improving cost effective air quality forecasting update 9/25/2008:

23
Improving Cost Effective Air Improving Cost Effective Air Quality Forecasting Quality Forecasting UPDATE 9/25/2008: UPDATE 9/25/2008: Revised NAQFS Analysis Revised NAQFS Analysis Addition of 2007 NAQFS data Addition of 2007 NAQFS data Prelim. Background research Prelim. Background research 2007 Model Evaluation 2007 Model Evaluation MD/VA Free Ride Program numbers MD/VA Free Ride Program numbers

Upload: dena

Post on 13-Jan-2016

34 views

Category:

Documents


0 download

DESCRIPTION

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 - PowerPoint PPT Presentation

TRANSCRIPT

Page 1: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Improving Cost Effective Air Improving Cost Effective Air Quality ForecastingQuality Forecasting

UPDATE 9/25/2008:UPDATE 9/25/2008:

•Revised NAQFS AnalysisRevised NAQFS Analysis•Addition of 2007 NAQFS dataAddition of 2007 NAQFS data•Prelim. Background researchPrelim. Background research

–2007 Model Evaluation2007 Model Evaluation–MD/VA Free Ride Program numbersMD/VA Free Ride Program numbers

Page 2: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Revised NAQFS AnalysisRevised NAQFS Analysis

• Update Yorks, 2007 with 2007 NAQFSUpdate Yorks, 2007 with 2007 NAQFS

– 1hr and 8hr forecasts1hr and 8hr forecasts

• 2007 forecast data inconsistent with 04-062007 forecast data inconsistent with 04-06

• Reviewed methodsReviewed methods

• 8hr average forecast values used as 1hr!8hr average forecast values used as 1hr!

– 8hr averages 8hr averages 1hr average 1hr average

– 8hr averaged 8hr averages 8hr averaged 8hr averages 8hr average 8hr average

• 04-06 maximum forecasts appeared low04-06 maximum forecasts appeared low

Page 3: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Original NAQFS Analysis (Yorks, 2007)Original NAQFS Analysis (Yorks, 2007)

Page 4: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Updated NAQFS AnalysisUpdated NAQFS Analysis

Page 5: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Original NAQFS Analysis (Yorks, 2007)Original NAQFS Analysis (Yorks, 2007)

Page 6: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Updated NAQFS AnalysisUpdated NAQFS Analysis

Page 7: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Original NAQFS Analysis (Yorks, 2007)Original NAQFS Analysis (Yorks, 2007)

Page 8: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Updated NAQFS AnalysisUpdated NAQFS Analysis

Page 9: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Original NAQFS Analysis (Yorks, 2007)Original NAQFS Analysis (Yorks, 2007)

Page 10: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Updated NAQFS AnalysisUpdated NAQFS Analysis

Page 11: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model Evaluation (Eder, 2008?)NAQFS Model Evaluation (Eder, 2008?)

• Model UpdatesModel Updates– WRF-NMM WRF-NMM Meteorology (NCEP) Meteorology (NCEP)– CMAQ CMAQ Air Quality (EPA) Air Quality (EPA)– CONUS DomainCONUS Domain– Surface layer roughness length parameters Surface layer roughness length parameters

under stable conditionsunder stable conditions• Effects on AQ under investigationEffects on AQ under investigation

– Canopy uptake (plant respiration)Canopy uptake (plant respiration)• NOx and CONOx and CO

• Discrete EvaluationDiscrete Evaluation– RMSE, N/MB, N/ME, rRMSE, N/MB, N/ME, r

• Categorical EvaluationCategorical Evaluation– A, B, FAR, CSI, WSI A, B, FAR, CSI, WSI (Kang, 2007) (Kang, 2007)

Page 12: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model EvaluationNAQFS Model Evaluation

• Discrete Evaluation (Eder, 2006)Discrete Evaluation (Eder, 2006)

Page 13: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model EvaluationNAQFS Model Evaluation

• Categorical EvaluationCategorical Evaluation– Ozone threshold exceeded?Ozone threshold exceeded?– Was it forecasted to exceed?Was it forecasted to exceed?

ForecastForecast ThresholdThreshold

• a a MISSMISS HIGHHIGH• b b HITHIT HIGHHIGH• c c HITHIT LOWLOW• d d MISSMISS LOWLOW

Page 14: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model EvaluationNAQFS Model Evaluation

• Categorical Evaluation (Eder, 2006)Categorical Evaluation (Eder, 2006)

•AA Accuracy (%) Accuracy (%)

•BB Bias (<1, underpredict; >1 Bias (<1, underpredict; >1

overpredict)overpredict)

•CSICSI Critical Success Index (%) Critical Success Index (%)

•FARFAR False Alarm Rate (%) False Alarm Rate (%)

•WSIWSI Weighted Severity Index (%) Weighted Severity Index (%)

(Kang, 2007)(Kang, 2007)

Page 15: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model Evaluation (Eder, 2008?)NAQFS Model Evaluation (Eder, 2008?)

•High Accuracy misleading! Large number of correctly forecasted non-exceedence High Accuracy misleading! Large number of correctly forecasted non-exceedence daysdays

•High FAR, Low CSI, and ~40% increase on WSI over CSI High FAR, Low CSI, and ~40% increase on WSI over CSI Lots of mis-forecasted Lots of mis-forecasted days, all very close to thresholddays, all very close to threshold

•Bias >1 Bias >1 Forecast generally higher than observed Forecast generally higher than observed

Page 16: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model Evaluation (Eder, 2008?)NAQFS Model Evaluation (Eder, 2008?)

•Low correlation days associated with cloud cover daysLow correlation days associated with cloud cover days

Page 17: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

NAQFS Model Evaluation (Eder, 2008?)NAQFS Model Evaluation (Eder, 2008?)

Page 18: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

MD/VA Free Ride Code Red ProgramMD/VA Free Ride Code Red Program

• Free bus rides on forecasted “AQI Red” daysFree bus rides on forecasted “AQI Red” days

• Attempt to reduce emission of ozone Attempt to reduce emission of ozone

precursorsprecursors

• Promotes awareness of air quality issuesPromotes awareness of air quality issues

• Educates public on steps to improve their airEducates public on steps to improve their air

• Less health problemsLess health problems

Page 19: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

MD/VA Free Ride Code Red ProgramMD/VA Free Ride Code Red Program

• Northern VANorthern VA

– Funded 100% by Congestion Mitigation and Air Funded 100% by Congestion Mitigation and Air

Quality Program (CMAQ)Quality Program (CMAQ)

– 06 06 2 days, 07 2 days, 07 1 day, 08 (so far) 1 day, 08 (so far) 2 days 2 days

– Cost to the 8 transit providers…Cost to the 8 transit providers…• $120,000 per day$120,000 per day

– Count of riders taken by the driversCount of riders taken by the drivers

– Rider count up 4% on code red daysRider count up 4% on code red days• ~ 54,000 vehicle trips reduced~ 54,000 vehicle trips reduced

• NVTC estimates numbers are more like 7-8%NVTC estimates numbers are more like 7-8%

Page 20: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

MD/VA Free Ride Code Red ProgramMD/VA Free Ride Code Red Program

• MarylandMaryland

– DC does NOT take part in this programDC does NOT take part in this program

– MDOT funds 35-40%, local jurisdiction MDOT funds 35-40%, local jurisdiction

remainderremainder

– 06 06 2 days, 07 2 days, 07 1 day, 08 (so far) 1 day, 08 (so far) 2 days 2 days

– Cost to MDOT…Cost to MDOT…• $35,000 - $40,000 per day$35,000 - $40,000 per day

– Count taken again by driversCount taken again by drivers

– Rider count up 5-7% on code red daysRider count up 5-7% on code red days

– ~ 4,750 – 6,650 boardings~ 4,750 – 6,650 boardings

Page 21: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

MD/VA Free Ride Code Red Program MD/VA Free Ride Code Red Program EffectivenessEffectiveness

• 0.0752 tons of NOx reduced per day0.0752 tons of NOx reduced per day

• 0.026 tons of VOC reduced per day0.026 tons of VOC reduced per day

• Cost per ton of NOx and VOC reductionCost per ton of NOx and VOC reduction– $1,003,819$1,003,819

• Not too bad…more effective Transportation Emission Not too bad…more effective Transportation Emission

Reduction Measure (TERM)Reduction Measure (TERM)– RT-Bus schedule and info RT-Bus schedule and info $34,074 $34,074

– Neighborhood Circulator Busses Neighborhood Circulator Busses $129,576 $129,576

– 1000 more parking spaces @ commuter stations 1000 more parking spaces @ commuter stations $430,602 $430,602

• Less effective TERMLess effective TERM– Free off-peak bus service Free off-peak bus service $1,408,268 $1,408,268

– Free Bus/Rail Rail/Bus X-fer Free Bus/Rail Rail/Bus X-fer $1,994,610 $1,994,610

Page 22: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

Where to from here?Where to from here?

• Investigate why models over-predictInvestigate why models over-predict

• How can we incorporate said findings into How can we incorporate said findings into the model?the model?

• Use AirNow/CASTNet/other AQ data Use AirNow/CASTNet/other AQ data sources to help verify forecastssources to help verify forecasts

• Determine data neededDetermine data needed– Temporal rangeTemporal range

– Locations to look atLocations to look at

– Types of cost dataTypes of cost data

• Determine the best method of evaluating Determine the best method of evaluating the value of an AQ forecastthe value of an AQ forecast

Page 23: Improving Cost Effective Air Quality Forecasting UPDATE 9/25/2008:

ReferencesReferences

• Desimone, J., 2008; Memorandum: Desimone, J., 2008; Memorandum: Free Ride on Code Red Program.Free Ride on Code Red Program. Provided by Joan Rholfs, MWCOG.Provided by Joan Rholfs, MWCOG.

• Eder, B., D. Kang, R. Mathur, S. Yu, K. Schere, 2006. An Operational Eder, B., D. Kang, R. Mathur, S. Yu, K. Schere, 2006. An Operational Evaluation of the Eta-CMAQ Air Quality Forecast Model. Evaluation of the Eta-CMAQ Air Quality Forecast Model. Atmospheric Atmospheric EnvironmentEnvironment 4040, 4894 – 4905, 4894 – 4905

• Eder, B., D. Kang, R. Mathur, J. Pleim, S. Yu, 2008?. An Evaluation of the 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. National Air Quality Forecast Capability for the Summer of 2007.

• Kang, D., R. Mathur, K. Schere, S. Yu, B. Eder, 2007. New Categorical Kang, D., R. Mathur, K. Schere, S. Yu, B. Eder, 2007. New Categorical Metrics for Air Quality Model Evaluation. Metrics for Air Quality Model Evaluation. Journal of Applied MeteorologyJournal of Applied Meteorology. . 4646, 549-55, 549-55