Statewide Protocol:Regional Application

August 27, 2003

Air Resources BoardCalifornia Environmental Protection Agency

Luis F. Woodhouse

RegionalModeling

IntegratedResults

Risk Assessment

Mapping andVisualization

MicroscaleModeling

Emissions andMeteorology

Modeling Framework

HARP

CHAPIS

2

Objectives

• Simulate year 2000 air quality, including toxics, for all California using regional air quality models.

• Will provide background concentrations to be used with local dispersion modeling.

3

Previous Experience

• Modeled southern California for year 1998– Air quality models: CMAQ and CALGRID– Meteorological models: MM5 and CALMET

• Developed annual model performance metrics (toxics and other species)

• Conducted sensitivity tests to expedite simulation while minimizing error introduced

4

Challenges for Statewide Modeling

• Very large modeling domain

• Potentially long run time (meteorological and regional models)

• Storage and processing of very large input and output files

• Evaluation of models input/output data

• Apply lessons learned to optimize statewide modeling

• Double counting5

Optimizing Statewide Regional Modeling

• Apply lessons learned from previous experience modeling southern California– Shorter time periods to represent year– Use of subdomains to cover entire State

• Leveraging from other studies:– SCOS/CCOS/CRPAQS modeling and

databases

6

Statewide Protocol:Regional Modeling

• Establishes criteria for conducting air quality modeling for toxic air pollutants:– Modeling domain definition– Selection and evaluation of models– Selection of chemical mechanism– Preparation of annual emissions inventory– Initial and boundary conditions– Evaluation of models results

7

Modeling Domain

Ideal case• Uses large modeling

domain that includes entire state

• Resource intensive and requires long run time

8

Modeling Domain (cont.)

For statewide simulation:

• Use 4 subdomains to cover state (4-km x 4-km grid)

• Each subdomain modeled independently

• Less resource intensive and faster run times for each subdomain

• SCOS subdomain by early 2004

9

Air Quality Models Selection

• Sound scientific basis

• Reflect up-to-date-science

• Documentation

• Source code and technical documentation available to the public

10

Air Quality Models (cont.)

For statewide simulation:

• State-of-the-science models will be evaluated (such as Models-3/CMAQ, CAMx and CALGRID)

• Criteria defined in protocol to evaluate model performance for ozone and toxics

11

Chemical Reaction MechanismSelection

• Successful peer review

• Extensively tested

• Represents recent advances in science

• Publicly available

12

Chemical Reaction Mechanism (cont.)

For statewide annual simulations:

• Selected SAPRC99 reaction mechanism

• Added explicit reactions for selected toxics

13

Toxics– 1,3-butadiene

– Formaldehyde

– Acetaldehyde

– Acrolein

– Benzene

– Carbon tetrachloride

– Chloroform

– Dichloromethane

– 1,2-Dichloroethane

– o-Dichlorobenzene

– p-Dichlorobenzene

– Ethylene oxide

– Perchloroethylene

– Styrene

– Toluene

– Trichloroethylene

– Vinyl Chloride

– Xylenes

– Diesel PM10

– Other PM10 species: Arsenic, Beryllium, Cadmium, Hexavalent Chromium, Iron, Lead, Manganese, Mercury, Nickel, Zinc and elemental carbon

14

Meteorological Model Selection

• Peer review process

• Fully documented

• Reflects recent advances in science

• Publicly available

15

Meteorology Models (cont.)

• For statewide modeling two meteorological models will be applied:– MM5: A prognostic model that predicts from

first principles, mass and energy transfer equations

– CALMET: A diagnostic model that uses observational data

16

Meteorological Models (Cont.)

• Evaluation of input and output data from meteorological models:– Overall pattern for selected periods– Predictions vs. observations for selected

periods

17

Emissions• Point and areawide emissions

– Extrapolated from 1999 baseline– Use surrogates to allocate area sources to

individual grids

• On-road motor vehicle emissions– Latest versions of EMFAC and DTIM4

• require hourly temperature and relative humidity

• Biogenic emissions

• Weekday and weekend emissions by month

18

Emissions (cont.)

• Emissions will be evaluated before use in air quality models– Spatial pattern of emissions– Temporal patterns– Comparison of predicted and observed ratios of

HC/NOx and CO/NOx

19

Initial and Boundary Conditions

• Protocol suggests default initial and constant boundary conditions– Boundary conditions same as for SCOS-97

• Pristine over ocean (40 ppb O3, 0.001 ppb NOx, and 20 ppbC VOC)

• South Coast clean over land (40 ppb O3, 2 ppb NOx, 60 ppbC VOC)

20

Summary

• Developed criteria for:– model selection– input preparation– input evaluation– model output evaluation

• Optimization of statewide modeling

21