Verification of Near Real-time Estimates of Tornado Damage Probability

To date, operational tornado detection algorithms can be divided into those that retrieve circulation characteristics using radial velocity, and those that use dual-polarization variables to detect tornado debris signatures. However, no operational algorithm combines both into a single algorithm for tornado detection.

CoreLogic has developed a comprehensive tornado probability algorithm that combines patterns in radial velocity, dual-polarization data, and visual reports to generate realistic tornado probability paths within one hour of tornado occurrence. Using the National Weather Service Damage Assessment Toolkit, individual radar volumes were temporally matched with georeferenced damage surveys to create verification points.

This study will present an overview of the verification dataset which covers over 70 individual tornadoes and 600 radar volumes. Additionally, the performance of the final tornado algorithm will be explored using the verification dataset and spatial comparisons to the damage survey polygons.