Spatiotemporal Patterns and Driving Mechanism Analysis of Enterprise Illegal Land Use and Land Resource Misallocation Based on Multi‐Source Inspectorate Data

ABSTRACT Illegal land use by enterprises and the resulting misallocation of land resources have become critical challenges for sustainable regional development, particularly in ecologically sensitive dryland regions. This study investigates the spatiotemporal patterns and driving mechanisms of enterprise‐related illegal land use in China's Inner Mongolia Autonomous Region (IMAR) during the transition from the 13th to the 14th Five‐Year Plans (2017–2024), using multi‐source inspectorate data integrated with remote sensing analysis. To address classification uncertainty in complex landscapes such as grassland–desert ecotones and peri‐urban zones, an uncertainty‐aware land use/land cover (LULC) framework based on satellite embedding representations and a fivefold Random Forest ensemble was employed. The classification achieved robust performance, with overall accuracy ranging from 90.3% to 94.4% and Kappa coefficients ranging from 0.884 to 0.932. Spatial analysis revealed that illegal land use activities and resource misallocation were concentrated in transition zones characterized by high ecological sensitivity and rapid policy‐driven land conversions. Grasslands expanded by 71,399 km 2 (+20.8%), while deserts decreased by 87,866 km 2 (−20.0%), indicating large‐scale restoration; however, localized inconsistencies suggest potential regulatory gaps and unauthorized land conversions. Forests and croplands increased moderately (+5.3% and +8.3%), while urban land declined in restricted zones, reflecting stricter land governance. Uncertainty hotspots correspond closely with areas of high inspection frequency, highlighting regions where enforcement and monitoring remain critical. The findings demonstrate that integrating multi‐source inspectorate data with uncertainty‐aware remote sensing approaches can effectively diagnose illegal land‐use patterns and their underlying drivers, providing a scientific basis for optimizing land resource allocation, strengthening regulatory frameworks, and supporting sustainable land management policies.