Abstract: The existing coal seam modeling methods using discrete points or geological boundaries from multi-source data lack enough attention to the dynamic changes of coal seam in the mining process. A new 3D coal seam fine modeling method is proposed. Firstly， creating regional grid of the mining area and converted to triangulated irregular network (TIN)，inter polating attributes of TIN nodes (elevation and thickness)using ordinary Kriging method，and then a generalized tri-prism(GTP) model of coal seam will be built as the initial model. Secondly， the initial model is refined gradually by the dynamic data within the influence domain of these data. A prototype system is created and implemented in Qianjiaying coal mine， Tangshan， China. The result shows that this method can accurately reflect the change of 3D coal seam model from the dynamic data， thus the accuracy of the 3D coal seam model can be improved gradually.

Key words: multi-source data, coal seam modeling, ordinary kriging, fault simulation;variogram