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: August 2009 Archives

Speakers of English in my subject, glaciology, have never been afraid of borrowing good words from other languages. My all-time favourite glaciological word comes from Icelandic: jökulhlaup.

First things first. How do you pronounce it? Icelandic j is pronounced like English y. Then you stretch the ö into an uh noise, and emphasize the h, which should be like the ch in Scots loch. The au is roughly as the vowel in cow, or possibly slurp. You can listen to an Icelander saying it here.

Next, what does it mean? It translates literally into English as "glacier burst", which will be more informative for most readers but is nowhere near as much fun. A jökulhlaup is a large, sudden and usually unwelcome increase in the rate of flow of a stream draining a basin in which there is an ice-dammed lake.

Glacier ice can be very effective as a dam for its own meltwater. Unfortunately it is also untrustworthy for this job. Being less dense than the water it is damming, it is vulnerable to flotation. If the water depth reaches nine tenths of the thickness of the ice dam (the ratio of ice density to water density), the ice will float.

Flotation can be avoided if the water manages to tunnel beneath the ice. A subsurface channel forms and the water starts squirting out of the lake. The flow rate grows steadily because the water enlarges the channel steadily, melting its walls. This kind of jökulhlaup ends abruptly when the supply of water runs out, that is, when the lake has emptied. The glacier carries on deforming slowly, squeezing the channel shut over the course of the winter, and the same thing happens again next summer once the lake has refilled with meltwater.

The nastier kind of jökulhlaup is the one in which flotation is sudden and on a large scale. Huge volumes of water can begin to flow almost immediately. How long the jökulhlaup lasts depends on how long the supply of water lasts, how good it is at enlarging the channel and how long it can keep the dam afloat. The nastiness lies in the unpredictability of flood onset. If you live downstream, or have invested in valuable downstream structures such as bridges or oil pipelines, you get little or no warning of the arrival of an enormous wall of water.

The biggest jökulhlaup we know of is probably the one that emptied Lake Agassiz-Ojibway and led to the disconcerting cold snap 8200 years ago. But in modern times they happen on a smaller scale every year, in hundreds of glacierized drainage basins.

Icelanders learned to live with jökulhlaups long ago. One option, in sparsely inhabited terrain, is simply to stay away from the rivers. Nepal and other Himalayan countries don't have that option. There are many more people than in Iceland, and the rivers are too important as a resource sustaining agriculture. Here the jökulhlaups have come to be referred to as "GLOFs" – glacial lake outburst floods – which I think is not nearly as good as jökulhlaup but does cover the point that not all of these floods are due to the breaching of ice dams. Some of them come from the sudden collapse of moraine dams, and some from the drainage of lakes that are not proglacial (in contact with the glacier margin) but subglacial or supraglacial (on the glacier surface).

Whatever their etymological merits, GLOFs are a serious hazard, and have spurred the completion by ICIMOD, the International Centre for Integrated Mountain Development, of extensive glacier and glacier-lake inventories along the length of the Himalaya. Fears that the hazard is worse now than in former times, when glacier retreat was less rapid, are rational. Glacier retreat creates space for the impounding of water between the glacier and the moraine it left behind at its position of maximum extent. There is more meltwater nowadays, and more scope for it to pond in whatever embayments result from the changing relationship of the glacier to its confining walls. Call them what you will, jökulhlaups or GLOFs are worth all of the attention they are beginning to get.

About 6250 BC, there was a drop in temperature, followed by recovery to what was normal for the time, over perhaps 200 years. Cooling was as much as –5°C at the summit of the Greenland Ice Sheet, and signs of a cooler, or in some places a drier or a dustier, atmosphere can be seen across much of the Northern Hemisphere.

Palaeoclimatologists call this cold snap "the 8.2 ka" (ka being thousands of years before the present, present being defined as 1950 AD). It is a much lesser phenomenon than the 15,000-year course of deglaciation, and lesser even than some of the other short-term anomalies we can see during deglaciation, but the 8.2 ka is nevertheless a clearly-defined, short, sharp blip in the record.

Why did the 8.2 ka start, and why, having started, did it stop? We now think we know the answer to the first of these questions. The simplest explanation is one which also explains, in a rather satisfying way, some of the other cold spells during deglaciation. It rests on how the bulk of the meltwater from the Laurentide Ice Sheet, covering northern North America, found its way to the ocean, and what it did when it got there. The four largest outlets for Laurentide meltwater were the Mississippi, the Mackenzie, Hudson Strait (draining all of the region now occupied by Hudson Bay) and the St Lawrence. As the ice sheet shrank, the volume of meltwater varied but so too did the path it followed to reach the Atlantic.

We can connect evidence from earlier times during deglaciation quite confidently with switches from the Mississippi to the St Lawrence as the main Laurentide meltwater outlet, and from the St Lawrence to the Mackenzie.

Both of these switches were followed by hemisphere-wide cold spells, but the 8.2 ka has a more dramatic precursor than either. It began with, or at least followed, the final dismemberment of the ice sheet into two parts, one east and one west of Hudson Bay. The dismemberment was due to a colossal flood. At the time, meltwater was ponded between the ice margin and the higher ground to the south, as the long-vanished but very large Lake Agassiz-Ojibway. Apparently the lake water was able to force open a subglacial channel beneath the dwindling neck of the ice sheet, draining in one fell swoop (or possibly two) towards Hudson Strait over a time believed to be a year or less. The level of the world ocean would have risen by something like 100–200 millimetres in each swoop, but more significantly the catchment delivering fresh water to the Atlantic via Hudson Strait would thereafter have been close to its present-day extent.

Why should it matter how the meltwater gets where it is going? The key to this question is in the adjective "fresh". In oceanography, fresh means not salty, and not salty means less dense. Make the surface layers of the north Atlantic less dense and you make them less likely to sink, which is bad news for the meridional overturning circulation or MOC. If you think the Laurentide Ice Sheet was big, you should check out the MOC, which is a major player on the global climatic playing field.

The 8.2 ka is one signal from the past for which I can't think of an immediate near-future angle. There are plenty of worrying ice-dammed lakes in the modern world, but there is no chance at all of a repeat of the 8.2 ka in modern times because there are no stores of ice-dammed water anywhere near the size of Lake Agassiz-Ojibway.

However, we don't know the answer to the second question: why did the 8.2 ka stop? Evidently it wasn't big enough to switch off the MOC, and according to the most authoritative recent assessment such an event is "very unlikely" in the foreseeable future. But it would be nice if we could be more confident about such assessments. It would help a lot if we could find out whether the 8.2 ka was a near miss or just a mildly interesting blip in the climatic record.

Even the smallest glacier is too heavy to weigh, at least by the classical method of lifting your object on to a set of scales and measuring the force with which it deflects a spring, or a known weight on the other arm of the scales. But that is not the only way to weigh something.

What we really want to know about the glacier is its mass balance, that is, the change in its mass over a stated period of time. Various ideas have evolved for measuring the mass balance, but until recently the list did not include what would be the obvious idea – repeated weighing – if the things were not too heavy and unwieldy.

That changed in 2002, when the U.S. National Aeronautics and Space Administration and the German Research Agency for Air and Space Travel launched the GRACE satellite mission. GRACE, short for Gravity Recovery And Climate Experiment, is revolutionizing the measurement of glacier mass balance.

GRACE is actually two satellites in the same orbit, one 200 km behind the other. Each continually measures its separation from the other with radar. Subtle differences in this distance are due to equally subtle differences in gravity as experienced by the two satellites – the two arms of the scales. These differences in the force of gravity are dominantly due, once a long list of corrections has been made, to the distribution (and redistribution) of mass in the solid and liquid Earth beneath the satellites.

Late last year Anthony Arendt and colleagues, and Scott Luthcke and colleagues, showed in impressive detail what GRACE was able to make of four years in the evolution of the mass balance of glaciers in southern Alaska. They were able to resolve changes every ten days, and to show that GRACE can see changes, if they are big enough, within regions as small as 200 km across.

Ten days is much better time resolution than offered by traditional methods, which are expensive, time-consuming, hazardous and sparse. GRACE's weak points for our purpose are its poor spatial focus and the fact that it needs the signal of mass change to be strong. The Alaskan glaciers were losing mass at an average rate of 20.6 gigatonnes per year, give or take 3.0 gigatonnes. A gigatonne is an awful lot of ice, but here the most interesting number is the error-bar number. If GRACE can resolve changes as small as 3.0 Gt/yr, what are the prospects for the GRACE follow-on mission for which glaciologists and others are already slavering?

Estimates of the global average glacier mass balance involve a lot of interpolation, which is a fancy word for guesswork. For example the number of annual mass balances that have ever been measured by traditional methods in the Karakoram and western Himalaya is exactly four, and they required a good deal of unsatisfactory corner-cutting. My interpolated estimates for this region suggest an annual loss in the neighbourhood of 3.0 Gt/yr. In other words the present GRACE would have a hard time seeing the glaciological signal from these remote and poorly-covered mountains. But if the GRACE follow-on had sharper focus and better sensitivity it would give invaluable answers, and there are several other mountain ranges and high-latitude archipelagos where it would do equally well or better.

Technically, the improvements now being sketched by GRACE specialists will come mainly from switching from radar to laser interferometry for measuring satellite separation, reducing drag on the satellites, and lowering their orbit. They won't amount to a complete solution of the problem of undersampling of glacier mass balance. There will always be glaciers too small for GRACE to notice, they will continue to contribute a significant proportion of the meltwater flowing into the sea, and we will still need to do small science if we want to understand glacier mass balance. But three cheers for the big science of the GRACE follow-on nevertheless.

A recent study, summarized here, described the Gamburtsev Mountains in the heart of East Antarctica. This buried landscape exhibits many of the classical results of alpine glaciation, including cirques – great bowl-shaped hollows carved out of mountainsides by glaciers – and overdeepened valleys.

The work may have begun as early as 34 million years ago, when records from elsewhere show that ice began to accumulate in Antarctica in significant amounts. But the bowl-shaped hollows are still there. Ice sheets are about 2000 km across, and they don't carve bowl-shaped hollows that are only about 5 km across. So the cirques were probably shaped more than 14 million years ago, which is when we think the ice in Antarctica grew to continental proportions. If this is right, the ice must have shifted from carving up the bedrock surface to protecting it.

We don't know when the Gamburtsev Mountains were first lifted up. The dates just given are from indirect reasoning. On the other hand, the glaciation of Antarctica had to start sometime, somewhere, and a preglacial mountain range not far from the South Pole sounds like a good nucleus. But there is more missing from the story than just the age of the Gamburtsevs.

First, whether mountainous or not, there has been an extensive landmass over the South Pole for much longer than 34 million years. Motions due to plate tectonics brought Antarctica to roughly its present position almost 100 million years ago, yet it seems to have enjoyed a benign climate for the first 60 or more million years of that span. The switch from benign to cool and then frigid could well have been triggered by the uplift of the Gamburtsevs, or possibly of the more extensive Transantarctic Mountain Range, but in the one case we have no evidence as yet and in the other the uplift has been going on for even longer than 100 million years.

Second, this is a good excuse for me to tell you about the widely-unread paper I published 25 years ago about the subglacial topography of Antarctica. Developments since then have not altered the main conclusion: if you take away the ice that now covers East Antarctica, and allow the bed to rebound from the load of 3 to 4 km of ice, you get a rather unusual preglacial continent. This ice-free East Antarctica of the geomorphologist's imagination is a full 700 m higher than all of the continents we know today (except that it is only 500 m higher than Africa – but that is another story). We are not talking about a single mountain range here. This is the whole continent, or in other words a broad plateau cooler than a normal continent would have been by perhaps 4°C.

Unfortunately, we don't know when Antarctica became an elevated plateau, any more than we know when the Gamburtsev Mountains first appeared. There are far too many ifs in the story of Antarctic topography and glaciation. That is a strong argument for reducing the number of ifs, but lurking in the background there is a familiar friend: the greenhouse effect.

Less greenhouse gas in the atmosphere would account for all of the evidence that Antarctica has been getting colder for several tens of millions of years. The evidence that the greenhouse effect has been diminishing for a long time is in fact extremely good. One, or to be more accurate Bob Berner of Yale University, does a detailed accounting of all the carbon in the rocks, and uses the book-keeping to drive calculations of how the carbon would have passed to and fro between the various stores, such as the Earth's deep interior, the biosphere and the atmosphere. The atmospheric stock, nearly all of it carbon dioxide, was about five times its present size 100 million years ago. (Why so? That is yet another other story.)

The more diverse the facts that a hypothesis succeeds in explaining, the more do we respect it. The long-term cooling of Antarctica is not as remote from our 21st-century concerns as it sounds. In fact the same explanation holds for the climate of Antarctica over the past 100 million years as for the temperatures we have measured over the past 100 years and the temperatures we expect over the next 100 years. Greenhouse gas makes our home warmer.

A couple of years ago, I was asked to help in the writing of a report from the U.S. Climate Change Science Program on Abrupt Climate Change. The 460-page report appeared some months ago, but mercifully a short summary was also provided. I was pleased to see that the lead authors defined "abrupt" carefully and clearly – I will come to that definition later – but working out what "abrupt" means caused me a good deal of trouble when I was just getting started.

Abrupt climate change is the name of a new branch of climatology, or rather of palaeoclimatology. I kept asking the palaeoclimatologists what the word meant, and got a variety of answers that were not entirely satisfactory. Eventually one of them said, laconically, "faster than the forcing", and I decided that that was the answer I was looking for.

You need to understand here that "forcing" is scientists' jargon for "cause", a word that we don't like because it is philosophically very shaky. Loosely, the forcing is the input to the system and, in the case of the atmosphere, "climatic change" is the output. Carbon dioxide from fossil fuels is an input. Warming is one of the expected outputs, and so is sea-level rise. In their present-day configuration, glaciers (excluding the ice sheets) are transferring about an additional half a gigatonne of water to the ocean, per year, for every gigatonne of carbon dioxide added to the atmosphere.

The current rate of addition of carbon dioxide to the atmosphere is about 28 Gt/yr. You are free to think either that this is or is not rapid. It is consistent with the calculation that the rate of transfer of meltwater, currently about 400 Gt/yr, is growing at about 12 Gt/yr. Again, you may or may not consider this rapid.

My original problem was that I think "abrupt" has to be more serious than "rapid". If you don't think the present-day changes are rapid, just wait a bit. If you want abrupt change, you may have to wait a bit longer, and could well be disappointed, but you can find lots of examples by looking back rather than ahead. Dansgaard-Oeschger transitions are examples of abrupt warming, well-documented over the course of the last ice age. There was a disconcerting cold snap at about 6250 BC – disconcerting to us, although our Mesolithic forebears had so little capital invested that they may have shrugged it off or even failed to notice it. There is evidence of still bigger abrupt changes further back in the past, up to tens of millions of years ago.

The definition settled on for our report on abrupt change was "a large-scale change in the climate system that takes place over a few decades or less, persists (or is anticipated to persist) for at least a few decades, and causes substantial disruptions in human and natural systems". In other words, "inconveniently rapid for the next generation or two of human beings". You may prefer this to "faster than the forcing" because of its greater immediacy, but it doesn't tell you as much about how things work.

Whichever definition you choose, it is important to realize that "rapid" does not merge smoothly into "abrupt" as the forcing grows more intense. The point about "abrupt" is that it is not what the forcing would lead you to expect. Could it happen to us, or to our grandchildren? There doesn't seem to be any reason why not, even though we can't assign a probability to it with any confidence.