Performance evaluation of ERA5 precipitation estimates across Iran

This paper aims to evaluate the European Centre for Medium-Range Weather Forecasts (ECMWF) Reanalysis 5th Generation (ERA5) reanalysis precipitation data for the 2006–2015 period against ground-truth data over Iran. The ERA5 was first evaluated at daily, monthly, seasonal, and annual time scales, and then, the quantile mapping bias correction method was applied to the regional precipitation time series of Iran’s homogeneous precipitation regions. A set of categorical indicators and statistical indices were used to evaluate the ERA5 precipitation product against observations. The results showed that ERA5 enjoys favorable performance in most homogeneous precipitation regions of Iran, except in G3 and G6. The categorical indicators results indicated better performance of ERA5 in the rainy months with probability of detection (POD) > 0.7 and critical success index (CSI) > 0.4 than in the dry months. It is concluded that the ERA5 can accurately estimate precipitation amount but has moderate performance in estimating the number of rainy days and the days without rainfall. The RMSE (root mean square error), NSE (Nash–Sutcliffe efficiency), and mean absolute error (MAE) measures also indicate that the bias correction has significantly reduced the estimation errors. The results showed that, after bias correction, the MAE in G3, G6, and G8 regions decreased by 75%, 25%, and 25%, respectively. Moreover, the RMSE in G3, G6, and G8 regions decreased by 79%, 20%, and 12%, respectively. We found that after bias correction, the NSE metrics in the G3 region improved by approximately 200% (from − 0.8 to 0.85). In general, the ERA5 precipitation dataset can be considered as an alternative to ground-truth data over the whole territory of Iran.