The impact of parametric uncertainty and topographic error in ice-sheet modelling

Abstract

Ice-sheet models (ISMs) developed to simulate the behaviour of continental-scale ice sheets under past, present or future climate scenarios are subject to a number of uncertainties from various sources. These sources include the conceptualization of the ISM and the degree of abstraction and parameterizations of processes such as ice dynamics and mass balance. The assumption of spatially or temporally constant parameters (such as degree-day factor, atmospheric lapse rate or geothermal heat flux) is one example. Additionally, uncertainties in ISM input data such as topography or precipitation propagate to the model results. In order to assess and compare the impact of uncertainties from model parameters and climate on the GLIMMER ice-sheet model, a parametric uncertainty analysis (PUA) was conducted. Parameter variation was deduced from a suite of sensitivity tests, and accuracy information was deduced from input data and the literature. Recorded variation of modelled ice extent across the PUA runs was 65% for equilibrium ice sheets. Additionally, the susceptibility of ISM results to modelled uncertainty in input topography was assessed. Resulting variations in modelled ice extent in the range of 0.8-6.6% are comparable to that of ISM parameters such as flow enhancement, basal traction and geothermal heat flux.