Automatic inference of ice models from postglacial sea-level observations: Theory and application to the British Isles

Late Pleistocene and Holocene sea level change is caused mostly by melting of ice sheets since the Last Glacial Maximum and the Earth's delayed uplift in response to changes in the surface load. Records of sea level change therefore contain information on Earth rheology and ice sheet history and have been used to constrain ice and Earth models. A method is presented which finds an ice model that simultaneously fits the sea level record and the constraints on ice elevation. It infers the ice model which maximizes the smoothness of the surface and thus produces realistic ice models with the minimum detail required to fit the observations. This is a significant improvement on previous work in which the ice model was iteratively improved by ad hoc adjustments resulting in a model that is dependent on the initial model. The method is applied to the British Isles and a model with a spatial resolution of 50 km and temporal resolution of 1000 years is obtained for the late glacial period. Glaciological data determining the maximum elevation of the ice provide an important constraint on the model as does information on the timing of formation of such trimline data. It is possible to infer ice volumes up to a few thousand years before the earliest sea level data, and therefore the availability of early sea level data is most important to constraining ice models for the time of maximum glaciation