Study On 3D Geological Modeling And Spatial Analysis For Prediction Of Buried Ore Bodies

The prediction and evaluation of mineral resources gradually enter a new stage of quantitative prediction and evaluation using GIS techniques, while GIS has become core techniques of quantitative prediction and evaluation of mineral resources. Traditional two-dimensional GIS can only represent information based on face. For three-dimensional spatial information of deposits in the prediction of buried ore bodies at large-scale, 2D GIS can not represent and analysis very well. This shortcoming in 2D GIS can be overcome by three-dimensional GIS and 3D geological modeling system, in which geological objects can be true 3D represented. Under these true 3D spatial data environment, however, the procedure of prediction and evaluation of mineral resources needs the efficient and robust techniques of 3D geological modeling and 3D spatial analysis to support.First the techniques of 3D geological modeling in the prediction of buried ore bodies were studied. The spatial data representations of geological objects were discussed. The paper gives a solution in which the geometric contours of geological body were built by surface modeling while geological body were represented by voxel model, and stored in linear octrees. In order to meet the needs of high-resolution modeling of massive geological bodies, moreover, a hardware accelerated algorithm which can generate linear octrees dynamically from surface representation models was presented. In the algorithm, depth buffers were used to determine whether a voxel is inside or outside a 3D object, a stack structure were used to store the decomposition status of octree, and the codes are exported during the stack operation so the sort and compression of nodes is avoided. The depth buffers are further used to divide the object. Based on this division, the construction of linear octree can be accelerated. The result shows the process of linear octree generation can be run fairly fast under high resolution.To meet the needs of quantitative extraction of ore-controlling indicators, the paper studied some techniques on voxel model based spatial analysis in the prediction of buried ore bodies. A distance analysis method is proposed which is suitable in concrete three-dimensional geological space. In this method Euclidean distance transform is employed to generate Euclidean distance field. Moreover, the paper presents a morphological filtering based algorithm which can get trend shape of geological body. Based on Euclidean distance field and mathematical morphology, the method for morphological character of geological body was studied. By using the method given in study, the undulant range of morphological changes of geological body can be sebquential extracted. Furthermore, employing dilation operation in mathematical morphology to generate raster buffer of an objects, the paper presents a GPU based algorithm for raster buffer generation. Based on these techniques for spatial analysis, the study also focused on modeling the geological ore-controlling actions, which included modeling ore-controlling actions of geological bodies, the contact zone of rock and country rocks, and modeling the angles of contact surfaces.