Simulating ice processes using the finite element, unstructured, adaptive model fluidity

• Main Author: Mouradian, Simon
• Other Authors: Piggott, Matthew, Engineering and Physical Sciences Research Council
• Format: Doctoral or Postdoctoral Thesis
• Language: unknown
• Published: Imperial College London 2015
• Subjects: Greenland; Alfred Wegener Institute; glacier; Ice Sheet; Sea ice
• Online Access: http://hdl.handle.net/10044/1/24846; https://doi.org/10.25560/24846

The cryosphere impacts global climate in various ways. Snow and ice have a higher albedo than land or the open ocean and therefore affect the total reflectivity of the earth. Sea ice forms an insulating layer over the polar oceans controlling both heat and water vapour fluxes between the atmosphere and polar ocean. Ice sheets hold around 77% of Earth's freshwater reserves, and recent increases in ice loss from the Earth's ice sheets are cause for concern. This thesis develops numerical tools that can be used for the study of various ice processes such as ocean -- sea ice interaction, ice sheet and glacier dynamics. A coupled ocean - sea ice model is developed, using the open source, unstructured, adaptive ocean model Fluidity and a finite element sea ice model developed at the Alfred Wegener Institute of Polar and Marine Research, FESIM. The tightly coupled model is verified and validated through a series of tests, demonstrating its dynamical capabilities. The sea ice dynamics are a model of Elastic-Viscous-Plastic rheology, as described in Hunke and Dukowicz. The thermodynamic parameterisation is similar to the 1D simplest model of Parkinson and Washington which is based on the zero-layer approach of Semtner. Furthermore, a new computational framework for carrying out ice sheet simulations is presented. A thermo-mechanical, non-linear, full-Stokes model is used to carry out the exercises of the Ice Sheet Model Intercomparison Project for higher--order models (ISMIP--HOM). The results presented here show that Fluidity compares favourably with other ice sheet models. Further tests are performed to demonstrate the use of dynamic adaptive remeshing in lowering the computational cost of models compared to their structured, fixed-mesh counterparts, by focusing resolution only where and when required. Finally, initial simulations of the full Greenland ice sheet are performed demonstrating the potential utility of adaptive meshes for large-scale, full-Stokes modelling. Open Access