From mapping to decision making: a hybrid rule-based and machine learning framework for spatial land-use zoning

The rapid conversion of land use in coastal regions necessitates advanced decision support frameworks that bridge the gap between mapping and operational zoning. This study introduces the Dual-Logic Spatial Zoning Model (DLSZM), a hybrid framework designed to translate socio-ecological indicators into four planning regimes: Strict Conservation, Managed Use, Development Guidance, and Restoration. Applied to the Antalya region in Türkiye at a 30-meter grid resolution, the results demonstrate a high degree of regional convergence between expert-driven and machine learning pathways. Quantitative evaluation via an area-weighted confusion matrix shows that both methods produced identical classifications for Managed Use zones across approximately 7,630 square kilometers. While Managed Use remains the dominant classification, occupying landscapes with moderate ecological value, significant structural divergences were identified in transitional coastal belts. Alluvial transition analysis reveals that the machine learning model, driven by non-linear interactions captured in SHAP analysis, reassigned significant land areas from Strict Conservation and Development categories into the Restoration zone. Specifically, the machine learning framework identifies approximately 2,046 square kilometers of Restoration area, indicating a substantially higher sensitivity to cumulative stressors and degradation signals compared to the expert-derived logic. These findings suggest that while expert systems provide normative clarity, the machine learning pathway offers a more intervention-oriented spatial interpretation, effectively capturing the complex vulnerability dynamics of rapidly transforming coastal environments.