Two rain events at Matanuska Glacier illustrate how subglacial drainage system development and snowpack conditions affect hydrologic response at the terminus. On 21 and 22 September 1995, over 56 mm of rain fell in the basin during a period usually characterized by much drier conditions. This event caused an 8-fold increase in discharge and a 47-fold increase in suspended-sediment concentration. Peak suspended-sediment concentration exceeded 20 kg m —3, suggesting rapid evacuation of stored sediment. While water discharge returned to its pre-storm level nine days after the rain ceased, suspended- sediment concentrations took about 20 days to return to pre-storm levels. These observations suggest that the storm influx late in the melt season probably forced subglacial water into a more distributed system. In addition, subglacially transported sediments were supplemented to an unknown degree by the influx of storm-eroded sediments off hillslopes and from tributary drainage basins.

A storm on 6 and 7 June 1997, dropped 28 mm of rain on the basin demonstrating the effects of meltwater retention in the snowpack and englacial and subglacial storage early in the melt season. Streamflow before the storm event was increasing gradually owing to warming temperatures; however, discharge during the storm and the following week increased only slightly. Suspended-sediment concentrations increased only a small amount, suggesting the drainage system was not yet well developed, and much of the run off occurred across the relatively clean surface of the glacier or through englacial channels.

The co-isotopic composition, both in δDand in δ18O, of interbedded debris-rich and clear ice layers, thought to have been formed at the grounding line of Hells Gate Ice Shelf, indicates freezing by a double-diffusion effect between continental meltwater and sea water within a subglacial sediment. A source of meltwater is required to sustain the process, since the temperature of the bed is below the freshwater melting point. The most likely source is a surficially frozen meltwater lake.

Rock outcrops, kept mostly snow-free by the action of katabatic winds, absorb solar radiation so local production of liquid water becomes possible in an area with year-round subfreezing air-temperature conditions. The meltwater accumulated in a marginal lake can eventually reach the subglacial substratum near a pinning point where the ice is relatively thin and fractured.

Entrainment of debris by cold-based glaciers having basal temperatures as low as — 17°C can be observed in the Dry Valleys of south Victoria Land, Antarctica. The classical models developed to explain debris incorporation at the glacier base are inappropriate in such cases, since the basal temperature is well below the freezing point. An alternative model, based on the presence of ice-marginal lakes, has recently been proposed by one of the authors (S. F.). In this model, transient wet-base conditions can occur as ice flows onto the unfrozen sediments of the lake bottom, creating conditions favorable to the entrainment of sediments and to ice accretion by water freezing.

Here we describe a situation where this model is consistent with an ice-composition study of the basal part of Suess Glacier, Taylor Valley. The stable isotope composition indicates that water freezing, most probably lake water, plays a major role in the formation of the basal ice layers. Total gas content of this basal ice is considerably depleted when compared to meteoric glacier ice, in accordance with a rejection mechanism during freezing. Its gas composition, strongly enriched in CO2, is also indicative of the presence of a former liquid phase.

We have developed a new surface-processes model incorporating large-scale fluvial processes, local hill-slope processes and glacial erosion. Ice thickness and velocity are calculated under a shallow-ice approximation. Simulation experiments in fast-growing orogens comparing the efficiencies of fluvial and glacial erosion, where the two are operating simultaneously over several glacial cycles, show that: glacial landscapes can support greater ice masses than fluvial landscapes; glacial valley and lake shapes create a disequilibrium between landform and land-forming process that leads to pulses of high erosion at the end of glacial periods; glacial erosion rates can reach a constant value in a uniformly growing orogen; and glacial erosion is capable of eroding drainage divides when the ice is moderately thick.