Tracking and nowcasting of convective cells using remote sensing data from radar and satellite

At the Deutsches Zentrum fur Luft- und Raumfahrt (DLR) an algorithm named "Cb-TRAM" has been developed to identify, track, and nowcast thunderstorm clouds in different development stages (ZINNER et al., 2008). The tracking of detected clouds is based on the so-called "pyramidal image matcher". Applying this tracking algorithm to radar data resulted in the development of the radar tracker Rad-TRAM. Rad-TRAM is introduced here and used in parallel with Cb-TRAM for three cases of thunderstorm occurrence in Central Europe. The purpose of the study is threefold. Firstly, to test the ability of both trackers to track these features. Secondly. to compare position and tracks of cloud and radar cells in a visual and a statistical analysis by overlaying cell structures of both systems. Thirdly, to test the nowcasting performance of both trackers against extrapolations based on Lagrangian persistence. It is found that both trackers are able to detect and track the thunderstorms. For all observation times a percentage of about 70 % of the satellite detected clouds in development stage "mature" overlap with radar cells when applying a minimum overlap criterion, i.e. one pixel of both cell types. Furthermore, it is found that Cb-TRAM as well as Rad-TRAM nowcasts for 15, 30, 45, and 60 minutes outperform extrapolations based on persistence. The results are discussed taking into account the different data bases used and specific thresholds in both algorithms.