Remote sensing of fractional cover of vegetation and exposed bedrock for karst rocky desertification assessment

摘要

The fractional cover of vegetation (PV) and exposed bedrock are key ecological indicators of the extent and degree of land degradation in karst regions. In this study, we suggested and compared new methodology for direct and objective estimation of key indicators of karst rocky desertification with hyperspectral and multispectral imagery. The results showed that the Hyperion estimated fractional covers of PV had good correlation with the field surveyed fractional covers and the R2 (coefficient of determination) and RMSE (root mean square error) for PV was 0.91 and 0.05, respectively; while for exposed bedrock was not so good, 0.53 and 0.11, respectively. It demonstrated that hyperspectral imagery was able to directly estimate the key ecological indicators of karst rocky desertification, which was in a heterogeneous landscape. As for the ASTER imagery, the results were not so accurate. It showed that multispectral imagery could not be used to effectively estimate the fractional cover of PV and exposed bedrock. Our study indicates that it could use hyperspectral imagery to directly and effectively estimate the fractional cover of PV and exposed bedrock for karst rocky desertification assessment in a heterogeneous landscape of karst ecosystem.