Antarctica crustal model by means of the Bayesian gravity inversion

The Antarctica crustal structure is still not completely unveiled due to the presence of thick ice shields all over the continent which prevent direct in situ measurements. In the last decades, various geophysical methods have been used to retrieve information of the upper crust and sediments distribution however there are still regions, especially in central Antarctica, where our knowledge is limited. For these kind of situations, in which the indirect investigation of the subsurface is the only valuable solution, the gravity data are an important source of information. After the recent dedicated satellite missions, like GRACE and GOCE, it has been possible to obtain global gravity field data with spatial resolution and accuracy almost comparable to those of local/regional gravity acquisitions, paving the way to new geophysical applications. The new challenge today is the capability to invert such gravity data on large areas with the aim to obtain a 3D density model of the Earth crust. This is in fact a problem characterized by intrinsic instability and non-uniqueness of the solution that to be solved requires the definition of strong constrains and numerical regularization.

In this work the authors propose the application of a Bayesian inversion algorithm to the Antarctica continent to infer a model of mass density distribution. The first operation is the definition of an initial geological model in terms of geological horizons and density. These two variables are considered as random variables and, within the iterative procedure based on Markov Chain Monte Carlo methods, they are adjusted in such a way to fit the gravity field on the surface. The test performed show that the method is capable of retrieving an estimated model consistent with the prior information and fitting the gravity observations according to their accuracy.