Self-Constrained Magnetization Vector Inversion in the Data Space
摘要
The interpretation of magnetic data in geophysical exploration faces significant challenges when attempting to recover accurate three-dimensional magnetization models, particularly in the presence of remanent magnetization. Traditional magnetic susceptibility inversion methods, which assume magnetization is solely induced and aligned with Earth’s field, fail to account for remanent components that are prevalent in igneous rocks, metamorphic rocks, and ore deposits. This study presents a novel self-constrained magnetization vector inversion (MVI) framework implemented in the Cartesian coordinate system that addresses these limitations through a two-step approach utilizing data-space inversion methodology. In the initial step, magnetic data are inverted to determine the magnetization intensity distribution using unconstrained MVI. Subsequently, this distribution is employed to spatially constrain the final MVI model using a weighted regularization function that assigns greater weight to dominant vector components. The method was validated using synthetic models, including a single dipping body and a complex three-body system with overlapping anomalies, demonstrating superior recovery of subsurface geometry and magnetization parameters compared to conventional approaches. Application to real magnetic data from the McFaulds Lake area, Ontario, specifically the Thunderbird Vanadium-Titanium deposit, confirms the method’s effectiveness in producing geologically meaningful results with sharper boundaries and improved compactness. The self-constrained approach successfully addresses the underdetermined nature of MVI problems while eliminating the need for extensive a priori geological information, making it particularly valuable for early-stage exploration programs in geologically complex environments.