Joint inversion of 3-D magnetotelluric and ambient noise dispersion data sets with cross-gradient constraints: methodology and application

The integrating technique of multigeophysical data sets has been treated as a feasible way to improve the imaging resolution of the subsurface structure. Based on the limited memory quasi-Newton approach, we obtain the electrical and velocity structure through the individual inversions of magnetotelluric (MT) and ambient noise dispersion data, and propose an alternating joint inversion strategy for integrating the MT and dispersion data with the cross-gradient structure constraints. A synthetic model has been applied to test the reliability and effectiveness of the joint inversion scheme. Some significant improvements have been achieved and shown in the joint inversion maps: the resolution of deep electric features is improved, especially that of the structure beneath the conductive layer; a finer velocity model is obtained than that of the ambient noise tomography. Based on the complementary advantages of the two geophysical methods, resistivity and velocity models with higher geometrical resemblance can be obtained. We also apply the method to the field data from the Tanlu fault zone, Eastern China. Structure similarities have been observed in the main features of the joint inversion models, which further validates the effectiveness and practicality of the proposed joint inversion strategy.