This site uses cookies. By continuing to use this site you agree to our use of cookies. To find out more, see our Privacy and Cookies policy.
Skip to the content

IOP A community website from IOP Publishing

Powered by Movable Type 4.34-en

In from the cold: March 2010 Archives

Holes in glaciers are of interest because of the work they do, transferring meltwater through the glacier. They are also interesting in their own right, although crawling into them is an acquired taste. They are dark and cold, you don't know where you are going, and you do know that the journey will get harder and even more dangerous the further you go. But this is just what attracts one group of people, the cavers or speleologists.

Two recent papers illustrate the (to me) doubtful pleasures of glacier speleology, while also offering valuable insights into how the meltwater finds its way into the hole and, once there, keeps on doing its job (which is to obey the forces of gravity and pressure, while also conforming to the constraints of thermodynamics).

Doug Benn and co-authors describe caving in three glaciers of diverse type, in Svalbard, Nepal and Alaska, while Jason Gulley has more detail about the ice caves of Benn's Alaskan glacier, Matanuska Glacier.

Even cavers draw the line, usually, at plunging into water-filled holes, but some of these caves are only partly full. Others are dry. The most favourable time for exploring ice caves is after the summer meltwater has gone, but before the ice has had time to squeeze them shut.

The photographs alone are worth it. Some of the dry caves show evidence of former water levels in the shape of sills. These are shelves of ice protruding from the cave wall. They record episodes in which the water began to refreeze at its contact with the overlying air, but then drained away.

Others have keyhole-shaped cross-sections, recording a transition from enlargement of a water-filled hole, by net melting all around the cross-section, to deepening of just the cave floor by a lesser flow of water.

Two of the caves have scalloped walls. The scallops are dish-shaped, and roughly dish-sized, indentations covering all surfaces around the cave wall. Their size reflects the vigour of interaction between the turbulent water flow and the wall, which is vulnerable to erosion by melting. Scallops are common features in many caves in limestone, where the wall is vulnerable to erosion because the rock is slightly soluble in water.

Veins of clear ice are common, running through the milky-white ice making up most of the cave wall. The veins record the refreezing of meltwater in a now-vanished crack. Refreezing yields clear ice. The milkiness of the milky-white ice derives from the abundant tiny bubbles that get left over because you can never squeeze all of the air out as you turn snow into ice.

Probably the most significant item of photographic evidence is that every one of these holes preserves some evidence of its origin as a much narrower crack. Sometimes you can see the crack in the cave roof, sometimes in the floor.

So the unifying theme of this work is hydrofracturing. Suppose your glacier already has a crack in it. If the crack is big enough we call it a crevasse. Whatever its size, it tells us that the body of the glacier is under a tensile stress exceeding the fracture toughness of ice. Whether the crack grows depends on the balance of forces at its tip, the point (in a two-dimensional diagram; in the glacier it is a line) where its width becomes negligible. This is where meltwater comes in. It is a thousand times denser than air, and may well be under the pressure of still more meltwater arriving from the glacier surface. The force balance is very different when the matter pushing against the wall of the crack is water instead of air.

It is now widely accepted that hydrofracturing — crack enlargement facilitated by meltwater — is a crucial piece of several glaciological puzzles, notably the disintegration of ice shelves.

I am glad to have these studies of holes because they increase my understanding of how surface meltwater gets to the bed of the glacier, where it spells trouble, or at least complexity. But it takes all sorts to make a world. No doubt my caving glaciological colleagues, in addition to being curious in general about glacier meltwater, are also glad to have the holes just so they can crawl into them.

Thank goodness for isotopes. The conventional wisdom about the history of glacier ice used to allow for four ice ages in the geologically recent past. Then, in the 1960s and 1970s, oxygen-isotope records from ocean sediments obliged us to increase the number of ice ages, to eight in the past million years (Ma) and many more in the past three or so Ma. It also became clear — although clear should be in quotation marks — that there has been an ice sheet's worth of ice in Antarctica, apparently continuously, since about 14 Ma. More recently, it has become clear that major glaciation began in Antarctica around 34 Ma.

But there is an increasingly persuasive argument that clear should still be in quotation marks. Kenneth Miller and co-authors argue that the isotope records suggest episodic withdrawals of water from the ocean as far back as the later Cretaceous, 100 Ma ago. The only place to put the implied amounts of withdrawn water is into ice sheets.

The argument is appealing because it does away with a long-standing puzzle. Why hasn't Antarctica been glacierized ever since it first drifted into place over the South Pole, where it has been sitting for the past 100 Ma? Miller's answer is simple: it has.

Roughly two oxygen atoms in every thousand are of the heavier 18O isotope, with two more neutrons and therefore 2/16ths more mass than the lighter, more abundant 16O. (The superscript to the left of the elemental symbol O is the mass number, or number of neutrons and protons in each atom, of the isotope.)

Different isotopes of the same element are chemically indistinguishable. But any process that moves stuff around, such as evaporation, is likely to be sensitive to mass. It takes more energy to move heavier objects. Water molecules with an 18O instead of a 16O tend to lag behind in the liquid reservoir. Technically, they have a lower vapour pressure. So they also tend to condense out of the vapour phase more readily.

One result of this fractionation, on the scale of global glaciology, is that an ice sheet, necessarily fashioned out of ocean water, must be isotopically light (more 16O) and the ice-age ocean correspondingly heavy (more 18O). By coring the ice sheet, or the sediment that accumulates on the ocean floor, we get highly accurate and detailed records of — what?

There is a serious complication. Fractionation depends on the temperature as well as the isotope masses.

In ice-core records, the dominant influence is temperature. In ocean sediment cores, the signal due to sequestration of ice tends to be stronger. Unlike the ice sheets, the ocean does not suffer the preferential condensation of heavy oxygen that goes on as the evaporated water makes its way, cooling as it goes, to the site of snowfall. Ocean temperature is still a major confounding factor, however. The story is preserved in fossil micro-organisms, the shells of which are usually assumed to have the isotope abundances of the water in which they lived. But when it is colder the micro-organisms prefer water (and carbonate) molecules with more 18O.

Without information from some other source, therefore, Miller and his colleagues are tackling a problem that is underdetermined, with more unknowns than equations. They draw on several such sources, but the most important is a record of sea-level changes from New Jersey. Of course one sea-level record does not establish a case, but as they cannot resolve the lowest sea-level stands very well their estimates are conservative. The extra independent data turn the exercise into a kind of intellectual, if still speculative, triangulation: a heavy-isotope excursion is likely to be glacial if it coincides with a sea-level fall, and thermal if it does not.

I put the argument for Cretaceous glaciation of Antarctica on the persuasive side of the persuasive/convincing borderline. So many factors contribute to the way the world used to be — palaeogeography, greenhouse gas concentrations and short, sharp changes of sea level are just a few — and hardly any of the evidence remains. But work like Miller's is a fine demonstration of how tantalizing the frontier of knowledge can be. And without the isotopes we would never get anywhere at this speculative frontier.

What with new observations from space of the flow of water beneath the Antarctic Ice Sheet, and against a backdrop of long-standing knowledge that glaciers go faster in the summertime, holes in glaciers have seen an upsurge of interest. We are not talking about cracks in glaciers — crevasses — although the holes usually owe their genesis to pre-existing cracks. These holes, or moulins, are tubes of crudely circular cross-section, they are made by meltwater, and all the evidence suggests that we need to know a lot more about them if we want to understand glacier flow properly.

One way to find out more about natural holes in glaciers is to drill artificial ones. Among the more useful but fundamentally simple technologies in glaciology is borehole video. You lower a camera down your borehole and shoot. You may find, as did Luke Copland and his fellow-workers on Haut Glacier d'Arolla in Switzerland some years ago, that your own hole has intercepted a hole made by the glacier itself. That is, there is a hole in the wall of your own hole.

The forces at work inside glacier ice are varied, but only one can produce this kind of hole: transfer of thermal and mechanical energy from flowing meltwater. The borehole video is showing us conduits. It is reasonable to conclude that the water is coming from the glacier surface. But where is it going?

One thing we have learned from holes in the walls of boreholes, a few centimetres in diameter at most, is that some of the conduits of the englacial drainage system are small. We also know that some are not so small, because boreholes sometimes penetrate larger voids, and sometimes with disconcerting results. If you tap into a void that is filled not just with water but with water under pressure, you get a geyser.

So boreholes in general, and borehole video in particular, are showing us fragments of a complicated system that conveys meltwater through the glacier. Presumably the water sometimes ends its journey by refreezing, but if it can transfer enough heat to the conduit wall it will keep the conduit open and may even enlarge it. If that happens, the water will eventually reach the margin of the glacier or, more interestingly, the bed.

Finding englacial meltwater conduits the size of your finger is a pretty impressive feat, but there is a limit to the number of boreholes we can drill, and finger-sized holes are not physically impressive. A recent study by Catania and Neumann of holes in the Greenland Ice Sheet is impressive as to both technology and physical scale.

They used ice-penetrating radar to image the holes remotely. The ice was 400-600 m thick and they made a number of assumptions, such as that the holes, with diameters of about a metre, are vertical cylinders. In their radar traverses the holes show up as strong diffractors, visually striking hyperbolical shapes, superimposed on the well-defined layering that represents the history of accumulating ice.

The layers are downwarped in association with two of their holes. They argue persuasively that this is because the meltwater delivered by the holes keeps on melting the basal ice, releasing gravitational potential energy as it flows away. Further, to achieve observable downwarping the holes must be persistent. The system of conduits is embedded in a medium that is flowing slowly downhill. Any one hole ought to be short-lived, getting pinched off as the ice carries it away from its source of surface meltwater, or simply squeezing it shut. The two persistent holes appear to have supplied meltwater to the bed for long enough, one to a few decades, to achieve about 30 m of further basal melting in one case and 15 m in the other.

Yet again we have fragments of an evidently complicated picture: two long-lived holes, several more short-lived ones (no layer downwarping), and part of the study area in which the diffractors are so numerous that they obscure the layering completely. The simplest explanation of the numerous diffractors is that they are more closely spaced, smaller holes, perhaps on the same scale as the finger-sized ones seen directly by Copland.

Evidently we still have a lot to learn about holes in glaciers.

The shrinking extent of sea ice in the Arctic has been a cause of concern for some decades, and the record low extent measured with passive-microwave radiometers in September 2007 gathered a good deal of publicity. The September minimum was 7.11 million km2 on average during 1979-1998. In 2007 it was 4.30 million km2. The two minima since then have each been greater. 2008 saw the second lowest and 2009 the third lowest extent.

You can check out the state of Arctic sea ice at the National Snow and Ice Data Center in Boulder, Colorado. Thus far during the present winter, 2009-2010, the extent has been tracking pretty closely the course followed in 2007, so two successive years of increased minimum annual extent do not justify us in concluding that the ice pack might be recovering. Equally, there is no sign of an impending catastrophe at the top of the world, but we would still like to understand why 2007 was a record-breaker. There have been several interesting attempts to explain it.

A particularly interesting attempt by Ron Kwok and colleagues appeared last month. They focus on a detail of the bigger picture, the outflow of sea ice through Nares Strait, the narrow gap between northwest Greenland and Ellesmere Island. To put this study in context, we need to step up from thinking about sea-ice extent to thinking about sea-ice mass.

Very roughly, the ice is 3 m thick on average, for a total mass each average September of about 19 million gigatonnes (but only 12 million Gt in 2007). The mass of the ice pack is the result of a balance between freezing, melting and export. The exported bergs and floes eventually melt, but not within the Arctic Ocean.

Most of the export, about 2000 Gt/yr, is through Fram Strait, between Greenland and Svalbard. Of the other possible outlets, the channels between the islands of the Canadian arctic archipelago contribute little. Apart from being narrow, they are most often blocked at their northward ends by plugs or "arches" of immobile ice. The arches form during the winter and persist until the end of summer, so that for much of the year there is effectively no southward ice export.

Kwok and colleagues found that no arch formed in 2007 at Nares Strait, which was therefore an open passageway for the full 365 days. Between 1998 and 2006, the open-channel state prevailed for only 140 to 230 days per year. From 2004 to 2006, when ice thickness measurements are available from the ICESat laser altimeter, the mass export was about 80-85 Gt/yr, but in 2007 it was 230 Gt/yr.

Why worry about such tiny amounts? The export through Nares Strait in 2007 was only 10% of that through Fram Strait, and insignificant in comparison with the total mass of the pack. The answer is that the ice in the Lincoln Sea, just north of Nares Strait, is some of the thickest, at about 5 m on average, in the whole Arctic Ocean.

The decline of Arctic sea ice is usually discussed in terms of its extent, but that is mainly because we have lots of information about extent. Measurements of thickness are harder to come by, and therefore so are estimates of total mass (area times thickness, multiplied by 900 kg m-3, the density of ice). But one of our main concerns about Arctic sea ice is that apart from shrinking in extent it is also getting thinner.

The impact on ice extent of the free evacuation of thick Lincoln Sea ice in 2007 was small, but it depleted the thick end of the frequency distribution of ice thickness. An ice pack with relatively more thin ice is less likely to survive the summer melt season, so the non-formation of ice arches in Nares Strait constitutes a positive feedback, magnifying the vulnerability of the ice pack as a whole. And it may be a positive feedback in another sense. Presumably arching, that is, blockage, is more likely when the supply of thick chunks of ice is greater. If an episode of free outflow decreases that supply, future episodes of free outflow become more probable.

Palaeogeography is a seductive subject. The appeal of conjuring a vanished landscape out of a few strands of evidence and a good deal of restrained conjecture is irresistible. We know that the reconstructed world is imaginary, but with right treatment of the evidence, under the right constraints, we also know that it must represent a realistic approximation to the way things were.

That is what Douglas Wilson and Bruce Luyendyk appear to have done for West Antarctica at the end of the Eocene, about 34 Ma (million years ago). Records of oxygen isotope ratios in the microfossils preserved in deep-sea sediment suggest that glacier ice began to accumulate in Antarctica at about that time. But models of the palaeoclimate have trouble simulating as much ice on East Antarctica as the isotopes suggest. East Antarctica, palaeoclimatically interesting in itself but not our focus here, is the larger, higher part of the continent.

West Antarctica is not big enough to house the missing ice. Most of it is below sea level. The prevailing wisdom is that you can't grow a marine ice sheet — that is, one with its bed below sea level — from scratch, at least not quickly. So, assuming we are reading the isotopes aright, where was the missing ice?

Enter Wilson and Luyendyk . Their geography of West Antarctica as it may have been at 34 Ma offers a persuasive answer: most of it wasn't below sea level back then.

The first step in the palaeogeographic reconstruction, starting from a map of the modern ice thickness, is to remove the ice and allow the underlying lithosphere to recover from the removed load, rebounding and flexing. You have to juggle with the calculated new surface elevations because in some parts, where the new surface is below sea level, the place of the ice load is taken by a new load of ocean water. This requires care, but it is not a deep problem. On the other hand the result isn't much help. Most of West Antarctica remains under water.

The second step is to account for thermal contraction of the lithosphere. As constrained by palaeomagnetic and other measurements, the main feature of the tectonic-plate system around here until about 28 Ma, the West Antarctic Rift System, was the surface expression of the rising limb of a convection cell in the Earth's mantle. The convection stretched the overlying lithosphere while causing the two sides of the rift to spread apart. Now too high for the fluid mantle rock supporting it, the lithosphere subsided gradually. Wilson and Luyendyk estimate that the subsidence since 34 Ma has varied from 200 to 500 m across West Antarctica.

Apart from correcting for the subsidence, you also have to undo the stretching, moving the Pacific side of the rift some tens of kilometres back towards the East Antarctic side.

A lot of erosion can happen in 34 million years. How do you restore a landscape that has ceased to exist? The answer is that we know, first of all, that the erosive products go downhill, and second that they have to end up somewhere. Most of them end up as sediment offshore, and not too far away. Wilson and Luyendyk rely on sediment thickness estimates for this, their third step. Not all of the sediment is due to erosion, a good deal of it being the fossils of marine microorganisms, and the eroded rock would have been more dense than the deposited sediment. Nevertheless, their approach is very conservative. The volume they restore is only 13% of the volume of offshore sediment.

This step also requires corrections for rebound and subsidence. Shifting loads of sediment are just like shifting loads of water and ice when it comes to the response of the underlying lithosphere.

In the end, Wilson and Luyendyk found another 1.5 million km2 of land, turning the West Antarctica of 34 Ma from an archipelago into a landmass. This plausible landmass, imaginary as it is, is consistent with all the evidence and is constrained by basic principles of physics. In turn it makes what the isotope records imply, that there was quite a lot of ice at 34 Ma, more plausible.

Getting beyond plausibility is a challenge, but one that would be worth rising to because it would allow us to move on to the next questions. Why so much ice? And why then and not earlier or later?