Sign in Subscribe
Global warming

Machine Learning and SHAP-Based Analysis of Deforestation and Forest Degradation Dynamics

Strategic Insights

— 1 min read

Researchers from the University of Duhok have developed a new method for analyzing deforestation and forest degradation using machine learning and SHAP (SHapley Additive exPlanations) techniques. This study used paired remote sensing and machine learning methods to simulate forest dynamics and identify the primary drivers of deforestation and forest degradation along the Iraq-Turkey border. The results showed a significant decrease in forest canopy cover, mainly due to illegal deforestation, road network expansion, and conflict-induced fires.

Key Takeaways:

  • The researchers used seven machine learning models to evaluate the effectiveness of XGBoost in predicting forest dynamics, with XGBoost outperforming the others and yielding predictive accuracies of 0.903 (2015), 0.910 (2019), and 0.950 (2024).
  • The study found that climatic factors primarily influenced vegetation cover in 2015, whereas anthropogenic drivers such as forest fires, road construction, and soil exposure intensified by 2024, accounting for up to 12% of the observed forest loss.
  • The forest canopy cover decreased significantly, from approximately 630 km² in 2015 to 577 km² in 2024, mainly due to illegal deforestation, road network expansion, and conflict-induced fires.
  • The study highlights the effectiveness of an ML-driven RS analysis for geoinformation needs in geopolitically complex and data-scarce regions.
  • The researchers conclude that their findings underscore the urgent need for robust, evidence-based conservation policies and demonstrate the utility of interpretable ML techniques for forest management policy optimization.
  • The methodological approach developed in this study provides a reproducible blueprint for future ecological assessment.

Statistics:

  • The forest canopy cover decreased by 53 km² between 2015 and 2024 (577 km² - 630 km²).
  • The percentage of forest loss attributed to anthropogenic drivers increased to 12% by 2024.
  • The XGBoost model achieved predictive accuracies of 0.903 (2015), 0.910 (2019), and 0.950 (2024).
  • The study found that up to 12% of the observed forest loss was attributed to anthropogenic drivers by 2024.

Sources:

  • Machine Learning and SHAP-Based Analysis of Deforestation and Forest Degradation Dynamics Along the Iraq-Turkey Border. Earth, 2025, 6(2): 49.
  • DOI: 10.3390/earth6020049
  • Publisher: MDPI AG
  • Available at: https://doi-org.sdpl.idm.oclc.org/10.3390/earth6020049

Read more