Transdimensional geometrical inversion: Application to undercover imaging using gravity data

We present and apply a 3D trans-dimensional geometrical inversion method for sampling the space of geophysical models and apply it to gravity data. This method builds on a multiple level set inversion approach to model the location of interfaces between rock units. To account for the unknown number of rock units that may be required by geophysical data, we use a birth and death process which inserts or removes rock units from an existing model. The algorithm also inverts for the geometry of rock units and their densities, while ensuring a parsimonious solution. The method is applied to field data from the prospective Boulia region (Queensland, Australia) to image rocks under sedimentary cover. In this field application, an implicit geological model derived from the interpretation of 2D seismic lines, borehole data, and geological rules is used to define prior geological information constraining the inversion. Preliminary results using the proposed algorithm indicate that up to 2 dense rock units that were not initially identified by geology alone may need to be added to the model. This shows the ability of our method to infer the presence of unseen geological features such as intrusions or facies variations, and that it could potentially be used as a tool to assist exploration geoscientists.