WSR-88D radar, tornado warnings, and tornado casualties

The impact of the installation of Weather Surveillance Radar-1988 Doppler (WSR-88D) radars in the 1990s on the quality of tornado warnings and occurrence of tornado casualties is examined. This analysis employs a dataset of tornadoes in the contiguous United States between 1986 and 1999. The date of WSR-88D radar installation in each National Weather Service Weather Forecast Office is used to divide the sample. Tornado warnings improved after the installation of Doppler radar; the percentage of tornadoes warned for increased from 35% before WSR-88D installation to 60% after installation while the mean lead time on warnings increased from 5.3 to 9.5 min and the false alarm ratio fell slightly. A regression analysis of tornado casualties, which controls for the characteristics of a tornado and its path, reveals that expected fatalities and expected injuries were 459% and 40% lower for tornadoes occurring after WSR-88D radar was installed in the NWS Weather Forecast Office. This analysis also finds that expected casualties are significantly lower for tornadoes occurring during the day or evening than late at night throughout the sample, which provides indirect evidence of the life-saving effects of tornado warnings.