Comparative analysis of cloud fraction obtained from Moderate-resolution Imaging Spectroradiometer （MODIS） and ground observations shows the existence of considerable deviations. Five correction methods

based on cloud fraction observed from ground meteorological stations

have been proposed in this paper

including difference correction

ratio correction

difference-mixed correction

ratio-mixed correction and normalization-mixed correction of MODIS cloud product. The results show that all of the five correction methods are workable. Comparatively

the ratio correction method has the highest precision and is easy to be implemented. Hence it is the best recommended correction method for MODIS cloud product. After correction

the cloud fraction of MODIS is identical with that of ground observations

both in spatial distribution and quantitative analysis. Furthermore

intensive observations test show that monthly Mean Absolute Bias Error （MABE） of cloud fraction obtained from MODIS after ratio correction is less than 5%. This study gives a reference for relative researches using ground observation data to correct the related satellite products

which is of great significance to the related investigations of plateau

mountains and desert with sparse stations.