A leading-order viscoelastic model for crevasse propagation and calving in ice shelves
We use a leading-order viscoelastic model for crevasse evolution, in which a purely viscous model for the deformation of the domain couples with linear elastic fracture mechanics models through a viscous pre-stress. The fracture mechanics model conversely couples with the viscous model by inserting...
| Main Authors: | , , |
|---|---|
| Format: | Article in Journal/Newspaper |
| Language: | English |
| Published: |
Copernicus Publications
2023
|
| Subjects: | |
| Online Access: | https://doi.org/10.5194/egusphere-2023-807 https://noa.gwlb.de/receive/cop_mods_00066970 https://noa.gwlb.de/servlets/MCRFileNodeServlet/cop_derivate_00065440/egusphere-2023-807.pdf https://egusphere.copernicus.org/preprints/2023/egusphere-2023-807/egusphere-2023-807.pdf |
| Summary: | We use a leading-order viscoelastic model for crevasse evolution, in which a purely viscous model for the deformation of the domain couples with linear elastic fracture mechanics models through a viscous pre-stress. The fracture mechanics model conversely couples with the viscous model by inserting cracks into the domain, which viscous flow subsequently pulls apart. By contrast with prior work, we solve the fracture mechanics problem on the actual domain geometry using a boundary element method, coupled with a finite element solution of the Stokes equations describing the viscous flow. We study a periodic array of surface and basal crevasses on an ice shelf being stretched at a prescribed rate. We find that calving can either occur instantly for large enough stretching rates or sufficiently high surface water levels or through feedbacks between partial fracture propagation and subsequent viscous deformation and adjustment of the viscous pre-stress acting on crack faces. Our results show that purely stress-based calving laws cannot robustly describe the process of calving, since they cannot account for the gradual evolution of local crevasse and surface geometry, which can be understood at the large scale as being more akin to the evolution of a damage variable. Future work will need to coarse-grain the type of process model we describe here in order to make it applicable to ice sheet simulations. |
|---|