A testbed for geomagnetic data assimilation

Abstract

Geomagnetic data assimilation merges past and present-day observations of the Earth’s magnetic field with numerical geodynamo models and the results are used to initialize forecasts. We present a new ‘proxy model’ that can be used to test, or rapidly prototype, numerical techniques for geomagnetic data assimilation. The basic idea for constructing a proxy is to capture the conceptual difficulties one encounters when assimilating observations into high-resolution, 3-D geodynamo simulations, but at a much lower computational cost. The framework of using proxy models as ‘gate-keepers’ for numerical methods that could/should be considered for more extensive testing on operational models has proven useful in numerical weather prediction, where advances in data assimilation and, hence, improved forecast skill, are at least in part enabled by the common use of a wide range of proxy models. We also present a large set of systematic data assimilation experiments with the proxy to reveal the importance of localization and inflation in geomagnetic data assimilation.