Pattern-Based Multiple-point Geostatistics for 3D Automatic Geological Modeling of Borehole Data

Urban 3D geological modeling is an essential approach for quickly understanding the underground geological structure of a city and guiding underground engineering construction. Modeling methods based on multiple-point geostatistics can provide probabilistic results regarding geological structure. The traditional multiple-point geostatistics modeling approach is characterized by low efficiency and typically relies on data from geological sections or conceptual models; therefore, it cannot be well applied to practical geological exploration projects that are based primarily on borehole data. In this paper, we propose a pattern-based multiple-point geostatistics modeling method PACSIM (pattern attribute classification simulation). This method uses borehole data as the primary data. First, geological structural information is extracted based on the borehole data to establish a training image database. Next, based on the distribution patterns of geological structures, a method for establishing attribute-based pattern databases is proposed to enhance modeling accuracy. Finally, a probability constraint strategy is introduced to address the distribution of complex strata and filter out grids with high certainty, thereby further improving the modeling accuracy. Based on the aforementioned strategies, a multiple-point geostatistics modeling workflow specifically targeting underground geological structures in urban areas was designed and subjected to practical verification. The results indicate that the proposed method required less time than the PSCSIM method, and improved the modeling efficiency by 72.87% while ensuring the accuracy of the modeling results. It can accurately identify relationships among complex strata and match the stratum distribution patterns revealed by borehole data, providing a reference for high-precision geological modeling in cases with high uncertainty.