Evaluation of the reliability of ERA5-Land data in assessing minimum and maximum daily temperatures in Iran (1991-2020)

In scientific research, the use of station data is often hindered by significant amounts of missing data and insufficient density of observation stations. This lack of comprehensive and representative data can significantly undermine our understanding of the phenomena under investigation. However, network-based databases have emerged as a promising solution to this challenge. One notable network database is the ERA5-Land, the latest reanalysis dataset from the European Center for Medium-Range Weather Forecasts (ECMWF). With a higher resolution than its predecessors, ERA5-Land has the potential to provide a more detailed and accurate representation of climatic conditions. Nevertheless, the reliability of this dataset in estimating Iran's maximum and minimum temperatures remains an open question. In this study, we aimed to evaluate the performance of the ERA5-Land climate database by comparing it to observation data from 143 stations across Iran during the normal climate period of 1991-2020. We calculated a range of statistical indices, including root mean square error (RMSE), Nash-Sutcliffe efficiency coefficient (NS), bias index, Mean absolute error (MAE), and Kling-Gupta coefficient (KGE), for both the maximum and minimum daily temperatures at each station. This comprehensive analysis allowed us to assess the spatial variations in the accuracy of the reanalyzed data. Furthermore, we delved deeper by analyzing the averages of these indicators in six regions that represent the diverse climatic conditions found within Iran. This approach enabled us better to understand the dataset's performance in different geographical settings. Our findings revealed that the ERA5-Land reanalysis data demonstrates good accuracy in estimating maximum and minimum temperatures in Iran. However, the performance is not uniform across the country. Daily maximum temperature data are generally more accurate than daily minimum data, with the latter being underestimated in most stations. Interestingly, the accuracy of the temperature estimates varies based on the climatic conditions. The data show the best performance in the warm cluster, while the poorest accuracy is observed in the cold cluster, particularly in the northwest of the country. This disparity highlights the importance of considering regional variations when evaluating the reliability of reanalysis data. Seasonal accuracy also emerges from the analysis. The MAE values for spring, summer, autumn, and winter seasons of maximum temperatures are 2.31, 2.16, 1.99, and 2.19, and for minimum temperatures are 2.6, 2.4, 2.4, and 2.3 respectively. The largest errors in estimating maximum and minimum daily temperatures occur during the spring season, while the lowest errors are observed in the autumn for maximum temperatures and the winter for minimum temperatures. Examining the annual accuracy trend reveals an interesting pattern. The annual MAE related to maximum temperature has increased from 1991 to 2020, with the highest error observed in 2020 across all climatic clusters. This suggests the need for continued monitoring and improvement of the reanalysis data to ensure its reliability in the face of evolving climate conditions.