Using GF-2 Images to Detect Tamarix Chinensis Community within a Vegetation Patch

Abstract

Abstract The quasi-circular vegetation patches (QVPs) are mainly composed of suaeda salsa, Tamarix chinensis, and Phragmites australis in the Yellow River Delta, China. The previous studies indicate that the shrubs within a vegetation patch play an important role in the establishment and disappearance of vegetation patch. Therefore, in this work, the method based on the tasselled cap brightness and greenness components derived from the April image and the August GF-2 image acquired after water replenishment with the decision tree classifier was developed for quickly detecting the Tamarix chinensis community. The detection rate was 73.5%, and the misclassification rate was 12.1%. In the future, more samples of plant community and multi-seasonal images such as the late fall-early winter leaf-off data needs to be applied to further improve the detection accuracy.