Abstract: Based on the 105 national meteorological stations over Xinjiang during 1981-2020,the 2 meters temperature errors of （CRA40）data were analyzed by using the mean error （ME）,root mean square error（RMSE）and correlation coefficient（r）.The characteristics of spatio-temporal distribution of the errors show that the 40-year CRA40 reanalysis of the daily average temperature are significantly correlated with the observed station data,the corresponding correlation is 0.973,and passes the 0.001significance test.The CRA40 reanalysis daily average temperature is consistent with the observed data,the annual average,maximum and minimum values of daily average temperature differ by 0.12,0.02and 1.94℃，respectively,and the maximum and minimum daily average temperatures appear on the same date.The annual inspection results of ME_CRA40,RMSE_CRA40,and r_CRA40show that the quality of CRA40 temperature data in Xinjiang from 2005 to 2020 has improved,and it is closer to ERA5 than before 2005.Before 2004,ME_CRA40was negative,and after 2005,it was positive.The monthly ME_CRA40and RMSE_CRA40inspection results show that the maximum temperature error in Xinjiang appears in the winter.In March,October and November,the temperature data error of CRA40 is slightly better than that of ERA5.The daily inspection results show that the number of days with the absolute value of ME_CRA40within 1.0℃account for 80.8%of the year,and the number of days with RMSE_CRA40less than3.0℃account for 72.3%of the year.Taking RMSE as an indicator,the data quality of CRA40 is better than that of ERA5 for 79 days,mainly in March-April and October-November.Using the correlation coefficient as an indicator,there are 22 days of CRA40 data quality better than ERA5,mainly in March.The results of the meteorological observation station-by-station inspection show that 49 of the105 stations in Xinjiang have better ME_CRA40,38 stations have better RMSE_CRA40,and 51 stations have worse r_CRA40.On the whole,the air temperature data quality of ERA5 is slightly better than that of CRA40,but according to different indicators,the stations with superior air temperature data quality of CRA40 account for 36.2%-48.6%respectively.In areas with an altitude of less than 1 500 m,the temperature data quality of ERA5 is superior.In mountainous areas with an altitude of 1 500-2 000 m,the temperature data quality of CRA40 is better than that of ERA5.In mountainous areas with an altitude of more than 2 000 m,the quality of CRA40 temperature data is slightly worse,close to ERA5.