Thunderstorms: Life cycle analyses and nowcasting based on multi-source data

Abstract

The nowcasting of the remaining lifetime of an existing thunderstorm can be improved by the use of lifetime characteristics. For this purpose parameters from geostationary satellite, ground-based radar, lightning, and model data are investigated. These parameters describe the microphysical state and the environmental conditions of the thunderstorm. A thunderstorm is defined using the Cb-TRAM algorithm. In total, more than 2000 thunderstorms over Germany are analyzed. The mathematical method fuzzy logic is used to combine the multi-source parameters. By adding the information of historical data, the remaining lifetime is then predicted. The nowcasting quality was determined by means of cross-validation. Compared to the extrapolation methods currently used, an increase in the probability of prediction can be obtained. Depending on the size of the tolerance range around the predicted value, the improvement varies between approx. 20% and 80%.