Given the magnitude of simulated permafrost thickness, horizontal extent and duration, it seems likely that permafrost had a significant impact on subglacial water flow and processes at the ice-bed interface.

Because permafrost was at least a few tens of metres thick up to the LGM, water flow through unlithified sediments was probably blocked and channelized flow was absent. As the Green Bay Lobe approached its maximum extent, the combination of the available potential gradient, the probable hydraulic conductivity (ca. 10-6ms-1), and the thickness of the remaining unfrozen lithi-fied aquifer (ca. 100 m) was inadequate to convey all basal melt-water from thawed-bed areas upstream. Indeed, only by raising simulated hydraulic conductivity to close to that of gravel (10-3 m s-1) could all subglacial meltwater drain. The presence in Wisconsin of tunnel channels and large boulders in tunnel channel fans has been linked to release of subglacially stored water (e.g.

Cutler et al., 2002). The model results support this linkage. The absence of such features in landsystems of the neighbouring Lake Michigan Lobe (Colgan et al., 2003) suggests that permafrost was less important there. This may have been due to the presence of proglacial lakes.

By locally preventing permafrost development, proglacial lakes had an impact on the basal heat balance, subglacial drainage and, consequently, lobe dynamics. Cutler et al. (2001) added a proglacial lake parameterization to the model and qualitatively matched geomorphological evidence of differences in ice extent, timing of ice advance, LGM ice-surface profile, and likely subglacial hydrological conditions between the Green Bay Lobe and its neighbouring lobes.

The foregoing discussion illustrates the value of time-dependent numerical models in the glacial geologist's toolkit. Models complement other investigative techniques, enabling researchers to test and quantify (perhaps with order-of-magnitude-scale uncertainty) the role of potentially important parameters in glacial environments.

0 0

Post a comment