Road-river synergies driving irrigation expansion in the Mu Us desert, China (1986–2023)

Irrigation expansion is transforming China's arid landscapes, yet how road networks and rivers jointly shape clustering patterns in the Mu Us Desert, and whether infrastructure outweighs climate as the primary driver, remain unquantified. A multi-index remote sensing framework integrating GCVI, NGI, EVI, and NDWI is developed here to map irrigation from Landsat imagery (1986–2023), achieving 92% overall accuracy (Kappa = 0.84). Kernel-density and buffer-ratio analyses reveal a structural shift from river-dependent irrigation (9.46% river-buffer dependence in 1986) to infrastructure-guided clusters (1.20% in 2023), accelerating after 2018 with center-pivot proliferation. As Mann-Kendall tests show no significant precipitation increase (p > 0.05), the > 17% cropland expansion during 2018–2023 was primarily policy- and engineering-driven, not climate-induced. These findings provide spatially explicit, infrastructure-resolved evidence for optimizing land-use planning and groundwater management in arid oasis ecotones.