A petrophysical joint inversion of magnetotelluric and gravity data for enhanced subsurface imaging of sedimentary environments.

Abstract

An emerging field in geophysics is that of joint inversions, in which multiple technique data sets are analysed and inverted simultaneously. This helps to integrate the complementary data sets and reduce model ambiguity, common in single technique inversions. In this thesis a new implementation of a magnetotelluric (MT) and gravity 2D joint inversion scheme is developed based on a petrophysical approach. In sedimentary rock environments, electrical conductivity (which underpins the MT technique) can be approximated by Archie’s Law, whereas density (which underpins the gravity technique) can be derived from the porosity-density relationship. Since both expressions are themselves dependent on porosity, this petrophysical property provides the crucial link exploited by the 2D joint inversion. The 2D joint inversion approach devised here inverts directly for a porosity model, which is converted to resistivity and density models through Archie’s Law and the porosity-density relationship, then constrained (fitted) by the MT and gravity data. Thus, a single porosity model is produced that satisfies both data sets. By means of synthetic data inversions, it was established that the joint inversion is more effective in reproducing the true subsurface model than can be achieved by an MT or gravity inversion alone. Models produced by the joint inversion show improved placement of subsurface features and a greater accuracy of reconstructing the original subsurface (physical property) values. For optimal joint inversion results, broadband MT data should be used in favour of long period MT data, and the number of gravity stations should be greater than or equal to the number of MT stations. The joint inversion is particularly useful in extracting coherent information from noisy MT data when combined with good quality gravity data. While evaluating the MT and gravity compatibility, a new method was developed for evaluating the information contained in the MT Jacobian (sensitivity) matrix. The Renmark Trough in South Australia is an area of current geothermal interest for which multi-technique data (seismic, gravity, MT) exists. These field data were used to demonstrate and verify the effective use of the joint inversion in a practical real-world example. The Renmark Trough is a half graben structure with the Hamley Fault delineating the north-east boundary. At the Hamley Fault, the base of the trough is 3.5 km deep and rises gradually in a south-west direction. The inversion of the MT data alone produced a model inconsistent with seismic knowledge of the basement depths and geometries. In contrast, the joint inversion yielded a more geologically accurate image of the trough and faithfully reconstructed the basement depths and geometries. In the process of developing the joint inversion scheme, a 2D gravity inversion algorithm, based on the Occam maximum smoothness approach, was produced. This inversion algorithm demonstrated the inherent non-uniqueness of gravity interpretation by only placing strong density contrasts at the surface. Attempts to improve the gravity inversion results, such as the use of depth weighting functions and fixing structure locations in parts of the model, were not as effective as the joint inversion in producing an accurate representation of the subsurface.