Navigating the NOAA/ESRL TEXAQS-II Model Verification Web Site.

Overview:

The purpose of the NOAA/Earth Systems Research Laboratory (ESRL) TEXAQS-II web site is to provide in real-time an evaluation of the performance of air quality models that are providing forecasts during the field program. The web site is updated on an hourly basis. As soon as a new model simulation is available, the time series of the entire simulation is plotted. Observations are then added in real-time each hourly hour.

An emphasis of the verification is to include as much information as possible on the meteorological skill of the models, as changes in the meteorology are largely responsible for the day-to-day variations in ozone and PM. A unique aspect of the evaluation is that it also includes ensemble forecasts of ozone and PM2.5 that are created from the mean of all of the models, as well as bias-corrected ensemble ozone and PM2.5 forecasts.

Models:

The choice of models is made by clicking on any of a row of grey boxes that runs across the top of the web page. At the present time, eight models are available on the web site:

• AURAMS, using GEM/ADOM-II, provided by the Meteorological Service of Canada at 28 km horizontal resolution.
• CHRONOS, using GEM/ADOM-II, also provided by the Meteorological Service of Canada, at 21 km horizontal resolution.
• NMM/CMAQ, provided by NOAA/NWS at 12 km resolution.
• WRF-36, the WRF-Chem model using ARW/RADM2, provided by NOAA/ESRL.
• WRF-12, the same model but at 12 km resolution.
• BAMS-15, using MM5/CBM-4, provided by Baron AMS at 15 km resolution.
• BAMS-5, the same model at 5 km resolution.
• STEM-12, using MM5/SAPRAC-99, provided by the University of Iowa at 12 km resolution.

Model forecasts are initialized at different times for different models, and can be 00, 06, 12 or 18z. Some models provide one forecast per day, while others provide two, initialized at different times. After clicking on any model box at the top of the page, the available model initialization times for the model become visible in the top box on the left side of the web page. If two initialization times are visible, click on the one desired.

Observation Sites/Variables:

There are two distinct types of observation sites, Profiler and Chemistry. The Profiler sites focus on meteorology, with surface ozone included at a few of the sites, as well as model vertical time-height ozone cross-sections that can be compared to the profiler meteorology. The Chemistry sites focus on surface chemistry, but also include surface meteorology.

After clicking on either the Profiler or Chemistry button on the left of the page, a drop-down list of sites appears directly below. There are 11 profiler sites, and 14 Chemistry sites.

Not all of the same variables are available at all sites. If either an observation or a model value is missing, it either will simply not be included in the plot, or the entire plot may say “No Data Available”.

The variables listed below can be obtained by clicking on the row of blue boxes that are present directly below the model names boxes near the top of the web page.

Profiler Sites

• Wind profiles (wind barbs with SNR color background that indicates the presence of precipitation)
• RASS temperature profiles (Color background, with high resolution wind barbs)
• Boundary layer depth (black dots shown on both of above plots)
• Ozone profiles (model values only)
• Surface winds, temperature, humidity, pressure, solar and net radiation, precipitation. (Large open symbols are hourly model values, small plus signs are 5 min observations)
• Surface ozone (Open symbol is model value, plus is hourly observation).

Chemistry Sites (CAMS)

• Surface Meteorology (wind speed and direction, temperature, humidity, pressure, precip, solar and net radiation)
• Surface Chemistry (CO, O3, PM2.5, NOx, SO2).
• Ensemble O3
• Ensemble PM2.5

The ozone and PM2.5 ensembles are created by calculating the mean of all of the model predictions that are available for each hour. Information that is contained on the ensemble plots includes:

• Models included in ensemble (names shown in blue at top of plot are included)
• Models included in ensemble at each hour (straight horizontal lines under plot)
• Ensemble mean (thick blue line)
• Max and min ensemble member (thin blue lines)
• Standard deviation of the ensemble members (vertical blue lines)
• Observed ozone or PM2.5 (red dots)

Below the ensemble plot is a similar plot, but for the bias-corrected ensemble. Two types of bias-correction are included. Information that is contained on the bias-corrected ensemble plots includes:

• 7-day bias corrected forecast (solid blue line). Here the ensemble forecast is corrected using the mean bias of each model found over the previous 7 days.
• Kalman Filter forecast (green line). Here a Kalman Filter is used to calculate the bias over the previous 7 days. The Kalman filter is based on an adaptive linear recursive algorithm and estimates the systematic component of forecast errors.
• Number of days of model data that were actually available over the previous 7 days for inclusion in the bias calculation. Warmer colors indicate more days have been available.


Data Archive:

A calendar can be found on the left side of the web page. Data for any day can be obtained by clicking on the day desired. The month can be changed using the blue arrows at the top of the calendar.


References:

Delle Monache, L., X. Deng, Y. Zhou, and R. Stull (2006): Ozone ensemble forecasts. Part I: a new ensemble design. Accepted in J. Geophys. Res.

Delle Monache, L., T. Nippen, X. Deng, Y. Zhou, and R. Stull (2006): Ozone ensemble forecasts. Part II: a Kalman-filter predictor bias correction. Accepted in J. Geophys. Res.

Wilczak, J., S. McKeen, I. Djalalova, G. Grell , S. Peckham, W. Gong, V. Bouchet, R. Moffet, J. McHenry, J. McQueen, P. Lee, Y. Tang, G. R. Carmichael (2006): Bias-corrected ensemble and probabilistic forecasts of surface ozone over eastern North America during the summer of 2004. Submitted to J. Geophys. Res.