Categorical forecasts of convective events using upper air predictors

Abstract

AbstractAerological data from the European area and Czech ground data were used to derive predictor vectors that would be efficient for making categorical forecasts of convective events over the Czech Republic. A series of simulated forecasts were performed with sets of predictor vectors. The Critical Success Index (CSI) was chosen to be the measure of the forecast accuracy. The convective precipitation and thunderstorms that occurred during the periods <0000,0600>, <0600,1200> and <0000,1200> UTC were forecast with the use of the data from 0000 UTC. Likewise, the data from 1200 UTC were used to predict event occurrences during periods <1200,1800>, <1800,2400> and <1200,2400> UTC. The resultant maximum CSI values corresponding to the different predictands are discussed, and their dependence on the event frequency is described. The Heidke Score and the Equitable Score determined for the predictor vectors with maximum CSI are indicated. The diagnostic predictor vectors can be viewed as the representation of the prognostic information under the Perfect Prog assumption.