Quantitative evaluation of mining geo-environmental quality in Northeast China: comprehensive index method and support vector machine models

Abstract

The long-term mining of mineral resources has contributed to damaging the geo-environment, drawing significant attention to the evaluation of geo-environmental quality. This paper presents an original system for evaluating mining geo-environmental quality in the Changjitu economic zone, Northeast China. The proposed evaluation framework considers five criteria, namely basic mining information, the geo-environmental background of the mining area, mining-related geological problems, the importance of the evaluation area, and the difficulty of geo-environmental recovery. Objective weighting methods, such as the variation coefficient, entropy, and Kantiray weighting methods, and subjective weighting methods such as the analytic hierarchy process are developed to determine the comprehensive weights of the elements and indicators. A common comprehensive index method and a new support vector machine (SVM) model are then proposed and compared to evaluate mining geo-environmental quality. The findings show that the accuracy of the linear SVM model is 93.10 %, demonstrating that the SVM is appropriate for the evaluation of mining geo-environmental quality. Compared with existing common methods, the SVM model, which classifies mining geo-environmental quality into multiple groups, adopts the structural risk minimization principle. The evaluation results also show that mining geo-environmental quality tends to rank as level II in the study area, accounting for 75.86 % of the total eligible mines compared with 2.59 and 21.55 % for levels I and III, indicating that most mining geo-environments are moderately affected by mining activities.