Key points are not available for this paper at this time.
Flood forecasting requires accurate estimates of the spatial and temporal rainfall rates over a catchment. Depending on the Z–R relationship and threshold values, processing of WSR-88D reflectivity data affects the rainfall estimates used in flood prediction. How much rainfall the catchment receives during an extreme event and the resulting hydrograph response depends on catchment characteristics that transform rainfall into runoff. A case study of an extreme meteorological and hydrologic event occurring 17–18 October 1994 in south Texas is examined. After the event, a review of the National Oceanic and Atmospheric Administration, National Weather Service (NWS) operations raised concerns over low estimates of accumulated rainfall by the WSR-88D radar compared to gauge accumulations. At the time of the review, the Z–R relationship in use during the event, Z = 300R1.4, and a 53-dBZ reflectivity threshold were suspected of having caused poor performance of the radar accumulation estimates. Because of default occultation parameters in use at the time, the WSR-88D precipitation processing subsystem used reflectivities from the 2.5° and 3.5° elevation angles out to 100 km, then 0.5° and 1.5° thereafter. The present analysis uses WSR-88D reflectivity solely from the lowest tilt at 0.5° over the study area, which lies within 23 km of the radar. Using a Geographic Information System, maps of radar-derived rainfall estimates are superimposed over the local scale of a catchment to evaluate the accuracy of this technique compared to rain gauges and catchment outflow for flood forecasting. In the authors’ recalculation of rainfall from WSR-88D reflectivity data, truncating reflectivities above 53 dBZ had no effect on catchment-averaged rainfall for this event and catchment investigated. The NWS Operational Support Facility has recommended the use of this tropical relationship under appropriate conditions since 1995. Applying the tropical Z–R relationship Z = 250R1.2 to the reflectivity data results in a better estimation of rainfall. Comparing slopes of the best fit regression lines of each Z–R relationship to daily rain gauge accumulation shows that Z = 300R1.4 estimates rainfall accumulations at less than 55% of that estimated by Z = 250R1.2. The underperformance using Z = 300R1.4 in our recalculation is consistent with reported radar accumulations of 38.1- versus 50.0-cm rain gauge accumulations, or 31% during the event using Z = 300R1.4. Using Z = 250R1.2 and comparing that resultant rainfall estimates to rain gauge amounts, daily rainfall accumulations were biased by −6% and −15%, for the 2-day period, respectively. The WSR-88D radar reflectivity can be used to accurately estimate rainfall for flood forecasting provided an appropriate Z–R relationship is used. Further improvements in operational flood forecasting can be made if rain gauge accumulations are used to calibrate the radar rainfall estimates.
Vieux et al. (Mon,) studied this question.