Two-dimensional joint inversion of seismic and gravity data with cosine similarity structural constraint

摘要

We introduce a new structural coupling constraint for joint inversion with an application of seismic traveltimes and gravity data. We call the new constraint 'Cosine Similarity' as it calculates the cosine inner product between the model vectors or the gradients of models. Compared with the other common-used constraints, this new method is more feasible and reasonable, and it can reduce the uncertainty in using the rock's physical relationship, the computing complexity by cross-gradient, and avoid the artificial a priori setting for the model variation directions required by the gradients inner product constraint. We test the algorithm on synthetic models with gravity and seismic data and prove it is feasible and effective.