Numerical modelling of subglacial erosion and sediment transport and its application to the North American ice sheets over the Last Glacial cycle

摘要

Present-day sediment distribution offers a potentially strong constraint on past ice sheet evolution. Glacial system models (GSMs), however, cannot address this constraint while lacking appropriate representations of subglacial sediment production and transport. Incorporating these elements in GSMs is also required in order to quantify the impact of a changing sediment cover on glacial cycle dynamics.Towards these goals, we present a subglacial process model (hereafter referred to as the sediment model) that incorporates mechanisms for sediment production, entrainment, transport, and deposition. Bedrock erosion is calculated by both Hallet's and Boulton's abrasion laws separately, and by a novel quarrying law parametrized as a function of subglacial cavity extent. These process-oriented erosion laws are compared against a simple empirical relationship between erosion rate and the work done by basal stress. Sediment entrainment is represented by Philip's law for regelation intrusion and soft-bed deformation is included as a subglacial sediment transport mechanism. The model is driven by the data-calibrated MUN (3D) GSM and a newly developed subglacial hydrology module.The sediment model is applied to the last North American glacial cycle and predicts sediment thickness and cumulative erosion patterns. Results are obtained in the context of a sensitivity analysis and are compared against the present-day distribution of glacigenic sediment and geological estimates of Laurentide Ice Sheet erosion. Given plausible ranges for the sensitivity parameters, chosen a priori based on available literature or on heuristic arguments, the calculated erosion depths overlap with the geological estimates of Laurentide erosion. Most of the runs in the sensitivity set produce unrealistically thick and continuous moraines along the eastern, southern and western margins of the North American ice complex, which suggests that the model overestimates sediment entrainment and thus englacial transport. A realistic sediment distribution is only obtained when the entrainment rate is capped at the average basal melting rate, which suggests that modelled entrainment and basal melting rates should be of the same order of magnitude.