Shipborne electromagnetic measurements of Antarctic sea‐ice thickness

We present a study of Antarctic sea‐ice thickness estimates made using a shipborne Geonics EM31 electromagnetic (EM) instrument, based on both 1D and 3D models. Apparent conductivities measured in the vertical coplanar (VCP) geometry are shown to be the measured quantity most sensitive to changes in the height of the instrument above seawater. An analysis of the effect of instrument orientation on the measured VCP apparent conductivity shows that the effects of pitch and roll on the calculated sea‐ice thickness can be neglected except in the case of very thin sea ice. Because only a single (quadrature) component of the magnetic field is measured at a single frequency, interpretation of shipborne EM31 data must necessarily be based on very simple models. For a typical sea‐ice bulk conductivity of ∼60 mS/m, a uniform half‐space model representing conductive seawater is appropriate for interpretation of VCP EM31 measurements over level sea ice up to ∼2.5 m thick. For thicker, more conductive sea ice, the interpretation model must account for the effect of the finite sea‐ice conductivity. Simultaneous acquisition of EM data at several frequencies and/or transmitter–receiver geometries permits interpretation of the data in terms of multilayered models. A synthetic example shows that 1D inversion of single‐frequency in‐phase and quadrature data from two transmitter–receiver geometries can yield reliable estimates of sea‐ice thickness even when the ice contains thin, highly conductive brine layers. Our 3D numerical model calculations show that smoothing the measured response over the system footprint means that the sea‐ice thickness recovered over multidimensional sea‐ice structures via half‐space inversion of apparent conductivity data yields a highly smoothed image of the actual keel relief. The dependence of footprint size on the height of the system above seawater results in the interpreted sea‐ice thicknesses being dependent on the deployment height of the instrument. Sea‐ice thickness data acquired using an EM31 equipped with a hardware processing module can be transformed to apparent conductivity and then inverted assuming a conductive half‐space model. For EM system heights >4.5 m above seawater, corresponding to large altitude and/or thick sea ice, inversion assuming a conductive half‐space model yields an improved estimate of the true sea‐ice thickness compared to that obtained using the processing module. However, the noise level in the estimated depth to seawater is relatively large (±0.1 m) in comparison with typical Antarctic sea‐ice thicknesses, and thickness estimates made using the shipborne system may be significantly in error over thin ice.