Evaluating reanalysis datasets as meteorological input for estimating reference evapotranspiration in Africa and Southwest Asia: Twenty-first century hydrological challenges and opportunities in Africa

Recent developments of higher-resolution and lower-latency reanalysis data allow mapping reference evapotranspiration (ETo) over large areas in a near real-time manner. This study evaluates the ERA5, AgERA5 and GEOS5 reanalysis datasets for meteorological input in Africa and Southwest Asia by comparing between data products and with 174 in situ sites. The inter-comparison reveals non-stationary differences between datasets and highlights temporal inconsistencies in the GEOS5 data. When evaluated against in situ measurements, GEOS5 demonstrates lower accuracy compared with ERA5 and AgERA5. Additionally, while all datasets accurately estimate air temperature and pressure, they overestimate windspeed and solar radiation, and underestimate vapour pressure. The propagation of uncertainty estimates of ERA5 through the FAO56 ETo equation shows particularly high uncertainty in the tropics. This study emphasizes the importance of applying multiple uncertainty assessment methods for better-informed use of reanalysis data, especially in data-scarce regions.