Application of Hyperion data to land degradation mapping in the Hengshan region of China

Hyperspectral remote sensing is a promising tool for analysing and evaluating the risk of increasing land degradation. The aim of this paper was to introduce a methodology for mapping land degradation using Hyperion data. Salinization and wind erosion (desertification) are alarming signs of land degradation in Hengshan County, in the northern Shanxi Province of China. The Soil-Adjusted Vegetation Index (SAVI), Desertification Soil Index (DSI), Soil Organic Material Index (SOMI), Soil Ferric Oxide Index (SFOI) and Normalized Difference Water Index (NDWI) were chosen as the evaluation factors to set up the assessment system, and the extent of land degradation in the study area was divided into four grades: none, slight, moderate and high. Samples of each grade were used for overlay analysis in ArcGIS software using the five bands of the hyperspectral indices as inputs. Then, the land degradation maps were generated using support vector machine (SVM) and artificial neural network (ANN) approaches. We found that the SVM results were better than the ANN, with an overall accuracy of 0.93 for the SVM as compared to 0.88 for the ANN. The main advantage of the SVM is attributed to the transformation of the complicated learning problem into a high-dimensional simplified linear problem, which can overcome some flaws in other mechanical learning methods, such as the ‘over-study’ problem for ANN. It is concluded that this type of procedure for estimating the extent of land degradation is feasible and the SVM algorithm is an efficient method for mapping land degradation.