Implementation of dynamic subglacial processes in a high-order ice sheet model

Current predictions of ice sheet mass balance and discharge variability including sea-level rise, are severely limited because subglacial processes are not reliably implemented in large-scale ice sheet models. One of the most limiting factors is the challenging nature of simulating the fast and transient flow of ice streams that drain the interior and discharge large volumes of ice into the polar ocean. This difficulty is partly due to uncertainties in the nature of subglacial processes, but also due to the complexity in including local basal processes at the scale computationally required to run a whole ice sheet model. We aim to improve the predictive capability of Glimmer-CISM, the Community Ice Sheet Model by (i) implementing higher-order ice flow physics and (ii) introducing dynamic subglacial processes that include hydrologically controlled shear strength evolution in a till layer with Coulomb plastic rheology. Here, we focus on the development and use of the latter within the large-scale model. The basal processes model was originally developed for simulations of the Siple Coast ice streams, in order to explain their observed flow variability as a result of processes taking place at the ice-till interface. Coupling between the subglacial processes model and the flow model occurs through the basal melting (or freezing) rate and through the determination of the till yield strength, which is a function of water availability within the till layer. Basal melting (freezing) weakens (strengthens) the till layer, reduces (increases) the basal resistance to flow, and increases (decreases) the sliding speed. So far, experiments have been conducted on a simple, idealized ice stream domain. This allows us to do sensitivity tests on a range of model parameters (e.g., sediment type, thickness, and distribution) and to tune the model before applying it to the entire West Antarctic Ice Sheet. We compare our model results to a case using a linear-viscous till rheology, where the sliding parameter is defined as a function of the subglacial water depth. Preliminary results show that the flow regimes and periodic behavior are distinctly different for the two cases. Additional modelling developments will include the generalization of the subglacial processes model by coupling pore water pressure in the till layer to a regional hydrology model. We anticipate that such a model will be necessary to explain and reproduce both the observed thinning along the Amundsen Coast, and the slow down and stagnation of ice streams in the Ross Sea sector.