RGI19 (Antarctica): current status and options

[fmaussion]

RGI19 currently has no changes to RGI6, except for Kerguelen where we will use an improved inventory.

Here is what we have done:

• we have considered the inventory by Hubert et al available in GLIMS, but although the outlines may be of better quality, the vector data quality is worse than the existing one (see examples below).
• we have had some discussions in the steering committee about including glaciers on the continent (dry valleys) but this was put aside in order to prioritize other tasks in RGI7

Here is what needs to be done:

• a new inventory (maybe based on Hubert et al)
• a discussion with the ice sheet community to avoid double counting
• ?

Hubert et al dataset image examples

Screenshots from @burnout-stephl .

[ezwelty]

@fmaussion Have you decided that glaciers on the peninsula proper be left out of RGI7, or were these glaciers not included for technical or time reasons?

The outlines in Huber et al. 2017 used a rock outcrop dataset originally produced automatically from imagery, hence the pixelated appearance. But with some filtering, these polygons should be ready to go.

[fmaussion]

Antarctic Mainland is not part of RGI - we only do Islands periphery.

The Huber 2017 dataset was considered as replacement for the Peninsula's islands, but it is just too messy:

[ezwelty]

Antarctic Mainland is not part of RGI - we only do Islands periphery.

In that case I'll drop off this conversation, but my thoughts before I leave. Huber et al. 2017 used the glacier basin outlines from Cook et al. 2014 as a base:

cook-2014.zip

Then cut out an earlier version of these automatic rock outcrops extracted from Landsat 8. These are, as you point out, very messy compared to the smoother but less accurate/complete traced rock outcrops. You can compare both at https://add.scar.org/ (see Rock outcrop and Rock outcrop (from Landsat 8)).

So I think the path forward is to improve the map of rock outcrops. Perhaps by dilating, smoothing the more complete but very "holey" automatic rock outcrops.

traced outcrops in orange, automatic outcrops in pink

[fmaussion]

Thank you! Yes I think that UZH is thinking about it, and that it would be a good addition to RGI7.1!