An evaluation of WSR-88D severe hail algorithms along the northeastern Gulf Coast

Software build 9.0 for the Weather Surveillance Radar-1988 Doppler (WSR-88D) contains several new or improved algorithms for detecting severe thunderstorms. The WSR-88D Operational Support Facility supports testing and optimization of these algorithms by local National Weather Service offices. This paper presents a new methodology for using Storm Data in these local evaluations. The methodology defines specific conditions a storm cell must meet to be included in the evaluation. These conditions include cell intensity and duration, population density along the cell track, and any previous severe reports in the county where the storm is located. These requirements avoid including storm cells that may have produced severe weather where reports would be very unlikely. The technique provides a more accurate picture of algorithm performance than if Storm Data is used with no special considerations. This study utilizes the new methodology with data currently available for the Tallahassee, Florida, county warning area (TLH CWA). It describes the performance of two algorithms used for detecting severe hail. The first is the Probability of Severe Hail (POSH), a component of the build 9.0 Hail Detection Algorithm. The second is the algorithm that calculates vertically integrated liquid (VIL). Early results show that the recommended POSH threshold of 50% appears appropriate for the TLH CWA. This suggests that the height of the freezing level provides a reasonably good estimate of the best severe hail index (SHI). However, early results also indicate that the average wet-bulb temperature from 1000 to 700 mb (low-level wet-bulb temperature) might produce an even better indication of the SHI threshold. Similarly, the threshold for VIL is highly correlated to the low-level wet-bulb temperature. Finally, the VIL algorithm is found to perform as well as the POSH parameter if the best VIL threshold can be determined in advance. Since the database used in these evaluations was relatively small, these findings should be considered tentative.