GESTAR II Seminar Series, September 27th at 1:00pm

Satellite precipitation programs often emphasize the need for (a) validating products and (b) reporting uncertainties that are essential for the proper adoption of these products. This has resulted in a long list of validation efforts and campaigns dating back to the earliest VIS/IR products and generally compare in-situ precipitation to the equivalent satellite product for a specific mission. When these studies are viewed not as individual efforts or campaigns, but in their totality, however, it becomes fairly obvious that satellite products have very different validation statistics in different regions and time periods.  That is, satellite products have a large dependence on the atmospheric state that is neither random nor systematic but varies with meteorological regimes.  This is the direct result of the under-constrained nature of cloud and precipitation retrievals that must be dealt with to project validation results from one region or one regime to another. While sensors with greater information content and algorithms that more fully exploit all available information can ameliorate this problem, they cannot eliminate it. This talk will illustrate these points, and make a case for a new, state-dependent validation approach that can lead to credible predictions of products and their uncertainties in areas that lack conventional data.