Dynamic Change Monitoring and Assessment for Sandy Land Based on Quantitative Remote Sensing

Abstract

Because of climatic change and human activities, sandification is becoming a serious threat to the sustainability of human habitation. The aim of this study, therefore, was to propose a method for sandy land detection based on mixed pixel decomposition; the dynamic change monitoring and assessment was then conducted. Results showed that the pixel purity index is a viable indicator for endmember extraction for sandy land detection via remote sensing by linear mixed pixel decomposition methods. Results showed that when the endmember proportion of sandy land accounted for > 50% of the total (except for the vegetation), a pixel would be detected as sandy land. The extraction accuracy was verified to be 86.42% by field data. Early-middle August was believed to be the most reasonable time to assess sandy land coverage based on vegetation coverage. The sandy land areas in 2005 and 2014 were 5524 km2 and 4109 km2 respectively, reduced by 25.6%. Under the governance of sandy land in the last ten years, the sandy land area declined continually, but some areas were still degraded to a worse status and need special care to protect.