oa Remanent magnetisation inversion

Remanent magnetisation is an important consideration in magnetic interpretation. In some cases failure to properly account for remanence can lead to completely erroneous interpretations. In general the strength and orientation of remanence are unknown. Two main strategies have been pursued for “unconstrained” inversion of large data sets. One strategy is to invert quantities, such as total magnetic gradient (3D analytic signal), which are insensitive to magnetisation direction. The inverted property is then magnetisation amplitude. Another strategy is to invert for the magnetisation vector, allowing its three components to vary freely. These approaches are useful, but the resulting magnetisation models are highly non-unique.

When interpreting magnetic data in tandem with geological modelling there is greater potential to infer remanence parameters. Non-uniqueness is reduced if the shape of magnetic domains is constrained, especially if the susceptibility is known and if remanence can be assumed uniform. Accordingly, inverting for the remanent magnetisation of individual homogeneous geological units of arbitrary 3D shape is the subject of this paper. Our remanent magnetisation inversion (RMI) approach can be regarded as a generalisation of parametric inversion of simple geometric bodies.

If susceptibility is known, the optimal remanent magnetisation vector within each selected unit is determined via iterative inversion. Sensitivity to change in magnetisation is determined in the x-, y-, and z-directions, and the perturbation vector is found via the method of steepest descent. If the susceptibility is unknown, the optimal susceptibility of each unit (subject to bounds) can be determined via a similar inversion procedure. The geological units can carry remanent magnetisation, but it is fixed during this stage. The susceptibility and/or remanence inversions can be repeated, if necessary, to refine the magnetic parameters. Self-demagnetisation and interactions are taken into account when susceptibilities are high.

The application of the RMI algorithm is illustrated in examples for both known and unknown susceptibility.