External current source modeling and simultaneous source-conductivity inversion in global induction studies

Global geomagnetic induction studies at longer periods have long relied on local transfer functions under a highly simplified P10 dipole source model. With such an approach data must be limited in period range, latitude extent,and/or geomagnetic conditions where such assumptions may be tenable. Even then, possibilities of source contamination are a serious concern. To overcome these challenges and improve global images of electrical conductivity (especially in the upper mantle) we propose to simultaneously estimate source and conductivity structures from the data. Our approach involves three steps: 1 Recently developed tools for frequency domain principal components analaysis with large amounts of missing data are applied to modern and historical geomagnetic observatory data to estimate the dominant modes of spatial variability at periods from 6 hours to 100 days. This analysis results in estimates of 3 components of the total (external + induced) magnetic field variations associated with unknown source currents. 2 Field components from step 1 are fit to external current source models, accounting for induction in a 1D Earth with a variable conductance thin-sheet at the surface. 3 Invert the field components from step 1 with a 3D spherical code, using the range of possible source models obtained from step 2, and explore trade-offs between deep Earth conductivity structure and source complications using an alternating iterative approach. We will present Earth conductivity and source models from application of this approach to longer period variations, as well as progress on source modeling for daily variation periods.