Evaluation of fine resolution gridded rainfall datasets over a dense network of rain gauges in Niger

This study evaluates the reliability of gridded rainfall products with fine spatial resolution using 29 gauges of a dense observation network in Niger. Four gridded datasets were considered, viz. the African Rainfall Climatology version 2.0 (ARC2), Climate Hazards Group InfraRed Precipitation with Stations (CHIRPS), Climate Prediction Center morphing technique (CMORPH) and Tropical Applications of Meteorology using Satellite data and ground-based observations (TAMSAT). The datasets were quantitatively compared to the gauge rainfall at different timescales and their rainfall detection skills were assessed using four validation metrics and three contingency indices, respectively. Furthermore, their potentials for trend analysis was assessed using a modified Mann-Kendall test; and finally, their capability of capturing damaging rainfalls was evaluated using data of five historical damaging rainfall amounts on a daily basis. The CMORPH product showed a good performance in detecting rainfall events and exhibited the best linear relationship with the gauge rainfall; however, it displayed high errors and relative bias (RB > 100%)-showing its inaccuracy in rainfall quantification. At daily, dekadal and monthly timescale, CHIRPS showed the lowest errors, followed by the ARC2 and TAMSAT, whereas ARC2 exhibited the lowest RB-between-12% and 22%. All of the four datasets were unable to unveil the significant trends found in ground observations and were inaccurate in sensing the depths of damaging rainfall amounts. The overall results revealed that the 3-hourly CMORPH dataset is a good sub-daily rainfall product-whereas daily, dekadal and monthly CHIRPS and ARC2 rainfalls have good potentials for hydro-climatological applications in Niger.