It is one of the interesting facts of modern glaciology that despite decades of research there is
still no general consensus on the most appropriate form of the basal boundary condition to be used
in modelling larger ice masses. Although it seems clear that a mixed (Robin) type boundary
condition, where basal motion is described as a function of basal stresses, is needed, there is
little agreement on what function should be used. Currently the most used mixed-type boundary
conditions is the Weertman sliding law which is a power-law. Review of the literature shows that
values for the stress exponent in Weertman sliding law used in numerical modelling cover the entire
range from 1 to infinity. This situation appears puzzling. If the stress exponent is an important
model parameter, one would expect that it's value could be accurately determined through model
optimisation. In this paper I will address the question of how the strength of nonlinear basal
processes, as measured by the value of the Weertman stress coefficient, can be estimated from
surface data. By using Bayesian inverse methods conditions on data quality and data density required
for extracting information about non-linear basal processes are investigated. |