Verification of ERA5 Hourly Air Temperature Data over Turkey

In this study, produced by European Centre for Medium-Range Weather Forecasts (ECMWF), ERA5 reanalysis hourly 2 m surface air temperature is validated over Turkey and 7 geographic regions using hourly temperature observations collected between 2011 and 2020 over 1773 ground-based meteorological observation stations operated by Turkish State Meteorological Service. In the analyses, error statistics were calculated for hourly, daily and monthly average temperatures. The results show ERA5 simulates air temperatures at different temporal resolutions with a bias of -0.01 - 0.06 °C, while the mean absolute error and mean error standard deviation are 3.3 - 4.0 °C and 1.6 - 3,4 °C, respectively. Although there is no significant hourly variation in mean error statistics during the day and monthly variation during the year, the error variability was found to be higher in winter compared to the other seasons. The correlations of ERA5 and station temperature data are very high (0.93 - 0.99). Compared to the high temporal correlation, low spatial correlation (0.36) suggests that spatially variable correction coefficients will be needed before reliable use of ERA5 temperature data in different regions. Mean error statistics are consistent with validation studies in the literature in different regions. Accordingly, the results obtained in this study suggest that ERA5 datasets can be reliably used in various applications in Turkey, especially focusing on regions with sparse meteorological station networks.