Category 2: Ice-Stream/Ice-Shelf

From Interactive System for Ice sheet Simulation
Jump to: navigation, search

The ice-stream/ice-shelf models are intended to both inform and refine the forcing fields for various whole ice sheet experiments, as well as intended to use output from whole ice sheet experiments to generate more precise regional views of dramatically changing regions.

These regional models can achieve a higher spatial resolution and impose a more rigorous set of physical conditions on the simulated ice dynamics. The people who have volunteered to participate in this effort bring models that span a range of dimensionality and offer a variety of geophysical complexities. They are (without detailed descriptions):

2D or greater
• Hulbe
• Goldberg
• Bassis (2.25-2.5D (an inverse approach)
• Price/Payne (GLAM)
• Dupont
• Parizek (no thermodynamics)
• Price/Payne (GLAM run as vertical slice in the x,z plane)


Set Up and Initialization

To satisfy the first function of informing the whole ice sheet models by refining the experiments’ forcing fields, it is necessary that these regional models capture the t0 state of the whole ice sheet models as closely as possible. This will probably be best accomplished by interpolating coarse grid values from the whole ice sheet models to the finer regional model grids followed by possible fine-grid-scale adjustments to minimize non-physical transients. In this process, it may also be possible to assimilate other data fields, such as rates of surface elevation and ice thickness change, to represent the t0 state more accurately.

To satisfy the second function (spatial refinement of whole-ice-sheet response), these models might be best used in a nested fashion, where the whole ice sheet model provides a temporal series of forcing fields as boundary conditions to time-dependent runs of the regional model. Here assimilation is less useful, because the regional model is being used as a higher spatial substitute for part of the whole-ice-sheet model. Alternatively, these regional models could be used in a static mode, where the goal is to provide spatial detail of a snapshot condition.

Thanks to significant field data collection efforts, there are new detailed maps of the surface and bed geometry, as well as the surface velocity fields of many critical areas of both ice sheets including Jakobshavns Isbrae, Helheim and Kangerlussuaq Glaciers in Greenland and Pine Island and Thwaites Glacier catchments in Antarctica. These make excellent candidates to study with regional models in a “stand-alone” mode, where the connection to a whole ice sheet model.

The regional models need not be viewed exclusively in terms of their links to the whole-ice-sheet models. For some very active basins of the ice sheet, the regional models can provide distinct estimates of sea-level contributions. However, the broader impact of regional changes requires that the changes be incorporated into a whole-ice-sheet model and many of the necessary parameters for regional models (e.g., internal temperatures, geothermal heat flux and basal conditions) are probably best drawn from whole-ice-sheet models. It is hard to overcome the advantages that result from a tying the regional models together with the whole-ice-sheet models to produce a consistent representation of ice sheet dynamics and response.

Control Run

To assist the whole ice sheet models in their control runs, it may be necessary to iterate conditions within certain regions with these regional models to ensure consistent geometric, dynamic and thermodynamic states between models. This is only relevant when the regional model will be used to prescribe forcing fields for particular future climate experiments. And even in those cases, this consistency may be satisfied by using the t0 state of the whole ice model as the initial state of the regional models that produce the forcing fields.


As discussed earlier, a tempting boundary condition for a whole ice sheet model attempting to simulate a change imposed at the margin, is to prescribe the discharge flux. Yet, to do so undermines the independence of the model-produced ice mass loss. Regional models can help solve this paradox by translating a particular scenario of grounding line retreat, or calving evolution into a series of geometric, dynamic and theromodynamic changes that are easy to impose on the whole ice sheet model.

Ice shelf removal

The details of a wholesale and sudden removal of an ice shelf is an extreme scenario described earlier designed to examine the upper limits of rapid ice loss. The specific time scale of shelf removal remains undetermined and may possibly be bounded by model stability.

Unlimited warm water

Allowing unlimited amounts of the warmest water observed next to an ice sheet is one means of setting an upper bound for oceanic forcing of ice sheet mass loss. The resulting ice shelf thinning translates into a significant increase of ice discharged from the attached ice stream. With the current generation of regional models, this experiment can be performed and evolved to produce the necessary forcing fields for the whole ice sheet models.

Cold water ocean

The third category of models addresses the connection between the ocean circulation, driven primarily by surface winds, and the evolution of the ice shelf. A first-order bracketing of the range of effects of different water temperatures on the ice shelf can be accomplished by pairing the above experiment driven by unlimited amounts of warm water reaching the ice shelf, by a companion experiment where the water reaching the ice shelf is the coldest that has been observed.

Ice shelf regeneration

Once the upper bound of sea level contribution has been explored and as the coupling between the regional models and whole ice sheet models become stronger and better able to handle more complex scenarios, it will be appropriate to begin introducing possible stabilizing factors to increase the realism of the experiments and produce increasingly reliable predictions. One such effect is allowing ice shelves to regenerate. This will require some calving parameterization. The stabilizing influence of this effect has been hypothesized, but it has never been quantitatively examined.