What can be predicted

We know the root causes of the change (warmer climate) and how it exerts its control (ice front buttressing and sliding at the bed), but many details elude observations and understanding. The first map of Greenland showing glacial velocities was published in 2006. The first revised map of ice thickness was assembled in 2001, with many glaciers not yet surveyed, i.e. we do not know how deep the beds are near the coast of many of these glaciers. We do not know the distribution of water at the glacier beds, where it accumulates, where it drains, and how it affects ice flow, although it certainly does. Human and financial resources dedicated to the ice sheet problem are woefully inadequate. Closing the gap between numerical models and reality is glaciology's next challenge. In the last few years, we learned two lessons: first, existing numerical models are not reliable enough to predict the future contribution to sea level from Greenland; and second, ice sheet changes are taking place on much shorter time scales than previously thought.

Numerical models of ice sheet evolution do not include ice shelf buttressing as a control (Huybrechts and Wolde, 1999). In Greenland, the rapid speed up of large glaciers cannot be explained by these models because they do not include glacier dynamics. The reasons for that are multiple:

• The physics is not adequately understood (for instance, what controls glacier sliding? where is the dominant resistance to flow? how subglacial water controls flow and so on).

• The mathematical treatment is difficult, requiring full 3-dimensional nonlinear Navier-Stokes equations with non-linear rheology and computer resources that are not typical of what is usually available to glaciologists.

• Observations of glacier changes (for example, glacier thickness) required to constrain the models have only been available recently and are still incomplete.

Existing numerical models excel at modeling the surface mass balance of ice sheets and ice sheet flow (under the shallow ice approximation) over time scales of millennia (Huybrechts and Wolde, 1999; Hanna et al, 2005). The real world revealed by satellites, however, shows a more dynamic landscape and time scales of variability measured in decades, years, months and even hours — all much closer to our own time scale. Such abrupt changes are not well understood or treated in numerical analyses.

Was this article helpful?

0 0

Post a comment