operational evaluation and model response comparison of camx and cmaq for ozone & pm2.5 kirk...

Operational Evaluation and Model Response Comparison of CAMx and CMAQ for Ozone & PM2.5

Kirk Baker, Brian Timin, Sharon PhillipsU.S. Environmental Protection Agency, Research Triangle Park, NC

Presented at the 2008 CMAS Conference

2

Operational Evaluation & Response Comparison

• Comprehensive Air Quality Model with Extensions (CAMx4) and the Community Multiscale Air Quality Model (CMAQ), treat the physical processes and chemistry that form ozone and PM2.5

• SIPS are submitted using either model

• Model performance is typically evaluated on an operational basis; comparing base year predictions to observations

• Since the modeled attainment demonstration includes modeling the relative change between current and future year emissions it is important to have confidence that modeling systems will predict ozone and PM2.5 concentrations consistently when emissions change

3

Modeling Background

• CMAQ v4.6 (aero4)• CAMx v4.5

• CB05 gas phase chemistry• RADM aqueous phase

chemistry• ISORROPIA inorganic

chemistry• Variations in secondary

organic chemistry between models

• Base year = 2002• Future year = 2020 • 12 km sized grid cells

• 14 vertical layers up to 15 km; • 30 m thick first layer

4

Domain Total Emissions (tpd)Domain Total Annual Average Emissions (TPD)

0

400000

800000

1200000

1600000

2000000

SO2 NOX RHC NH3

2002 2020

5

PM2.5 Bias Metric by Quarter for 2002

6

Annual PM2.5 Model Response

• Models are used in a relative sense for regulatory modeling

• Relative response factor (RRF) = estimated concentrations in future year / estimated concentrations in base (current) year

• RRFs are applied to observed design values to estimate future year design values– RRF*base design value = future design value

• RRFs are calculated for each chemical component of PM2.5

• RRFs estimated using the MATS software tool

7

Annual PM2.5 Future Year Design Values and

Speciated RRFs

8

Quarterly Nitrate Concentration (left) and RRFs (right)

9

specie r2 CV r2 CVSO4- 0.977 2.90 0.985 2.72NO3= 0.562 15.00 0.997 7.00NH4+ 0.885 3.64 0.992 2.83OC 0.918 1.13 0.999 1.22EC 0.988 1.79 0.997 2.00CRUSTAL 0.975 1.29 0.999 1.29

RRF FYDV

Relationships between CAMx and CMAQ estimated RRF and FYDV by specie

10

Bias of Daily 8-hr Ozone Maximum

11

8-hr O3 Future Year Design Values and RRFs

12

8-hr O3 Model Response

• Examined future year design values, RRFs, and number of days for RRF calculating by 4 bins of model estimate 8-hr ozone– 85+– 75 to 85– 65 to 75– 55 to 65

• Assess how model response changes based on predicted concentrations

13

8-hr O3 RRFs by Model Prediction Bin (4 bins)

CMAQ CAMx

14

8-hr O3 RRFs by Model Prediction Bin (4 bins)

15

8-hr O3 number of days in RRF calc. by Model Prediction Bin (4 bins)

16

8-hr FYDV, RRF, and days used for 85+ ppb bin

17

Remarks

• CMAQ and CAMx modeling systems consistently predict ambient concentrations of 8-hr ozone and PM2.5, which is encouraging since they are used to support modeled attainment demonstrations for NAAQS

• The relative response factors and future year design values of 8-hr ozone and annual PM2.5 are very similar using both CAMx and CMAQ even though there are differences in base year model predictions

• Larger 8-hr O3 reductions are seen at higher model predicted concentrations

• Using a different modeling system should give similar predicted future year design values when inputs and key physics options are consistent

18

Future Work

• 24-hr PM2.5 model response

• A more thorough dynamic evaluation is needed to determine if these modeling systems appropriately respond to emissions changes

19

END

20

21

PM2.5 Model Performance

CMAQ CAMx