Data Mining Techniques for Pattern Recognition: Tornado Signatures in Doppler Weather Radar Data

Abstract

The US Doppler radar network installed software contains over 40 algorithms that sample the near-radar environment and generate diagnostic and predictive information. There is a smaller group of algorithms whose role is to alert forecasters of potentially severe weather. Of these algorithms, two particularly are powerful: the tornado detection algorithm (TDA) and the mesocyclone detection algorithm (MDA). These algorithms are based on empirically determined thresholds. The goal of this study is to uncover physically meaningful predictive patterns in weather radar data that alert to severe weather before the severe weather occurs.