In order to study the scale error of low resolution meteorological satellite cloud detection and its impact on the calculation of downlink radiation, cloud detection using high resolution stationary satellite GF-4 data and error analysis were carried out. Firstly, the cloud detection of GF-4 data is carried out by using visible channel threshold method and time series method, and the error of cloud detection results of Himawari-8 and FY-2 (FY-2G, FY-2E) is analyzed based on the results of GF-4 cloud detection.In the study area, FY-2G, FY-2E and Himawari-8 cloud images could distinguish the clouds and clear sky. The main reason for the error was the scale effect produced by different spatial resolution satellites(the differences caused by cloud detection algorithms are not discussed here).Most of the errors occurred in the areas of thin clouds and broken clouds.High resolution data could detect broken clouds, while low resolution data lead to false and missed detection. On this basis, the error of remote sensing calculation of short wave radiation was analyzed,and it was found that the error of the actual cloud amount in the pixel would bring significant error to the estimation of the downward radiation.The relative error of the instantaneous downward radiation in the selected test area was -173.52%, and the maximum relative error of shortwave radiation was -20.20%.The results show that the high resolution stationary satellite data can significantly improve the estimation accuracy of the downlink shortwave radiation in the regions with more broken clouds.