An analysis of ionospheric versus oceanic tidal magnetic signals

Observed tidal geomagnetic variations are due to a combination of electric currents in the ionosphere, ocean, and their induced counterparts. Using ocean tides to constrain subsurface electrical conductivity in oceanic regions is a promising frontier; however, removing the ionospheric tidal components from the observed data is critical for this. To achieve this goal, we compare the M2 lunar semidiurnal tidal signals estimated from global observatories to physics-based forward models of the ionospheric M2 magnetic field and the oceanic M2 magnetic field. Our global analysis used hourly data from 65 non-polar geomagnetic observatories to directly fit for the M2 amplitudes. Our study focuses on the recent deep solar minimum (May 28-Aug. 28, 2009), when the magnetic disturbance was minimal and ionospheric tidal signals were expected to be strongest in the northern hemisphere. We simulate the EM signals due to the oceanic and ionospheric M2 tidal flow using the frequency-domain numerical solution described in Kuvshinov [2008]. The oceanic electric current source was prepared using the global tidal model HAMTIDE, laterally-variable seawater conductivity, and the World Magnetic Model. Meanwhile, the ionospheric electric current source was created using the Whole Atmosphere Community Climate Model climatological lunar tidal simulation as a lower boundary (32 km altitude) condition on the Thermosphere Ionosphere Mesosphere Electrodynamics General Circulation Model. At ground level, predicted ionospheric M2 signals are strongest in the horizontal components whereas the predicted oceanic signals are strongest in the vertical direction. We find that the ionospheric and oceanic forward model closely describes the components of the observed M2 tidal signal. Typical observed amplitudes range from 0.02 nT to 3.83 nT, whereas the observations' standard error ranges from 0.07 to 0.55 nT. The observed Z-component amplitudes have both the lowest median standard error value (0.14 nT) and the lowest median amplitude strength (0.43 nT). Using a chi-square goodness of fit test, we find that the components have reduced chi-squared values of 2.8 for X, 6.1 for Y and 14.4 for Z. Overall, the agreement between the physics-based model predictions and the observations is very encouraging for electromagnetic sensing applications.