Multidimensional analysis of desertification sensitivity in the horqin sandy land based on an improved MEDALUS − ESA model and an optimal geographical detector

• The improved MEDALUS model more effectively detects desertification sensitivity. • Quality indices displayed similar sensitivity distribution and quantity patterns. • Desertification sensitivity in the Horqin Sandy Land reversed from 2000 to 2020. • Multiple factors outweigh single factor in driving desertification sensitivity. Desertification is a global ecological and environmental problem that seriously threatens ecosystems and regional socioeconomic development. Focusing on the Horqin Sandy Land in Northeast China, this study constructs and optimizes a land desertification sensitivity index system via the improved MEDALUS − EAS model. It covers five quality indices: soil, climate, vegetation, land management, and socio-economy. The analytic hierarchy process (AHP) determines factor weights. We analyzed the dynamic changes in desertification sensitivity from 2000 − 2020and used the optimal parameter geographical detector for the driving factors. The findings show that the vegetation quality index has the highest weight (0.4197), followed by the climate (0.3319) and soil (0.1309) indices, with lower weights for land management (0.0674) and the socioeconomic index (0.0502). The Horqin Sandy Land’s desertification sensitivity index ranged from 0.0695 to 1.0034 from 2000 − 2020, reversing this trend. The proportions of extremely, highly, and moderately sensitive areas decreased, whereas the proportions of slightly sensitive and nonsensitive areas stabilized or increased. Spatial differentiation detection indicates that net primary productivity (NPP) and vegetation coverage are the main drivers of significant interactive enhancement. Spatiotemporal change detection reveals that vegetation coverage and wind speed dominate the evolution of sensitivity, with strong interactions. Multifactor consideration is key for accurate assessment, offering important references for regional ecological governance.