Here’s my 2 cents.
In short, BedMachine v3 bathymetry, ice shelf draft, and grounding line (via the mask) is probably your best first bet at the continental scale. You may need to “correct” or smooth some variables in that dataset, depending on mesh resolution and how you interpolate. But it’s a pretty good leading order estimate for the variables you’re interested in. They update regularly though, so worth keeping an eye on the NSIDC website.
** Really important: I’m sure you’re aware but people use different polar stereographic projections, some referenced to -70 and others to -71, so it’s important to check. Also check whether data are reported in ice equivalent elevation or actual elevation. E.g., BedMachine corrects the REMA surface elevation to ice equivalent using an assumption of firn density (i.e., it converts the snow and firn layers to ice equivalent), which will naturally impact ice thickness calculation using hydrostatic equilibrium – see more below.
The community does rely a lot on BedMachine, but for high-resolution regional applications might draw on regional studies that have more complete datasets, particularly if time series (e.g., of grounding line migration) are available.
More details:
Bathymetry. For ocean bathymetry, BedMachine integrates what’s in the latest IBCSO with datasets derived from other sources, e.g., gravity inversions. It doesn’t have all the latest data, but I’d say it’s a product that’s more regularly updated than other bathymetries, so you might consider using that for the ice shelf cavities. There’s also a recent whole-Antarctic gravity dataset (Bathymetry of the Antarctic continental shelf and ice shelf cavities from circumpolar gravity anomalies and other data | Scientific Reports) that I haven’t used, but would recommend considering as the most complete dataset for gravity-derived cavity bathymetry. Other regional gravity inversions might be higher resolution, so it could be a case of doing a bit of a trawl through the literature if you find areas that are particularly sensitive where BedMachine or Raphaelle’s datasets don’t cut it.
Ice shelf draft. BedMachine calculates ice shelf draft assuming hydrostatic equilibrium. That’s a good leading order approximation and I’d probably recommend that as a first start. If ice sheet modellers don’t use BedMachine shelf draft, they would probably still assume hydrostatic equilibrium to calculate ice shelf thickness, e.g., using REMA surface elevations (REMA is available at 8 m; BedMachine uses REMA but interpolates to 500 m). Hydrostatic equilibrium can be a bit iffy: in the smaller embayed ice shelves; anywhere there are ice shelf interactions with topography that’s not resolved (e.g., a sub-ice shelf pinning point that’s not mapped in the bathymetry); in the ~10-20 km near the grounding zone; anywhere there’s significant damage (crevassing). So you may need to think a bit more here. You can sometimes get better estimates of shelf draft from radargrams (all the available datasets, ungridded, are summarised in the Bedmap3 dataset here), but it’s not guaranteed if you’re in an area of heavy crevassing where the radar might struggle. Bedmap3 might provide a good sanity check if you’re really struggling with difficult areas where the BedMachine estimates don’t make sense, but it’s also important to consider the impact of interpolation scheme when using gridded data.
Grounding line. Recommend using BedMachine mask in the first instance. But keep in mind that grounding zones can be super transient (100s of m of grounding line migration associated with variable ocean forcing on sub-annual timescales). Continent-wide satellite-derived estimates (e.g., from Rignot et al., 2016; MODIS; ASAID) outdate very quickly in regions that are evolving quickly, and similarly to bathymetry, the regional studies might provide the most accurate and high-resolution characterisation of the grounding line, but they’re not available everywhere. There are some machine learning-derived products (e.g., Mohajerani et al., 2021) which are reasonably good, but also not necessarily continuous.
