Detection of Changes in Land Degradation in Northeast China from Landsat Tm and Aster Data

This study aims to determine the accuracy level at which different forms of land degradation can be mapped from medium-resolution satellite data, and to assess how accurately degraded land can be detected from multi-temporal satellite images. Land degradation in the form of salinization and waterlogging in Tongyu County, western Jilin Province of Northeast China was mapped from Landsat TM and ASTER images at 30 m using supervised classification, together with several other land covers. These land covers have been mapped at an overall accuracy of 80% from the TM image with the accuracy for individual covers ranging from 75% to 100% except settled areas. At 80.0%, the accuracy for barren land is higher than that for degraded farmland. The overall classification accuracy was achieved at 75.3% from the ASTER data. The accuracy for degraded farmland rose marginally to 76.7%. It is concluded that moderately degraded land can be mapped from both ASTER and TM data at over 75%. Severely degraded land can be mapped more accurately over 80%. Between 1989 and 2004 grassland decreased from 282.9 km2 to 79.8 km2 while healthy farmland increased by well over 120%. On the other hand, fallow land increased by 125.2% due to excessively high soil salinity. Besides, degraded farmland and barren land rose by 19.1% and 33.1%, respectively. Thus, inappropriate land reclamation and cultivation are blamed for soil degradation inside the study area.