Verification of ERA5 and ERA-Interim precipitation over Africa at intra-annual and interannual timescales

Abstract

Over data sparse regions such as Africa, reanalysis datasets can provide a useful alternative to enable climate investigations across a range of temporal scales. Here, a detailed comparison of ERA5 and ERA-Interim precipitation over Africa is done at annual and seasonal temporal scales, over various subregions, to identify biases. Additionally, the representation of interannual variability in the reanalyses is evaluated with an emphasis on the ENSO teleconnection to Africa. Overall, there have been several substantial improvements in ERA5 compared to ERA-Interim. ERA-Interim's wet bias in both the annual and seasonal rainfall over tropical Africa has been substantially reduced in ERA5, which is also evident in the area-averaged bias over Africa (ERA5 = 39.097, ERA-Interim = 91.734). ERA5's representation of the annual cycle has improved over most selected subregions, notably over the southwestern Cape (rERA5=0.993, rERAI=0.988) and Cape South Coast (rERA5=0.917, rERAI=0.859) of South Africa, as well as over the Southern Tropics (rERA5=0.998, rERAI=0.971). ERA5 shows higher correlations with observations in the spatial patterns of interannual variability across all seasons, but particularly DJF (Spearman's rERA5=0.504, rERAI=0.475) and MAM (Spearman's rERA5=0.488, rERAI=0.435). Finally, both models capture the ENSO signal remarkably well, although the spatial extent of positive/negative anomalies could be improved in some seasons and regions.