Global 3-D imaging of mantle conductivity based on inversion of observatory C-responses - II. Data analysis and results

The global 3-D electrical conductivity distribution in the mantle (in the depth range between 400 and 1600 km) is imaged by inverting C-responses estimated on a global net of geomagnetic observatories. Very long time-series (up to 51 years; 19572007) of hourly means of three components of the geomagnetic field from 281 geomagnetic observatories are collected and analysed. Special attention is given to data processing in order to obtain unbiased C-responses with trustworthy estimates of experimental errors in the period range from 2.9 to 104.2 d. After careful inspection of the obtained C-responses the data from 119 observatories are chosen for the further analysis. Squared coherency is used as a main quality indicator to detect (and then to exclude from consideration) observatories with a large noise-to-signal ratio. During this analysis we found thatalong with the C-responses from high-latitude observatories (geomagnetic latitudes higher than 58 degrees)the C-responses from all low-latitude observatories (geomagnetic latitudes below 11 degrees) also have very low squared coherencies, and thus cannot be used for global induction studies. We found that the C-responses from the selected 119 mid-latitude observatories show a huge variability both in real and imaginary parts, and we investigated to what extent the ocean effect can explain such a scatter. By performing the systematic model calculations we conclude that: (1) the variability due to the ocean effect is substantial, especially at shorter periods, and it is seen for periods up to 40 d or so; (2) the imaginary part of the C-responses is to a larger extent influenced by the oceans; (3) two types of anomalous C-response behaviour associated with the ocean effect can be distinguished; (4) to accurately reproduce the ocean effect a lateral resolution of 1 degrees x 1 degrees of the conductance distribution is needed, and (5) the ocean effect alone does not explain the whole variability of the observed C-responses. We also detected that part of the variability in the real part of the C-responses is due to the auroral effect. In addition we discovered that the auroral effect in the C-responses reveals strong longitudinal variability, at least in the Northern Hemisphere. Europe appears to be the region with smallest degree of distortion compared with North America and northern Asia. We found that the imaginary part of the C-responses is weakly affected by the auroral source, thus confirming the fact that in the considered period range the electromagnetic (EM) induction from the auroral electrojet is small. Assuming weak dependence of the auroral signals on the Earths conductivity at considered periods, and longitudinal variability of the auroral effect, we developed a scheme to correct the experimental C-responses for this effect. With these developments and findings in mind we performed a number of regularized 3-D inversions of our experimental data in order to detect robust features in the recovered 3-D conductivity images. Although differing in details, all our 3-D inversions reveal a substantial level of lateral heterogeneity in the mantle at the depths between 410 and 1600 km. Conductivity values vary laterally by more than one order of magnitude between resistive and conductive regions. The maximum lateral variations of the conductivity have been detected in the layer at depths between 670 and 900 km. By comparing our global 3-D results with the results of independent global and semi-global 3-D conductivity studies, we conclude that 3-D conductivity mantle models produced so far are preliminary as different groups obtain disparate results, thus complicating quantitative comparison with seismic tomography or/and geodynamic models. In spite of this, our 3-D EM study and most other 3-D EM studies reveal at least two robust features: reduced conductivity beneath southern Europe and northern Africa, and enhanced conductivity in northeastern China.