A rapidly converging initialisation method to simulate the present-day Greenland ice sheet using the GRISLI ice sheet model (version 1.3) - Archive ouverte HAL Accéder directement au contenu
Article Dans Une Revue Geoscientific Model Development Année : 2019

A rapidly converging initialisation method to simulate the present-day Greenland ice sheet using the GRISLI ice sheet model (version 1.3)

Résumé

Providing reliable projections of the ice sheet contribution to future sea-level rise has become one of the main challenges of the ice sheet modelling community. To increase confidence in future projections, a good knowledge of the present-day state of ice flow dynamics, which is critically dependent on basal conditions, is strongly needed. The main difficulty is tied to the scarcity of observations at the ice-bed interface at the scale of the whole ice sheet, resulting in poorly constrained parameterisations in ice sheet models. To circumvent this drawback, inverse modelling approaches can be developed to infer initial conditions for ice sheet models that best reproduce available data. Most often such approaches allow for a good representation of the mean present-day state of the ice sheet but are accompanied with unphys-ical trends. Here, we present an initialisation method for the Greenland ice sheet using the thermo-mechanical hybrid GRISLI (GRenoble Ice Shelf and Land Ice) ice sheet model. Our approach is based on the adjustment of the basal drag coefficient that relates the sliding velocities at the ice-bed interface to basal shear stress in unfrozen bed areas. This method relies on an iterative process in which the basal drag is periodically adjusted in such a way that the simulated ice thickness matches the observed one. The quality of the method is assessed by computing the root mean square errors in ice thickness changes. Because the method is based on an adjustment of the sliding velocities only, the results are discussed in terms of varying ice flow enhancement factors that control the deformation rates. We show that this factor has a strong impact on the minimisation of ice thickness errors and has to be chosen as a function of the internal thermal state of the ice sheet (e.g. a low enhancement factor for a warm ice sheet). While the method performance slightly increases with the duration of the minimisation procedure, an ice thickness root mean square error (RMSE) of 50.3 m is obtained in only 1320 model years. This highlights a rapid convergence and demonstrates that the method can be used for computation-ally expensive ice sheet models.
Fichier principal
Vignette du fichier
gmd-12-2481-2019.pdf (13.91 Mo) Télécharger le fichier
Origine Fichiers éditeurs autorisés sur une archive ouverte
Loading...

Dates et versions

hal-02328234 , version 1 (08-09-2020)

Identifiants

Citer

Sébastien Le Clec'H, Aurélien Quiquet, Sylvie Charbit, Christophe Dumas, Masa Kageyama, et al.. A rapidly converging initialisation method to simulate the present-day Greenland ice sheet using the GRISLI ice sheet model (version 1.3). Geoscientific Model Development, 2019, 12 (6), pp.2481-2499. ⟨10.5194/gmd-12-2481-2019⟩. ⟨hal-02328234⟩
106 Consultations
56 Téléchargements

Altmetric

Partager

Gmail Mastodon Facebook X LinkedIn More