Differences between tangential geostrophy and columnar flow.

Core flows inverted from time-dependent geomagnetic field models image the geodynamo at the top of its generation region, the Earth’s outer core. Physical assumptions incorporated in these inversions affect the resulting flows. Based on rapid rotation dominance, two assumptions similar in form yet different in essence have been proposed: tangential geostrophy (TG, LeMouël 1984) and columnar flow (CF, Amit & Olson 2004). We recall that CF is theoretically consistent with the quasi-geostrophy (QG) theory for an incompressible fluid with spherical solid boundaries whereas TG is not. As such, we highlight the importance of applying the CF assumption when inverting geomagnetic data for interior core (columnar) flows that can be used in kinematic dynamo and thermal convection models in the Boussinesq approximation. Next we evaluate the non-uniqueness associated with CF flows. The areas of ambiguous patches at the core surface where invisible TG or CF flows reside are roughly comparable. The spatial distribution of ambiguous patches for both TG and CF is quite asymmetric about the equator, so assuming equatorial symmetry is expected to reduce the non-uniqueness significantly. In fact, for assumed equatorial symmetry, the only possible non-unique flows will be those along hypothetical ζ-contours in the opposite hemispheres that their equatorial plane projections are parallel. TG flows exhibit a strong Atlantic/Pacific hemispheric dichotomy and a well-defined eccentric gyre whereas in CF flows the dichotomy between these two hemispheres is weaker and the gyre is less clear suggesting that the eccentric gyre might not conserve mass. Both TG and CF upwelling/downwelling patterns are strongly localized in the equatorial region. In addition, in both cases upwelling/downwelling is correlated with equatorward/poleward flow respectively, as expected for QG convection. CF upwelling is more intense than TG upwelling but the magnitude ratio is smaller than the factor 2 distinguishing the analytical expressions of the two assumptions. This smaller magnitude ratio is due to the fact that presently observed geomagnetic secular variation features are mostly explained by magnetic field advection by toroidal core flow in the frozen-flux approximation. Robust upwelling features below India/Indonesia may be viewed as geomagnetic evidence for whole core convection. [ABSTRACT FROM AUTHOR]