Blending multi-satellite, atmospheric reanalysis and gauge precipitation products to facilitate hydrological modelling

Satellite-retrieved and atmospheric reanalysis precipitation can bridge the spatiotemporal gaps of in-situ gauging networks, but estimation biases limit their reliable applications in hydrological monitoring modelling. To correct occurrence intensity simultaneously, this study develops a three-stage blending approach to integrate three multi-satellite datasets (IMERG Final, TMPA 3B42V7 PERSIANN-CDR), ERA5 product gauge dataset within dynamic framework. Firstly, systematic four members were individually corrected by combining local scaling ratio bias correction methods. Then, “state weights” used for determining wet/dry events optimized evaluating score function bias-corrected members. Thirdly, “intensity using cuckoo search (CS) algorithm Bayesian Model Averaging (BMA) method, respectively. The produced weights varying both spatially temporally, performance was thoroughly evaluated over mainland China. Results show that scheme performs better than individual two-stage methods terms all eight statistical metrics, CS outperforms BMA method third stage. By randomly sampling validation sites K-fold experiments, developed also demonstrates superior ungauged regions. After interpolating normalizing parameters gauges entire domain ordinary kriging, new blended with daily 0.25° scale produced. Four models are forced primary precipitations 238 catchments China, further confirming facilitate modelling demonstrated improving Kling-Gupta efficiency simulated streamflow 12–35%.