ABSTRACT
ABSTRACT: The present study deal with the recognition in vegetation of Songhua dam water reserves based on the decision tree classification approach. ASTER images were used as sources of information for the study area to extract every vegetation type’s spectral characteristic curve. The results showed that the threshold of NDVI to distinct vegetation and non-vegetation is 0. The threshold of NDVI to distinct forest and non-forest is 0.3. The accuracy assessment showed that: model based on decision tree classification total accuracy was 88.27%, and the Kappa coefficient was 0.85. Vegetation coverage rate was 80.76%, of which the forest was 318.59 km2, accounting for 62.64% of all.
