An Object-Based Tornado Debris Signature Detection Algorithm

An experimental version of the Hydrometeor Classification Algorithm (HCA) used in the WSR-88D radar network has been modified to detection the Tornado Debris Signature (TDS). The operational HCA is designed to provide a “best guess” of the primary source of scattering (e.g., rain, hail, snow, biological targets, etc.) within a radar volume. Within the HCA, each bin is processed independently of all other bins, which means that it is not possible to explicitly spread information regarding the spatial relationships and characteristics of different features (in fact, the idea of a “feature” does not exist in the HCA). This work presents a verification of the TDS-enabled HCA described in Snyder et al. (2015). Although the algorithm was originally designed with the goal of minimizing false alarms, data quality deficiencies, radar artifacts, and the gate-by-gate nature of the HCA results in a large number of false alarms for TDS events, though subjective assessments of these events by those familiar with radar observations of convective storms make apparent that such detections are relatively easy to reject. An object-based algorithm has being developed in an effort to get around some of the restrictions of the operational HCA. The object-based algorithm constrains detections on clusters of range gates classified as TDS using several criteria (including size, aspect ratio, and maximum aggregation value), and it allows the data to be easily viewed atop other radar fields. This object-based algorithm, by using a set of constraints, has a significantly lower false alarm rate and may prove more useful to operational meteorologists.