This research program focuses on the changes of the Earth’s magnetic field (the geomagnetic field) which occur on interannual to millennial time scales, and on their physical causes, which are related to the dynamics of the Earth’s fluid outer core, where the geomagnetic field is generated through dynamo action. The goal of this project is to better constrain and understand the physical mechanisms governing the geomagnetic secular variation, in light of the available data. To this end, we will resort to the techniques of data assimilation, which are well-established in the fields of meteorology and oceanography, and have just recently come to the fore in the field of geomagnetism. The primary goal of this research program is thus to implement and validate numerical tools able to combine in an optimal fashion the information contained in the database of geomagnetic observations and the constraints imposed by dynamical models, by adjusting the model trajectory to provide an adequate fit to the data.

Two physical models will be considered in parallel, one relying on the so-called quasi-geostrophic approximation, the other fully modeling three-dimensional, buoyancy-driven magnetohydrodynamic processes. The quasi-geostrophic model is developed over a pre-existing dynamo field, and makes it possible to introduce some numerical simplifications. It is well-suited to describe the rapid (interannual to decadal) changes occurring in the Earth’s core. The three-dimensional model has the advantage of fully describing self-consistent dynamo action. But its ability to describe rapid geomagnetic changes is hampered by numerical limitations. It is better suited to the description of longer term (secular to millennial) geomagnetic variations. From this perspective, the two models we propose to use are complementary.

Each model will be put at the heart of an assimilation system able to perform retrospective and prospective analyses of geomagnetic data. These systems will be extensively validated and their robustness tested using synthetic data. Upon validation, they will be used to assimilate real geomagnetic measurements covering the past two millennia, on a series of nested time intervals. Over the past 2,000 years indeed, geomagnetic observations form a highly heterogeneous catalog, in terms of data distribution and quality, which both decrease when going backward in time. The reanalysis of this catalog by means of the tools we will have developed will make it possible to produce more homogeneous and dynamically consistent pictures of Earth’s core over these time periods. In addition, comparison of the products of the reanalyses obtained using the two different physical models will shed light on the processes governing the geomagnetic secular variation, and their connection with the underlying background dynamo mechanism. We finally intend to take advantage of the predictive power of our two systems, inherited from the dynamics they rely on, and propose candidate models to contribute to the next generation of the international geomagnetic reference field model, in 2015, which calls for a predictive secular variation over the period 2015-2020.