Hutchings, Jennifer Katy; (2000) On modelling the mass of Arctic sea ice. Doctoral thesis (Ph.D), UCL (University College London).

Text On_modelling_the_mass_of_arcti.pdf Download (19MB)

Abstract

Sea ice has been highlighted as a climate change indicator [IPCC, 1995]. Models are useful tools to study Arctic sea ice on decadal and longer time scales [Vinnikov et al., 1999] and the viscous-plastic model has been identified as the best model available to simulate ice motion [Kreyscher et al., 2000]. Here we investigate whether a stand-alone viscous-plastic model reproduces observed ice thickness. The Kiel sea ice model and the UCL model, documented in this thesis, are compared to ERS radar altimeter estimates of ice freeboard. Compared to the observations, ice thickness is over estimated by 2 m in both models. Near the Canadian Archipelago this thickness difference increases to 3 m. We investigate whether a large thickness error in the model can be explained by errors in the model force balance. A sensitivity study of ocean and wind drag coefficients and maximum ice strength shows that the model ice thickness cannot be improved by only varying the maximum ice strength. The viscous-plastic model is computationally expensive to solve accurately, which hinders its use in GCMs. We revise the numerical solution, introducing a new overrelaxation method which guarantees the stress solution is always within the yield criterion. The convergence of the velocity vector is much improved compared to the iterative scheme of Zhang and Hibier [1997]. Finally a numerical error is identified in the traditional velocity correction scheme which accounts for up to 0.5 m of the model ice thickness error. An efficient algorithm is designed and implemented to ensure the fully coupled mass-momentum solution is found to numerical accuracy. In this thesis we find that the viscous-plastic model over estimates Arctic ice thickness in the late 1990s. Up to 25% of this error may be attributed to unresolved mass-momentum coupling, and we suggest other errors may lie in thermodynamic modelling.

Download activity - last month

Export as JSONXMLCSV

Download activity - last 12 months

Export as JSONXMLCSV

Downloads by country - last 12 months

Export as CSVXMLJSON

Archive Staff Only

View Item