Abstract: Using the data of CLDAS 5 km hourly visibility products and visibility observation stations in Sichuan Province,the intelligent grid hourly visibility products were interpolated to 156 national visibility instrument observation stations in Sichuan Province by bilinear and neighboring interpolation methods.The correlation coefficient,root mean square error,mean value error,mean absolute error and other parameters of the two were compared,and the data of CLDAS 5 km hourly visibility products were evaluated and the data availability were analyzed. The correlation coefficient of the hourly visibility products in the Sichuan Basin and the central part of the Panxi area was high,generally above 0.7,the error value was small,and the mean value error was generally negative,indicating a small underestimation and good data availability; However,the correlation coefficient was small and the error value was large in the western Sichuan Plateau and the southern part of the Panxi area. The mean value error was generally positive,indicating that most of them were highly estimated and the data availability was average. The grading evaluation shows that as the visibility level decreases,there were more and more missing data in the plateau area,and the high valuation in the basin also increases significantly. The availability of data with visibility levels of "excellent" and "good" was better.