Assessment Method For Mining Geo-Environment Usingweights-Of-Evidence Modeling

Assessment of mine geological environment is the main basis for mine geological environment supervision and management. The study of assessment method is an important content in the field of mine geological environment. Using objective, quantitative and reproducible assessment methods is helpful to improve the reliability and precision of assessment results, and provides scientific basis for reasonable utilization and protection of mineral resource. Many researches on assessment method have been carried out deeply. However there are some limitations inherent in existing methods. Limitations refer to:(a) the problem of the scale effects was neglected too often. The assessment results obtained at a certain scale was improperly used as a substitute for other scale assessment results. And there is short of researches on multi-scale and optimal scale for assessment of mine geological environment;(b) Selection of evaluation index and calculation of weight-values of the evaluation index remain largely subjective;(c) classification of mine geological environment is more likely to be influenced by subjective factors, and the results are often not reproducible.The aim of this paper is to overcome the limitations from existing assessment methods. We select the rare-earth mine located in the south of Jiangxi Province as study area. Relevant thematic maps representing various factors that are related to mine geological environment were generated using spatial statistics, RS and GIS techniques. In presenting a multi-scale assessment of mine geological environment, an optimal scale selection method was proposed based on the spatial variation quantitative parameters of feature in the study area. Then, the weights-of-evidence modeling was adapted from the mineral resource assessment field to mine geological assessment. The model is a data-driven model which based on a log-linear form of Bayes’ rule. The spatial relationships and the distribution of evidential theme was used by Weights-of-evidence modeling for describing and interpreting the relative importance of evidence. Weights-of-evidence modelling was applied to calculate weights and probability theme, which is beneficial in avoiding the influence form subjective factors. The prior probability and the raises in Cumulative Area-Posterior Probability(CAPP) curve are used to define the number and position of geological environment classes. Based on field data, the accuracy of the model was evaluated by using the success rate curve and prediction rata curve. This work presents an objective, quantitative and reproducible framework for evaluation of the mining geo-environment quality. The conclusions can be summarized as follows:(1) High resolution satellite images contain detailed information to extract mine ground objects precisely. We present an analysis of multi-scale geological environment assessment with a linear model for rare-earth mine located in the south of Jiangxi Province. The analysis suggests that spatial heterogeneity play significant role in the scale effects. And the spatial heterogeneity consists of heterogeneity of land surface and spatial distribution patterns of evidence weights. Based on geo-statistics theory, both mine geological feature locations and feature values are used simultaneously to measure mean variance and global Moran’s I for a series of distances. The mean variance reflects the homogeneity of unit cells, and the global Moran’s I measures the heterogeneity between unit cells. A model evaluates the optimal scale based on measure mean variance and global Moran’s I simultaneously was proposed. The results we obtained demonstrate that the spatial scale of 202 m is the optimal scale for assessment mine geological environment in the study area. The homogeneity of unit cells and the heterogeneity between unit cells reach optimal equilibrium point on that distance.(2) The decision to accept or reject evidential themes as predictors is assisted by considering the contrast values and the results of conditional independence test. Then predictor themes were combined to create the posterior probability map. The relationship between mine geological environment and evidence themes were investigated using the Weights-of-evidence modeling. The results showed that the most important factors governing the geological environment are mining development and vegetation fraction. Based on the CAPP curves and the prior probability, the mine geological environment have been classified into three classes: good quality(area proportion is 4.94%), moderate quality(7.11%) and poor quality(87.95%).(3) The efficiency curves were used to validate the model. The validation showed a high success rate(88.7%) and prediction rate(92.2%).The analysis indicates that the weights-of-evidence modeling can avoid the influence of subjective judgment effectively. The modeling has a good reproducibility because of its objective nature and is useful in the assessment of mine geological environment.