290 likes | 431 Views
Evaluation of the VISTAS 2002 CMAQ/CAMx Annual Simulations. T. W. Tesche & Dennis McNally -- Alpine Geophysics, LLC Ralph Morris -- ENVIRON Gail Tonnesen -- UC Riverside Patricia Brewer -- VISTAS Technical Coordinator James Boylan – Georgia Dept of Natural Resources Models-3 CMAS Conference
E N D
Evaluation of the VISTAS 2002 CMAQ/CAMx Annual Simulations T. W. Tesche & Dennis McNally -- Alpine Geophysics, LLC Ralph Morris -- ENVIRON Gail Tonnesen -- UC RiversidePatricia Brewer -- VISTAS Technical Coordinator James Boylan – Georgia Dept of Natural Resources Models-3 CMAS Conference 18-20 October 2004 Chapel Hill, NC
Outline • VISTAS objectives • Model set-up for initial Phase II runs • Highlights of CMAQ/CAMx evaluations • Operational, Comparative, Diagnostic, Mechanistic • Some findings from diagnostic studies • Suggestions
VISTAS AQ Modeling Objectives • Phase I: • Evaluate suite of models for episodic and annual simulation of Regional Haze & PM2.5 on 36/12 km US grid • Phase II: • Select and evaluate preferred model(s) for 2002 annual period via detailed model performance and sensitivity evaluations • Evaluate emission control strategies for regional haze, particularly for VISTAS region. • Support VISTAS states responsible for upcoming PM2.5 attainment demonstrations.
Model Set-up for Initial 2002 Annual Run • 36/12 km grid, 19 layers • CMAQ v4.3 and CAMx v4.0 • MM5 (Pleim-Xiu_ACM8 36/12 km) • 2002 Emissions for VISTAS states (WRAP and CENRAP updates; NEI 1999 V2 for rest of U.S.) • CMAQ (CB4, SORGAM); CAMx (CB4, SOAP) • BCs from 2001 Seasonal GEOS-CHEM • Models run in 4 quarters with 15 day spin-up • VISTAS Phase II Modeling Protocol followed • For reports, results, presentations…. http://pah.cert.ucr.edu/vistas/vistas2/reports
Operational Evaluation • Focus on • Visibility-related PM species • Identify needed improvements before final 2002 basecase simulations begin (next week…!) • Use suite of 15 metrics and graphical tools • Evaluate by month and monitoring network • Multiple evaluation teams • ENVIRON, UCR, Alpine, VISTAS-TAWG, GA-DNR
Monitors in VISTAS 12 km MPE Domain Yorkville, GA
Sulfate Fractional Bias and Error: CMAQ(note scale: 0-100%) IMPROVE Data for VISTAS States:12 km grid
Nitrate Fractional Bias and Error: CMAQ(note scale: 0-200%) IMPROVE Data for VISTAS States:12 km grid
Bias as Function of Concentration: CMAQ Data for VISTAS States:12 km grid
Operational Evaluation Summary for CMAQ & CAMx • Good: SO4 and EC • Good-Fair: PM2.5 and PM10 • Fair: NH4 • Fair-Poor OC and CM • Poor NO3 and Soils
Comparative Evaluation • Inter-compare CMAQ V4.3 and CAMx V.4 • Use identical SMOKE/MM5 inputs & VISTAS evaluation protocol • Examine reasons for similar and divergent behavior • Gas phase and aerosol species • Wet and dry deposition patterns • Conduct sensitivity experiments to elucidate similar and divergent behavior in CMAQ and CAMx
CMAQ/CAMx Fractional Bias: 12 km EC/CM “Flip-Flop”
Comparative Evaluation Summary In general: CMAQ and CAMx respond consistently for most gas-phase and PM species Winter: Large over-predictions of NO3 and CM Summer: Large under-predictions of NO3 (but concentrations are quite small) All Seasons: Soils over-predicted; OC under-predicted (understated primary OC emissions?)
Diagnostic Evaluation • Examine PM and gas-phase species by network • Evaluate effects of grid resolution, model response by sub-region, and range of time scales • Examine differences in CMAQ/CAMx response • Synthesize CMAQ/CAMx model evaluation results to elucidate possible sources of model bias and error (e.g. formulation, inputs, …)
Seasonal & Annual Average AerosolBias and Error: CMAQ IMPROVE Data for VISTAS States:12 km grid
CMAQ Spatial Mean Nitrate: VISTAS vs. MANE-VU VISTAS: Jan ‘02 MANE-VU Jan ‘02 VISTAS: May ‘02 MANE-VU May ‘02
CMAQ Spatial Mean Sulfate: VISTAS vs. MRPO VISTAS: Jan ‘02 MRPO Jan ‘02 VISTAS: May ‘02 MRPO May ‘02
Spatial Mean EC Dry Deposition CMAQ-Jan ’02 CAMx-Jan ’02 CMAQ-Jul ’02 CAMx-Jul ‘02 CMAQ dep >CAMx depfor EC
Spatial Mean CM Dry Deposition CMAQ-Jan ’02 CAMx-Jan ’02 CMAQ-Jul ’02 CAMx-Jul ‘02 CMAQ dep <<CAMxdepfor EC
Yorkville NO3, Temp & Mixing Ratio Time Series (Jan ’02) Temperature Mixing Ratio NO3
SEARCH Hourly Nitrogen Species at Yorkville, GA: Jan ‘02 NO NO2 NOy HNO3
Diagnostic Evaluation Summary -CMAQ and CAMx consistent for most species across all domains and time scales. - EC/CM bias ‘flip-flop’ due to different dry deposition algorithms in CMAQ/CAMx • OC bias differences in CMAQ/CAMx, in part, attributed to • Different SOA chemistry formulations • Different environmental chamber data sets and parameterizations.
Mechanistic Evaluation: CB4 vs SAPRC99 for Jan ’02 & Jul ’01 Episodes • Very Similar Base Case Performance for SO4, NO3 and OC: • Differences between 36 and 12 km grid larger than differences between CB4 and SAPRC • SAPRC exhibits slightly improved performance for ozone compared to CB4 • Generally Similar Response to 30% Controls, except: • SO4 sensitivity to NOx controls • SAPRC approximately twice as sensitive • Tied to H2O2 and O3 sensitivity to NOx controls • O3 sensitivity to VOC • SAPRC more sensitive than CB4
Three Suggestions • Devote greater emphasis to the diagnostic component of MPE (consider range of time and space scales, super-site data sets) • Utilize the extensive 2002 aircraft data base for aloft model evaluation (probe ‘regional transport’ issue) • Employ corroborative models to explore key uncertainties in • Input data base development • Base case model performance • Reliability of model response to emission controls