Title: Just how bad is the air we breathe? The Air Quality Forecasting System (AQF)
1Just how bad is the air we breathe?The Air
Quality Forecasting System (AQF)
- AQF developed by NOAA and the EPA,
- based on NCEPs WRF model and
- EPAs CMAQ modeling system
- 5x experimental (CONUS),
- 3x operational (eastern U.S.)
- 48-h forecasts, 06Z, 12Z, 1-h and 8-h ave,
- daily max, 1-h gt 125 ppb, 8-h gt 85 ppb
2(No Transcript)
3MDLs Evaluation of CMAQ Ozone and Aerosol
Forecasts
Jerry Gorline, Wilson Shaffer, Michael Schenk,
Valery Dagostaro, Arthur Taylor
MDL Air Quality Verification web site
http//slosh.nws.noaa.gov/aqverifUsername
RestrictedPassword Restricted
- 3x Eta vs. 3x WRF comparison, end Eta June 20,
2006 - Verify predictions over Houston area in Early
June 06
November 29, 2006
4Goals Air Quality Verification
- Compare 5x model performance on 3x domain to 3x
model. - Performance metrics
- H (Fraction Correct), TS, POD, FAR,
- MAE, ME/Bias.
- Use spatial maps to complement performance
measures. - Preliminary 1-h aerosol spatial maps and
statistics. - Caveat 5x Isoprene correction Aug. 4, 2006.
5Definitions of Statistical Terms
6Timing for 8-h 12Z Ozone Concentrations 2006
Season
Observations from Tanked BUFR files
Model Begins 12Z Day 1
Midnight Day 3
Midnight Day 2
04Z 00 16
11Z 07 47
12Z 08 24
05Z 01 17
04Z 00 40
03Z 23 39
UTC EDT EDT CMAQ Projection
24-h Obs Data 03Z 11Z Day3
24-h Obs Data 04Z 12Z Day2
The 11Z observed 8 hour average spans 03Z -
11Z. 12Z model output at projection 47 spans
projection 40 - 47.
The 12Z observed 8 hour average spans 04Z -
12Z. 12Z model output at projection 24 spans
projection 17 - 24.
24 Hr Model Data Proj 24 47
7Timing for 8-h 06Z Ozone Concentrations 2006
Season
Observations from Tanked BUFR files
Model Begins 06Z Day 1
Midnight Day 3
Midnight Day 2
UTC EDT EDT CMAQ Projection
21Z 17 39
04Z 00 46
05Z 01 23
22Z 18 16
24-h Obs Data 20Z 04Z Day2
24-h Obs Data 21Z Day1 - 05Z Day2
The 04Z observed 8 hour average spans 20Z - 04Z,
end Day 2 06Z model output at projection 46
spans projection 39 - 46.
The 05Z observed 8 hour average spans 21Z -
05Z. (First 8-h proj with Day 2 data) 06Z model
output at projection 23 spans projection 16 - 23.
24 Hr Model Data Proj 23 46
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 Preliminary Aerosol Verification
- Bar charts
- Scatter plots
- MAE/bias/rmse plots
- Spatial maps, daily 1-h max
- Daily statistics
33 34 35 36 37 38 39 40 41 42 43Comments, 2006
- Lower POD and FAR for 5x compared to 3x, lower
5x TS during high ozone episodes. - 5x MAE/bias lower than for 3x.
- 5x model on CONUS grid Consistent
under-prediction, especially in California. - 1-h aerosol results Under-prediction,
forecasted diurnal drop not observed in summer
but starting to show up in recent statistics.
44Case Study Surface ozone reduction associated
with thunderstorms, 2005
458-h observations for four stations in band of
predicted exceedances
46Surface observations for Tuesday, July 12, 2005,
2300 UTC