Basic clustering algorithms including K-means, fusion c-means and C4.5 decision tree have been applied in cloud
detection with FY-2C data. With the limitation that no reliable template is available, supervised classification algorithm
is utilized to test the credibility of non-supervised classification algorithms, which contribute to generate a cloud
classification model with high credibility. It is proved that cloud classification product distributed by National Satellite
Meteorological Center enjoys a high credibility and stability. It is also demonstrated that FY-2C data is eligible to
classify cloud into 4 types as cumulonimbus, cirrostratus, dense-cirrus and low and middle cloud.