Back to the matter of closed sub-glacial troughs -- how they are eroded and how ice manages to extract billions of tonnes of rock from them and to transport all this debris uphill and over "exit thresholds" before releasing it somewhere near the outer coastline.
Several things have come up in my recent reading. First, some troughs have beds which are divided up into a series of connected basins. According to Julian Dowdeswell and others the bed of Nordvestfjord is like this, with a series of deep basins (over 1200m deep) separated by sills between 600m and 900m deep. I have not seen the long profiles, and so we can but speculate as to whether the sills coincide with outcrops of highly resistant rocks (on the basis of lithology or structure) and whether the basins coincide with pulses or additions to glacier discharge derived from tributary glaciers. In the case of Hardangerfjord, the deepest part of the biggest basin is more than 30 km downstream from the nearest tributary input, and the sharp sill which then follows (about 300m high) is apparently unrelated to any route suitable for diffluence or discharge reduction. So something else is going on -- and a simple model will clearly not do.
Bedrock characteristics must have something to do with the very complex long profile of the bedrock floor of Hardangerfjord -- and the same must apply to the long profiles of many other outlet glacier troughs and fjords studied with the aid of new sounding techniques over the past decade or so. Two other factors now come into play: isostatic rebound and the relationship outlet glaciers to ice shelves.
1. Isostatic rebound
Isostatic depression of a land mass in response to the loading imposed by an ice sheet or ice cap, and then isostatic recovery associated with ice melting, must play a considerable part in determining what processes will operate within a glacial trough used by an outlet glacier. In Scandinavia, it appears that in the centre of the ice sheet there have been about 750m of isostatic depression, followed by almost the same amount of recovery, during the Devensian glaciation. Possibly, isostatic responses were even greater in the Anglian and other glacial episodes. In the fjord region of Norway, an isostatic rebound of approx 200m has been recorded in the inner reaches of the fjords, and about 100m in the outer reaches.
Because of the differential rates of isostatic response to crustal loading and unloading, stresses are exerted within the near-surface crust. These stresses express themselves as seismic events such as small earthquakes; and many such have been recorded around the fringes of Scandinavia. In practical terms, old faults are regenerated or re-activated, and rock fracture is a consequence. On what scale does this occur? And does it happen in some strata more than in others? Whatever the answers are, it seems probable that over many millions of years, each glacial event will be followed by a "rock fracturing" episode which might facilitate easier erosion when the next glacial episode comes along. Another factor, on the floor and flanks of troughs, will be pressure release and "rock bursting" in response to rock removal or the replacement of metres or tens of metres of rock by ice or even by water.
When David Sugden and I were working in Greenland and in the South Shetland Islands, we began to suspect (from our mapping of raised beaches) that some of the isostatic responses in recently deglaciated areas were very localised indeed. In some recent work by David and his colleagues around the Shackleton Range in Antarctica, it appears that isostatic adjustments to rock loss (by glacial erosion) can have major effects on the routes followed by ice discharging from the Antarctic ice sheet but also on the speed of flow and the capacity for enhanced erosive activity. In other words, there is a positive feedback effect.
Sugden, D,E et al, Emergence of the Shackleton Range from beneath the Antarctic Ice Sheet due to glacial erosion. Geomorphology · March 2014
In this false-colour map of bedrock elevations we can see the Shackleton Range standing up as a series of nunataks, flanked by the outlet glacier troughs of the Slessor and Recovery Glaciers. These are remarkably deep -- as shown by the blue colour. Each trough has a base well over 2,000m beneath sea level, and each trough is closed, with a wide threshold (shown in green) in the pro-montane zone to the west of the trough exits. The threshold is particularly spectacular in the case of the Recovery Glacier trough, as we can see in the specked green and yellow markings on the map. Also, if we examine the blue-coloured areas in detail, we can see that each trough is made up of a series of connected basins and sills -- as already described for Nordvestfjord and Hardangerfjord. So glacier ice is clearly excavating vast amounts of bedrock and moving it uphill and away. But it is also moving through one basin after another, excavating and then rising, doing it again, and then doing it again. Is there any significance in this? There must be, since ice always obeys the laws of physics. There must be some very complex interactions going on, far too deep beneath the ice surface for anybody to observe them.
This is all very interesting, although of course it tells us nothing at all about the precise mechanics operating on the glacier bed.
What's going on here is by no means unusual. If you look at this bedform map of Antarctica, and search for the blue-coloured areas, you will see that there are indeed closed troughs all over the place. Rather a lot of material is being moved uphill.............
The Cryosphere, 7, 375–393, 2013
© Author(s) 2013. CC Attribution 3.0 License.
Bedmap2: improved ice bed, surface and thickness datasets for Antarctica
P. Fretwell et al
2. Interactions with ice shelves
Returning to Sognefjord and the other big fjords of western Norway, the isobase map shows us that at the peak of the Devensian glaciation, the land surface was depressed by the weight of ice by something like 150m in the west coast zone. That is approximately the same as the known eustatic depression of sea-level at the time. So the relationships of land and sea around 20,000 year ago would have been approximately the same as they are today. That means that if you had taken away all the ice of the Sognefjord outlet glacier, the sea would have flooded in at more or less the same relative level as today. This is an important point, since it enables us to look at the current relationships between Antarctic ice shelves and outlet glaciers, and maybe learn some lessons from them.
It is widely assumed that at the peak of each Pleistocene glaciation the edge of the Scandinavian ice sheet was some distance to the west of the current Norwegian coastline. Did all of the ice discharged westwards calve straight into deep water, or was there a fringing ice shelf? Ice shelves thrive where a lot of glacier ice is disgorged into embayments or concavities in the coastline -- this provides anchoring points on the flanking headlands. Out at sea, to the west of Norway, there would have been no anchoring headlands, but an ice shelf might have been supported on the skerries or strandflat which has exercised geomorphologists for many years. What might the subglacial dynamics have been in that scenario? With sea-level in more or less its current relative position, there might have been a lifting of the shelf in places, a moving grounding line, and a consequent reduction in bed erosion. If saltwater penetrated well inland of the ice shelf edge, there might have been enhanced melting on the base of the shelf from contact with warm oceanic water, and such water might well have penetrated intermittently over the thresholds and into the fjords. Might there have been an "enhanced buoyancy" effect? Maybe -- especially following the retreat of the ice edge to the vicinity of the present coast.
There is a vast literature on Antarctic ice shelves and their dynamics -- and great work currently in progress, given the importance of shelf behaviour as an indicator of climate change. There is no point, just now, in exploring this literature in detail. But here is one interesting article:
Rapid submarine ice melting in the grounding zones of ice shelves in West Antarctica
Ala Khazendar et al,
Nature Communications 7, Article number: 13243 (2016)
The authors have studied the Dotson and Crosson ice shelves and their main feeders, namely Smith Glacier, Pope Glacier and Kohler Glacier. Their main interest is in discharge changes and accelerated melting, but their radar sounding work is of immediate relevance. Here are the results of many radar sounding runs (mostly in 2002, 2004 and 2009):
There are considerable differences between the bedforms of the three glaciers, but the long profiles are very irregular. In the case of the Kohler Glacier, there is an apparent transverse channel on the seaward side of the threshold; the threshold is much more marked in the case of the Smith Glacier, with a reverse slope rising from -2000m to -800m over a distance of c 30 km; and the bed of Pope Glacier shows no strong reverse slope at all, but instead a gradual rise of c 750m over a distance of c 25 km. Note the different scales on the diagrams. Another interesting feature of the Pope Glacier and Kohler Glacier profiles is the lateral variation in bedrock profiles, revealed when radar runs are parallel but slightly offset one from another.
Clearly these and most other outlet glaciers around the Antarctic coast have overdeepened sections and thresholds as they approach their exit points. They also have rather irregular floors with various closed basins connected by sills. The smooth bed of Sognefjord might be something of an anomaly......
Is most of the work of overdeepening done by ice, or by meltwater and slush working under very high hydrostatic pressure? More to come......