Climatological Applications of Daily Practically Perfect Severe Weather Hindcasts

Climatological distributions of U.S. tornado, large hail, and damaging convective wind gust occurrence are relatively well known despite inherent problems with the severe weather reporting database. This study presents and examines a modern climatology of U.S. tornado, hail, and convective wind gust report frequency using a kernel density estimate in a framework known as practically perfect hindcasts. Through the context of daily kernel density surfaces of severe report probabilities, one can systematically examine the spatiotemporal dimensions of severe weather risk. This work provides a climatology of these kernel density surfaces at various, well-established, probability thresholds used by the Storm Prediction Center. These results affirm previous research on the topic, as well as add information about event clustering and spatial footprints. In addition, explorations of major events, identified by their spatial and temporal extremes, are discussed. Finally, future directions and applications of these data are presented. This includes pairing of the kernel density estimates with environmental information to create machine learning models for various forecasting applications.