Comparative evaluation of the accuracy of re-analysed and gauge-based climatic data in Iran

Due to the low density of meteorological stations and the short statistical period of data recording in new stations, limited temporal and spatial information is available in many regions of Iran. On the other hand, the dependence of agricultural production on temperature, precipitation, and other climatic variables increases the importance of analysing the spatial-temporal changes of these variables. One way to overcome this limitation is to use global or regional gridded datasets. This study aimed to compare the accuracy of precipitation data, minimum temperature, and maximum temperature of climate datasets of CPC Global, CRU TS, ERA-interim, ERA5, and MERRA-2 with the data of synoptic stations in a 30-year period (1989–2019) on a monthly scale. Also, an indirect comparison through the calculated values of potential evapotranspiration (ET 0 ) by the FAO Penman–Monteith equation using the data of datasets and stations was another objective of the study. The climatic data of Ahvaz, Aligudarz, Babolsar, Kerman, Kuhrang, Qazvin, Rasht, and Saqqez stations were used for this purpose. Data analysis was performed using R 2 , RMSE, MAE, and MBE criteria. The results showed that the highest correlation ( R 2 = 0.999) belonged to the maximum temperature data of the Ahvaz station and ERA5 dataset, while the lowest correlation ( R 2 = 0.171) belonged to the precipitation data of the Qazvin station and MERRA-2 dataset. Moreover, based on the RMSE values, ET 0 data estimated with the ERA5 dataset at the Rasht station (0.258) and precipitation data of the MERRA-2 dataset at the Kuhrang station (144.583) had the highest and lowest estimation performance, respectively. For the precipitation, maximum temperature, minimum temperature, and ET 0 variables, ERA5, CPC Global, ERA5, and ERA-interim datasets exerted the highest estimations performance, respectively. The overestimation of ET 0 was mainly due to the overestimation of wind speed and the underestimation of dew point temperature. The findings of this study generally indicated that in most climatic regions of Iran, the ERA5 dataset had a better performance than other datasets in estimating weather variables.