Elektrine lite

← Feed

@pkuipersmunneke@mastodon.nl

2026-08-17 09:55 UTC

For the next IPCC report, the ice-sheet community is working on improved estimates for Antarctic ice loss for different warming scenarios and a wide range of climate models. To estimate when atmospheric warming leads to sufficient surface meltwater for Antarctic ice shelves to collapse, we developed a machine learning emulator. It estimates when the snow and firn cover over ice shelves has gone and meltwater can start to destabilize an ice shelf (see picture for the process of firn depletion due to increased melt, and subsequent drainage into fractures leading to collapse). The emulator was developed using XGBoost. It can be used in combination with any climate model as long as surface melt estimates are reliable. It is planned to be part of an Antarctic ice-shelf collapse protocol for future sea-level scenarios. The preprint is now available here: https://egusphere.copernicus.org/preprints/2026/egusphere-2026-3453/

Replies (0)

No replies.