Research on Detailed Lithological Models and Inversion for Coal Mines Under Multi-Source Data Constraints

Accurate characterization of coal seam thickness and the lithology of roof/floor stratum is a key aspect of Three-dimensional (3D) geological modeling in coal mines. Currently, one of the most effective lithology prediction methods is geostatistical inversion, although the accuracy of geostatistical inversion mainly depends on the precision of the initial stratigraphic framework model. This paper compares three methods of stratigraphic framework construction based on geological statistical inversion and proposes a method that primarily uses seismic data and incorporates multi-source data using radial basis function (RBF). The results show that the stratigraphic framework model constructed using RBF improves the accuracy by approximately 0.2 m compared to the kriging-based model and by approximately 1 m compared to the model based solely on seismic data. This method allows for the precise representation of stratigraphic sequences in 3D space, significantly enhancing the accuracy of the stratigraphic framework model. The inversion results of the stratigraphic framework based on RBF show an improvement of approximately 0.2 m in accuracy compared to the kriging-based inversion model and an improvement of approximately 0.8 m compared to the inversion model based on seismic data alone. A reliable stratigraphic framework model is crucial for improving the accuracy of inversion results. High-precision stratigraphic framework models greatly enhance the accuracy of the final geological lithology model built from inversion. The constructed geological lithology model can more accurately depict the thickness and spatial distribution of various lithologies, providing significant guidance for intelligent and unmanned coal mining operations.