Precise inversion of soil water-salt and management zoning for oasis cropland in arid regions

Water scarcity and soil salinization critically constrain sustainable cropland development in arid oasis regions. Conventional homogeneous management overlooks spatial variability, reducing resource-use efficiency. To address this, we developed a framework for delineating precision management zones (MZ) through the joint inversion of topsoil (0-10 cm) moisture and salinity, integrating multi-source environmental variables. Using Google Earth Engine and Sentinel-2 imagery, cropland in the Wei-Ku Oasis was extracted with a Random Forest classifier (91.59% validation accuracy). Among five modeling strategies compared, the optimal models achieved for electrical conductivity of the saturated paste extract (ECe), the R²_test was 0.765, R²_cv was 0.68, and the standard deviation (SD) was 0.07; for soil water content (SWC), the R²_test = 0.747, R²_cv = 0.67, and SD = 0.08. Subsequently, eight key variables (ECe, SWC, clay, silt, sand, bulk density, pH, and Digital Elevation Model (DEM)) were reduced by principal component analysis (PCA). Both K-means and fuzzy C-means clustering divided the cropland into four MZs, showing similar spatial patterns but different areal proportions. Distinct variations among the zones were observed in soil water-salt characteristics, particle-size composition, and crop growth conditions. Based on these zones, tailored management measures were proposed, including windbreak and sand fixation, water-saving and salinity control, precision fertilization, and soil fertility improvement. This study provides critical spatial information on soil moisture and salinity and data-driven decision support for precision management of cropland in arid oasis regions, offering valuable guidance for salinization mitigation, differentiated irrigation, and sustainable agricultural development.