Testing capability of successive correction method and Bayesian spatial model to fill gaps over the radar network

Radar rainfall estimates are critical input products in a distributed hydrologic prediction and flash-flood forecast models. In this study, capability of using the Successive Correction Method (SCM) with a Bayesian spatial model to produce a multi-source rainfall data by combining radar, satellite, rain-gauge and PRISM (Parameter-elevation Regressions on Independent Slopes Model) products over an artificially created radar gap in Oklahoma is evaluated. Real gap areas over radar network cannot be used as a test-bed due to lack of availability of radar rainfall required for validation of generated multi-sources product. In mountainous regions, radars suffer from signal blockage which leads to gaps in the radar rainfall field. Even though satellite rainfall products are available without geographic limitations, their accuracy over land is questionable. Since mountainous regions are hydrologically sensitive areas for prediction, the need for a rainfall product with better accuracy than satellite-based products is critical. Daily and 4 X 4 km satellite, radar, rain-gauge, and monthly PRISM precipitation products for the year 2006 were used for this study. Rainfall products from satellite IR based Hydro-Estimator and radar Stage-II are merged using the SCM so that the artificially created gap over the radar network could be filled. The satellite-radar product from SCM is further combined with rain-gauge and climatological PRISM precipitation products in a Bayesian spatial model. The satellite-radar-gauge-PRISM combined precipitation product is evaluated using three evaluation criteria: coefficient of correlation, absolute bias and Nash-Sutcliff efficiency. Generated multi-source rainfall product using this method produced a better product than Hydro-Estimator when it is evaluated using independent rain-gauges. The present study implies that using the available radar pixels surrounding the gap area, rain-gauge, PRISM and satellite products, a radar like product is achievable over radar gap areas that benefits and has huge impacts on hydrological simulations and prediction purposes.