Evaluation of operational weather radar rainfall estimates in flood-prone areas of California

The flood-prone river basins in many of the mountainous regions of California's coastal areas are believed to lack adequate operational radar coverage. This study evaluates quantitative precipitation estimation (QPE) from operational WSR-88D radar measurements in the vicinity of the Russian River basin, which sometimes floods during severe precipitation events caused by landfilling atmospheric rivers. A vertical profile of reflectivity (VPR) correction was devised for use with the WSR-88D data to improve QPE results. The QPE assessment was performed using multi-year observations from a ground-based site operated by the NOAA's Hydrometeorology Testbed (HMT) Program. HMT data included high resolution gauge measurements and S-band profiler observations, which provided high-resolution information on the vertical structure of precipitation. It is shown that the WSR-88 measurements detect most of the bright band (BB) rainfall, which contributed over half of the total precipitation. For this rain type the VPR corrected hourly QPE results showed relatively small mean absolute biases (<10-15%) and errors of about 50-60%. About one third of the total rainy hours with mostly shallow non-bright-band (NBB) rain, which generally was lighter than BB rain and provided approximately 15% of total precipitation, were not detected by the operational radar. The accuracy of radar-based QPE for the detected fraction of NBB rain was rather poor with biases of approximately -50% - -60% and characteristic errors around 80%. On some occasions, the radar provided false rain accumulations when detecting high clouds, which did not precipitate at the ground or coexisted with shallow rain (~ 6% of total accumulation). For heavier rainfall with significant total accumulations, which provided significant flood hazards, radar-based QPE results for precipitation event totals showed relatively good agreement with gauge data.