重磁异常三维物性反演随机子域法方法技术

3-D gravity and magnetic inversion for physical properties using stochastic subspaces

本研究针对三维物性反演中存在的困难和问题,提出三维物性反演的随机子域方法技术,首先是将正反演中保持不变的几何格架分离计算并存储,避免重复计算,从而提高正反演计算速度;其次是利用对称性等实现等效计算,明显降低格架计算和存储要求;再通过随机子域方式,降低反演的维数问题;另外,通过概率方式控制子域生成的分布,实现约束新机制.模型和实例计算表明了方法技术的效果,为大面积重磁数据的三维反演提供了有效的途径.

Focused on the existing difficulties and problems in 3-D inversion for physical properties, the paper brings forward stochastic subspace methodology for 3-D inversion for physical properties. Firstly, it computes separately and saves the geometric trellis which keeps unchanged during forward simulation and inversion to avoid repetitive computation so as to increase the speed of forward simulation and inversion computation. Secondly, it uses symmetry to realize equivalent computation, which distinctly lowers the requirement of trellis computation and storage. And thirdly with stochastic subspace inversion method it reduces the number of dimensions of inversion. In addition, it controls the distribution of the subspaces generated through probability method to realize the new mechanism of constraint. The computations of model and field data demonstrate the effect of the methodology which is hopeful to be of practicality-oriented 3-D inversion for physical properties of large scale gravity and magnetic data and meets the requirement of explanation of 3-D inversion.