3-D joint inversion of multi-array data set in the resistivity method based on MPI parallel algorithm

Because of interference from local anomaly bodies in the shallow subsurface, the inversion accuracy of ground-based resistivity method is limited. Although such interference can be reduced by virtue of borehole data in the inversion, due to the limitation of drilling position, the amount of data is not guaranteed, thus precision of inversion can be reduced. In this paper, a set of three-dimensional joint inversion algorithms that can simultaneously invert surface, surface-borehole, borehole-surface and borehole-borehole array data is developed. Firstly, forward modeling is conducted by the finite different method, and restoration of the resistivity structure is made by the Nonlinear Conjugate Gradient method (NLCG). Secondly, the Message Passing Interface (MPI) function library is used to accelerate forward modeling and sensitivity matrix calculation. The maximum speedup ratio of 4. 51 is achieved by the inversion program when the 12 processes are opened up. Finally, the validity of the inversion algorithm is proved by numerical examples of synthetic data and measured data, and it is also proved that the inversion results of the combined data set of multiple arrays are obviously superior to those of a single surface array data set.