Observing system simulation experiments in geomagnetic data assimilation

Abstract

Observing System Simulation Experiments (OSSEs) have been carried out for a geomagnetic data assimilation system. The purpose of these experiments is to demonstrate how new algorithms for assimilating geomagnetic observations into a geodynamo model can be evaluated. A “nature” run is carried out to establish a “true” evolution of the Earth's core state, from which synthetic surface geomagnetic field observations are created. These observations are then assimilated into the geodynamo model using an optimal interpolation (OI) scheme. Model error is simulated by using a different Rayleigh number in the nature and model runs. Because the true core state evolution is known completely, the assimilation results can be evaluated in terms of any state variable and at any point in the computational domain. In this work we focus on the poloidal (observed) and toroidal (unobserved) components of the magnetic field throughout the outer core. Experiments are carried out using observations of different degrees and varying forecast error correlation length scales. We also investigate the impact of model error on the assimilation and on the accuracy of geomagnetic forecasts. Assimilation runs lasting about 90% of the magnetic free decay time show a positive impact on both components of magnetic field deep within the outer core. Forecasts of the surface magnetic field show much lower error growth, indicating that the initial condition used for the forecasts has been substantially improved through assimilation.