Glacier and ice sheet sensitivity experiments such as those described above highlight regions of the world that are most sensitive to expected changes in temperature or precipitation. More realistic, geographically and temporally explicit scenarios are required for quantitative forecasts of glacier and icefield response to climate change. Studies of this type require coupling of glacier and climate models, and have been carried out to assess the transient response of various ice masses to model-based climate change scenarios (e.g. Schneeberger et al., 2001; Huybrechts et al., 2002). Studies to date have all involved one-way or 'offline' forcing, where time-dependent glacier or ice-sheet simulations are driven by reanalysed or modelled climate fields. Future forecasts use monthly, annual or decadal average climatology to estimate mass-balance fields, and integrate the cumulative mass balance and ice dynamics response for a given ice mass.
Numerous studies have analysed the likely response of the Greenland Ice Sheet to future climate change using this methodology (e.g. Huybrechts et al., 2002). Plate 32.2 presents an illustrative simulation for the year 2200, using climate forecast fields from a coupled ocean-atmosphere simulation with the NcAR ccSM v2.0 (B. Otto-Bliesner, personal communication, 2003). These climate fields have been used to drive a three-dimensional dynamical ice-sheet model through a 200-yr transient run, using an ice sheet model that is spun up through a prior simulation of the last glacial cycle (Marshall & cuffey, 2000). Differences from the reference climatology of Plate 32.1 are plotted to illustrate the spatial patterns of ice-sheet response to the simulated climate change.
The degree-day mass balance model used to generate these results is oversimplified and the AGcM representation of modern-day temperature and precipitation patterns is poor in some regions of the ice sheet (cf. Ohmura et al., 1996; Fig. 32.1). There are therefore significant uncertainties in this type of forecast. High-resolution AGcM simulations focusing on the polar regions show continually increasing promise for modelling snow accumulation patterns over ice sheets (Bromwich et al., 1995, 2001; Ohmura et al., 1996; Wild & Ohmura, 2000; Wild et al., 2003). With appropriate 'downscaling' or interpolation strategies to link between AGcM and ice-sheet model grids (of the order of
20km for continental ice sheets), ice-sheet-scale mass balance can be simulated with reasonable accuracy (Pollard & Thompson, 1997; Glover, 1999). There are nevertheless many outstanding challenges in climate field downscaling, particularly for regional-scale icefields. I discuss this further in section 32.5.
coupled glacier and climate models that address future climate change impacts require an initial glacier distribution to represent present-day conditions. This initial distribution shapes the surface topography and it provides the reference mass balance state. The resulting mass-balance sensitivity tests should be considered only as snapshots of glacier distribution and mass balance for a particular climatic and glacierized state. This neglects ice dynamical feed-backs and mass balance feed-backs that will result from changing glacier geometry (hence surface topography and local energy balance/boundary-layer meteorology feed-backs).
These studies also implicitly treat the initial glacier distribution as if it is in an equilibrium state with the reference climatology, which certainly is not the case. Even small glaciers take decades to respond to climatic changes, while continental ice sheets take tens of thousands of years. Many contemporary ice masses are in a state of negative balance for modern reference climatologies (e.g. 1961-1990), and a long integration with a perturbed climate scenario would compound the effects of both the disequilibrium reference state and climate change scenario. These simulations can be run out to equilibrium for pedagogic purposes, with or without glacier dynamics feed-backs, but some of the glacier distributions that are predicted in such a future scenario will simply reflect the disequilibrium of modern-day ice masses with the reference climatology. Experiments of this type are therefore best interpreted as 'instantaneous' mass balance perturbation experiments, with the resultant mass balance fields differenced from those of the reference state.
One tactic to develop improved glacier and ice sheet initial conditions is to run global-scale 'spin-up' simulations using historical or modelled climatology. For alpine glaciers and icefields (small glacial systems) or those with a rapid climatic/dynamic response time (e.g. Icelandic ice caps), a simulation from 1800 to 2000 would give a reconstruction of present-day ice masses that is reasonably in tune with the historical climatic forcing. Polar icefields and continental-scale ice sheets have much longer dynamical response times. The Greenland and Antarctic Ice Sheets are still responding to late Pleistocene climate and the Pleistocene-Holocene transition, with englacial temperatures and impurity concentrations reflecting Pleistocene conditions. This influences ice rheology and the basal temperature distribution of the ice sheets (hence, areas where basal flow is possible). For these ice masses, spin-up simulations typically extend for at least 120 kyr, encompassing the last glacial cycle (e.g. Huybrechts et al., 1991,2002).
Was this article helpful?