| چکیده انگلیسی مقاله |
Evaluation of ERA5-Land Data in Estimating Monthly Extreme Temperature Indices in Iran
An investigation was undertaken to ascertain the suitability of the ERA5-Land reanalysis dataset, a recent product of the European Centre for Medium-Range Weather Forecasts (ECMWF) offering enhanced spatial resolution relative to its predecessors, for the accurate estimation of monthly extreme temperature indices across the diverse climatic landscape of Iran. Recognizing the inherent uncertainties associated with the application of global reanalysis data at regional scales, particularly in estimating extreme events, the current study rigorously evaluated the performance of ERA5-Land against observational records from a network of 143 meteorological stations spanning the standard climate normal period from 1991 to 2020. The evaluation framework incorporated a suite of established statistical metrics, including the Pearson correlation coefficient (R) to assess the strength of linear association, the bias that reflecting the tendency to overestimate or underestimate a quantity due to inherent inaccuracies, the root mean square error (RMSE) to quantify the magnitude of overall error, the Nash-Sutcliffe efficiency coefficient (NS) to determine the predictive power relative to the mean of observations, relative bias to identify systematic over- or underestimation, mean absolute error (MAE) to measure the average magnitude of errors, and the Kling-Gupta efficiency coefficient (KGE) to provide a comprehensive assessment of correlation, variability bias, and mean bias for the monthly extreme temperature indices: maximum monthly daily maximum temperature (TXx), maximum monthly daily minimum temperature (TNx), minimum monthly daily maximum temperature (TXn), and minimum monthly daily minimum temperature (TNn). Furthermore, a detailed analysis of the spatial distribution of these performance metrics was conducted to elucidate regional variations in ERA5-Land's accuracy, followed by a comparative assessment of the average index values across six distinct climatic regions within Iran, thereby providing a comprehensive understanding of the dataset's capabilities and limitations in representing extreme. The results showed that the average RMSE across all stations for the four Monthly extreme temperature indices studied, TXx, TNx, TXn, and TNn, are 2.79, 2.57, 2.23, and 3.78, respectively, indicating that the ERA5-Land reanalysis data generally have good accuracy in estimating monthly extreme temperatures across Iran. In most stations, the monthly extreme values of the ERA5-Land data are often underestimated compared to observational data. The correlation coefficients for the three indices TXx, TXn, and TNx, are all greater than 0.85 across all stations; however, for the TNn index, some stations in cluster four (very hot cluster), including Ahvaz, Bostan, and Abadan, exhibit very low correlation coefficients. Therefore, it can be concluded that the ERA5-Land reanalysis data can accurately estimate the trends of TXx, TXn, and TNx at all stations, but using these reanalysis data for the TNn index in the southwest region is not appropriate. The Nash coefficient values for most stations are acceptable, although in some cases, this value has turned negative, indicating that the use of ERA5-Land reanalysis data is not suitable for estimating the desired index. Using the average value of observational data as an estimate yields higher accuracy than that of the reanalysis data. The Nash coefficient for the TXx index is negative at Yasuj, Konarak, and Chabahar stations, for the TNx index at Kerman station, and for the TNn index at Safiabad, Bostan, Ahvaz, Masjed Soleiman, Ramhormoz, Abadan, Kerman, and Minab stations. The KGE index, which simultaneously measures mean bias, variance bias, and correlation, shows acceptable values for estimating TXx, TXn, and, TNx at all stations (greater than 0.55, 0.49, and 0.33, respectively). However, for estimating the TNn index, the KGE is negative at 37 stations, most of which are in clusters one and two. The analysis of KGE values revealed that the negativity arises from the bias in the mean of the reanalysis data compared to the mean of the observational data. Therefore, ERA5-Land data should be used with caution for estimating TNn in stations located in the northwest and northeast regions. Overall, the accuracy of the reanalysis data in estimating monthly extreme temperature indices varies across different regions of Iran based on the index in question. The estimation of TXx from the reanalysis data shows the highest accuracy in the warm cluster and the lowest in the coastal warm cluster, whereas ERA5-Land data shows the best accuracy for estimating TNx in the coastal warm cluster and the lowest accuracy in the very cold cluster. For the TXn index, the highest accuracy is observed in the warm cluster and the lowest in the very cold cluster, while the results for the TNn index indicate that the estimation accuracy is highest in the temperate humid cluster and very low in the very warm cluster.
Keywords: ERA5-Land, Iran, Monthly Extreme Temperature Indices |