Enhancing desertification risk mapping with entropy-based weighting and machine learning: Insights from Iraq

Area of Science:

• Environmental Science
• Geospatial Analysis
• Machine Learning Applications

Background:

• Desertification is a major environmental threat in arid and semi-arid regions.
• Iraq faces increasing land degradation due to climate variability and human activities.
• Traditional desertification risk assessments often use fixed-weight indices, limiting accuracy and regional specificity.

Purpose of the Study:

• To refine desertification risk assessment in Iraq.
• To develop an enhanced model integrating entropy-based weighting, machine learning, and uncertainty quantification.
• To improve the accuracy and spatial differentiation of desertification risk mapping.

Main Methods:

• Integration of an enhanced MEDALUS model with entropy-based weighting.
• Utilization of satellite-derived vegetation indices (NDVI, SAVI, FVC), soil properties, climatic data, and human activity indicators.
• Application of machine learning (XGBoost) and uncertainty quantification (Monte Carlo simulations) for refined risk classification and interpretation (SHAP analysis).

Main Results:

• Nearly 50% of Iraq's land area is classified within moderate to extreme desertification risk zones.
• Climatic aridity and vegetation health were identified as the primary drivers of land degradation.
• The machine learning approach achieved high accuracy (94.46%, κ=0.9261) in risk categorization.
• Entropy-weighted approach enhanced spatial precision in risk mapping.

Conclusions:

• The developed hybrid Desertification Risk Index (DRI) provides a more adaptive and regionally specific risk classification.
• The study highlights the limitations of traditional fixed-weight indices.
• The research offers a robust, data-driven framework for land management and policy to mitigate desertification in Iraq and similar regions.