The performance of a high-resolution satellite-derived precipitation product over the topographically complex landscape of Eswatini

Abstract

The study evaluated the use of Climate Hazard Group InfraRed Precipitation with Stations (CHIRPS) data for monitoring rainfall data in Eswatini. Various statistical metrics such as Bias, correlation coefficient (r), mean absolute error (MAE) and root mean square error (RMSE) were used to evaluate the CHIRPS 2.0 data against 14 rain gauge observations acquired during 1981–2020. CHIRPS 2.0 rainfall agrees well with rain gauge precipitation at monthly (r = 0.73, Bias = 1.02, RMSE = 50.44 and MAD = 31.44), seasonal (r = 0.77, Bias = 1.01, RMSE = 36.99 and MAD = 24.15) and annual scales (r = 0.65, Bias = 2.46, RMSE = 500.78 and MAD = 468.06). Moreover, areas characterized by complex topography and land use, and areas in transition zones (to a different agroecological zone) had generally poor correlations. Nonetheless, CHIRPS 2.0 captures well the spatial distribution of rainfall in the different agroecological zones of Eswatini, even in areas with no rain gauge data. In conclusion, CHIRPS 2.0 can be a very valuable tool in filling gaps created by poor spatial coverage of ground-based rain gauges, especially in the developing world where this is often the case.