230 likes | 356 Views
Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:. Revised NAQFS Analysis Addition of 2007 NAQFS data Prelim. Background research 2007 Model Evaluation MD/VA Free Ride Program numbers. Revised NAQFS Analysis. Update Yorks, 2007 with 2007 NAQFS 1hr and 8hr forecasts
E N D
Improving Cost Effective Air Quality ForecastingUPDATE 9/25/2008: Revised NAQFS Analysis Addition of 2007 NAQFS data Prelim. Background research 2007 Model Evaluation MD/VA Free Ride Program numbers
Revised NAQFS Analysis • Update Yorks, 2007 with 2007 NAQFS • 1hr and 8hr forecasts • 2007 forecast data inconsistent with 04-06 • Reviewed methods • 8hr average forecast values used as 1hr! • 8hr averages 1hr average • 8hr averaged 8hr averages 8hr average • 04-06 maximum forecasts appeared low
NAQFS Model Evaluation (Eder, 2008?) • Model Updates • WRF-NMM Meteorology (NCEP) • CMAQ Air Quality (EPA) • CONUS Domain • Surface layer roughness length parameters under stable conditions • Effects on AQ under investigation • Canopy uptake (plant respiration) • NOx and CO • Discrete Evaluation • RMSE, N/MB, N/ME, r • Categorical Evaluation • A, B, FAR, CSI, WSI (Kang, 2007)
NAQFS Model Evaluation • Discrete Evaluation (Eder, 2006)
NAQFS Model Evaluation • Categorical Evaluation • Ozone threshold exceeded? • Was it forecasted to exceed? Forecast Threshold • a MISS HIGH • b HIT HIGH • c HIT LOW • d MISS LOW
NAQFS Model Evaluation • Categorical Evaluation (Eder, 2006) • A Accuracy (%) • B Bias (<1, underpredict; >1 overpredict) • CSI Critical Success Index (%) • FAR False Alarm Rate (%) • WSI Weighted Severity Index (%) (Kang, 2007)
NAQFS Model Evaluation (Eder, 2008?) • High Accuracy misleading! Large number of correctly forecasted non-exceedence days • High FAR, Low CSI, and ~40% increase on WSI over CSI Lots of mis-forecasted days, all very close to threshold • Bias >1 Forecast generally higher than observed
NAQFS Model Evaluation (Eder, 2008?) • Low correlation days associated with cloud cover days
MD/VA Free Ride Code Red Program • Free bus rides on forecasted “AQI Red” days • Attempt to reduce emission of ozone precursors • Promotes awareness of air quality issues • Educates public on steps to improve their air • Less health problems
MD/VA Free Ride Code Red Program • Northern VA • Funded 100% by Congestion Mitigation and Air Quality Program (CMAQ) • 06 2 days, 07 1 day, 08 (so far) 2 days • Cost to the 8 transit providers… • $120,000 per day • Count of riders taken by the drivers • Rider count up 4% on code red days • ~ 54,000 vehicle trips reduced • NVTC estimates numbers are more like 7-8%
MD/VA Free Ride Code Red Program • Maryland • DC does NOT take part in this program • MDOT funds 35-40%, local jurisdiction remainder • 06 2 days, 07 1 day, 08 (so far) 2 days • Cost to MDOT… • $35,000 - $40,000 per day • Count taken again by drivers • Rider count up 5-7% on code red days • ~ 4,750 – 6,650 boardings
MD/VA Free Ride Code Red Program Effectiveness • 0.0752 tons of NOx reduced per day • 0.026 tons of VOC reduced per day • Cost per ton of NOx and VOC reduction • $1,003,819 • Not too bad…more effective Transportation Emission Reduction Measure (TERM) • RT-Bus schedule and info $34,074 • Neighborhood Circulator Busses $129,576 • 1000 more parking spaces @ commuter stations $430,602 • Less effective TERM • Free off-peak bus service $1,408,268 • Free Bus/Rail Rail/Bus X-fer $1,994,610
Where to from here? • Investigate why models over-predict • How can we incorporate said findings into the model? • Use AirNow/CASTNet/other AQ data sources to help verify forecasts • Determine data needed • Temporal range • Locations to look at • Types of cost data • Determine the best method of evaluating the value of an AQ forecast
References • Desimone, J., 2008; Memorandum: Free Ride on Code Red Program. Provided by Joan Rholfs, MWCOG. • Eder, B., D. Kang, R. Mathur, S. Yu, K. Schere, 2006. An Operational Evaluation of the Eta-CMAQ Air Quality Forecast Model. Atmospheric Environment40, 4894 – 4905 • Eder, B., D. Kang, R. Mathur, J. Pleim, S. Yu, 2008?. An Evaluation of the National Air Quality Forecast Capability for the Summer of 2007. • Kang, D., R. Mathur, K. Schere, S. Yu, B. Eder, 2007. New Categorical Metrics for Air Quality Model Evaluation. Journal of Applied Meteorology. 46, 549-55