Evaluation of Radar-derived Polarimetric Precipitation Estimates for Extreme Cases Using a Dense Network of Rain Gauge Data

Radar-derived quantitative precipitation estimation (QPE) has become a key factor for many meteorological and hydrological analyses and applications. In hydrology, radar-based QPE products are commonly used as main input to rainfall-runoff models to simulate and predict streamflow, and their errors significantly contribute to the uncertainty in streamflow simulations. The QPE method for the U.S. Weather Surveillance Radar-1988 Doppler (WSR-88D) radars has evolved along with their hardware upgrades (e.g., polarimetry) since their deployment in the early 1990s. Many QPE studies during the last decade have focused on utilizing advantages of polarimetric observations and assessing a variety of approaches. The results from those studies demonstrated the improved accuracy and potential of polarimetric estimates in different conditions and cases. In this study, the authors use ground observations from a dense network of 171 rain gauges in Kansas City and evaluate precipitation estimates derived from two polarimetric QPE algorithms based on specific differential phase and specific attenuation for recent extreme rainfall cases observed from radars in Missouri and Kansas. The performance of these estimates is also compared with that of the conventional algorithms derived from radar reflectivity. This comparison demonstrates that the two polarimetric algorithms outperform the reflectivity based one. The authors also examine estimation performance of the QPE algorithms in terms of the spatial variability and other relevant factors (e.g., presence of hail) for each extreme storm case.