Correlation Based Geomagnetic Field Modeling

Abstract:

We present a new method for determining geomagnetic field models. It is based on the construction of an a priori correlation structure derived from our knowledge about characteristic length scales and sources of the geomagnetic field. The magnetic field measurements are then seen as correlated random variables too and the inversion process amounts to compute the a posteriori correlation structure using Bayes theorem. We show how this technique allows the statistical separation of the various field contributions and the assessment of their uncertainties.