Due to the combined effects of global warming and human activities, the ecological environment of the Yellow River source area has undergone profound changes and desertification has become increasingly prominent. In this study, an optimal desertification monitoring index based on feature space was proposed for the Yellow River source area, and constructed using Landsat images. Then, the spatial and temporal variation of desertification in the Yellow River source area and its driving mechanism were studied using Geodetector. The main conclusions are as follows: (1) The newly proposed feature space-based desertification monitoring index has good applicability in the study area. The best inversion accuracy of the point-to-point Albedo-NDVI feature space model was 88.4%. (2) Desertification in the eastern and southern regions of the Yellow River source area has a tendency to increase, while the desertification situation in the central region is relatively stable. (3) From 1995 to 2015, there was a significant improvement in desertification in the study area, as evidenced by a decrease in desertification intensity. (4) As the intensity of human disturbance increases, the influence of natural factors on desertification gradually diminishes. The interaction of natural and anthropogenic factors has greater explanatory power for desertification than that of individual natural or anthropogenic factors. The research results can be used as a reference for decision-making on desertification control in the Three-River Source Region.
Keywords: Desertification; Feature space; Geodetector; Source region of Yellow River; Spatial and temporal pattern.
Copyright © 2021 Elsevier B.V. All rights reserved.