2D data-space joint inversion of multiple geophysicaldata with cross-gradient constraints

Each of the various exploration methods can only provide information on the physical properties of an underground medium from a single aspect, and the resolution varies significantly according to each method. To achieve comprehensive interpretation of complex subsurface bodies, joint inversion of multiple physical parameters is required. The traditional joint inversion method is mainly based on physical relationship coupling and structural similarity coupling. The above joint inversion methods are performed in model-space when the amount of physical parameters for inversion increases from two to multiple or the regional grid division of inversion is large. Such inversions demand extremely high computing times and memory usage. In this study, we apply a 2D model-space joint inversion of magnetotelluric (MT), gravity, magnetic, and seismic travel-time data with cross-gradient constraints. Based on the model-space method, we then derive the data expression in the data-space and develop a 2D data-space cross-gradient joint inversion algorithm of MT, gravity, magnetic, and seismic travel-time data. Finally, we design a complex synthetic model and invert the synthetic data using separate, model-space, and data-space inversion methods. Irrespective of model-space or data-space, results of the joint inversion are found to be superior to those of separate inversion. Furthermore, model-space and data-space inversion yield results that reflect the real model more closely. Nevertheless, the data-space method has lower memory requirement and higher calculation speed than the model-space method. Therefore, the data-space method is expected to be the preferred method for future research involving joint inversion processing of large amounts of data.