Machine learning assessment of illegal mining (Galamsey) impacts on forest vegetation: a case study of Wassa Amenfi East District, Ghana

Abstract

This study addresses the persistent land management challenge of galamsey (illegal mining) in Wassa Amenfi East by exploring the impact of these activities on vegetation through advanced machine learning techniques. A comparative analysis was conducted using four machine learning algorithms Support Vector Machine (SVM), Random Forest (RF), Artificial Neural Network (ANN), and Maximum Likelihood Classification (MLC) to assess their effectiveness in detecting and analyzing vegetation changes due to galamsey operations. The results highlight the Random Forest (RF) algorithm as the most effective, with overall accuracy scores of 88% for 2015 and 87% for 2023, and Kappa Coefficient values of 0.84 and 0.82, respectively, demonstrating its consistent superiority over the other methods. Findings of the study reveal significant vegetation and forest cover loss due to galamsey, driven by poverty, unemployment, poor policies, livelihood pursuits, and quick financial gains by traditional authorities. This study stresses the need for targeted interventions to mitigate the environmental impact of galamsey and suggests the adoption of advanced machine learning techniques for more accurate and effective land management strategies.