Abstract: Using the hourly meteorological observation data of national stations and regional stations, the data quality of ground automatic stations of Sichuan data integration business operation for nearly five years is analyzed and evaluated. The results show that the overall data quality of Sichuan national stations is at an excellent level. With the transformation of automatic observation in Sichuan in April 2020, there is a significant mutation in data availability rate from high to low, and then gradually increases. The data quality of regional stations is lower than that of national stations, but there is a significant upward trend overall. From the perspective of seasonal fluctuations, the main reason for the decrease of availability rate is data missing, the availability rate of national stations is low in the flood season and high in the non-flood season, while that of the regional station is the opposite. From the perspective of daily variation, the data availability rate of regional stations is greater during the day than that at night, and after reaching a peak at 17 o'clock, it shows a downward trend until the early morning. From the perspective of spatial distribution, the data availability rate of Sichuan Basin is the highest, and the data availability rate of the western Basin and the plateau transition area is mostly low. From the perspective of the quality of each element of the regional stations, the availability rate of wind direction and wind speed is the highest, the availability rate of precipitation is the lowest and the fluctuation of each month is large. The data missing rate of each element is well correlated, mainly due to the missing of the whole station caused by communication failure or software failure. In general, the data quality of ground automatic stations has obvious seasonal fluctuations, daily changes and spatial distribution characteristics, but the overall performance is relatively stable and shows a certain upward trend.