Refining models of the glacial isostatic adjustment process

grantor: University of TorontoSurface observables associated with glacial isostatic adjustment (GIA) can be employed to infer mantle viscosity structure and changes in ice mass distribution. These parameters play a key role in, respectively, our understanding of flow processes within the mantle and long-time-scale changes in the Earth's climate. The work presented in this thesis explores the importance of a number of 'second order' aspects of GIA models that are commonly neglected in the inference procedure. Observations of sea-level change comprise the most effective data set employed in GIA modeling studies to date. For this reason, much of the work presented in this thesis focuses on a number of approximations inherent in a widely adopted GIA sea-level theory. In particular, assumptions relating to the redistribution of the water load associated with the melting of the last great ice sheets are considered. Results show that sea-level predictions are sensitive to whether or not the Earth's shorelines (which are known to have migrated by many kilometers in some areas throughout the postglacial period) are treated as 'fixed' or time dependent. The results of GIA analyses based on sea-level data obtained from regions far removed from the ancient ice masses (the 'far field') may be significantly biased if this mechanism is not incorporated into the sea-level model. Also, the sea-level theory most commonly adopted in GIA studies is shown to predict an incorrect water load redistribution in regions once covered by marine-based ice sheets (e.g., Hudson Bay). This leads to an error in the predicted relative sea-level (RSL) curves that is large enough to bias inferences of viscosity and/or ice thicknesses based on data obtained in these ('near-field') regions. To date, most GIA sea-level analyses have assumed a non-rotating Earth model. That is, the component of sea-level change induced by GIA perturbations to the Earth's rotation vector is ignored. The first gravitationally self-consistent predictions of this sea-level component are presented in this thesis. These results show that previous estimates of the magnitude of the rotation-induced signal are too large by a factor of 3-5. In contrast to conclusions based on these previous studies, the present results show that this signal is of too small a magnitude to be significant in RSL analyses based on near-field data. However, estimates of Late Holocene melting events derived from modeling studies of far-field data can be significantly biased if this sea-level contribution is not accounted for. Recent inferences of mantle viscosity based on convection-related data (e.g., the long-wavelength geoid) include a thin (~100-200 km) low viscosity layer at the base of the upper mantle. Such fine-scale viscosity structure is commonly neglected in coarse (3-layer) parameterizations of mantle viscosity variation with depth. The sensitivity of a suite of GIA observables to the viscosity and thickness of such a layer is explored. In contrast to previous suggestions, the present results show that predictions of both RSL and secular change in the geoid are sensitive to this viscosity structure. Therefore, future analyses that employ these two data types to infer either mantle viscosity or changes in ice sheet mass distribution should adopt a viscosity-depth parameterization that can account for the possible existence of fine-scale, low-viscosity structure at the base of the upper mantle.Ph.D