Assessment of ground subsidence using GIS and the weights-of-evidence model

The weights-of-evidence model, one of the Bayesian probability models was applied in evaluating a ground subsidence spatial hazard near abandoned underground coal mines (AUCMs) at Magyori area, Samcheok City in Korea using GIS. Using ground subsidence location and a spatial database containing information such as mining tunnel, borehole, topography, geology, and land use, the weights-of-evidence model was applied to calculate each relevant factor's rating for the Magyori area in Korea. Seven major factors controlling or related to ground subsidence were determined from the probability analysis of the existing ground subsidence area; depth of drift and distance from drift from the mining tunnel map, slope gradient obtained from the topographical map, ground water level and permeability from borehole data, geology and land use. Tests of conditional independence were performed for the selection of factors, allowing 6 combinations of factors to be analyzed. For the analysis of mapping ground subsidence spatial hazard, the contrast values, W+ and W−, of each factor's rating were overlaid spatially. The results of the analysis were validated using receiver operating characteristic (ROC) with the previous ground subsidence locations. In the case of all factor used, the area under the ROC curve (AUC) showed 0.9667, which corresponds to an accuracy of 96.67%. In the case of the combinations, the case of distance from drift, depth of ground water and land use used, showed the 90.71% (AUC: 0.9071) accuracy which is the best result produced in this analysis. The results can be used for hazard prevention and land-use planning near AUCM areas.

Research highlights
► In this paper, we have used factors controlling or related to ground subsidence as input to weights-of-evidence (WOE) model for ground subsidence hazard spatial hazard near abandoned underground coal mines (AUCMs) at Magyori area, Samcheok City in Korea using GIS. The weights-of-modeling has not been employed for such purpose. There is not enough research done in these areas which is quite obvious from the less number of articles published in recent years. This paper would significantly add to the existing literature on geospatial data and its use in the model in ground subsidence hazard mapping.
► The weights-of-evidence model, one of the Bayesian probability models was applied in evaluating a ground subsidence spatial hazard. Seven major factors controlling or related to ground subsidence were determined from the probability analysis of the existing ground subsidence area; depth of drift and distance from drift from the mining tunnel map, slope gradient obtained from the topographical map, ground water level and permeability from borehole data, geology and land use. The results of the analysis were validated using receiver operating characteristic (ROC) with the previous ground subsidence locations. In the case of all factor used, the area under the ROC curve (AUC) showed 0.9667, which corresponds to an accuracy of 96.67%. In the case of the combinations, the case of distance from drift, depth of ground water and land use used, showed the 90.71% (AUC: 0.9071) accuracy which is the best result produced in this analysis.