ESRL/PSD analog probabilistic precipitation forecasts
Many forecast users desire reliable, skillful high-resolution ensemble predictions, perhaps for such applications as probabilistic quantitative precipitation forecasting or hydrologic applications. Our reforecast dataset is comparatively low resolution (T62, or about 250 km). However, by downscaling the forecasts through analog techniques a high-resolution probabilistic forecast can be produced.
The basic idea is this: if we have a long time series of high-resolution analyses, then we can examine today ensemble forecast, look back to our reforecasts and find days in the past where the old forecasts were similar to the current forecast, and note the analyzed conditions associated with those forecasts. With knowledge of the dates of the similar forecasts, we can collect an ensemble of high-resolution analyzed conditions. The precipitation analyses used for this procedure are the 32-km grids from the North American Regional Reanalysis , downscaled to 5-km resolution using the 'mountain-mapper' technique .
ESRL/PSD probabilistic precipitation forecast maps for the Lower Colorado River Basin can be accessed here and for the state of Colorado here.