Multiple-Point Geostatistics-Based Three-Dimensional Automatic Geological Modeling and Uncertainty Analysis for Borehole Data
Jiateng Guo,Zhaoxue Wang,Chaoling Li,Fengdan Li,Mark Walter Jessell,Lixin Wu,Jiangmei Wang
DOI: https://doi.org/10.1007/s11053-022-10071-6
IF: 5.609
2022-05-15
Natural Resources Research
Abstract:The three-dimensional characterization of geological structures is important for determining the distribution of subsurface mineral resources. However, geological structures and geological phenomena have great stochasticity and uncertainty at the microscopic level. Traditional multiple-point geostatistics use mostly three-dimensional conceptual models or two-dimensional sections as training images, while simulations using direct low-dimensional borehole data are lacking. In this paper, we propose a new multiple-point geostatistical method to automatically reconstruct three-dimensional geological models directly from borehole data, which can significantly reduce the complexity of intermediate manual operations. First, the geological structure characteristics in the borehole data are extracted, and then an initial prior model is constructed based on geological constraints. Next, for the non-stationary problem, the mobile local scan approach is proposed to make the simulation nodes scan in a certain range of scaled area to simultaneously achieve the zonal simulation effect and eliminate the discontinuity problem between zonal boundaries. Based on this solution, the whole modeling workflow is designed. Finally, the algorithm is validated using actual plains area geological survey data, compared to other modeling methods, and evaluated for model uncertainty. The results show that the proposed 3D geological modeling method can effectively expose the stratigraphic structural morphology, stratigraphic attributes and overburden relationships. It will provide decision support for resource exploration and reduce exploration costs.
geosciences, multidisciplinary