Gravity inversion for modelling of salt structures in offshore Abu Dhabi using trans-dimensional parametrization

This study demonstrates a nonlinear inversion of gravity data to image the Infracambrian Hormuz salt structure of Ghasha oilfield offshore Abu Dhabi, UAE. Bayesian approach with a trans-dimensional (trans-D) parametrization of the subsurface model is carried out without any constraints and regularization, bringing more objective inversion outputs. The trans-D parametrizations represent a discretization of the salt structure by an irregular grid of Voronoi nodes. Both the number and positions of Voronoi nodes are unknown parameters estimated from gravity data. A parallel tempering technique is applied to achieve an efficient rjMCMC sampling and increase acceptance rates. We apply a predefined basement from depth around 8 km to 10 km in the model space based on prior geological information. Moreover, we define a depth of 3.5 km as the lower bound of depth in our model space. As a result, Voronoi nodes place into the truncated model space that does not include the basement and depths shallower than 3.5 km. This discretization results in a salt model consistent with the geological information in Ghasha. This study also provides new insight into the existence and shape of oil reservoirs associated with the underlying Ghasha by introducing a piercing and a non-piercing salt structure.