Skip navigation

Featured Science Paper

Initialization of ice-sheet forecasts viewed as an inverse Robin problem

To begin any forecast of how the ice sheets of Greenland and Antarctica will contribute to sea level, we first need to know the present-day shape and flow of the ice. This provides a starting point for computer models that predict how the ice will flow in future. This approach is similar in spirit to short-term weather forecasting, where an assessment of the present-day weather patterns forms the basis for prediction over the next few days. Of course, ice flows much more slowly than air. Nevertheless, including slower components of the climate system, such as ice sheets, within a climate model makes predicting 21st Century climate mathematically more like short-term weather forecasting — accurate initial conditions for the ice sheets should improve simulations of changes to ocean currents over the coming centuries, as well as sea level.

This paper provides a simple method to reveal the flow of ice throughout the whole thickness of the ice sheet, using observations from satellites and aircraft. Once these ice-flow patterns have been found, they can be used as the starting point for the predictive simulation, either for sea-level prediction, or within a climate model. In this theoretical paper, we show that satellite observations of the flow at the surface of the ice sheet can be extended to its interior using mathematical methods borrowed from medical imaging. The key is to estimate the viscosity of the ice and the slipperiness of the underlying sediment that agree best with the available observations. The method performs well in tests, and is already being applied to a variety of existing ice sheet models.

Find link to the full paper in the NERC Open Research Archive


Arthern, R.J, Gudmundsson, G.H. 2010


Journal of Glaciology, 56, No. 197, 527-533