An Assessment of GF-1 Fused Multispectral Images in Different Months of Spring for Mapping Quasi-Circular Vegetation Patch

Abstract

Abstract The vegetation phenological information derived from multi-seasonal imagery is helpful for mapping vegetation dynamics. The previous studies indicate that the spring imagery is considered as an optimal data to map the quasi-circular vegetation patches (QVPs) in the Yellow River Delta, China. In this work, the GF-1 images acquired in different months of spring was compared for mapping the QVPs using the decision tree classifier and watershed image segmentation technique based on the brightness and greenness components of tasselled cap transformation. The result of this study recommended that the April GF-1 image with only circle-like detection rules could result in higher detection accuracy compared with that of the March and May images. In the future, more effective image segmentation algorithms and quantitative influence of image quality on detection accuracy should be considered.