To address the issues of large anomalous triangulations, invalid interpolations, and uneven boundary interpolations in kriging interpolation, we propose research on a coal seam modeling construction method based on improved kriging interpolation. The work methodology assumes that by introducing kriging interpolation and analyzing its problems, we improve the interpolation method via a local interpolation algorithm for large anomalous triangulations, an optimization algorithm for locally redundant interpolation points, and a nonuniform boundary adaptive local interpolation algorithm. These improvements allow the interpolation method to better reflect the variability and realistic nature of coal seams. The research results indicate that applying this method to the construction of the Dananhu No. 2 open-pit mine coal seam model has improved the issue of coal seam transition stiffness, such as abnormal large-area triangulation in areas with significant elevation differences. This approach appropriately reduces the memory space usage without altering the coal seam morphology (which saves approximately 27,000 KB of memory, equivalent to the space occupied by 4 out of 21 coal seams). It has also prevented inaccuracies in boundary line positioning and transitions caused by too low a density of points on the coal seam reserve boundary line, resulting in smoother model transitions at the boundaries that better align with the actual coal seam change trends, the error rate in coal quality estimation decreased by 62.69%. This study provides data support for mining planning and reduces costs. This method can be extended to the construction of all mine models.