High-resolution geomagnetic field modeling and forecasting

We use geomagnetic observatory data, survey data and satellite data from the CHAMP, Oersted, MAGSAT, DE-2 and POGO missions to derive two time-dependent spherical harmonic models of Earth's magnetic field at the core-mantle boundary: one for the years 1957-2009 and the other for the years 2001-2009 (in order to investigate the limits of core field resolution with the most recent, highly accurate data). We pay particular attention to observatory and satellite data analysis and to spatial and temporal data distributions in order to separate external and internal fields. Our approach is to produce models with varying spatial roughness and to examine them with respect to correlations with known structures of core and crustal fields. The final models are consistent with other main field models in their general structure, but show differences predominantly in places where main field features are known to be complex (e.g. the South Atlantic Anomaly). Thus, the models reveal a more detailed spatial and temporal structure of the magnetic field at the core-mantle boundary. Such high-resolution models can be used to infer small-scale core surface flows and core dynamics.;We use the 1957-2009 geomagnetic field model to derive time-dependent core flow models and produce hindcasts of the Earth's main magnetic field. The goal of this study is to explore whether we can accurately forecast changes in geomagnetic secular variation by advecting core-surface flows forward in time and accounting for torsional oscillations. We compare hindcasts produced over different time intervals and computed from steady and time-varying core flow models, and also consider differently parametrized core flows (such as steady flow, steadily accelerated flow and steadily accelerated flow with torsional oscillations). We find that the steadily accelerated flow plus torsional oscillations is able to accurately reproduce changes in the Earth's magnetic field for short-term (5 years) and medium-term (13 years) hindcasts and over time intervals characterized by both slower and faster secular variation. We also find that hindcasts are strongly dependent on the accuracy of the core flow models, and that hindcasts can be improved by properly accounting for non-steady flow acceleration in addition to torsional oscillations.