CRC NARSTO-Northeast Modeling Study

G:\crca24\CMAS_workshop\CMAS_CRC_NARSTO.ppt

CRC NARSTO-Northeast Modeling Study

Ralph E. Morris, Edward Tai, and Greg Yarwood

ENVIRON International Corporation

101 Rowland Way

Novato, California 94945

Models-3 User’s Workshop hosted by CMAS

October 21-23, 2002

Research Triangle Park, North Carolina


Acknowledgements

• Coordinating Research Council (CRC) Atmospheric Impacts Committee (Brent Bailey, Coordinator)

• Work performed with Alpine Geophysics, STI, and PSU

• NARSTO acknowledged for data collection and analysis activities and comments

• MM5 data provided by Nelson Seaman at PSU• NARSTO Data Analysis Study directed by Till

Stoekenius at ENVIRON


Purpose

• Investigate sensitivity of photochemical grid models to inputs and options using the extensive July 1995 NARSTO-Northeast database:– Photochemical Grid Model (CMAQ vs CAMx)– Meteorological Model (MM5 vs RAMS)– Level of FDDA in MM5– VOC and/or NOx Emission Reductions– Grid Resolution (12-km vs 4-km)– Chemical Mechanism (CB-IV vs SAPRC97)– Advection Solver (Bott vs PPM vs Smolarkiewicz)


Overview of Approach

• Minimize any interpolation of met data– LCP grid for MM5 met; PSP grid for RAMS met

• MM5 run with 3 levels of FDDA– Analysis Nudging (AN) only– AN and Observation Nudging (ON)– AN, ON, and VAD nudging (FOBS)

• CMAQ and CAMx sensitivity runs– 36-km, 12-km, and 4-km nests



Emissions Processed by EMS95

VOC Emission 234,026 TPD

Area 12%

Point 4%

Mobile 5%

Biogenic 79%

NOx Emissions 65,440 TPD

Area18%

Point46%

Mobile26%

Biogenic10%


Difficulties Encountered/Lessons Learned

• Emissions: Numerous delays were encountered waiting for the latest emissions that were not substantially different from the draft values.

• CMAQ QSSA Chemistry Solver: QSSA chemistry solver is slow, inaccurate, and may go unstable so should not be used.

• MAQSIP Modeling System: STI had difficulties getting MAQSIP to work which was believed to be due to not using the MAQSIP version of MM5.

• Layer 1 Vertical Diffusivities: Both CMAQ (MCIP) and CAMx (MM5CAMx) modeling systems were updated during then study with a minimum layer 1 KV.


Ozone Hourly Time Series in NYC SubdomainCAMx/MM5, CAMx/RAMS, CMAQ/MM5


Daily Maximum Ozone -- July 14, 1995 CAMx/MM5 CMAQ/MM5 CAMx/RAMS


Daily Maximum Ozone -- July 15, 1995 CAMx/MM5 CAMx/RAMS CMAQ/MM5


12-km vs 4-km Grids-- CMAQ July 14, 1995


12-km vs 4-km Grids -- CAMx July 14, 1995


12-km vs 4-km Grids -- CMAQ July 15, 1995


12-km vs 4-km -- CAMx July 15, 1995


Comments on Grid Resolution Sensitivity

• July 12-14 CMAQ and CAMx 4-km ozone estimates more like each other than their 12-km parent

• 4-km ozone estimates generally lower than 12-km

• CMAQ 4-km ozone smoother due to KH inversely proportional to grid spacing

• July 15, 1995 very different response to 4-km grid:– MM5 4-km grid explicitly resolves convective activity

with downdrafts etc.– Many other desirable features in MM5 4-km fields


MM5 12-km vs 4-km Wind Field 07/15/95

1044 1080 1116 1152 1188 1224 1260 1296 1332 1368 1404 1440-144

-108

-72

-36

0

36

72

108

144

180

216

252

288

324

360

396

10 m /s

C AM x Layer 1 W ind F ie ld from 4km M M 5 D ata w ith O zone (ppb)July 15, 1995. H our 17

1044 1080 1116 1152 1188 1224 1260 1296 1332 1368 1404 1440-144

-108

-72

-36

0

36

72

108

144

180

216

252

288

324

360

396

10 m /s

C AM x Layer 1 W ind F ie ld Interpolated from 12km M M 5 D ata w ith O zone (ppb)Ju ly 15, 1995. H our 17


Emission Reduction Sensitivity Tests

• Investigate sensitivity of ozone responses to 50% NOx and/or 50% VOC anthropogenic emission reductions:– Photochemical model: CMAQ vs. CAMx– Grid resolution: 12-km vs. 4-km (NOx only)– Chemical mechanism: CB-IV vs SAPRC97– Advection Solver: Bott vs Smolarkiewicz– Low-level vs Elevated NOx

• Design Value Scaling for Four NE Sites


50% Anthropogenic VOC Reduction 1995 1-Hour Max Ozone with 50% VOC Controls Scaled to the Design Values

0

20

40

60

80

100

120

140

160

180

New Haven, CT Cecil, MD Camden, NJ Bronx, NY

Ozo

ne

(pp

b)

EPA95-97 Design Value

CAMx/FOBS 36/12

CAMx/FOBS 36/12 Smolar

CAMx/FOBS 36/12 SAPRC

CAMx/RAMS 36/12

CMAQ 12

124ppb


50% Anthropogenic NOx Reduction1995 1-Hour Max Ozone with 50% NOx Controls Scaled to the Design Values

0

20

40

60

80

100

120

140

160

180

200

New Haven, CT Cecil, MD Camden, NJ Bronx, NY

Ozo

ne

(pp

b)

EPA95-97 Design Value

CAMx/FOBS 36/12

CAMx/FOBS 36/12 Smolar

CAMx/FOBS 36/12 SAPRC

CAMx/RAMS 36/12

CMAQ 12

CAMx/FOBS 36/12/4

CAMx/FOBS 36/12/(4emis)

CMAQ 4

124ppb


Emission Reduction Sensitivity Conclusions

• NOx reductions result in wide-spread decreases in ozone except in urban cores where increases sometimes occur

• VOC reductions result in ozone reductions in urban cores but little change elsewhere

• CAMx/MM5 is more VOC sensitive than either CMAQ/MM5 or CAMx/RAMS

• SPARC97 chemistry is more VOC sensitive than CB-IV

• Use of a 4-km grid increases the NOx disbenefits in CAMx/MM5 but has less effects in CMAQ/MM5


Overall Conclusions• QSSA chemistry solver is slow and inaccurate and

should not be used• Smolarkiewicz advection solver is overly diffusive

and should not be used• CMAQ horizontal diffusion coefficient

parameterization that is inversely proportional to grid resolution negates many of the benefits of high resolution grids

• Meteorological modeling of convective activity should be studied for air pollution modeling

• Ozone Design Value scaling can be very different using different models

Documents

CRC NARSTO-Northeast Modeling Study