• Citer

The details of how to implement and apply sub-shelf melt parametrizations in standalone ice sheet models - Does it matter for Antarctic sea level projections?

par Zipf, Lars ;Coulon, Violaine ;Pattyn, Frank
Référence 36th Forum for Research into Ice Shelf Processes (FRISP 2023) (19-22 June 2023: Stalheim, Norway)
Publication Publié, 2023-06-21

Poster de conférence

• Détails
• Contenu

Résumé :

Sub-shelf melting is the main driver of current Antarctic mass loss. However, the ocean processesgoverning ice shelf melting are not well understood and sub-shelf melt is currently themain uncertainty source within Antarctic sea level projections. Several parameterizations of differentcomplexity exist and can be used by standalone ice sheet models to link oceanic propertiesto sub-shelf melt.The various melt parameterizations already contribute significantly to the uncertainty of futuresimulations of the Antarctic ice sheet. In addition, several approaches exist to determine thetuning parameters used within the sub-shelf melt parameterizations, adding another layer of uncertainty.Furthermore, how the parameterization is integrated into the initialization procedureof the ice sheet model additionally contributes to the uncertainty of the projections.Here, we use Kori, a vertically integrated hybrid (SSA-SIA) ice sheet–ice shelf model (the successorof f.ETISh with improved initialization procedure strongly reducing the model drift), toinvestigate how the different possibilities of implementing the sub-shelf melt parameterizationsin a standalone ice sheet model contribute to the uncertainty of Antarctica’s future contributionto global mean sea level rise. In particular, the different possibilities of integrating the sub-shelfmelt parameterization into the initialization procedure of the ice sheet model are evaluated inmore detail. To minimize the feedback of different ice geometries after the initialization, a similarinitialization procedure, based on data assimilation of the ice geometry and aiming for a steadystate at the end of the initialization, is performed for all experiments.