Geomagnetic Core Field Secular Variation Models

We analyse models describing time changes of the Earth’s core magnetic field (secular variation) covering the historical period (several centuries) and the more recent satellite era (previous decade), and we illustrate how both the information contained in the data and the a priori information (regularisation) affect the result of the ill-posed geomagnetic inverse problem. We show how data quality, frequency and selection procedures govern part of the temporal changes in the secular variation norms and spectra, which are sometimes difficult to dissociate from true changes of the core state. We highlight the difficulty of resolving the time variability of the high degree secular variation coefficients (i.e. the secular acceleration), arising for instance from the challenge to properly separate sources of internal and of external origin. In addition, the regularisation process may also result in artificial changes in the model norms and spectra. Model users should keep in mind that such features can be mis-interpreted as the signature of physical mechanisms (e.g. diffusion). Finally, we present perspectives concerning core field modelling: imposing dynamical constraints (e.g. by means of data assimilation) reduces the non-uniqueness of the geomagnetic inverse problem.