MERGING RADAR-ONLY QPE AND RAIN GAUGE DATA VIA CONDITIONAL BIAS-PENALIZED OPTIMAL ESTIMATION
MetadataShow full item record
A new technique for merging radar precipitation estimates and rain gauge data is developed and evaluated to improve multisensor quantitative precipitation estimation (QPE). Various types of linear and nonlinear techniques have been used to combine rain gauge and radar data. Linear cokriging and its variants, for example, may be considered as the best linear unbiased estimators which minimize the error variance in the unconditional sense. They are, however, subject to conditional biases (CB) that may be unacceptably large for estimation of heavy-to-extreme precipitation. In this work, I develop, apply, and evaluate conditional bias-penalized cokriging (CBPCK) for spatial estimation of precipitation using weather radar and rain gauge data which explicitly minimizes Type-II CB. The proposed CBPCK is a bivariate version of extended conditional bias-penalized kriging (ECBPK) which was developed for gauge-only estimation of heavy-to-extreme precipitation. To evaluate the proposed method, CBPCK is comparatively evaluated with a variant of ordinary cokriging (OCK), the currently used algorithm in NWS’s Multisensor Precipitation Estimator (MPE), via cross validation and visual examination of merged fields. The analysis domain is about 560 x 560 〖km〗^2 in the North Central Texas region and the analysis period is from 2002 to 2011. The radar data used is from the reanalysis of the radar-only National Mosaic and multisensor QPE (NMQ/Q2). The rain gauge data used is from the Hydrometeorological Automated Data System (HADS). The results show that CBPCK significantly reduce CB for estimation of heavy-to-extreme precipitation at subdaily scales of accumulation, and that the margin of improvement over OCK is larger when the fractional coverage of rainfall is high, i.e., when it is precipitation over most of the area over the ungauged location. CBPCK may be used in reanalysis or in real-time analysis for which accurate estimation of heavy-to-extreme precipitation is of particular importance.