Using Real-time Gauge Observations to Improve Satellite Precipitation Estimation: A New Fusion Method

Real-time precipitation data with high spatial and temporal resolutions play a crucial role in improving hydrological forecasting. In order to improve the spatiotemporal resolutions of satellite precipitation data, a new precipitation fusion method that combine the geographically weighted regression method (GWR) and mixed geographically weighted regression method (MGWR) with four weighting functions (Threshold (TH), Inverse distance (IDS), Gauss (GAU), BI-square (BI)) were formulated in this study. The proposed fusion schemes were further tested for fusing the gauge precipitation and CMORPH (CPC MORPHing technique) precipitation product in the Ziwuhe Basin of China. Our analysis shows that the combination of mixed geographically weighted regression and BI-square function (MGWR-BI algorithm) achieves the best fusion performance. The hourly 1 km precipitation produced by the MGWR-BI algorithm was much more accurate than the original satellite precipitation data. The developed fusion method not only improves the spatial resolution but also enhances the data quality, which is important and valuable for hydrological modeling and other applications.