The PQPF data represent high resolution (1/8 degree) probabilistic forecasts at various thresholds. These forecasts are based on the GEFS Reforecast V2 and Climatology-Calibrated Precipitation Analysis datasets. Raw (and frequently biased) ensemble forecasts from the GEFS are transformed into reliable predictive probability distributions for various precipitation accumulations. A nonhomogeneous regression model, with regression parameters fit to training data, is used to derive probabilistic forecasts that are sufficiently reliable at various lead times, and have better resolution than forecasts obtained by state-of-the-art analog approaches. This is especially true for forecasts of extreme events, where analog methods can struggle.

The Extreme Forecast Index (EFI) is an index designed to identify situations where the GEFS mean forecast is predicting an extreme solution relative to a reference climatology. The reforecast dataset is used to provide the context in how usual a situation is. The EFI is currently being produced for temperature, 10-m winds, and accumulated precipitation. The EFI will scale from 0 to 100. The percentile shown represents where the current forecast value falls within quantiles created from the reforecast database climatology. Values toward 0 and 100 are very near or beyond the reference climate distribution. The reference climatology for a given month includes all forecasts from that month and the two surrounding months.

How to use the table:  
Click on a value (e.g., 92) to plot an image for that valid time and parameter. Click on a column header (e.g., P) to loop images for that parameter at all forecast hours.

Availability:  
These forecast products will usually, but not always, be updated around 16Z each day. This is an experimental forecast product and may not always be available (No after hours or weekend support).

Considerations:  

• GEFS is under-dispersive
• EFI temperatures are significant near 0 and 100 (cold/warm). The EFI "T" column will only show whichever percentile (cold/warm) is most significant.
• A '-' symbol in the table indicates that data is not available for that time period.

We will do our best keep this page up updated, but please be aware that this is not operationally supported (no weekend support).

For more information:  
ESRL Hi-Res PQPF based on GEFS Reforecasts and CCPA
Extreme Forecast Index (EFI) based on GEFS Reforecasts

Description of Extreme Forecast Index (EFI)

STID Forecast Confidence Toolkit

If you do notice any issues or would like to provide feedback to improve this page, please contact the page contributors (see below).

Page Contributors:
Chad Kahler (Western Region STID)
Mike Staudenmaier (Western Region STID)

Data Provided By:
Tom Hamill (ESRL PSD)
Michael Scheuerer (CIRES)
Gary Bates (CIRES)