A Self-Calibrating Blended Satellite Algorithm for Estimating Heavy Precipitation

The availability of accurate and timely estimates of rainfall is crucial for accurate diagnosis and prediction of flooding. However, raingauges and radar both have limits in their ability to provide the necessary information. Raingauges provide insufficient sampling in space and time, while radar estimates have superior sampling but are constrained in some regions by range effects and beam block.

Rainfall estimates from satellite data can supplement these sources of precipitation information. The Geostationary Operational Environmental Satellite (GOES) offers high-resolution, uniform coverage over the continental United States and the surrounding waters, allowing rainfall estimates to be made for regions where raingauge and/or radar data are not available, such as flash flood situations in regions of complex terrain and heavy rainfall from tropical systems approaching land.

Numerous algorithms for estimating rainfall from satellite data have been developed during the past quarter century, but the highly variable relationship between satellite radiances and surface rainfall rates has limited their effectiveness. In order to improve the quality of GOES-based rainfall estimates, a technique has been developed in which two available estimates (the Auto-Estimator and the GOES Multi-Spectral Rainfall Algorithm) and the GOES brightness temperature data are used to produce a new estimate that is calibrated against radar and raingauge data in real time. This allows the new estimates to reflect regional and seasonal variations in the relationship between brightness temperature fields and surface rainfall rates. The usefulness of this approach will be demonstrated using several case studies of heavy rainfall/flash flooding events.