Modified Kriging Spatial Interpolation Algorithm Based on Octree Mechanism
摘要
Neighborhood search is an important step in the spatial interpolation algorithm. Whether the neighborhood range is properly selected has a great impact on the interpolation efficiency and accuracy. Aiming at the problem that there were few studies on neighborhood search of interpolation algorithm, a neighborhood search strategy based on octree considering the spatial distribution of discrete points was proposed in this paper. Firstly, the minimal enclosing box of the sampling points was constructed and divided with octree, and the sampling points were grouped into each divided box. Then, the spatial distribution of the interpolating points was constrained by defining the point density. Finally, the above neighborhood search strategy is applied to the ordinary Kriging interpolation model. In order to verify the feasibility of the proposed method, in true 3D geological modeling, the proposed algorithm of this paper, the conventional Kriging interpolation based on fixed distance and fixed number strategy, and the inverse distance weighted interpolation were all used to calculate respectively, and the geological model was constructed from the data obtained from the interpolation. By comparing the method of this paper with the traditional spatial interpolation methods, it was concluded that the proposed method was superior to the traditional method in interpolation accuracy and efficiency when the same number of points were obtained under the same conditions, except for the fixed number method of 30 sample points. Although the fixed number method of 30 sample points had a slight advantage in accuracy, its calculation time is 6.6 times longer than that of the proposed method. In addition, the proposed method improved the time by 20% compared with the traditional method under the same conditions. Compared with the traditional method, the proposed method reduced the redundancy by nearly 1/3 when using the interpolated data to construct the geological model at the same level