AQMOS � Air Quality Model Output Statistics
Beta Air Quality Model Output Statistics Beta
Sonoma Technology Inc.
Home  Verification  Contact Us  FAQ  Tutorial  Sign in 

Frequently Asked Questions

  1. How do I get an account for AQMOS?
  2. What are the data sources for AQMOS?
  3. How does AQMOS work?
  4. How can I view the predictions for my area?
  5. What is the format of the output screen?
  6. Can I see historical AQMOS predictions?
1) How do I get an account for AQMOS?
AQMOS accounts are not available at this time.
Back to Top
2) What are the data sources for AQMOS?
AQMOS uses numerical model data and observations. Current- and next-day numerical model predictions come from the 6Z and 12Z runs of the NOAA air quality model ( and the 0Z run of the experimental BlueSky Gateway air quality model ( The NOAA model provides peak daily 8-hr ozone predictions and the BlueSky Gateway model provides peak daily 8-hr ozone and daily 24-hr PM2.5 predictions. Observed peak daily air quality concentrations are obtained from the AIRNow Gateway system (
Back to Top
3) How does AQMOS work?
AQMOS keeps a running database of historical model predictions and observations to calculate a regression equation to determine the model bias. Each day, AQMOS uses this database to compute a regression equation for each city, parameter, model, forecast period, and season. This regression equation is then applied to the current model predictions to get bias-corrected model prediction for today and tomorrow. For example, if the NOAA 12Z ozone predictions are always higher than the observed concentrations, AQMOS will adjust the model prediction down. If the model average error is zero, then there will be no adjustment to the model prediction.

To better capture model performance on high pollution days (which are of greatest concern to forecasters), AQMOS calculates the regression in one of two ways, depending on whether the model prediction for a city is above or below a threshold value:
  1. When the prediction is above the city-specific threshold (e.g., 70 ppb for 8-hr ozone), AQMOS limits the historical data used to calculate the regression to only days when the model previously predicted above the threshold. A minimum of 5 historical days above the threshold are needed to calculate the regression.
  2. For areas without sufficient days above the threshold and for all areas with model predictions below the threshold, all historical model and observed data for the season will be used in the regression calculation.

For more detailed information on the methodology behind AQMOS, please watch the Tutorial video.
Back to Top
4) How can I view the predictions for my area?
After logging in, use the drop-down boxes to filter results based on a particular air quality model, state, reporting area (city), and/or pollutant. Any combination of the filters may be used. If no filter is used, all results will be shown. Press “Go” to see the predictions.
Back to Top
5) What is the format of the output screen?
Each row includes the Model, Model initialization time, Forecast Area, and Pollutant followed by a data grid: the first three columns of the grid show information for the current-day predictions while the last three columns show the next-day predictions. For each day, the Model column shows the original numerical model-predicted pollutant concentration, color-coded by the corresponding AQI category. Ozone predictions are shown in parts per billion (ppb) and PM2.5 predictions are shown in micrograms per cubic meter (µg/m3). The Model Corrected column shows the bias-corrected predicted concentration produced by AQMOS, also color-coded by AQI category. The Image column shows a graphic display of the original model predictions when the image icon is clicked on.
Back to Top
6) Can I see historical AQMOS predictions?
Yes, archived predictions are available back to April 1, 2009. To view archived predictions, type the desired date (mm/dd/yyyy) into the editable textbox (i.e., the second textbox) at the top of the page and press “Go” or click on the calendar icon and select the desired date. The current date is shown by default.
Back to Top