/tgt/BLD/GC_DIFF. EXT;
/work/yoj /tgt/BLD/GC_DDEP . EXT;
/work/yoj /tgt /BLD/GC_DEPV . EXT ;
/work/yo j/tgt/BLD/GC_ADV. EXT;
/work/yoj /tgt/BLD/GC_CONC . EXT;
/work/yoj /tgt /BLD/GC_G2AE . EXT;
/work/yo j /tgt/BLD/GC_G2AQ . EXT;
/work/yo j/tgt/BLD/GC_SCAV. EXT;
/work/yo j /tgt /BLD/GC_WDEP . EXT ;
/work/yoj /tgt /BLD/AE_SPC. EXT;
/work/yoj /tgt/BLD/AE_EM!S . EXT;
/work/yoj /tgt/BLD/AE_ICBC . EXT;
/work/yoj /tgt/BLD/AE_DIFF . EXT;
/work/yoj/tgt/BLD/AE_DDEP.EXT;
18-15
-------
EPA/6QO/R-99/030
Example Configuration File
(6) include SUBST_AE_DEPV
(6) include SUBST_AE_ADV
<6) include SUBST_AE_CONC
(6) include SUBST_AE_A2AQ
(6) include SUBST_AE_SCAV
(6) include SUBST_AE_WDEP
{6} include SUBST_NR_SPC
<6) include SUBST_NR_EMIS
(6) include SUBST_NR_ICBC
(6) include SUBST_NR_DIFF
(6) include SUBST_NR_DDEP
(6) include SUBST_NR_DEPV
(6) include SUBST_NR_ADV
(6) include SUBST_NR_N2AE
(6) include SUBST_NR_N2AQ
(6) include SUBST_NR_SCAV
(fi) include SUBST_NR_WDEP
C6) include SUBST_TR_SPC
(6) include SUBST_TR_EMIS
(6) include SUBST_TR_ICBC
(6) include SUBST_TR_DIFF
(6) include SUBST_TR_DDEP
(6) include" SUBST_TR_DEPV
(6) include SUBST_TR_ADV
(6) include SOBST_TR_T2AQ
(6) include SUBST_TR_SCAV
(6) include SUBST_TR_WDEP
/work/yo j /tgt/BLD/AE_DEPV . EXT ;
/work/yo j / tgt/BLD/AE_ADV . EXT ;
/work/yo j /tgt/BLD/AE_CONC . EXT ;
/work/yo j /tgt/BLD/AE_A2AQ . EXT ;
/work/yo j/tgt/BLD/AE_SCAV. EXT;
/work/yo j /tgt/BLD/AE_WDEP . EXT;
/work/yo j /tgt/BLD/NR_SPC . EXT;
/work/yoj /tgt/BLD/NR_EMIS , EXT;
/work/yo j /tgt/BLD/NR_ICBC . EXT ;
/work/yo j /tgt/BLD/NR_DIFF . EXT;
/work/yoj /tgt/BLD/NR_DDEP . EXT;
/work/yoj /tgt/BLD/NR_DEPV. EXT;
/work/yoj / tgt /BJDD/NR_ADV . EXT ;
/work/yoj /tgt/BLD/NR_N2AE . EXT ;
/work/yoj /tgt/BLD/NR_N2AQ . EXT;
/ work/yo j/tgt /BLD/NR_SCAV. EXT;
/work/yo j /tgt/BLD/NR_WDEP . EXT;
/work/yoj /tgt/BLD/TR_SPC . EXT;
/work/yoj /tgt/BJDD/TR_EMIS . EXT ;
/work/yo j/tgt/BLD/TR_ICBC. EXT, •
/work/yo j /tgt/BLD/TR_DIFF . EXT;
/work/yoj /tgt/BLD/TR_DDEP . EXT;
/work/yoj /tgt/BLD/TR_DEPV. EXT;
/work/yoj /tgt/BLD/TR_ADV, EXT;
/work/yoj /tgt/BLD/TR_T2AQ . EXT;
/work/yo j/tgt/BLD/TR_SCAV. EXT;
/work/yo j /tgt/BLD/TR_WDEP . EXT ;
// Process Analysis / Integrated Reaction Rates processing
<6) include SUBST_PACTL_ID
(6) include SUBST_PACMN_ID
(6) include SUBST_PADAT_ID
(7) module ctm release ;
(7) module init release ;
/work/yoj /tgt/BLD/PA_CTL . EXT;
/work/yoj /tgt/BLD/PA_CMN . EXT;
/work/yoj /tgt/BLD/PA_DAT . EXT;
// options are denrate and adjcon_noop
(7) module denrate release ;
(7) module gencoor release ;
// options are hbot and hadv_noop
(7) module hppm release ;
// options are vbot and vadv_noop
{7) module vppm release ;
// options are phot and phot_noop
(7) module phot release ;
// options are ping_qssa,
ping_smvgear and ping_noop
18-16
-------
EPA/60G/R-99/030
Example Configuration File
(7) module ping_qssa release ;
// options are qssa, smvgear and chem_noop
(7) module qssa release ;
// aerosols
(7) module aero release ;
// aerosol dep vel
(7) module aero_depv release ;
// options are eddy and vdi££_noop
(7) module eddy release ;
// options are const and hdiff_noop
(7) module unif release ;
// options are cloud_radm and cloud_noop
(7) module cloud_radm release ;
// options are pa and pa_noop, which requires the replacment of the three
// global include files with their pa_noop counterparts
(7) module pa release ;
(7) module util release ;
Legend:
(1) - Model name definition. This string is what the model executable will
be named.
(2) - The C pre-processor, Fortran compiler and link flags. The flags
specified indicate that the model is to be compiled with the option to input
emissions in the vertical diffusion processing, and to compile and link on a
Sun Spare, Ultra-30 workstation.
(3) - I/O API and netCDP object libraries to be linked. The format is
virtually identical to that of a UNIX make command.
(4) - mSbld flag: one of many options. verbose indicates report all actions.
Other options are:
• compile_all - force compile, even if object files are current
* clean_up - remove all source files upon successful completion
» no_compile - do everything except compile
• no_link - do everything except link
» one_step - compile and link in one step
« parse_only - checks config file syntax
• show_only - show requested commands but doesn't execute them
(5), (6) - global include files. The syntax is:
include internal-s trinq full-path-name.
Every routine that requires a specific include file must contain a Fortran
include statement using the internal-string- (see Section 18.3 for examples) .
(5) - " fixed" global include files. The fixed include files have been
constructed outside of the Models-3 framework and contain global data not
directly related to the domain or chemical mechanism such as modeling
18-17
-------
EPA/6QQ,'R-99/030
Example Configuration File
constants and file logical names,
(6) - The Models-3 framework automatically generates all the include files
labeled (6). The user supplies information through the Models-3 Science
Manager that determines the data in these include files. These data define
the complete problem domain for a particular CMAQ application. See Chapter
15 for a complete discussion.
{7) - The module name to extract from the model code repository and an
optional revision flag to select a particular module version. The syntax is;
module module name revision flag.
For additional information on this implementation, the reader is referred to the Model Building
Tool described in Fine et al. [3].
18.6 How a Model is Executed
In order to run a model executable, various UNIX environment variables must be set in the shell
that invokes the execute command. Generally, these involve the modeling scenario start date and
time, the run duration, the output time step interval, various internal code flags that differ among
the models, and all the input and output logical, or symbolic file names (see Section 15.4.3 in
Chapter 15). There are various ways by which external file names can be referenced in the code,
but the most general across all UNIX platforms is to link them by means of environment
variables. There are I/O API utility functions that allow a user easy access to these variables
within the code, making such accesses generic and portable.
An_qdditiona|Jeaturejhat is provided through the I/O API is to declare a file "volatile" by
a -v flag in the shell's declaration for the environment variable1 By doing this, the I/O
%vi!l cause the netCDF file to update (sync) its disk copy after every write and thereby
updUte the netCDF header. Otherwise, netCDF file headers are not updated until the files are
closed. This is useful, for example, to allow a user to analyze an open netCDF file using
visualization tools while the model is executing. It is also useful in case of a system crash. A
CCTM model can be restarted at the scenario time step after the last successful write using the
aborted output file as the input initial data.
18.7 Using the Models-3 Framework
The Models-3 framework simplifies the model building and model execution tasks, especially if
key objects have been pre-defined. Thus if a user is conducting a study that uses the same
domain or chemical mechanism, but uses different code options, e.g., then rebuilding and re-
exeeuting using the framework makes conducting the study very straightforward. Other
examples where the framework greatly simplifies operational tasks would be where a model is to
be applied on different domains, or for control strategies in which the only thing that changes are
the emissions input files. A knowledgeable user is still able to build and execute model
applications outside the framework. In fact, one can easily run model applications using
executables and UNIX scripts generated by the framework.
18-18
-------
EPA/600/R-99/030
In general, in order to build a model within the framework, the coordinate system, horizontal grid
and vertical layer definitions, and chemical mechanism must be specified. This is done within
the Models-3 Science Manager component. If objects required for a particular application, such
as a coordinate system specification, are already defined, then the user needs only to select those
objects. The Configuration File Manager, an additional subcomponent within the Science
Manager, allows a user to control the module definitions and fixed 4 include files that go into the
configuration file used to build the model. Also the user can specify compiling and linking flags
related to a specific platform on which the model is to be executed.
With all required Models-3 objects defined, the task of building a model is made simple using
the Models-3 component, Model Builder. It's important to note that as users develop models, the
set of objects will grow to become a rich pool from which to select for a particular application.
The Models-3 framework was designed to allow extensive re-use of objects. Model Builder
generates all the "non-fixed" include files, then builds a model, linking in the include files
including the user-specified fixed ones. See the section above for the include file definitions.
Perhaps one of the most useful features of the Models-3 framework is die ability to graphically
set up and execute single or multiple model executions with automatic data registration. This
allows a user to carefully control the order of model executions and the management of all the
input and output data. The Models-3 component that performs this function is the Study Planner.
Study Planner allows a user to specify input data sets, executables, and UNIX environment
variables, and to initiate (possibly multiple) model runs.
18.8 Conformant Code
18.8.1 Thin Interface
As mentioned above in section 18.5.1 , the Models-3/CMAQ system was designed to be robust
and flexible with respect to the interchange of modules and the elimination of cross-module data
dependencies. Consequently, the concept of a "thin interface" has been employed in the design,
which applies principally to the class-drivers. At the least, the thin interface implementation
implies the following requirements:
• Eliminate global memory references (across modules). This implies no common blocks
across modules, no hidden data paths, no "back doors";
• Each module reads and interpolates its required data independently. The I/O API helps to
ensure this kind of data independence; and
• Standardized argument list (CGRJD, Date, Time, TimeStep) for the class-driver, as
described in the section above.
These requirements attempt to incorporate the object-oriented idea of encapsulation in the
Models-3/CMAQ design. The following quotation is from Rumbaugh et al. [7]:
Encapsulation (also information hiding) consists of separating the external aspects
of an object, which are accessible to other objects, from the internal
4not generated by the framework
18-19
-------
EPA/600/R-99/030
implementation details of the object, which are hidden from other objects.
Encapsulation prevents a program from becoming so interdependent that a small
change has massive ripple effects. The implementation of an object can be
changed without affecting the applications that use it [emphasis added].
The encapsulation design makes the CMAQ system safer and enables the transaction processing,
plug&play capability. This design also makes it easier for a user to trace data and usage within a
module, particularly at the class-driver level.
18.8.2 Coding Rules
In order to maintain the object oriented concepts implemented in the CM AQ system design, we
have established a small set of coding rules that apply for those that develop CMAQ science and
affect the low-level design of the models. We have developed standards to control data
dependencies at the class-driver level, but we have not propagated these coding standards to the
sub-module level.
1. The models are generally coded in Fortran (Fortran-77 conventions are used). It is
possible to link in subroutines written in the C language, although this has not been done
within the current CMAQ implementation.
2. The modules must be controlled by a top-level class-driver routine, whose calling
arguments must be the computational concentration grid array, CGRID, the current
scenario data and time, and the controlling time step vector. See the section above.
3. The class-driver is also responsible for any temporal integration required within the
module. (The time steps for process integration at the module level are usually shorter
than those of the CCTM synchronization time step.)
4. Any reads and writes for the module should be done at the level of the class-driver
routine. Although not absolutely necessary, this is strongly suggested because it is
usually much easier to control the timing of the data accesses at the highest level of the
module where the current scenario date and time are known.
5. Use the IMPLICIT NONE Fortran declaration to maintain some control on typographic
errors and undefined variables. Although not standard Fortran-77, the use of IMPLICIT
NONE forces the developer to declare all internal variables.
6. Use the global include files for domain definitions, chemical mechanism data, and other
data where available.
7. Use the I/O API for external data references where appropriate. For an illustration of
these rules, the reader is referred to the code template below.
At the sub-module level there are no strict I/O or coding standards. Here it is envisioned that
individual researchers/programmers use their own coding styles for their algorithms. However
the following suggestions are offered to facilitate the potential incorporation of a module into the
CMAQ system:
18-20
-------
EPA/600/R-99/030
• It is expected that MKS units are used for input and output variables, as these units have
been standardized throughout the CMAQ system. Within a sub-module subroutine
whatever units are most convenient can be used. However, the developer must be
responsible for any unit conversions to MKS for input and output, and thus avoid any
potential errors.
• For efficiency and performance considerations, operations may need to be done on
groups of grid cells (a block of cells) at a time. If there are N cells in the block and the
entire domain contains M cells, then the entire domain can be decomposed into M/N
blocks. We have used #=500, For operations in the horizontal (x,y), the cell constraint
becomes Xx. Y< N, where X= number of cells in the x-direetion, and Y= number of cells
in the y-direction. For operations in both the horizontal and vertical, the constraint
becomes Xx 7x Z< N, where Z= number of cells in the z-direction. There may be some
operations, such as for some horizontal advection schemes, where this decomposition
into blocks becomes more difficult or impossible.
18.8.3 Science Process Code Template
The following demonstrates what a science process class-driver Fortran 77 subroutine might look
like. We recommend that a code developer follows this template, where appropriate, to get
maximum benefit from the design concepts implemented in the Models-3/CMAQ system. This
template is generic and attempts to show most, if not all the features available. Some class-
drivers and most other sub-programs within a module may not have, nor require, most or any of
these features. (The numerals at the left-hand margin are for the legend and are not part of the
text, and the text within "< >" indicates code not included.)
Example of Science Process Class-driver
( 1) SUBROUTINE VDIFF SUBST_GRID_ID { CGRID, JDATE, JTIME, TSTEP !
/ 2) f~™™—™_m«™».»,™«. — «.««. — •• — — — -__-. — « — _ — »—».«.
S 2S C Function:
( 2)
( 2! C Preconditions:
{ 2}
(2) C Subroutines and Functions Called:
( 2)
( 2} C Revision History:
( 2} C
( 3} IMPLICIT NONE
( 45
C 4)
( 5)
< 5}
( 5)
{ 5)
{ 5)
( 5)
( 5)
( 5)
( 5)
( 5)
t 5}
( 5)
INCLUDE SOBST_HGRD_ID
INCLUDE SUBST_VGRD_ID
INCLUDE SUBST_GC_SPC
INCLUDE SOBST_GC_EMIS
INCLUDE SUBST_GC_DEPV
INCLUDE SUBST_GC_DDEP
INCLUDE S0BST~GC_DIFF
INCLUDE SUBST_AE_SPC
INCLUDE SUBST_AE_DEPV
INCLUDE SUBST_AE_DDEP
INCLUDE SUBST_AE_DIFF
INCLUDE SUBST_NR_SPC
INCLUDE SOBST_NR_EMIS
INCLODE SUBST NR DEPV
! horizontal dimensioning parameters
! vertical dimensioning parameters
! gas chemistry species table
! gas chem emis surrogate names and map table
I gas chem dep vel surrogate names and map table
I gas chem dry dep species and map table
I gas chem diffusion species and map table
! aerosol species table
! aerosol dep vel surrogate names and map table
! aerosol dry dep species and map table
! aerosol diffusion species and map table
! non-reactive species table
I non-react emis surrogate names and map Cable
1 non-react dep vel surrogate names and map table
18-21
-------
EPA/600/R-99/03Q
Example of Science Process Class-driver
( 5)
( S)
( S)
f S)
( S)
< S)
( 5)
( 6)
( 6}
( 6)
( 6}
( 6)
( 7!
( 7)
( 7)
( 7)
( 7)
( 7J
( 4)
INCLUDE SUBST_NR_DDEP
INCLUDE SUBST_NR_DIFF
INCLUDE SUBST_TR_SPC
INCLUDE SUBSTJTRJSMIS
INCLUDE SUBST_TR_DEPV
INCLUDE SUBST_TR~DDEP
INCLUDE SUBST_TR_DIFF
Hifdef etnls_vdi£
INCLUDE SUBST EMPR VO
INCLUDE SUBST EMPR CH
ftendif
INCLUDE
INCLUDE
INCLUDE
INCLUDE
INCLUDE
INCLUDE
INCLUDE
INCLUDE
SUBST_
SUBST_
SUBST"
SUBST_
SUBST"
SUBST"
SUBST"
SUBST
^PACTL_ID
CONST
,FILES_ID
tIOPARMS
IOFDESC
.IQDECIj
XSTAT
COORD_ID
! non-react dry dep species and map table
I non-react diffusion species and map table
I tracer species table
I tracer emis surrogate names and map table
! tracer dep vel surrogate names and map table
! tracer dry dep species and map table
! tracer diffusion species and map table
I emissions processing in vdif
! emissions processing in chem
! PA control parameters
! constants
! file name parameters
! I/O parameters definitions
I file header data structure
1 I/O definitions and declarations
I M3EXIT status codes
1 coordinate and domain definitions (req IQPARMS)
(8) C Arguments:
( *)
( 8)
8)
8)
(
(
( 8)
( 8)
( 8)
( a)
( 9)
( 9)
(10}
(10}
(10)
(10)
(11)
(11)
{12}
(12)
(13)
(13)
(13)
(13)
(13)
(13)
(13)
(13)
(13)
(13)
(131
(13)
(13)
(13)
(13)
REAL
INTEGER
INTEGER
INTEGER
CGRIDt NCOLS,NROWS,NLAYS,* )
concentrations
JDATE
JTIME
TSTSPC 3
I current model date, coded YYYYDDD
! current model time, coded HHMMSS
I time step vector SHHMMSS)
! TSTEP{15 = local output step
! TSTEPC2) = sciproc sync, step (chem)
I TSTEPO) « advection time step
C Parameters:
< >
C External Functions not previously declared in IODECL3.EXT:
INTEGER
EXTERNAL
SECSDIPF, SEC2TIME, TIME2SEC
SECSDIFF, SEC2TIME, TIME2SEC
(13)
(13)
(135
(14)
C Pile variables:
< >
C Local variables:
< >
IF ( FIRSTIME ) THEN
FIRSTIME * .FALSE.
LOGDSV » INIT3 ()
C Open the met files:
IF ( .NOT. OPEN3 ( MET_CRQ_3D, FSREAD3 , PNAME ) ) THEN
XMSG » 'Could not open '// MET_CRO_3D // ' file'
CALL M3EXIT( PNAME, JDATE, JTIME, XMSG, XSTAT1 )
END IF
< open other met files >
C Open Emissions files:
< do other intialization or operations that need be done only once >
END IF I if firstime
C set file interpolation to middle of time step
18-22
-------
EPA/600/R-99/030
Example of Science Process Class-driver
(14!
(14) MDATE = JDATE
(14) MTIME = JTIME
(14) MSTEP = TIME2SEC{ TSTEPi 2 } J
(14) CALL NEXTIME ( MDATE, MTIME, SEC2TIME ( MSTEP / 2 ) !
(15) C read&interpolate met data
U5)
(IBS VNAME = 'DENSA_J'
(15! IF ( .NOT. INTERP3 ! MET_CRO_3D, VNAME, PNAME,
(15) & MDATE, MTIME, NCOLS*NROWS»NLAYS,
(15) & RRHOJ ) ) THEN
(15) XMSG »•'Could not Interpolate DENSA_J from ' // MET_CRO_3D
(15) CALL M3EXIT( PNAMS, MDATE, MTIME, XMSG, XSTAT1 )
(15) END IF
(15)
(15) < do other reads >
(15)
(IS) C read&interpolate deposition velocities
(15)
(15) < do operations >
(15)
(15) C read&interpolate emissions
(151
(15) CALL RDEMIS ( MDATE, MTIME, NCOLS, NROWS, EMISLYRS, NEMIS, EMJTRAC,
(15) & VDEMIS )
(15!
(16! IF (LIPR) CALL PA_OPDATE_EMIS ( 'VDIF', VDEMIS, JDATE, JTIME, TSTEP )
< do other operations >
(17) CALL EDYINTB SUBST_GRID_ID ( EDDYV, DT, JDATE, JTIME, TSTEP( 2 ) )
(18! DO 345 R = 1, NROWS
(18! DO 344 C = 1, NCOLS
< do operations >
(19) DO 301 N = 1, NSTEPSi C,R )
< do operations >
(19) 301 CONTINUE I end time steps loop
(18! 344 CONTINUE ! end loop on col C
(18) 34S CONTINUE ! end loop on row R
(20) C If last call chis hour: write accumulated depositions:
(20)
(20) WSTEP = WSTBP + TIME2SEC! TSTEPS 2 ) )
(20! IF ( WSTEP .GE. TIME2SEC( TSTEP{ 1)1! THEN
(20) MDATE = JDATE
(20) MTIME = JTIME
(20) CALL NEXTIMEt MDATE, MTIME, TSTEP( 2 ) )
(20) WSTEP = 0
(20)
(20) IF ( .NOT. MRITE3( CTM_DRY_DEP_1,
(20) & ALLVAR3, MDATE, MTIME, DDEP ) S THEN
(20) XMSG = 'Could not write ' // CTM_DRY_DEP_1 // * file'
(20) . CALL M3EXIT( PNAME, MDATE, MTIME, XMSG, XSTAT1 )
(20! END IF
(20!
(20) WRITE{ LOGDEV, '( /SX, 3( A, :, IX ), 18, " :" , 16.6 )' )
(20) & 'Timestep written to', CTM_DRY_DEP_1,
(20) & 'for date and time', MDATE, MTIME
(20)
(16) IF (LIPR) CALL PA UPDATE DDEP ( 'VDIF', DDEP, JDATE, JTIME, TSTEP i
18-23
-------
EPA/600/R-99/030
Example of Science Process Class-driver
120)
(20) < do other operations >
(20)
(20) END IF
(21) RETURN
(21) END
Legend:
( 1) - Class-driver subroutine internal name. Note the calling argument list. It is fixed for
class-drivers. The string * SUBST__GSID_ID" is used foe developing run-time nesting applications,
which has not been fully implemented in the current version. By specifying subroutine names
associated with a particular nest and using corresponding domain include files for that
subroutine, it ia possible to set up a multi-nest application. The details have not been
di«cu»»ed because this kind of nesting has not been implemented in the current release of the
Model»-3/CMAQ system.
{ 2) - Header comments. Highly recommended for internal documentation.
( 3) - Highly recommended for Fortran. IMPLICIT NONE will catch typo's and other undeclared
variable errors at compile time.
( 4) - Domain array dimensioning and looping global variables, e.g. NCOLS, NROWS, NLAYS.
( 5) - Chemical mechanism array dimensioning and looping global variables.
( 6) - C Preprocessor flags that determine which emissions control dimensioning and looping
variables are compiled.
( 7} • Other global array dimensioning and looping global variables including those for the I/O
API, The logical variable LIPR is defined in the SUBST_PACTL_ID include file for use at lines
labeled (IS).
{ 8) - Declarations for the argument list (standardized).
{ 9) - Declarations 4U»d PARAMETER statements for local Fortran parameters.
(10) - Declarations for external functions not previously declared,
(11) - Declarations for arrays to hold external file data.
(12) - Declarations for local variables.
(13) - Code section for subroutine initialization and for any local data that need not be set at
ev«ry entry into the subroutine. Such data would require a SAVE statement in the declarations.
For example flRSTIHI is initialized to .TRUE, in the local variables section (125.
(14) - Illustrates using an I/O API function to set file interpolation time.
(IS) - Read accesses from I/O API files using the I/O API time-interpolation function.
(16) - Call to process analysis routine to obtain data for the optional integrated process rates
function.
(17) - Illustrates call to another science process within the module.
(18) - Main computational loop over the horizontal grid.
(19} - Tim« step loop over sub-synchronization time step intervals.
(205 - Illustrates writing to an I/O API file within a module.
(21) - Subroutine end
18.8.4 Robustness and Computational Efficiency
Scientists, while working in their discipline, are generally not interested in the physical situations
that don't significantly affect the phenomena they are studying and modeling. Typically, the
scientist works with box models, and uses specialized data to test hypotheses, parameterizations
and numerical schemes. The focus is to generate new results based on output from these
relatively simple models. In trying to integrate these codes into a more general grid model,
problems may arise when modeling simulations carry their formulations into regimes not
explored in initial studies, therefore not properly accounted for in the grid model. The scientist
who is developing codes that may go into a grid model should be cognizant of these issues and
should make every effort to make the codes robust.
18-24
-------
EPA/600/R-99/030
Another issue is code efficiency. In order to handle the pathological situations, codes like these
may become inefficient, spending many cycles dealing with the "off-problem" cases. A good
example is solving for the roots of a cubic polynomial whose coefficients, depending closely on
the locally modeled physics, may range over broad values. So, in addition to robustness,
attention should be paid to code performance. As the science improves in environmental
modeling, this usually translates into increased computational complexity. Although hardware
performance is continually improving, it cannot keep pace with the computational demands of
new science developments. Not typically the main interest of the science developer, software
performance is nevertheless a critical issue and must be addressed. Otherwise scientifically
sound models will not be of much general use if they take an inordinate amount of time to
execute.
18.9 Conclusion
The CMAQ system has been designed and implemented in such a way that its integration into
Models-3 allows access to the extensive functionality of the Models-3 framework. Users can
easily select important model options and can readily develop and execute models that apply to
their requirements.
We have integrated the CMAQ codes into the Models-3 system by following the basic set of
guidelines and rules presented in this chapter.
18.10 References
[1] http://interactivate.com/public/cvswebsites/cvs_toc.html, http://www.cyclic.com/cvs/, and
the UNIX man pages
[2] http://www.cs.cmu.edu/People/vaschelp/Archiving/Rcs/, and the UNIX man pages
[3] Fine, S. S., W. T. Smith, D. Hwang, T. L. Turner, 1998: Improving model development
with configuration management, IEEE Computational Science and Engineering, 5(1, Ja-Mr), 56-
65. and http://envpro.ncsc.org/pub_files/fme 1998b.html
[4] http://sage.mcnc.org/products/ioapi/
[5] ftp://ftp.unidata.ucar.edu/pub/netcdf/
[6] EPA Third-Generation Air Quality Modeling System, Models-3 Volume 9b, User Manual,
Appendices, June 1998, EPA-600/R-98/069(b)
[7] J. Rumbaugh, M. Blaha, W. Premerlani, F. Eddy, and W. Lorensen, 1991: Object-Oriented
Modeling and Design, Prentice Hall
[8] Grell, G. A., J. Dudhia and D. R. Stauffer, 1994: A description of the fifth-generation Penn
State/NCAR mesoscale model (MM5). NCAR Technical Note, NCAR/TN-398+STR, 122pp.
18-25
-------
EPA/600/R-99/030
This chapter is taken from Science Algorithms of the EPA Models-3 Community
Multiscale Air Quality (CMAQ) Modeling System, edited by D. W. Byun and J. K. S,
Ching, 1999.
4 U.S. GOVERNMENT PRINTING OFFICE: 1999 - 7SO_- 101 / 00082
18-26"
-------
United States
Environmental Protection Agency
Center for Environmental Research Information
Cincinnati, OH 45268
Official Business
Penalty for Private Use
$300
EPA/600/R-99/030
Please make all necessary phanges on the below label.
detach or copy, and return to the address in the upper
felt-hand comer.
If you do not wish to receive these reports CHECK HERE D;
detach, or copy this cover, and return to the address in the
upper left-hand comer.
------- |