ALBERT

Description: Monthly means of the magnetic field measurements at ground observatories are a key data source for studying temporal changes of the core magnetic field. However, when they are calculated in the usual way, contributions of external (magnetospheric and ionospheric) origin may remain, which make them less favourable for studying the field generated by dynamo action in the core. We remove external field predictions, including a new way of characterizing the magnetospheric ring current, from the data and then calculate revised monthly means using robust methods. The geomagnetic secular variation (SV) is calculated as the first annual differences of these monthly means, which also removes the static crustal field. SV time-series based on revised monthly means are much less scattered than those calculated from ordinary monthly means, and their variances and correlations between components are smaller. On the annual to decadal timescale, the SV is generated primarily by advection in the fluid outer core. We demonstrate the utility of the revised monthly means by calculating models of the core surface advective flow between 1997 and 2013 directly from the SV data. One set of models assumes flow that is constant over three months; such models exhibit large and rapid temporal variations. For models of this type, less complex flows achieve the same fit to the SV derived from revised monthly means than those from ordinary monthly means. However, those obtained from ordinary monthly means are able to follow excursions in SV that are likely to be external field contamination rather than core signals. Having established that we can find models that fit the data adequately, we then assess how much temporal variability is required. Previous studies have suggested that the flow is consistent with torsional oscillations (TO), solid body-like oscillations of fluid on concentric cylinders with axes aligned along the Earth's rotation axis. TO have been proposed to explain decadal timescale changes in the length-of-day. We invert for flow models where the only temporal changes are consistent with TO, but such models have an unacceptably large data misfit. However, if we relax the TO constraint to allow a little more temporal variability, we can fit the data as well as with flows assumed constant over three months, demonstrating that rapid SV changes can be reproduced by rather small flow changes. Although the flow itself changes slowly, its time derivative can be locally (temporally and spatially) large, in particular when and where core surface secular acceleration peaks. Spherical harmonic expansion coefficients of the flows are not well resolved, and many of them are strongly correlated. Averaging functions, a measure of our ability to determine the flow at a given location from the data distribution available, are poor approximations to the ideal, even when centred on points of the core surface below areas of high observatory density. Both resolution and averaging functions are noticeably worse for the toroidal flow component, which dominates the flow, than the poloidal flow component, except around the magnetic equator where averaging functions for both components are poor.

Description: We present a new model of the radial (1-D) conductivity structure of Earth's mantle. This model is derived from more than 10 yr of magnetic measurements from the satellites Ørsted, CHAMP, SAC-C and the Swarm trio as well as the global network of geomagnetic observatories. After removal of core and crustal field as predicted by a recent field model, we fit the magnetic data with spherical harmonic coefficients describing ring current activity and associated induction effects and estimate global C -responses at periods between 1.5 and 150 d. The C -responses are corrected for 3-D effects due to induction in the oceans and inverted for a 1-D model of mantle conductivity using both probabilistic and deterministic methods. Very similar results are obtained, consisting of a highly resistive upper mantle, an increase in conductivity in and beneath the transition zone and a conductive lower mantle. Analysis of the Hessian of the cost function reveals that the data are most sensitive to structures at depths between 800 and 1200 km, in agreement with the results obtained from the probabilistic approach. Preliminary interpretation of the inverted conductivity structure based on laboratory-based conductivity profiles shows that the recovered structure in the lower mantle either requires higher temperatures or the presence of material of high conductivity related to ponding of carbonate melts below the transition zone.

Description: We demonstrate that first differences of polar orbiting satellite magnetic data in the along-track direction can be used to obtain high resolution models of the lithospheric field. Along-track differences approximate the north–south magnetic field gradients for non-polar latitudes. In a test case, using 2 yr of low altitude data from the CHAMP satellite, we show that use of along-track differences of vector field data results in an enhanced recovery of the small scale lithospheric field, compared to the use of the vector field data themselves. We show that the along-track technique performs especially well in the estimation of near zonal spherical harmonic coefficients. Moreover, lithospheric field models determined using along-track differences are found to be less sensitive to the presence of unmodelled external field contributions and problems associated with the polar gap are ameliorated. Experiments in modelling the Earth's lithospheric magnetic field with along-track differences are presented here as a proof of concept. We anticipate that use of such along-track differences in combination with east–west field differences, as are now provided by the Swarm satellite constellation, will be important in building the next generation of lithospheric field models.

Description: We present a new technique for modelling the global lithospheric magnetic field at Earth's surface based on the estimation of equivalent potential field sources. As a demonstration we show an application to magnetic field measurements made by the CHAMP satellite during the period 2009–2010 when it was at its lowest altitude and solar activity was quiet. All three components of the vector field data are utilized at all available latitudes. Estimates of core and large-scale magnetospheric sources are removed from the measurements using the CHAOS-4 model. Quiet-time and night-side data selection criteria are also employed to minimize the influence of the ionospheric field. The model for the remaining lithospheric magnetic field consists of magnetic equivalent potential field sources (monopoles) arranged in an icosahedron grid at a depth of 100 km below the surface. The corresponding model parameters are estimated using an iteratively reweighted least-squares algorithm that includes model regularization (either quadratic or maximum entropy) and Huber weighting. Data error covariance matrices are implemented, accounting for the dependence of data variances on quasi-dipole latitude. The resulting equivalent source lithospheric field models show a degree correlation to MF7 greater than 0.7 out to spherical harmonic degree 100. Compared to the quadratic regularization approach, the entropy regularized model possesses notably lower power above degree 70 and a lower number of degrees of freedom despite fitting the observations to a very similar level. Advantages of our equivalent source method include its local nature, the possibility for regional grid refinement and the production of local power spectra, the ability to implement constraints and regularization depending on geographical position, and the ease of transforming the equivalent source values into spherical harmonics.

Description: A comprehensive magnetic field model named CM5 has been derived from CHAMP, Ørsted and SAC-C satellite and observatory hourly-means data from 2000 August to 2013 January using the Swarm Level-2 Comprehensive Inversion (CI) algorithm. Swarm is a recently launched constellation of three satellites to map the Earth's magnetic field. The CI technique includes several interesting features such as the bias mitigation scheme known as Selective Infinite Variance Weighting (SIVW), a new treatment for attitude error in satellite vector measurements, and the inclusion of 3-D conductivity for ionospheric induction. SIVW has allowed for a much improved lithospheric field recovery over CM4 by exploiting CHAMP along-track difference data yielding resolution levels up to spherical harmonic degree 107, and has allowed for the successful extraction of the oceanic M 2 tidal magnetic field from quiet, nightside data. The 3-D induction now captures anomalous Solar-quiet features in coastal observatory daily records. CM5 provides a satisfactory, continuous description of the major magnetic fields in the near-Earth region over this time span, and its lithospheric, ionospheric and oceanic M 2 tidal constituents may be used as validation tools for future Swarm Level-2 products coming from the CI algorithm and other dedicated product algorithms.

Description: The C -response is a conventional transfer function in global electromagnetic induction research and is traditionally determined from observations of magnetic variations in the vertical and horizontal components. Its interpretation relies on the assumption that the source of the variations is well approximated by a large-scale symmetric (magnetospheric) ring current, described by a single spherical harmonic. However, there is growing evidence for a more complex structure of this source. In this paper, we investigate the variability of C -responses due to sources different from the dominating large-scale symmetric ring current. We show that the effect is significant and persists at all periods. Describing the magnetospheric source by a single spherical harmonic coefficient thus injects substantial errors into the estimated responses. To overcome the problem, we introduce arrays of alternative transfer functions that relate the components of the magnetic variation to different spherical harmonic coefficients. These transfer functions can handle a complex spatial structure of the magnetospheric source. Compared to C -responses, we observe a significant increase in the coherencies relating input and output quantities of the new transfer functions, especially at high latitudes. This increases the usability of observatory magnetic data for the recovery of global 3-D mantle conductivity structure.