Sensitivity analysis of a viscoelastic parameterization for gravity wave dispersion in ice covered seas

Viscoelastic parameterization has been proposed for modeling gravity wave propagation under various ice covers. The resulting dispersion relation predicts wave speed change and attenuation, both depend on the viscoelastic parameters. Through direct measurements of wave characteristics, these parameters may be inversely determined. To effectively use an inverse method, the present study employs ANOVA to perform a sensitivity analysis on the dispersion relation. Two dispersion models are examined. Both models have qualitatively similar sensitivity to their parameters, but different parameterization. The statistical results show that the change of wave speed is most strongly dependent on the elasticity parameter, while attenuation is most strongly dependent on the wave period. The sensitivity analysis provides additional information. For example, for high frequency waves strong interactions exist between wave period, ice thickness, and viscoelastic parameters, which result in uncertainties to the inversely determined parameters. For low frequency waves, the viscoelastic parameters may be determined from attenuation data alone. If in addition the ice cover is thick and rigid, the accuracy of ice thickness is not crucial to the inversely determined elastic parameter. These findings are tested and validated using a field data set. The method presented in this study may be used to test other viscoelastic models. (C) 2015 Elsevier B.V. All rights reserved.