Based on the surface topographic altitude data, FY-2C infrared channel data and cloud cover data from observation stations during 30 April-01 May 2007 from 2000 to 0300 (UTC), an analysis of the relation between the topographic altitude and the clear IR radiance is performed. It is found that the IR brightness temperature of surface is influenced by the change of surface topographic altitudes at nighttime in clear areas. The IR brightness temperature in some clear areas at high altitude is lower than (or equal to) that in other cloudy areas at low altitude. This results in difficulty to process cloud detection using infrared channel data. Therefore, the standard database of clear surface IR brightness temperature at different topographic altitudes is established and used to separate satellite infrared observations into “clear” and “cloudy” categories as a method at nighttime. A case selected randomly shows that 88 percent of satellite observations can be correctly separated for an infrared imagery at nighttime. There are 4 percent of “clear” observations to be incorrectly separated into “cloudy” ones. In addition, 8 percent of “cloudy” observations are incorrectly separated into “clear” category. By the method 92 percent of satellite observations in clear areas can be correctly categorized, but 82 percent in cloudy areas. In order to analyze the operational use of the method the satellite infrared observations at 0000 UTC during 2-5 May 2007 are separated into “clear” and “cloudy” categories based on the standard database from the datasets at 0000 UTC 1 May 2007. The 92 percent of total satellite observations can be correctly separated, and the accuracy is close to, but better than that of cloud detection at 0000 UTC 1 May. The 93 percent of satellite observations in clear areas can be correctly categorized, which results in the better total accuracy, also 92 percent in cloudy areas. 