Dynamic monitoring of desertification in Naiman Banner based on feature space models with typical surface parameters derived from LANDSAT images

Abstract Naiman Banner is one of the most typical semi‐arid vulnerable ecological zones that is characterized by vegetation degradation and severe desertification. Previous studies mostly utilized a single feature space or only the linear correlation model to monitor desertification. In this study, the optimal monitoring model that fully considers the multiple feature spaces and the nonlinear relationship between surface parameters in Naiman Banner was determined, and then the spatial–temporal evolution patterns of the desertification from 1989 to 2017 were analyzed. The results showed that: (a) The albedo–normalized difference vegetation index point‐to‐point model was applicable to areas with high‐vegetation coverage, whereas the albedo–modified soil‐adjusted vegetation index (MSAVI) linear model had better applicability in areas with relatively low vegetation coverage. Considering the surface cover, the desertification monitoring index based on albedo‐MSAVI linear model had the best applicability. (b) During 1989–2017, the total desertification area percentage of Naiman County decreased from 90.85% in 1989 to 63.35% in 2017, showing a significant downward trend. (c) During the past 30 years, the desertification process was significantly influenced by human activities, such as ‘Three‐North’ Shelterbelt, Returning Farmland to Forestry, and Returning Farmland to Grassland (d) The desertification levels of unused land and grassland with low coverage was relatively high, whereas the desertification degree of cultivated and forest lands was low. These results can be used as an effective reference for the remote sensing monitoring and evaluation of desertification and can provide data and decision support for the ecological restoration of the region.