Land cover classification optimized to detect areas at risk of desertification in North China based on SPOT VEGETATION imagery

Abstract

Monitoring of desertification processes by satellite remote sensing is an important task in China and other arid regions of the world. We used a full year 2000 (1 January 2000–31 December 2000) time-series of SPOT VEGETATION images with 1 km spatial resolution to produce a land cover map with special emphasis on the detection of sparse vegetation as an indicator of areas at risk of desertification. The study area covered 2000×3500 km in North China extending from temperate forests to the Gobi desert to the Tibet high plateau. A classification approach for different land cover types with special emphasis on sparse vegetation cover was developed which was able to resolve problems related to seasonal effects and the highly variable natural conditions. The best classification results were obtained by exploiting seasonal effects detectable in time-series of optimized Normalized Difference Vegetation Index images calculated from 10-day composites. Compared to the Global Land Cover 2000 and MODIS Vegetation Continuous Field classification, more sparsely vegetated land was detected by this approach. The areas at risk of desertification were modelled, and the result suggests that 1.60 million km 2 are areas at risk of desertification. Due to the wide swath and sensitivity to vegetation growth SPOT VEGETATION imagery should be very useful to detect large-scale dynamics of environmental changes and desertification processes.