3D stochastic inversion and joint inversion of potential fields for multi scale parameters

In this thesis we present the development of new techniques for the interpretation of potential field (gravity and magnetic data), which are the most widespread economic geophysical methods used for oil and mineral exploration. These new techniques help to address the long-standing issue with the interpretation of potential fields, namely the intrinsic non-uniqueness inversion of these types of data. The thesis takes the form of three papers (four including Appendix), which have been published, or soon to be published, in respected international journals. The purpose of the thesis is to introduce new methods based on 3D stochastical approaches for: 1) Inversion of potential field data (magnetic), 2) Multiscale Inversion using surface and borehole data and 3) Joint inversion of geophysical potential field data.;We first present a stochastic inversion method based on a geostatistical approach to recover 3D susceptibility models from magnetic data. The aim of applying geostatistics is to provide quantitative descriptions of natural variables distributed in space or in time and space. We evaluate the uncertainty on the parameter model by using geostatistical unconditional simulations. The realizations are post-conditioned by cokriging to observation data. In order to avoid the natural tendency of the estimated structure to lay near the surface, depth weighting is included in the cokriging system.;Then, we introduce algorithm for multiscale inversion, the presented algorithm has the capability of inverting data on multiple supports. The method involves four main steps :i. upscaling of borehole parameters (It could be density or susceptibility ) to block parameters, ii. selection of block to use as constraints based on a threshold on kriging variance, iii. inversion of observation data with selected block densities as constraints, and iv. downscaling of inverted parameters to small prisms. Two modes of application are presented : estimation and simulation.;Finally, a novel stochastic joint inversion method based on cokriging is applied to estimate density and magnetic susceptibility distributions from gravity and total magnetic field data. The method fully integrates the physical relations between the properties (density and magnetic susceptibility) and the indirect observations (gravity and total magnetic field). As a consequence, when the data are considered noise-free, the inverted fields exactly reproduce the observed data. The required density and magnetic susceptibility auto- and cross covariance are assumed to follow a linear model of coregionalization (LCM).;In all the methods presented in this thesis, compact and stochastic synthetic models are investigated. The results show the ability of the methods to invert surface and borehole data simultaneously on multiple scale parameters. A case study using ground measurements of total magnetic field and gravity data at the Perseverance mine (Quebec, Canada) is selected and tested with the 3 approaches presented. The recovered 3D susceptibility and density model provides beneficial information that can be used to analyze the geology of massive sulfides for the domain under study.