A Data Fusion System for GPM Dual-frequency Radar Rainfall Estimation

The Ku and Ka band Dual-frequency Precipitation Radar (DPR) onboard the Global Precipitation Measurement (GPM) satellite Core Observatory is capable to provide detailed information of precipitation, including the precipitation types, size distributions, and precipitation intensity. The DPR measurements and products play an important role in understanding extreme weather events and the water circle on a global scale. This paper presents an innovative rainfall estimation system for GPM DPR by combining the dual-frequency observations. The critical design of this system includes two machine learning models, connecting rain gauges and ground radar, and then ground radar and DPR. In particular, the first model trains ground radar data with rainfall measurements from gauges to create ground radar-based rainfall estimation. The second model uses the ground radar-derived product to train DPR Ku and Ka band data together in order to obtain dual-frequency-based rainfall product. For demonstration purposes, ground radar data collected from the NEXRAD radar located in Melbourne, Florida are used. The ground radar data are processed onto Constant Altitude Plan Position Indicator (CAPPI) grids at multiple vertical levels with a spatial resolution of 1km by 1km. Rainfall measurements from 178 rain gauge stations in this area are gathered and considered ground references. In addition, GPM DPR data from coincident overpasses in the same area are aligned to match ground radar data using the methodology proposed by Bolen and Chandrasekar (2003). The derived dual frequency rainfall product is compared against the traditional single frequency-based estimates, which shows great potential of the dual-frequency measurements in radar rainfall estimation.