Global Precipitation Estimate Error Decomposition Analysis for 14 Passive Microwave Sensors

Abstract

Previous studies investigated the satellite precipitation error characteristics via error decomposition analysis over several land regions. This study, for the first time, decomposes the total error of precipitation estimates from 14 passive microwave (PMW) sensors in the Global Precipitation Measurement (GPM) PMW radiometer constellation into four independent error components (i.e., hit-positive, hit-negative, miss, and false) over both land and ocean, using GPM radar and radiometer Combined Precipitation rates as the reference. Results reveal that over ocean, cross-track scanning sensors exhibit larger miss and false error components than conical scanning sensors due to their lack of low-frequency channels and coarser spatial resolution. Over land, the dominant error component is the hit-negative (i.e., underestimation) error component for all sensors, which is particularly evident for the heavy precipitation intensities (>8 mm hr⁻¹). Geospatial distribution analysis reveals that the error components show clear regional dependence over both land and ocean.