Evaluating Rainfall Timing and Volume of Gridded Precipitation Products

Abstract

The accurate estimation of precipitation is important for hydro-meteorological applications. The evaluation of Gridded Precipitation Products (GPPs) must account for the zero-inflated nature of precipitation data and differences in spatial support between rain gauges and satellite grids. Traditional validation approaches in East Africa evaluate datasets in a temporally rigid manner, penalising them for minor spatial or timing displacements. This study evaluates the phase alignment, event detection, and precipitation volume of ERA5, IMERG, CHIRPS, and TAMSAT against the Trans-African Hydro-Meteorological Observatory (TAHMO) network across East Africa. First, we apply variance-stabilising transformations to address skewness and zero inflation in daily precipitation. Second, we utilise consecutive overlapping temporal windows and temporal neighbourhood verification to isolate timing accuracy from volume. Finally, we cluster raw diurnal precipitation cycles into regimes. Our results reveal that temporal phase misalignments are localised. At the diurnal scale, the GPPs and TAHMO assign different regimes at 58.1% of the stations. At the daily scale, IMERG exhibits superior phase alignment and event detection, whereas CHIRPS and TAMSAT preserve precipitation volume more accurately.