Based on the decision tree method, an integrated processing technique on the detachment of cloud from the earth's surface with FY-2C multi-channel satellite data is performed step by step using three different measures, including the threshold method of multi-channel, the threshold method of similar test on time series of composed cloud/surface map derived from clear sky, and the limited regional dynamic threshold method. Experiments are made to fulfill the detachment of cloud map from the earth surface in the area of 60 S～60°N, 45°E～165°E with FY-2C meteorological satellite data in July of 2005. In contrast with manual estimation, the detachment made by the computer automatically presents an accuracy rate over 90%. The analysis results show that, when taking the differences of the brightness temperature between the clouds and the earth surface (or the sea surface) into account as well as the characteristics of their temporal evolvements in different channels, the establishing and applying of multi methods on the cloud-surface detachment will be helpful to increase the quantity of information in detaching. These two approaches suggested in the paper—the similar test method which considers the varying characteristics with time of the clouds, and the stepwise test method which prefers to higher reliability under cloudy or non-cloudy circumstances—will be in favor of the detachment and its accuracy.