Generating Multi-Sensor Precipitation Estimates Over Radar Gap Areas

Generating a multi-sensor precipitation product over radar gap area is the objective of the present study. A merging approach is developed to improve Satellite-based Precipitation Estimates (SPE) by merging with ground-based Radar Rainfall (RR) estimates because remote satellites are the only source that can collect information from areas where are inaccessible by ground-based radar and/or rain gauge networks. The merging algorithm is capable of extending radar information from pixels with available RR to their neighboring pixels with no radar information by merging RR with SPE, which is, usually, available for all pixels. SPE is combined with RR using the weighting-based approach of Successive Correction Method (SCM) after local bias correction of SPE with respect to RR. High resolution satellite infrared-based rainfall estimates from the NESDIS Hydro Estimator algorithm (HE), at hourly 4 km × 4 km basis, is selected to be merged with radarbased NEXRAD Stage IV rainfall measurements to generate rainfall product for the radar gap areas. To be able to validate the generated rainfall against NEXRAD, different size areas with available radar rainfall are selected as radar gap regions. The developed merging technique is evaluated for several study cases in summer 2003 and 2004. The results show that generated rainfall for the radar gap areas are more correlated with RR (average 0.67) than original HE with RR (average 0.36) and the RMSE between merged and radar rainfall (average 2.8 mm) is less than the RMSE between satellite and radar rainfall (average 4.48 mm). And also, the pattern and intensity of the generated rainfall for radar gap area became more similar to the pattern and value of RR. In addition, the enhancement of the generated rainfall with respect to RR is more significant for high rainfall amounts. Key-Words: Merging, Radar, Gap area, Precipitation, SCM, Satellite, Rainfall.