Enhancing wind erosion risk assessment through remote sensing techniques

Area of Science:

• Environmental Science
• Geospatial Analysis
• Computational Science

Background:

• Wind erosion and dust storms pose significant environmental threats in arid and semi-arid regions.
• Effective prevention and management strategies are crucial for mitigating these impacts.

Purpose of the Study:

• To model, monitor, and predict wind erosion risk in Northeast Iran using remote sensing and machine learning.
• To identify key remote sensing indicators correlated with field data for accurate risk assessment.

Main Methods:

• Comprehensive literature review to select eight key remote sensing indicators.
• Application of machine learning models: Random Forest (RF), Support Vector Machine (SVM), Gradient Boosting Machine (GBM), and Generalized Linear Models (GLM).
• Ensemble modeling using a weighted average approach to reduce uncertainty and combine model predictions.

Main Results:

• The RF model performed best in 2008 (AUC=0.92), while GBM excelled in 2023 (AUC=0.95).
• The ensemble model accurately depicted elevated wind erosion intensity in Northeast Iran by 2023.
• A projected 23% increase in wind erosion intensity is anticipated by 2038 in central and southern areas, considering climate and land use changes.

Conclusions:

• Ensemble modeling effectively reduces uncertainty in wind erosion assessment.
• The findings provide a reliable basis for implementing targeted planning and management strategies.
• Proactive measures are essential to mitigate the escalating threat of wind erosion in the region.