Search

Temperature is a major variable in the Earth's climate system, which plays an important role in energy exchange interactions between the Earth's surface and the atmosphere. There are various sources for temperature estimation, including ground stations, satellite products, reanalysis datasets, and multi-source weighted-ensemble datasets. Reanalysis datasets are generated by combining different types of observational data for a certain time in numerical weather prediction models and using ground and satellite observations. The purpose of this research is to investigate the performance of the ERA5- Land and AgERA5 reanalysis datasets as well as the MSWX multi-source dataset to determine reliable datasets for estimating temperature in Iran. First, we evaluated the temporal variations of the three datasets against the station data. We used the air temperature from 98 stations for 30 years from 1991 to 2020. Three metrics including Root Mean Square Error, Bias, and Index of agreement were used to evaluate ERA5-Land, AgERA5, and MSWX datasets.Then, considering the seven main climate zones of Iran, the spatiotemporal quality of the annual mean temperature was evaluated in different climate zones. The results showed that all three datasets have less error and bias in the estimation of the minimum temperature of Iran. However, AgERA5 and MSWX significantly showed less error in the estimation of the maximum temperature with RMSE of 1.74℃ and 1.42℃, respectively. On the other hand, the ERA5-Land dataset shows overestimation (5.05℃) and high error values (5.07℃) over the country-averaged. The results showed that the MSWX dataset has a better performance in estimating Iran's temperature with an average bias of 1℃. The interannual variations and decreasing and increasing trends of temperature in three datasets with correlation above 0.86 in all climate zones of Iran show a high consistency with observational data.The RMSE in all three datasets reaches its maximum in the winter season in mountainous climate zones of the country. This may be caused by snow-albedo feedback in mountainous climate zones. The findings showed that performing bias correction and downscaling methods as they have been done in MSWX and AgERA5, significantly improved the reanalysis dataset compared to the direct model output. Nevertheless, in the southwest of the Caspian Sea, the bias-corrected MSWX and AgERA5 show more errors than ERA5-Land. In general, the values of bias and RMSE in all three datasets are affected by the physical schemes of the model, parameterization, and data assimilation system, or the downscaling and bias correction methods. However, sources of bias can be different in different seasons of the year. The monthly spatial distribution of temperature in Iran shows that the minimum temperatures are located in the middle of Alborz and the northwest mountains of the country, and the coldest month of the year is January with a temperature of -10.8℃. The maximum temperatures in Iran are located in the southwest of the country and the southern coasts, and the hottest month of the year is July with an average temperature of 42.38℃. [ABSTRACT FROM AUTHOR]