A Comparison of Three Radar-Based Severe-Storm-Detection Algorithms on Colorado High Plains Thunderstorms

Abstract

Volumetric reflectivity data collected with the National Center for Atmospheric Research's (NCAR) 10-cm Doppler radar during the Program for Regional Observing and Forecasting Services (PROFS) 1985 Real-Time Forecast Exercise (RT-85) have been used to evaluate the Next Generation Weather Radar (NEXRAD) hail algorithm, a modified version of the Radar Data Processor (RADAP) II-based Severe Weather Probability (SWP) algorithm, tuned for Colorado High Plains thunderstorms, and a Precipitation Type/Intensity (PTI) hail detection algorithm that uses dual-polarization of echo return. Comparative analyses of algorithm performance are presented using two statistical scoring techniques. The results are consistent for the two methods. The SWP algorithm exhibited the best performance when a 0.10 probability threshold was used for the hail/no hail decision criterion. Probability of detection, false alarm rate and critical success index scores for the SWP algorithm were 0.58, 0.71 and 0.23, respectively. Application of signal detection theory (SDT) analysis indicated SWP to be statistically better at forecasting hail than either the PTI or NEXRAD algorithms at the 0.95 confidence level. A discussion of SDT and its application to severe weather forecast verification and the sensitivity of SDT to sample selection is presented.