Item

Abstract

The core magnetic field of the Earth is generated by turbulent convection in the planet’s electrically-conducting fluid outer core (geodynamo). While the outer core cannot be directly observed, measurements of the magnetic field at and above the Earth’s surface provide clues to the state of the geodynamo (e.g., fluid flow and magnetic field in the unobserved deep interior). In recent years, there has been growing interest in using data assimilation (DA) with 3-D numerical geodynamo models to better understand the Earth’s outer core and forecast changes in the magnetic field. There remain however, significant challenges to overcome if such geomagnetic data assimilation (GDA) systems are to make optimal use of available models and data. We present a set of numerical experiments with NASA’s Geomagnetic Ensemble Modeling System (GEMS)—an ensemble Kalman filter (EnKF) based GDA system. The experiments focus on the development of “localization” and “inflation” techniques in GDA. While localization and inflation are common and often critical elements of DA systems in other applications (e.g. numerical weather prediction), localization has only recently begun to be systematically explored in GDA (Sanchez et al. 2019, 2020) and inflation has yet to be implemented. Determining an appropriate localization is difficult due to the spectral nature of the “data” being assimilated: low-degree spherical harmonics defining a magnetic potential near the core-mantle boundary. Our results show a significant improvement in forecasts of the magnetic field, particularly at smaller, less computationally demanding ensemble sizes.