Coal seam surface modeling and updating with multi-source data integration using Bayesian Geostatistics

A reliable coal seam surface model needs to reconcile all available geological data such as boreholes, cross-sections, and coal seam floor contour maps. In addition, the model should be updated when local geological information such as coal seam observations is available. This paper develops a Bayesian Geostatistical approach for coal seam surface modeling using multi-source geological data in different stages and at different scales. The proposed approach contains two major components: Bayesian maximum entropy (BME) nonlinear estimation method to incorporate boreholes, cross-sections, and coal seam floor contour maps obtained in geological survey stage, and Bayesian inference (BI) method to assimilate coal seam point observations and geological sketches of tunnels obtained in mining stage. Coal seam surface elevations and its uncertainties are first estimated using BME method. The regional estimates are then used in BI method as prior knowledge, and updated when coal seam observations at a local scale are available. This provides a systematic and rigorous framework to incorporate multi-source geological data, and an effective way to improve the accuracy of coal seam surface models. The proposed approach is illustrated through a case study of a 3D subsurface modeling of the Wang-feng-gang Coal Mine, China. The coal seam surface estimates are compared with those of Ordinary kriging and Bayesian kriging methods, and compared with observed values along two tunnels in mining process.