In from the cold: December 2009 Archives
None of us are perfect. Like Canon Chasuble, I myself, for example, am peculiarly susceptible to draughts. Worse than that: I once submitted several decades of revised data on the annual mass balance of White Glacier in northern Canada to the World Glacier Monitoring Service in Zürich - and managed to date all the balances wrong by one year. Fortunately, I noticed the blunder and was able to correct it but this story illustrates an important truth: blunders happen.
I am not the only glaciologist, nor the only scientist, to have blundered. One weakness of WGMS is that, being underfunded, it doesn't thoroughly check the data submitted to it because it can't afford to. In fact, my blunder would have been quite hard for anyone but me to detect. But for anything more intelligent than lists of raw numbers, science has a way of testing the validity of claims. It is called peer review, in which we insist that results not be published until they have been given a thorough going-over by more than one qualified colleague. This is far from a guarantee of correctness, but long experience shows that it is far better than doing nothing.
I recently came across a collection of blunders in no less a place than the second volume, known as WG II, of the Fourth Assessment Report of IPCC, the Intergovernmental Panel on Climate Change. WG II reproduced an unreviewed claim that Himalayan glaciers are very likely to disappear by 2035. The claim was first made in 1999 in a more restricted way, concerning only the central and eastern Himalayas, in a news story in New Scientist, and slightly earlier in an Indian electronic magazine Down to Earth. According to Google, it has been repeated hundreds of thousands of times, presumably in trusting good faith every time. One recent repetition was by Rajenda Pachauri, the Chairman of the IPCC.
It is a tangled story, but in essence the claim is wrong because the date 2035 seems to be a hasty misreading by somebody, now unidentifiable, of the date 2350 in an obscure report. In its context, 2350 was a reasonable date, but changing it to 2035 turns the claim about the disappearance of Himalayan glaciers into garbage.
There is no comfort for climate sceptics in this. A few years ago one of them, a botanist writing to New Scientist, tried to reason from the supposed fact that 555 out of 625 glaciers reported to WGMS were advancing. This claim provoked the not very scientific but understandable response from WGMS spokesperson Frank Paul that, "This is complete bullshit". Investigation showed that there were two levels of wrongness about the claim. Superficially, it was wrong because the sceptic had failed to press the Shift key: he should have typed 55%, not 555. More deeply, it was just plain wrong. The sceptical article was the start of a trail, traced impressively by George Monbiot, that leads back to a non-existent paper not published in Science in 1989.
Depending on what set of years you look at in the real WGMS records, the actual percentage of advancing glaciers is usually nearer to 5.5% or even 0.55% than 55%. It did approach 55% between 1975 and 1985, but, as Figure 5.1 in the WGMS summary of glaciological facts and figures makes clear, the global percentage is distorted by the over-representation of measurements from central Europe. A well known cooling peaked at about the period 1965-1970 in the Alps, which are much studied but host only a tiny proportion of the world's glaciers.
Like me and Canon Chasuble, peer review is not perfect, but there is no question that it reduces the risk of blunders happening and wasting everybody's time. "Everybody" includes concerned lay folk as well as the scientists. One partial antidote to the effects of blunders that is available to all, including those who have to take the experts' words on trust, is the first of Bertrand Russell's ten commandments for liberal thinkers: "Do not feel absolutely certain about anything."
Luckily for my country, Canada (glacierized area 201,000 km2), Antarctica (12.35 million km2) isn't a country at all, and many people have yet to get around to thinking of Greenland (1.76 million km2) as a country. So, if you will allow me Greenland, I live in the world's most heavily glacierized country, not excluding Russia (78,700 km2). But what about the bottom end of the list?
Mexico had 23 glaciers on three volcanoes, but the three on Popocatepetl got wiped out by an eruption a few years ago. (The secret is to stress the first e.) Venezuela is hanging on grimly but certainly unavailingly to its last five glaciers in the Sierra Nevada de Mérida. There are glaciers in all of the other Andean countries, and in Congo, Uganda, Kenya and Tanzania. The most poorly known of Indonesia's glaciers, which are all in New Guinea, disappeared between 1989 and 2003. The others are likely to go soon.
Spain still has a couple of dozen glaciers in the Pyrenees. There used to be a glacieret in the Sierra Nevada, in Andalusia. It was the southernmost glacier in Europe until it disappeared in or about 1913.
Turkey has about 20 km2, and Iran about 27 km2, of glacier ice on scattered mountains.
Would Mongolia surprise you? Depending on which source you consult, it has between 350 and 840 km2 of glaciers in the Mongol Altai and nearby ranges in western Mongolia. There is much more to Mongolia than the Gobi Desert and memories of Genghis Khan.
For a quarter of a century I puzzled over why the guidelines for the World Glacier Inventory included the two-letter country code BU for Burma. Then along came Google Earth, and I was able to zoom in on Hkakabo Razi, a 5,881 m peak near the northern tip of the country. (Sorry, I haven't got a trick for pronouncing this one.) Sure enough, there were the glaciers. This is another demonstration of how Fritz Müller, the architect of the World Glacier Inventory, was decades ahead of his time (or at least of me). As far as I can make out, the Burmese glaciers cover 10-15 km2. I even bought a digital version of a Soviet map of northernmost Burma so that I could inventory them. It is an excellent map, but maddeningly all of the contours are brown and it doesn't show the glacier outlines.
Australia makes a good trick question for trivia contests, because of Heard Island. Heard, with 231 km2 of glacier ice (in 2008; 288 km2 in 1947) and the highest Australian mountain (Mawson Peak, 2745 m), is 4000 km southwest of Perth, in the vastness of the Southern Ocean. But it is a bona fide territory of Australia, cobber.
Sadly, what I used to consider the most surprising glacierized country is now off the list. Marion Island is a part of South Africa about 1800 km southeast of Port Elizabeth. The summit ice on Marion Island, covering about 1 km2, was one of the world's most elusive glaciers. The only useful description in the scientific literature was by a passing ornithologist, and the only usable map is a crude sketch by him. Like the Soviet map of Hkakabo Razi, the South Africans' official map of their only glacier is excellent with the sole exception that it doesn't show the glacier. It does say "Ice Plateau", but that is it as far as glaciological content goes.
We will never know now what South Africa's glacier used to look like. Paul Sumner and colleagues reported recently that the summit ice on Marion Island disappeared during the 1990s, leaving only remnants protected by falls of volcanic scoria.
I admit to a weakness for useless trivia. Nevertheless the search for little glaciers in improbable countries is not a complete waste of time. Of the 131,000 glaciers in the incomplete World Glacier Inventory, 97,000 are smaller than 1 km2. They only account for 8% of the total inventoried area, but the true figure is probably greater. It is time-consuming to count and measure tiny glaciers, so they tend to be omitted from surveys. And that kind of percentage is not negligible if we want the best accuracy when estimating quantities such as the glacial contribution to sea-level rise.
One of the things we have been taught by Hinduism, the majority religion of India, is that everything is connected to everything else. Two recent papers about groundwater extraction in north India offer a fine illustration of the truth of this proposition.
Matthew Rodell and colleagues, writing in Nature, analyse data from the GRACE satellites to show that in three north-western states of India more water is being pumped out of the ground than is being put back. The net extraction rate over six years is 18 gigatonnes/yr, give or take five, which translates to an equivalent depth of water of 40 mm/yr. The qualifier "net" is important because on top of the natural processes of recharge during the monsoon months there must be some recharge from the glacier meltwaters originating in the Himalaya to the north. Rodell and colleagues estimate this recharge as 3 Gt/yr. I reckon it is likely to be somewhat more but the point is that we have to take them together. The conversion of the two non-renewable resources, glacier ice plus groundwater, to ocean water makes an even bigger problem.
Most of the water irrigates cropland, from which it evaporates or finds its way into the rivers. One way or the other, most of it ends up in the ocean – which is what makes it non-renewable.
VM Tiwari and colleagues published on the same subject, apparently independently but relying also on GRACE, in Geophysical Research Letters. They focussed on the whole northern subcontinent, and found an extraction rate of 54 Gt/yr, give or take nine, equivalent to 20 mm/yr of depth of water removed. Add to this most of the 13 Gt/yr of whole-Himalaya loss from glaciers and you get an awfully big problem, affecting 600 million people according to Tiwari and colleagues. (This glacier number is so rough that it doesn't even have an error bar, and it is not clear that the ice loss has been separated cleanly from the groundwater loss.)
One striking thing about these interconnections is that the two papers are the first firm evidence for significant contributions to sea-level rise from "terrestrial storage", that is, aquifers, dams and the like. Up to now it has been conjectured that the terrestrial source is small, but it looks as though we will have to rethink that.
600 million is an awful lot of people. Rodell and colleagues quote the New York Times to the effect that most middle-class residents of New Delhi do not have a dependable source of clean water, from which I infer that a) all upper-class residents probably do, and b) probably no lower-class residents do.
Whatever its source, the water we are adding to the ocean spells trouble for people living near sea level. This includes the members of the government of the Maldives, who now hold their cabinet meetings wearing scuba gear; the 60,000-odd inhabitants of the coral islands of the tiny Indian union territory of Lakshadweep, just north of the Maldives; and the many more inhabitants of the delta of the Ganga-Brahmaputra in Calcutta and neighbouring Bangladesh, where it is rainy and swampy enough that they don't have much call for irrigation but they inherit the problem of groundwater extraction in Rajasthan, Haryana and the Punjab by a roundabout route.
So another striking thing about the interconnectedness of Himalayan glacier mass balance, Indo–Gangetic extraction of groundwater, evaporation from irrigated fields, lack of clean water in New Delhi, and the impending submergence of the Maldives, Lakshadweep and the delta of the Ganga-Brahmaputra is that it also demonstrates that the Hindus, Moslems, Christians, Sikhs, Buddhists, Parsees and other adherents of the Indian subcontinent are all connected to each other – and to me.
All of us who live near enough to a pole, or high enough up a mountain, have done experimental work on the densification of snow. When I was six I once, to test a hypothesis I have now forgotten, posted a snowball through the letterbox outside my school. My hypothesis was unsound, and in the aftermath I was taught my first, painful lesson about the complexities of densification.
When I grew up, I learned a little more. To be specific, I learned about Sorge's Law. Sorge's name (pronounced zor-guh, with the g hard) is well known to glaciologists because in 1954 Henri Bader suggested that it be commemorated by assigning it to a surprising but very important physical relationship: when snow accumulates steadily and there is no melting, the density is always the same at any one depth beneath the surface. The name has stuck.
Density is mass per unit volume. New snow can have a density as low as 50 kilograms per cubic metre if it is particularly fluffy, but the snowflakes collapse rapidly and as time passes the new snow, densifying under its own weight, takes up less and less space. We call it ice once it reaches about 830 kg m-3, at which point the air spaces have pretty much closed off and turned into bubbles. Pure ice has a density of 917 kg m-3, but most glacier ice is a bit less dense because of the remaining bubbles and other voids.
Ernst Sorge's observations were made in Greenland in the winter of 1930–1931, in a hand-dug pit 16 metres deep. His Law actually has rather meagre observational support but, as Bader showed, it follows from some rather simple algebra. It means that, as new mass is added by snowfall, an equal mass of old snow turns into ice (that is, crosses the 830 kg m-3 borderline) at depth.
That gives us a valuable payoff. Provided that nearly the whole thickness is ice, not snow, a change in that thickness, or equivalently in the surface elevation, can be converted to a change of mass by multiplying by the density of the ice. We rely on this extensively when trying to convert volume balances into mass balances. The volume balance, in cubic metres, is thickness change (m) times area (m2). There is nothing wrong with it, except that the mass balance, in kilograms (volume change times density), is more useful for purposes like estimating sea-level change. But a volume balance is considerably easier to measure. You can do it remotely, with laser altimeters, and you don't have to measure the density in the field, expensively.
The savings from invoking Sorge's Law are extremely attractive. For example, we can extend its range of validity by assuming that the density profile also remains unchanged when the rates of melting at the surface and refreezing at depth are constant and equal. But what if this assumption, or the more basic assumption of steady accumulation, does not hold? We don't mind much if the snowfall rate varies from year to year, or with the seasons. (In interior Greenland, for example, more snow falls in summer than in winter.) We can handle that kind of flickering behaviour, as long as we know about it.
A more subtle problem is that the temperature may vary over the years, and that alters the compaction rate (slower when it is colder). A less subtle problem is that on many glaciers nowadays the density profile is being eaten away by continuing mass loss. In their middle reaches, the glaciers are losing stuff that would still be there, suffering compaction, if Sorge's Law applied. Here, Sorge gives too great a density for the lost mass.
What this adds up to is a pressing need to know more about how often Sorge's Law really holds on glaciers, and how to make accurate corrections where it doesn't. Remote sensing can deliver much more knowledge than laboriously digging holes. What would be really nice would be a way to measure not just the glacier surface elevation changes from space, but also the density profile. That is not around the corner, and until we get there we all need to remember that volume change is not the same as mass change. The density we adopt for the conversion is a source of uncertainty.
Alternatively you can browse posts for this category archived by month: