Close Window

North Carolina State University: "Detection of tornadic thunderstorms (Part 2)"

Final Report

This project is a continuation of the earlier work in which NWS forecasters were trained in the use of the Detection of Tornadic Thunderstorm (DOTT) technique developed by NCSU. The focus on this phase of the research was to address real-time situations by: 1) testing the DOTT shortly after a severe weather episode in order to compare conventional analysis with DOTT, and 2) extending the study to examine active convective situations which do not develop tornadoes. The real-time experiment ultimately included two tornado outbreak days (April 28 and May 11, 1992) and five non-tornado days. The tornadoes occurred in Oklahoma and Texas, and the parent thunderstorms were, in most cases, producers of enormously large hail (golf ball, baseball, to grapefruit size). The DOTT technique applied to the April 28 data identified 3 cells to be tornadic. In actuality, 2 of these were tornadic and 1 was a false alarm. Each of the remaining 12 cells were correctly identified as non-tornadic by the technique. Expected loss values (obtained from ordinal regression analyses) for a given cell indicated expected losses to be due to strong tornadoes for several cells, while field reports indicated only weak tornadoes or none at all, in the case of one cell. The investigators theorize that the weighting scheme may need to be adjusted when it is suspected that a cell is an HP supercell characterized by large hailstones. The analysis for May 11 yielded eight correct and one incorrect nowcasts. As for the nowcast intensities, mixed results were obtained for this case as well. The lead times provided by the DOTT for both cases were also varied, depending on the availability of images. If RISOP data were available at 5 minute intervals, the DOTT technique could be very useful in alerting WSR-88D radars as to which cells should receive close monitoring during a severe storm outbreak.