Contribution to 8 th Annual CMAS Conference Friday
Contribution to: 8 th Annual CMAS Conference Friday Center, UNC-Chapel Hill OCTOBER, 19 -21, 2009, Friday Center for Continuing Education, 100 Friday Center Dr Chapel Hill, NC 27517, United States Simulations with MM 5 -CMAQ and WRF/CHEM models of a High elevated PM 10 and PM 2. 5 episode in Germany during Winter, 2003 R. San José 1, J. L. Pérez 1, J. L. Morant 1 and R. M. González 2 Environmental Software and Modelling Group Computer Science School – Technical University of Madrid (UPM) Campus de Montegancedo – 28660 Madrid (Spain) http: //artico. lma. fi. upm. es 1 Department of Meteorology, Complutense University of Madrid (UCM) 2 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
DOMAINS PROJECTION: Lambert Conformal Conic Central Latitude: 50. 0 N Central Longitude: 10. 0 E METEOROLOGICAL DOMAIN: - MOTHER DOMAIN: 60*60 90 Km. RESOLUTION + 23 VERTICAL LEVELS - NESTING DOMAIN: 61*61 30 Km. RESOLUTION + 23 VERTICAL LEVELS - SIGMA VERTICAL LEVELS: 1. 00, 0. 99, 0. 98, 0. 96, 0. 93, 0. 89, 0. 85, 0. 80, 0. 75, 0. 70, 0. 65, 0. 60, 0. 55, 0. 50, 0. 45, 0. 40, 0. 35, 0. 30, 0. 25, 0. 20, 0. 15, 0. 10, 0. 05, 0. 00 - LANDUSE DATA: USGS 24 CATEGORIES CHEMICAL DOMAIN: - MOTHER DOMAIN: 53*53 90 Km. RESOLUTION + 23 VERTICAL LEVELS Lower-Left Corner LCC (-2385000, -2385000) - NESTING DOMAIN: 54*54 30 Km. RESOLUTION + 23 VERTICAL LEVELS Lower. Left Corner LCC (-765000, -765000) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
DOMAINS Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
GLOBAL METEOROLOGICAL DATA - NCEP/NCAR Reanalysis Data (NNRPDS 090) - Global Reanalysis Model is T 62 (209 km) - 28 vertical sigma levels. - Data is available at 6 hour intervals - Ptop = 10000 Pa. - Boundary Conditions: every 6 hours. - Grid Nudging at mother domain (90 Km) - two-way nesting (90 -30 Km) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
MM 5 CONFIGURATION -Physics Options in MM 5: • Cumulus Parameterization: Kain-Fritsch 2 • PBL Scheme and Diffusion: MRF PBL • Explicit Moisture Scheme : Schultz microphysics • Radiation Scheme: Cloud-radiation • Surface Scheme : Noah Land-Surface Model Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
SYSTEM DESCRIPTION - MM 5 PSU/NCAR Mesoscale Model. Version 3 -7 (December 23 2004) - MCIP Meteorology-Chemistry Interface Processor Version 3. 3 (August 2007) -CMAQ Community Multi-scale Air Quality Version 4. 6 (Octuber 2006) - Magerit clúster: • 1200 node (2400 processors) e. Server Blade. Center JS 20 • Power PC 2'2 GHz and 4 Gb de RAM • Myrinet network • IBM XL Fortran 10. 1 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
BOUNDARY & INITIAL CHEMICAL CONDITIONS - Mother domain boundary conditions: profile (fixed) - Mother domain initial conditions (0 hours) from air clean profile. - Nesting domain boundary conditions from mother domain - Nesting domain initial conditions (0 hours) from air clean profile. Air clean profile example: 6 LAYERS 1. 00 0. 98 0. 93 0. 84 0. 60 0. 30 0. 00 POLLUTANT (PPm) "SO 2 " 0. 300 E-03 0. 200 E-03 0. 100 E-03 0. 200 E-04 0. 100 E-04 "SULF " 1. 000 E-30 "NO 2 " 0. 167 E-03 0. 084 E-03 0. 000 E+00 "NO " 0. 083 E-03 0. 042 E-03 0. 000 E+00 "O 3 " 0. 350 E-01 0. 400 E-01 0. 500 E-01 0. 600 E-01 0. 700 E-01 "HNO 3 " 0. 500 E-04 0. 700 E-04 0. 100 E-03 "H 2 O 2 " 0. 100 E-02 0. 150 E-02 0. 100 E-02 0. 800 E-03 0. 200 E-03 "ALD " 0. 300 E-04 0. 350 E-04 0. 300 E-04 0. 200 E-04 0. 100 E-04 "HCHO " 0. 250 E-03 0. 200 E-03 0. 100 E-03 0. 500 E-04 "OP 1 " 0. 250 E-06 0. 200 E-06 0. 100 E-06 0. 500 E-07 "OP 2 " 0. 300 E-07 0. 350 E-07 0. 300 E-07 0. 200 E-07 0. 100 E-07 "PAA " 0. 300 E-04 0. 250 E-04 0. 200 E-04 0. 150 E-04 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CMAQ-CCTM SETUP - Advection scheme: global mass-conserving scheme (Yamartino ) - Vertical Difussion: Asymmetric Convective Model (ACM 2) - CB 05 chemical mechanism (Yarwood et al. 2005). - Euler Backward Solver (EBI) solver - CMAQ Aerosol : The 3 rd generation modal CMAQ aerosol model - DEPOSITION MODEL: Models-3 / Pleim (MCIP) - Aqueous/cloud chemistry ON. - KZMIN=TRUE (in TERRAIN IEXTRA=TRUE – extra data land use -) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CASE 1 KZMIN KZ : Vertical diffusivity coefficient (m 2/s) KZMIN = FALSE MINIMUN KZ = 1 m 2/s KZMIN = TRUE MINIMUN KZ = KZL + (KZUKZL) * UFRAC KZL = LOWEST KZ 0. 5 m 2/s KZU = HIGHEST KZ 2. 0 m 2/s UFRAC = URBAN PERCENTAGE NORMALIZED TO 1 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EMISSION DATA - TNO-UBA Emission Data: • SNAP Activities: 1 Energy sector, utilities, refinaries 2 Fossil fuels, small sources 3 Fossil fuels, industry 4 Process emissions 5 Mining 6 Solvent use, use of products 7 Road transport (SO 2 only) 71 Road transport gasoline 72 Road transport diesel 73 Road transport LPG 741)Road transport non-exhaust (volatilization) 752)Road transport non-exhaust (tire, break and road wear) 8 Non-road transport 9 Waste processing 1 10 Agriculture • Geographic coordinates. • Cell Size is 1/4 degree (Y) by 1/8 degree (X) (aprox. 15 km) • Pollutans: CH 4, CO, NH 3, NMVOC, NOx, SO 2, PM 10, PM 2. 5 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EMISSION DATA - Mother Domain include EDGAR (Emission Database for Global Atmospheric Research) : * Geographic coordinates. * Cell Size is 1 degree by 1 degree * Pollutans: CO, NMVOC, NOx, SO 2 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EMISSION DATA Geographic coordinates TO Lamber Conformal Conic CMAQ grid: * EMEP interpolation routine updated by UPM. (geog >>> lambert) (original polar >>> geog km (50)) -Time distribution of EURODELTA VOC SPLITING: * SPECIATE Version 4. 0 (EPA, US) (January 18, 2007): . 1594 compounds. VOC-to-TOG Conversion Factors * Lumping VOC : . EMITDB – Carter (Development of an Improved Chemical Speciation Database for Processing Emissions of Volatile Organic Compounds for Air Quality Models ) Data Base >>>>> Excel file Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CMAQ – EMISSION – TIME DISTRIBUTION EURODELTA DAILY FACTORS MISSING COUNTRIES: Albania, Croatia, Bosnia, Serbia Bulgaria Turkey Hungary Belarus, Ukraine, Moldova, Russian Romania Germany Federal Republic Germany MONTHLY FACTORS MISSING COUNTRIES Czech Republic Slovakia Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EMISSION: VOC SPLITING (EMIMO >>>> CMAQ) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
VOC SPLITING (EMIMO >>>> CMAQ) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
VOC SPLITING (EMIMO >>>> WRF/CHEM) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
BIOGENIC & DUST EMISSION DATA: BIOEMI-UPM (EMIMO >>>> CMAQ) - Biogenic Outputs: Isoprene, biogenic NOX and biogenic VOC - Biogenic Inputs: Land uses USGS 24 categories Temperature On-Line (MM 5 Output) - Dust Outputs: PM - Dust Inputs: Land uses USGS 24 categories to 5 categories Wind Speed 10 m On-Line (MM 5 Output) Rain / Snow (No Dust) On-Line (MM 5 Output) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
BIOGENIC & DUST EMISSION DATA: BIOEMI-UPM Environ, 1998 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
WRF-CHEM DOMAIN 90 KM 30 KM Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
From Grell et al. (2007) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
WRF/Chem CONFIGURATION • Cumulus Parameterization: GRELL-DEVENYI ENSEMBLE SCHEME • PBL Scheme and Diffusion: YONSEI UNIVERSITY (YSU PBL) • Explicit Moisture Scheme : LIN et al. SCHEME microphysics VERSION 1 WSM (WRF single-moment ) 5 -class microphysics VERSION 2 • Radiation Scheme: RRTM – Goddard radiation VERSION 1 RRTM - Dudhia radiation VERSION 2 • Surface Scheme : Noah Land-Surface Model • Cloud Chemistry: : ON VERSION 1 OFF VERSION 2 • Photolisis: FTUV Version 1 FAST-J Version 2 • Biogenic model: Gunther scheme • Chemistry: CBMZ+MOSAIC (4 sectional aerosol bin) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
Time profiles DAILY FACTORS (MISSING COUNTRIES): Albania, Croatia, Bosnia, Serbia Bulgaria Turkey Hungary Belarus, Ukraine, Moldova, Russian Romania Germany Federal Republic Germany MONTHLY FACTORS MISSING COUNTRIES Czech Republic Slovakia VOC TO TOC FACTOR: 3. 2 CLOUD CHEMISTRY ON AND OFF Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
MM 5 -CMAQ 30 KM SP. RES. GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. EDGAR EMISS. PM 10 DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
MM 5 -CMAQ 30 KM SP. RES. GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. EDGAR EMISS. PM 2. 5 DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
MM 5 -CMAQ 30 KM SP. RES. GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. EDGAR EMISS. EC DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EURODELTA DAILY FACTORS MISSING COUNTRIES + KZMIN=TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EURODELTA DAILY FACTORS MISSING COUNTRIES + KZMIN=TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
EURODELTA DAILY FACTORS MISSING COUNTRIES + KZMIN=TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
KZMIN = TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es PM 10 KZMIN = FALSE
KZMIN = TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es PM 25 KZMIN = FALSE
KZMIN = TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es EC KZMIN = FALSE
DAILY EMISSION FACTOR OK Environmental Software and Modelling Group http: //artico. lma. fi. upm. es PM 10 NO DAILY EMISSION FACTOR
WRF-CHEM 30 KM SP. RES. (global >>>30) GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. PM 10 DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
WRF-CHEM 30 KM SP. RES. (global >>>30) GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. PM 2. 5 DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
WRF-CHEM 30 KM SP. RES. (global >>>30) GERMANY JAN-APR, 2003 TNO 15 KM EMISSIONS CAFÉ-DELTA TIME EMISS. PROFILES. EDGAR EMISS. EC DAILY AVERAGES AND STATION AVERAGES Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
MM 5 -CMAQ: STATISTICAL RESULTS MM 5 -CMAQ AVG STATION PM 10 PM 25 EC OC NO 3 NH 4 SO 4 NO 2 SO 2 NH 3 Observed mean 32. 55 24. 26 3. 10 3. 00 7. 44 3. 87 4. 85 15. 56 4. 07 3. 11 52. 00 Calculated mean/Obs. mean 0. 90 1. 00 1. 29 1. 96 0. 63 0. 54 0. 25 1. 28 0. 46 0. 10 Observed STD 18. 55 15. 85 3. 21 2. 56 4. 12 2. 75 3. 41 6. 05 3. 19 1. 56 13. 48 Calculated STD/Obs. STD 0. 74 0. 75 0. 62 1. 28 0. 97 0. 67 0. 24 1. 67 0. 41 0. 28 Bias -3. 40 0. 01 0. 87 2. 82 -2. 78 -1. 75 -3. 63 4. 38 -2. 20 -2. 79 1. 91 Absolute Bias 8. 21 8. 11 1. 95 3. 22 3. 65 2. 16 3. 63 6. 88 2. 47 2. 79 7. 45 R 2 0. 72 0. 45 0. 29 0. 35 0. 38 0. 31 0. 44 0. 35 0. 12 0. 07 0. 60 RMSE 10. 46 11. 71 2. 80 3. 90 4. 50 2. 89 4. 65 9. 22 3. 70 3. 17 9. 21 RMSE/Observed Mead 0. 32 0. 90 1. 30 0. 60 0. 75 0. 96 0. 59 0. 91 1. 02 0. 18 Percentage within +/- 50 90. 00 72. 50 35. 00 17. 50 41. 25 37. 50 No of Data Sets 1440 0. 48 240 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es 80 80 400 1. 04 1. 01 67. 50 43. 75 1. 25 98. 75 80 80 400 1200
WRF/CHEM: STATISTICAL RESULTS WRF-CHEM AVG STATION PM 10 PM 25 EC OC NO 3 NH 4 SO 4 NO 2 SO 2 NH 3 Observed mean 32. 55 24. 26 3. 10 3. 00 7. 44 3. 87 4. 85 15. 56 4. 07 3. 11 52. 00 Calculated mean/Obs. mean 1. 13 0. 78 1. 33 2. 01 1. 34 0. 15 2. 74 2. 94 0. 24 0. 50 Observed STD 18. 55 15. 85 3. 21 2. 56 4. 12 2. 75 3. 41 6. 05 3. 19 1. 56 13. 48 Calculated STD/Obs. STD 0. 93 1. 08 0. 42 0. 96 2. 46 1. 31 0. 16 2. 39 2. 47 0. 50 0. 74 Bias 4. 30 9. 46 -0. 65 0. 96 Absolute Bias 7. 88 11. 51 1. 46 7. 57 2. 25 4. 10 27. 03 7. 95 2. 42 25. 99 R 2 0. 73 0. 58 0. 56 0. 76 0. 42 0. 53 RMSE 10. 67 14. 95 2. 41 2. 00 10. 13 3. 04 5. 09 29. 78 10. 08 3. 04 27. 87 RMSE/Observed Mead 0. 33 0. 66 1. 36 0. 79 1. 05 1. 91 2. 48 0. 98 0. 54 Percentage within +/- 50 75. 00 48. 75 58. 75 46. 25 22. 50 50. 00 7. 50 6. 25 35. 00 42. 50 No of Data Sets 1440 1. 39 0. 62 240 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es 0. 57 0. 78 80 80 7. 52 1. 29 -4. 10 27. 01 7. 91 -2. 35 -25. 99 400 80 400 0. 25 80 0. 42 0. 11 0. 44 80 400 1200
MM 5 -CMAQ & WRF-CHEM R 2 PM 10 PM 25 EC OC NO 3 NH 4 SO 4 NO 2 SO 2 NH 3 O 3 MM 5 -CMAQ 0. 72 0. 45 0. 29 0. 35 0. 38 0. 31 0. 44 0. 35 0. 12 0. 07 0. 60 WRF/CHEM 0. 73 0. 58 0. 57 0. 56 0. 76 0. 42 0. 53 0. 25 0. 42 0. 11 0. 44 Calculated mean/Obs. mean PM 10 PM 25 EC OC NO 3 NH 4 SO 4 NO 2 SO 2 NH 3 O 3 MM 5 -CMAQ 0. 90 1. 00 1. 29 1. 96 0. 63 0. 54 0. 25 1. 28 0. 46 0. 10 1. 04 WRF/CHEM 1. 13 1. 39 0. 78 1. 33 2. 01 1. 34 0. 15 2. 74 2. 94 0. 24 0. 50 PM 10 PM 25 EC OC NO 3 NH 4 SO 4 NO 2 SO 2 NH 3 O 3 Percentage within +/- 50 MM 5 -CMAQ 90. 00 72. 50 35. 00 17. 50 41. 25 37. 50 67. 50 43. 75 1. 25 WRF/CHEM 75. 00 48. 75 58. 75 46. 25 22. 50 50. 00 Environmental Software and Modelling Group http: //artico. lma. fi. upm. es 7. 50 98. 75 6. 25 35. 00 42. 50
CASE 2 MM 5 -CMAQ O 3 HOURLY VALUES AVERAGE STATION FROM 125 EUROPEAN BACKGROUND STATIONS (COST 602) Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CASE 2 WRF/Chem (VOC TO TOC 1. 14) O 3 HOURLY VALUES AVERAGE STATION FROM 125 EUROPEAN BACKGROUND STATIONS Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CASE 2 WRF/Chem (VOC TO TOC 1. 6) O 3 HOURLY VALUES AVERAGE STATION FROM 125 EUROPEAN BACKGROUND STATIONS Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CASE 2 MM 5 -CMAQ KZMIN = TRUE Environmental Software and Modelling Group http: //artico. lma. fi. upm. es KZMINFALSE - KZMINTRUE
CASE 1: CPU TIMES WRF-CHEM 90 918 minutes WRF-CHEM 30 1490 minutes MM 5_90_30 136 minutes CMAQ 90 553 minutes CMAQ 30 652 minutes TOTAL CPU TIME: MM 5 -CMAQ >>>>>> 1341 MINUTES TOTAL CPU TIME: WRF-CHEM >>>>>> 2408 MINUTES MAGERIT SUPERCOMPUTER 32 PROCESSORS, Power PC, 2'2 GHz and 4 Gb de RAM Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
CONCLUSIONS • KZMIN AND EMISSION TIME PROFILES HAVE IMPROVED SUBSTANTIALLY THE RESULTS • THE NEW VONFIGURATION OF WRF/CHEM HAS IMPROVED SUBSTANTIALLY THE CONCENTRATION RESULTS • WRF/CHEM IS BETTER THAN MM 5 -CMAQ FOR PM 10 AND PM 2. 5 CONCENTRATIONS • MM 5 -CMAQ IS BETTER THAN WRF/CHEM FOR OZONE CONCENTRATIONS • MM 5 -CMAQ IS HARDLY REPRODUCING THE HIGHEST PM 10 AND PM 2. 5 PEAKS (FEBRUARY AND MARCH, 2003) • WRF/CHEM IS MUCH MORE SENSITIVE THAN MM 5 -CMAQ AND CAPTURE THE HIGHEST PEAKS (FEBRUARY AND MARCH) MUCH BETTER THAN MM 5 -CMAQ Acknowledgements: COST 728 EU PROGRAMME Environmental Software and Modelling Group http: //artico. lma. fi. upm. es
- Slides: 44
