Scientific Rationale: Recent models of the Earth's magnetic field, derived from observatory and satellite data, present rapid temporal evolutions, on subdecadal time-scales, that have not been seen before. These rapid evolutions are particularly apparent in the secular acceleration, i.e. in the time derivative of the core field secular variation. What is the origin of this signal? How much is it distorted by the conductive mantle? What processes in the core can give rise to these fast evolutions? This team addresses these questions covering three specialties: Data analysis, Core flow modeling and Magnetohydrodynamics. Regarding the data analysis, the repeatability of the observed signal and its evolution in time has to be investigated; the impact of the conductive mantle estimated. As the magnetic field acceleration is now better resolved, the temporal evolution of the core flow models can be studied. The frozen-flux approximation and quasi-geostrophy assumptions need to be tested by, for example, characterizing under these assumptions hydromagnetic normal modes in the core that could be responsible for the observed fast changes in both the field and the flow.