Two-dimensional joint inversion of DC resistivity and seismic refraction traveltime data constrained by Gaussian mixture clustering

In the field of near-surface geophysical exploration, both direct current resistivity and seismic refraction methods are widely used. The observed data for these two methods are usually inverted and interpreted separately. The direct current resistivity method exhibits high sensitivity to high-conductivity anomalies, while the seismic refraction method excels in resolving vertical velocity changes at stratigraphic interfaces. By combining these two methods, we can achieve complementary benefits and reduce ambiguity, thus enhancing the inversion. Conventional joint inversion algorithms often rely on the direct relationship between rock physical properties or structural similarity. However, these algorithms may not be practical for all situations. To address this issue, this paper presents a novel joint inversion approach based on a Gaussian Mixture Model (GMM) constraint. By incorporating the GMM into the objective function, we can mutually constrain resistivity and velocity, enabling two-dimensional joint inversion of resistivity and seismic refraction methods. Synthetic examples demonstrate that the GMM is not only suitable for separate constrained inversion but also effective for joint inversion. When accurate prior statistical information of rock properties is available, employing the GMM for joint inversion yields the best results. Even when utilizing inaccurate prior GMM information, the GMM constrained joint inversion still outperforms separate inversion.