Estimating sea ice thickness from its freeboard

The complex relation between sea ice draft and freeboard

The European CryoSat-2 satellite is held as the future for basin-scale sea ice thickness monitoring, transforming the measured freeboard into ice thickness. This transformation is fraught with problems which have not been resolved, however.

It is now generally acknowledged that the reflecting surface is unconstrained – initially it was stated that the satellite radar would ‘see’ the ice-snow interface, but this has proven not to be the case, with reflection usually occurring from somewhere within the snowpack, depending on the properties of the snow cover (particularly any layering structure) and the surface roughness of the ice.

In areas with significant ridging or areas with mixed ice types, the altimeter thickness estimate is much lower than the real average footprint thickness, since the high backscatter magnitude
from the thinnest ice dominates the radar return. Ridges have a low backscatter and freeboard will be significantly underestimated. The error induced by this effect over mixed ice types can be massive, leading the altimeter to determine ice thickness as 1 m (thin FY ice) instead of its true mode at 3m (MY ice), for instance. Footprints that would be correctly rendered – those with one single ice type and no ridging - are uncommon.

These problems relate to the top surface (ice or snow). Even less constrained is the relation between features on the top surface to those beneath the water, which represent the majority of the ice volume. Recent work by the proposers as part of the DAMOCLES project (Wadhams & Doble, 2008; Doble et al, 2009; Doble et al., 2011) matched mosaics of ice draft from AUV upward-looking sonar to co-incident maps of freeboard from an airborne scanning laser profilometer, and demonstrated that the freeboard:draft ratio is highly variable. The R value (draft/freeboard) over 350,000 points was 3.8 ±2.4 (mode ±STD). These - highly variable - values are very different to the value of 7.9 taken as representative of thick MY ice (Wadhams et al., 1992). The investigations were conducted over relatively small areas (400 m radius) due to limitations of the Gavia AUV, however. It is crucial to examine the relation over larger areas, similar to a typical Cryosat-2 footprint (7000 x 250 m) to investigate whether the
cloud of possible thicknesses resulting from the highly variable R on the smallest scales converges to an acceptable error limit at larger scales, and if so, at what scale this occurs;
whether there are process-related breakpoints in the R-scale relation; how the area-averaged freeboard measurement actually relates to the true thickness distribution in a given region; whether this average is in fact a modal value or whether the deformed ice plays a significant role in the radar return; and thus whether the footprint-averaged return in fact biased in these terms.

Cryosat “Arctic ice volume estimates” (e.g. Laxon et al., 2013) currently assume a draft:freeboard relation for exclusively level ice.

Cal-val campaigns funded by the European Space Agency have notably omitted any underwater component to determine the actual ice draft within the footprint, relying on airborne electromagnetic induction (EM) to give ice thickness. Though EM does indeed give a reasonable level ice thickness, it is very poorly constrained over deformed ice, due to a combination of footprint effects (averaging measurements over the sensing toroid of c.30 m diameter) and the significant (water-filled) porosity of deformed ice features.

Correctly addressing this yawning gap in our conversion of space-based freeboard to ice thickness requires a concerted campaign by long-range AUVs fitted with upward-looking multibeam sonar. Track length needs to be order(100 km) which has been shown to satisfactorily characterise any given ice regime. Co-incident overflights of the same region with an EM bird will give the surface profiles (the latest birds are fitted with scanning laser profilometers) required to relate draft to freeboard as well as allowing us to determine the EM thickness relation over deformed ice, driving the EM method forward to become a practicable thickness measure over a heterogeneous sea ice cover.

We have begun work on this task under the DAMOCLES and now ACCESS EU programmes, though progress is slowed by a lack of availability of a suitable long-range AUV and its associated significant mobilisation costs. A national programme, for instance with the IFREMER
Explorer-class vehicles, would have a significant impact on our knowledge and understanding of sea ice thickness at this critical time.

Researchers involved: Martin Doble (LOV), Peter Wadhams (U. Cambridge), Christian Haas (York University), Stefan Hendricks (AWI)