Personnel at the WSR-88D Operational Support Facility (OSF) and the National Severe Storms Laboratory (NSSL) analyzed 34 cases containing 2134 volume scans (approximately 194 hours of radar data) representing 168 tornadoes from many different areas of the United States. The goal of this project was to optimize performance of the WSR-88D (Weather Surveillance Radar - 1988, Doppler) build 10 tornado detection algorithm (TDA). Three dimensional plots of algorithm performance (probability of detection, false alarm ratio, and critical success index) as a function of adaptable parameter values illuminate opportunities for algorithm optimization.
The new tornado detection algorithm has a probability of detection much higher than the previous WSR-88D tornadic vortex signature (TVS) algorithm. In comparison, the new algorithm also detects more high shear, gate-to-gate regions in the atmosphere not associated with tornadoes. These “false” tornadic detections are mitigated by adjusting TDA adaptable parameters.
Several adaptable parameter sets, combination of adaptable parameter threshold values, were developed to be used in various meteorological situations such as isolated supercells, squall lines, and tropical storms. As the build 10 TDA is tuned to eliminate false alarms, probability of detection also decreases. Forecasters will have the capability to choose an adaptable parameter set based on type of storms expected.