Fraction Vegetation Cover Extraction Using High Spatial Resolution Imagery in Karst Areas

Abstract

The karst region in southwestern China is characterized by a high heterogeneity in vegetation cover. This is due to the high intensity of anthropogenic disturbances and strong development of carbonate rock. Medium- and coarse-resolution satellite data have been used in regional-level vegetation studies, but they lack fine details and their quality suffers from terrain relief of the karst landform. In this study, we explored the feasibility of multispectral high spatial resolution ALOS (10 m) imagery for extracting fractional vegetation cover using the multiple end-member spectral mixture analysis (MESMA) approach. Field measurements were used to assess the accuracy of vegetation cover estimation and evaluate the effect of shadows on vegetation cover estimation. The overall accuracy of vegetation cover estimation was 77.2%, and the kappa coefficient was 0.71. Although the accuracy of estimated vegetation cover in the shaded area was lower than in the south-facing area, the average values were close. The results suggest that the MESMA approach in combination with high spatial resolution imagery could be applied to karst areas for fine-scale feature extraction.