Comparative Evaluation of Merging and Local Bias Correction for Radar-Gauge QPE

Improving the accuracy of quantitative precipitation estimates (QPE) directly enhances capabilities for flash flood forecasting and water prediction. Compared to radar only QPE approaches, radar-gauge merged QPE is able to provide an enhanced estimation in high-temporal resolution with both spatial continuity and local accuracy. We comparatively evaluate two radar-gauge QPE products from the real-time Multi-Radar Multi-Sensor (MRMS) system. The first is generated via conditional bias (CB)-penalized cokriging which merges hourly radar-only QPE and rain gauge data by minimizing the weighted sum of error variance and expectation of the Type-II CB squared. The second is from the local gauge correction (LGC) module in the operational MRMS. The LGC minimizes the interpolation errors of radar-gauge difference through an inverse distance weighting function. Evaluation is carried out for different types of precipitation events over the CONUS. For both methods, the MRMS radar-only QPE product and quality-controlled hourly gauge data are used as inputs. The resulting radar-gauge QPE products are compared against the daily CoCoRaHS gauge data for independent validation.