Inversion of gravity data for base salt

We develop an algorithm for inverting gravity data to construct estimates of the base of salt and investigate the sensitivity of the recovered model to four different sources of errors. The inversion algorithm is based on the approach of Tikhonov regularization.The prior information is typically shape and depth of the top salt and of a known part of base salt from seismic image. The error in such prior information governs the reliability of the recovered base of salt. It is therefore important to understand the influencing factors of the inversion: (1) input gravity data, (2) top of salt, (3) known part of the base salt, and (4) the background density profile. We use a synthetic model to illustrate the algorithm and find that the recovered model can be a good representation of the true model in the absence of errors. Moderate perturbation in any of these four factors could lead to large errors in the recovered model, leaving open the question as when gravity data can aid in improving the image of base salt.