Soils and Remote Sensing Indices, and Machine Learning for Sustainable Land Management in Bahariya Oasis, Egypt (2018–2026)

Arid and semi-arid regions face persistent challenges of salinity, sodicity, nutrient depletion, and geomorphic constraints that undermine sustainable agriculture. This study develops an integrated framework for Quantitative soil evaluation in Bahariya Oasis, Egypt, combining soil taxonomy, geomorphological mapping, remote sensing (RS) indices, and machine learning (ML). Soil indices; quality, fertility, salinity, degradation, resilience, and sustainability were constructed from normalized laboratory parameters and validated against RS-derived indices. Random Forest regression and classification models were applied to predict soil indices from RS inputs, achieving high accuracy (R² > 0.80) and identifying Normalized Difference Vegetation Index, Moisture Stress Index, and Salinity Index as the most influential predictors. This ML integration reduced redundancy by objectively classifying soils into sustainable hotspots (Alluvial Fans / Plains and Overflow Basin / Cultivated Plains), transitional rehabilitation zones, and unsuitable sabkha depressions. Multi-criteria evaluation using the Analytic Hierarchy Process produced composite suitability scores, which were operationalized into a Decision Support System (DSS). The DSS generated reproducible, spatially explicit maps with >80% agreement to field observations, enabling prioritization of reclamation, conservation, and agricultural expansion strategies. The framework demonstrates that sustainable agricultural expansion in Bahariya Oasis must be evidence-based, multidimensional, and policy-driven, focusing on resilient geomorphic units while conserving degradation hotspots. By embedding soil and RS indices within ML-enhanced DSS outputs, this research contributes to global efforts in sustainable land management, aligning with SDG 2 (Zero Hunger) and SDG 15 (Life on Land). The approach offers a transferable model for arid zone evaluation across North Africa and the Middle East, with future extensions incorporating climate scenarios, groundwater modelling, and socio-economic drivers to enhance adaptive management under uncertainty.