1. Technical Description of the CAMx Photochemical Model
Ralph Morris
ENVIRON International Corp
Novato, CA
T. W. Tesche
Alpine Geophysics, LLC
Ft. Wright, KY
Farmington, NM
16 July 2003

2. CAMx Overview
Simulates the physical and chemical processes governing the formation and transport of ozone in the troposphere
three-dimensional, Eulerian (grid-based) model
requires specification of meteorological, emissions, land-use, and other geographic inputs
output includes hourly concentrations of ozone and precursor pollutants for each grid cell within a (three-dimensional) modeling domain

3. CAMx Overview (continued)
Mathematically simulates the following processes:
emission of ozone precursors (anthropogenic and biogenic)
advection and diffusion (transport)
photochemistry
deposition

4. Surface Removal/Deposition
CAMx Formulation
Change in Concentration
= Advection by Winds
Turbulent Diffusion
+ Ri + Si + Li
Chemical Reaction
Emissions
Surface Removal/Deposition

5. CAMx Regulatory Applications
Regional-Scale Ozone/PM Policy Applications:
OTAG
Regional NOx SIP Call
Tier II Motor Vehicle/Fuel Standards Analysis
VISTAS
WRAP
SARMAP
CPRAQS
EPA SOx/NOx Regional Transport Rule

6. CAMx Regulatory Applications
8-hr Ozone Policy and EAC Applications:
Peninsular Florida Ozone Study (PFOS)
Missouri/Kansas/EPA (MoKan)
Denver EAC
San Juan EAC
Tulsa EAC
Texas EACs

7. Comprehensive Air-quality Model with eXtensions (CAMx)
3-D Eulerian tropospheric photochemical transport model
treats emissions, chemistry, dispersion, removal of gaseous and aerosol air pollution
scales range from individual point sources (< 1 km) to regional (>1000 km)
Unifies features required of “one atmosphere” “state-of-the-science” models
new coding of several industry-accepted algorithms
computationally and memory efficient
easy to use
modular framework permits easy substitution of revised and/or alternate algorithms
publicly available (

8. CAMx (continued) Technical Features: Two-Way Grid Nesting
horizontal and vertical nesting
supports multiple levels
variable meshing factors
flexi-nesting (any amount of fine grid inputs)
Plume-in-Grid (PiG) submodel
Multiple, fast and accurate chemical mechanisms
Mass conservative and mass consistent transport scheme

9. CAMx (continued) Multiple map projections
curvi-linear latitude/longitude; Universal Transverse Mercator; Lambert Conformal (MM5); Rotated Polar Stereographic (RAMS)
Ozone Source Apportionment (OSAT) and other “Probing Tools”
tracks source region/category contributions to receptor ozone concentrations
indicates if ozone formed in NOx or VOC-limited conditions
Ability to use historical air quality model databases developed for other models
OTAG, LMOS, COAST/Houston, Atlanta, Northeast Corridor

10. CAMx Technical Components
Overview
solves continuity equation for each species
time splitting operation
each process solved individually for each grid, each time step
time step size maintains stable solution of transport on each grid
multiple transport steps per master grid step required for
nested grids
multiple chemistry steps per transport step required
model developed to run on meteorological modeling grid
reduces error due to interpolation and averaging
multiple map projections available

11. CAMx Technical Components (continued)
Transport
advection and diffusion solvers are mass conservative
horizontal and vertical advection linked through the divergent compressible atmospheric continuity equation
mass consistency
order of east-west and north-south advection alternates each master grid step
three options available for horizontal advection solvers:
Smolarkiewicz (1983): diffusion-corrective forward-upstream scheme
Bott (1989): area-preserving flux-form solver (less diffusive, more accurate)
Piece-wise Parabolic Method (PPM)

12. CAMx Technical Components (continued)
Transport
vertical transport solved with a semi-implicit Crank-Nicholson scheme, accounting for:
resolved vertical velocity
mass exchange across variable vertical layer structure
vertical diffusion solved with an implicit scheme
dry deposition rates are used as the surface boundary condition
horizontal diffusion solved with an explicit scheme in two directions simultaneously

13. CAMx Technical Components (continued)
Pollutant Removal
dry deposition velocities for each species determined using resistance approach (Wesely, 1989)
dependent upon: season, land cover, solar flux, near-surface stability, surface wetness, species solubility and diffusivity
for aerosols: size spectrum dictates sedimentation velocity
wet scavenging based on Maul (1980)
exponential decay
decay rate dependent upon: rainfall rate, species solubility
species removed from entire grid column (all layers)

14. CAMx Technical Components (continued)
Photochemistry
CBM-IV (Gery et al., 1989)
3 variations available
SAPRC97 (Carter, 1990) and SAPRC99 (Carter, 1999)
chemically up-to-date
tested extensively against environmental chamber data
uses a different approach for VOC lumping
all mechanisms are balanced for nitrogen conservation
photolysis rates derived from TUV preprocessor
generates lookup table over: zenith angle, altitude, ozone column, albedo, turbidity
first two determined for each grid cell internally
last three provided by input files

15. CAMx Technical Components (continued)
photolysis rates affected by clouds
UAM-V approach: rates scaled by fractional cloud coverage only
RADM approach: rates scaled by optical depth and cloud coverage

16. CAMx Technical Components (continued)
Chemistry Solver
most “expensive” component of photochemical grid simulations
CAMx solver increases efficiency and flexibility
adaptive hybrid approach:
radicals (fastest reacting species) solved using implicitly steady state approximation
fast state species solved using second-order Runge-Kutta method
slow state species solved explicitly
“Adaptive” = number of fast state species changes according to the chemical regime

17. CAMx Technical Components (continued)
Chemical Mechanism Compiler (CMC):
pre-processing program that generates solver FORTRAN source code
allows quick, error-free coding of updates or new mechanisms
allows for several mechanisms to be available in a single model compilation

18. CAMx Technical Components (continued)
Aerosol Chemistry (PMCAMx version)
gas-phase mechanism (based on CBM-IV mechanism 2)
additional biogenic olefin and condensable organic carbon species
homogeneous transformation of SO2 to sulfate
production of nitric acid
production of condensable organic carbon
aerosol package calculates the following transformations:
aqueous transformation of SO2 to sulfate
condensable organic carbon to secondary organic aerosol
sodium nitrate formation
gaseous nitric acid to aerosol nitrate
gaseous ammonia to aerosol ammonium
NO3/SO4/NH4 equilibrium as a function of NH4, temperature, humidity

19. CAMx Technical Components (continued)
Plume-in-Grid (PiG)
fine resolution needed for chemistry/dispersion of large NOx plumes
tracks stream of plume segments (puffs) in a Lagrangian frame
each puff moved by winds in host cell
puff growth (dispersion) determined by diffusion coefficients in host cell
GREASD PiG: faster, conceptually simpler
reduced NOx chemistry set (NO-NO, NOx/ozone equilibrium, HNO3 production)
puffs leak mass according to growth rates and grid cell size
puffs terminated due to age or sufficiently dilute NOx
IRON PiG: slower full photochemistry (not released yet)
Incremental Reactions for Organic and NOx (IRON)

20. CAMx Technical Components (continued)
Ozone Source Apportionment (OSAT)
determines source area/category contributions to ozone anywhere in the domain
uses tracers to track precursor emissions and ozone production/destruction
also tracks contribution of initial and boundary conditions
estimates whether ozone is produced under NOx- or VOC-limited conditions
removes need to run model repeatedly to understand:
chemical regime
influences of various sources
HOWEVER: cannot quantify ozone response to NOx or VOC controls

21. CAMx Technical Components (continued)
Ozone Source Apportionment (OSAT)
utilizes CAMx routines for dispersion
ensures OSAT tracers and CAMx concentrations are consistent with each other
4 basic reaction tracers:
NOx emissions (N tracer)
VOC emissions (V tracer)
ozone production attributed to NOx (O3N)
ozone production attributed to VOC (O3V)
N and V decay based on weighted NOx and VOC decay rates

22. CAMx Technical Components (continued)
Ozone Source Apportionment (OSAT)
O3N and O3V accumulate a portion of total ozone production/destruction in each cell
VOC vs. NOx limited chemistry differentiated by (Sillman, 1995): P(H2O2)/P(HNO3) = 0.35
chemical allocation methodologies:
OSAT: standard approach
APCA: attributes ozone production to anthropogenic (controllable) sources only
GOAT: ozone is tracked based on where it formed, not where precursors were emitted
OPPAT: ozone formation is attributed to both VOC and NOX that participated

23. CAMx Input Requirements
Meteorology (Fortran binary)
3-D time-varying fields to define the state of the lower troposphere
affects transport, chemistry, surface removal, wet scavenging
Air Quality (UAM-IV Fortran binary)
initial and boundary conditions
Emissions (UAM-IV Fortran binary)
2-D time-varying gridded emission fields and individual
elevated point sources

24. CAMx Input Requirements (concluded)
Other
user control file (ASCII)
chemistry parameters file (ASCII)
time- and space-varying vertical grid structure (Fortran binary)
landuse / land cover (Fortran binary)
albedo / haze / ozone column (ASCII)
photolysis rates lookup table (ASCII)
OSAT source / receptor definition files (ASCII)

26. CAMx Model Output
3-D time-varying average concentration files (UAM-IV Fortran binary)
user-selected species
ppm (gasses) or g/m3 (aerosols)
optionally include surface layer or all layers
master grid file and fine grid file
3-D instantaneous concentration files (UAM-IV Fortran binary)
all species (mol/m3)
all layers
output last two hours of simulation for model restart

27. CAMx Model Output (continued)
PiG restart files (Fortran binary)
all relevant PiG information for model restart
Diagnostic files (ASCII)
repeat run control parameters and I/O file names
diagnostic messages and warning/error messages
CPU timing
mass budgets

28. CAMx Model Output (continued)
OSAT output files
master and fine grid instantaneous tracer files (Fortran binary)
master and fine grid layer 1 average tracer files (Fortran binary)
receptor file of tracer concentrations at discrete receptors (ASCII)

30. Computer Requirements
Dependent upon size of CAMx application
standard model vs. large OSAT configuration
number of nested grids needed
plan accounting for episode length and throughput for desired number of simulations
Computer platform used
Type (Linux, DEC, SUN, etc.)
Clock Speed (MHz)
Compiler (optimization)
Open MPI Multiprocessors

31. Computer Requirements

32. Computer Requirements (continued)
Minimal Recommended hardware – Unix/Linux workstation
256 Mb memory
standard OTAG simulation requires ~75 Mb
EPA SIP call OSAT configuration requires ~200 Mb
300 MHZ CPU speed
standard OTAG simulation requires ~2 hours/simulation day
EPA SIP call OSAT configuration requires ~8 hours/simulation day
10 Gb available disk space
standard OTAG simulation requires ~1.5 Gb inputs, 1 Gb output
EPA SIP call OSAT configuration requires ~2 Gb inputs, 8 Gb output
Graphics monitor and associated drivers
ODEQ Linux Computer System
Duel 2.2+ GHz Processors
2 Gb of RAM
Four 80 Gb drives in RAID system (240 Gb disk storage)

33. Computer Requirements (concluded)
Minimal Software Requirements
Fortran 77
PAVE (or some other graphics viewer)
ancillary software packages as necessary for pre-processing models
SAS (if using EMS-95)
Fortran 90 (MM5)
ODEQ Linux Computer System
PG Fortran compiler (F77/F90)
PAVE

34. Computer Requirements (concluded)

35. Comparison with Other Models

36. Application of CAMx CAMx Setup Base Case Development
Performance Evaluation
Future Base Case
Emission Control Scenario Evaluation

37. CAMx Domain Definition
Coverage
geographical/political issues
influence of boundary conditions
needs for fine resolution in key areas
depth
resource constraints
Resolution
master grid:
met model, coordinate projection, layer structures
nested grids:
where, how many, vertical nesting

38. CAMx Uses “Arakawa C” Grid

39. CAMx Domain Definition (continued)
Configuration of nested grids
horizontal
Vertical (not recommended)
Typical application
36/12/4-km grid system
Number of vertical layers TBD

40. CAMx Domain Definition (continued)
Rules for grid nesting:
1) “meshing factor” must be an integer
2) cell size of the finest grid must be a common denominator for all parent grids above it
3) the restriction in (2) does not apply to parallel fine grids of the same generation
4) fine grids cannot overlap, but can share a common boundary or edge;
5) fine grids cannot extend into a non-modeled area of the coarse grid;
6) four “generations” of nests allowed; 10 total grids allowed
7) the depth of all grids must exactly match
8) nesting vertical layers is allowed, but they must be a subset of the parent grid layers (vertical nesting not recommended)
9) there must be a matching fine grid layer interface at each parent grid layer interface

41. CAMx Domain Definition (concluded)
OSAT source/receptor areas
Source Area Map
Landuse
USGS database, and/or
developed from local data used for emission surrogates

43. CAMx Air Quality Inputs
Initial Conditions (ICs)
model spinup
role of ambient measurements
Boundary Conditions (BCs)
Lateral boundaries (time, space varying)
Concentrations aloft (time, space invariant)
Based on clean air background, observations+clean air, and/or continental model simulations

44. CAMx Chemistry Definition
“Chemical” Processes in CAMx
gas phase chemistry
mechanism, rate constants, photolysis rates
aerosol chemistry
mechanism
dry deposition
Henry’s law constant (KH), diffusivity, reactivity
wet deposition
Henry’s law constant

45. CAMx Chemistry Definition (continued)
Chemistry Parameters File: Contents
mechanism (choose from 1 through 5 or inert)
photolysis reaction parameters
gas phase species list
lower bound values
deposition parameters (KH, diffusivity, reactivity)
aerosol species list (mechanism 4 only)
gas phase reaction rate constants

46. CAMx Chemistry Mechanisms

47. CAMx Chemistry Definition (continued)
Photolysis Rates Input
two options:
directly input using a look up table, or
set by ratio to another reaction
must directly input at least one reaction
input files:
chemistry parameters (specify direct/ratio options)
photolysis rate file
albedo/haze/ozone column (ALHZOZ) file

48. CAMx Chemistry Definition (continued)
Photolysis Rate Input Files
look-up tables for all directly specified photolysis reactions
function of zenith angle, height, albedo, haze, ozone column
prepared using TUV light model from NCAR

49. CAMx Chemistry Definition (continued)
Albedo/haze/ozone column (ALHZOZ) file
categorize albedo, haze and ozone column into “bins”
bin ranges must match photolysis rate file
ALHZOZ contains gridded fields of bin indexes
ozone column from NASA satellite data (TOMS)
albedo based on USGS land use data
haze generally set to a constant default value

50. CAMx Meteorological Inputs
Objective analysis
interpolate measurements to grid points
independent fields of winds, temperature, humidity
old, rarely used – not recommended
Diagnostic models
adjust gridded fields for effect of terrain, stability, divergence minimization
some models provide mixing depth estimates and micromet variables
DWM, CALMET

51. CAMx Meteorological Inputs
Prognostic models
solve predictive equations of motion, thermodynamics, and continuity
hydrostatic models => simpler equations, but omit physics at small scales
incorporate measurement data to “nudge” model simulation
linked, consistent fields of winds, temperature, humidity, and turbulent energy
MM5, RAMS, Eta, HOTMAC, SAIMM

52. CAMx “Probing Tools” “Probing Tools”
A module within the photochemical grid model that extracts information on source-receptor relationships, chemical and physical processes, mass flux, etc.
Used to diagnose model to understand why it is getting the answer it gets and improve modeling performance
Used to better define optimal emission control strategies

53. Three “Probing Tools” in CAMx
Decoupled Direct Method (DDM) – first-order sensitivity coefficients of ozone by emission source groups (also in URM and CMAQ models)
Ozone Source Apportionment Technology (OSAT) – ozone source apportionment by source groups (source category + geographic region)
Process Analysis – output information on chemical processes and mass flux (also in CMAQ)

54. Decoupled Direct Method (DDM)
CAMx Probing Tools
Decoupled Direct Method (DDM)
Sensitivity coefficients provides information on the relationship of CAMx-estimated ozone (or other species) and sources of precursors (emissions, boundary conditions, and initial concentrations)
Information is useful for
Control Strategy Development
Model performance evaluation
Diagnostic analysis

55. CAMx Probing Tools (Concluded)
Process Analysis
Integrated Process Rates (IPR)
Integrated Reaction Rates (IRR)
Chemical Process Analysis (CPA)
Ozone Source Apportionment Technology (OSAT)
Ozone contribution by source category and source type
VOC vs. NOx limited estimates
Anthropogenic Precursor Culpability Assessment (APCA)
More control strategy relevant information\
Only allocates ozone to biogenic emissions when biogenic VOC interacts with biogenic NOx

56. Probing Tool Comparison

57. CAMx DDM - Ozone Sensitivity to + 20% Area Source VOC Emissions Changes (DDM) (Brute Force)

58. CAMx DDM - Ozone Sensitivity to + 20% Area Source NOx Emissions Changes (DDM) (Brute Force)

59. Process Analysis for Chemistry

60. Process Analysis of OH Budget

61. Process Analysis of Odd Oxygen (O3 + NO2) Budget

62. Process Analysis: Ozone Time Series Analysis

63. Process Analysis: Response of Ozone Production to Emissions

64. OSAT Ozone Source Apportionment: Total Ozone and Utility Contribution

65. OSAT Ozone Source Apportionment: Industrial Sources and Area+Non-Road Contributions

66. OSAT Ozone Source Apportionment: On-Road Mobile and Biogenic Sources Contributions

67. OSAT Source Apportionment by Receptor 2007 Base Case

68. OSAT Source Apportionment by Receptor 2007 NOx SIP Call