1. Introduction
Since ground-breaking work in the late 1960s and 1970s, both numerical models and field experiments have sought to quantify subglacial drainage under temperate valley glaciers. Following work by Röthlisberger, Shreve and Weertman, field experiments using dye breakthrough curves (e.g., Nienow et al.) suggested that subglacial drainage evolves through time with distance up-glacier as meltwater forces evolution from a distributed to a channelized drainage network. Yet, glaciers are extremely effective sediment-producing agents. Geological deposits reveal that they leave behind thick sequences of till and that these can appear to be water-worked. If the beds of glaciers comprise sediment, then fluvial processes and associated morphodynamics could also pay a critical role in the evolution of subglacial drainage. Here, we test this possibility.
2. Methodology
The focus of the work is the Haut Glacier d’Arolla, Switzerland. Collaboration with a hydropower company gave access to a high quality 2 minute discharge record. A Japanese hydrophone and turbidity probe, installed in the stream at the glacier snout, were calibrated using direct sampling. The morphology of subglacial channels was estimated using ground penetrating radar (GPR). Terrestrial laser scanning was used to measure glacier surface melt and hydraulic jacking (short term glacier surface uplift). A hydraulic model for sediment transport capacity (Lane et al., 2017, Geomorphology) was modified to deal with a dendritic network of subglacial conduits, driven by meltwater production.
3. Results
2.1 The morphology of subglacial streams
GPR data from close to the glacier margin suggest that subglacial channels are deeply incised into bed sediment, to greater degrees than they are eroded into the ice. They also reveal short, meter-scale distance, variability in bed slope (including reverse slopes) and subglacial channel cross-sectional area.
2.2 Bedload and suspended load export from the glacier
Data suggest that sediment export from underneath the glacier is threshold controlled. In the late melt season, efficient subglacial drainage overnight causes shear stress to fall below the critical threshold required for sediment entrainment and transport. During the morning rising limb of the hydrograph, suspended bed material transport was found to begin before bedload transport, and it also showed clockwise hysteresis, suggesting early morning flushing of finer bed material.
2.3 Hydraulic jacking at the wrong time of year
Terrestrial laser scanning in the near glacier margin zone showed late morning uplift of the glacier surface of up to 0.15 m late in the meltwater season. It implies a distributed layer of pressurised water at the glacier bed, and hydraulic jacking. Whilst this has previously been measured in the spring when flow at the glacier bed is likely to be distributed, a mechanism is required to explain its occurrence late in the meltwater season. One possibility is that falling sediment transport capacity overnight blocks channels with water forced laterally under pressure as discharge rises the following day.
2.4 Hydraulic modelling
At any one time, hydraulic efficiency considerations suggest that network bifurcation with distance up-glacier causes a more rapid decrease in subglacial sediment transport than in discharge. The hydraulic model showed that early melt season snowmelt leads to upstream propagation of a wave of rising sediment transport capacity. This could explain the observed: (1) incision of subglacial channels into bed sediment that propagates upstream through time; and (2) the associated development of a more efficient subglacial drainage network as snowline recession occurs. Later in the melt season, progressive decrease in the baseflow component of discharge occurs due to greater subglacial hydrological efficiency. Minimum flows become lower than those required for sediment entrainment and transport. Such conditions may lead to overnight sediment deposition on subglacial channel bed and the reductions in channel capacity necessary to create channel blockage and hydraulic jacking during the rising limb of the next day’s hydrograph.
4. Conclusions
This research provides circumstantial evidence for a very different model for transience in the development of subglacial drainage networks, which treats subglacial channels as rivers. River morphodynamics under glaciers have tended to be treated over simplistically. Here we present field data and hydraulic modelling that show that there could be progressive development of a drainage network by sediment erosion earlier in the melt season, of the sort inferred from dye breakout curves. We also show that the impacts of daily variability in discharge could lead to the channel blocking needed to explain measured hydraulic jacking.
Acknowledgments
Support from Ludovic Baron, the Canton de Vaud, Hydroexploitation SA, Alpiq SA, and Grande Dixence SA is acknowledged.