RealClimate logo

Worldwide glacier retreat

Filed under: — eric @ 18 March 2005 - (Français)

One of the most visually compelling examples of recent climate change is the retreat of glaciers in mountain regions. In the U.S. this is perhaps most famously observed in Glacier National Park, where the terminus of glaciers have retreated by several kilometers in the past century, and could be gone before the next century (see e.g. the USGS web site, here, and here). In Europe, where there is abundant historical information (in the form of paintings, photographs, as well as more formal record-keeping), retreat has been virtually monotonic since the mid 19th century (see e.g. images of the glaciers at Chamonix). These changes are extremely well documented, and no serious person questions that they demonstrate long term warming of climate in these regions. New work published in Science (“Extracting a Climate Signal from 169 Glacier Records”) highlights these results, and uses them to make a new estimate of global temperature history since about 1600 A.D., which agrees rather well with previous, independent temperature reconstructions.

Of course, as we frequently remind readers on this site, changes in one particular region do not necessarily translate to worldwide trends. That is why the work of such groups of scientists as the World Glacier Monitoring Service, which compiles observations on changes in mass, volume, area and length of glaciers, is important. From the compilations of WGMS (and many other groups and individuals), we know that glacier retreat is in fact an essentially global phenomenon, with only a few isolated (and well understood) counter-examples, such as western Norway. The figure at right shows an example from WGMS, as published in the 2001 IPCC report. (Click on the figure for details).The photos at left show South Cascade Glacier in Washington State in 1928 and 2000.
What causes glaciers to retreat like this? With the exception of glaciers that terminate in the ocean, and glaciers in the polar regions or at extreme high altitudes where the temperature is always below freezing, essentially just two things determine whether a glacier is advancing or retreating: how much snow falls in the winter, and how warm it is during the summer.

For typical glaciers in mid latitudes, the role of temperature is generally more important than winter precipitation. This is because a bit of extra heat in summer is a very efficient way to get rid of ice. A 1°C increase in temperature, applied uniformly across a glacier, is enough to melt a vertical meter of ice each year. For typical mid-latitude glaciers, winter snow accumulation is on the order of 1 m/year (ice equivalent — or about 3 m of snow). On balance then, a 1°C rise in summer temperature has roughly the same effect as a year in which no snow accumulates. Put another way, for every degree rise in summer temperature, an extra meter of ice-equivalent would be required to offset the extra loss. (This makes it clear why glaciers in coastal Norway are not as strongly influenced by temperature – at these locations, winter precipitation typically exceeds several ice-equivalent meters per year). To give another, more specific example, at a typical glacier on Mt. Baker, in Washington State, a summer temperature increase of 1°C translates to a ~150 m increase in the altitude of the equilibrium line (the point where annual ice accumulation = annual loss), and a resulting ~2 km retreat of the glacier terminus. The same change, if driven by winter precipitation, would require about a 25% decrease in local precipitation at this site.

What all this means is that glaciers comprise a rather nice “proxy” for climate change in general, and for temperature change in particular. Glaciologists have for many years used this fact to make estimates of temperature change from records of glacier change. This work received an important update in the journal Science, with the publication of a paper by J. Oerlemans, of Utrecht University. Oerlemans’s paper does three useful things. First, it provides a compilation of global trends in glacier terminus positions since 1600 A.D. Second, it uses this compilation to create a new estimate of global temperature change. Third it provides an estimate of uncertainties on the temperature estimates, taking into account plausible changes in winter precipitation.

Oerlemans’s reconstruction of global temperatures (largely from mid latitude glaciers) is entirely independent of the much talked about temperature records from other paleoclimate proxy data (e.g. Moberg and others, Mann and others, Crowley and others). Yet Oerlemans’s findings basically agree with the earlier results, as shown in the figure, below. Indeed, the reconstruction of temperature from glacier data is notable for having a rather distinctive “hockey stick” shape, the aspect of the original Mann, Bradley & Hughes reconstruction that seems to attract the most attention and criticism. This poses a substantial challenge to those who have dismissed the “hockey stick” as due to biases or errors. Some will of course quibble with this perspective, because the data prior to the 19th century are rather sparse. (Only a few records go back to the 17th and 18th centuries). However, the “hockey stick” shape is clearly in the data, from both the Northern and Southern hemispheres (see for example the data for Grindelwald, d’Argenti&#232re, and Franz Joseph in the figure at right).

Figure shows comparison of the Oerlemans reconstruction with those of Mann et al. 1999, and Moberg et al., 2005. Click on the figure for comparisons with other temperature reconstrutions.

A few comments:
First, the exact relationship between a glacier and temperature is a bit more complex than implied above, and also depends on the glacier geometry and aspect (which direction it is facing), and on radiative as well as sensible heat fluxes. (The difference between radiative and sensible heat fluxes may be thought of as the difference between the ambient temperature is, and how intense the sun is. We all have had the experience of feeling warmth when sitting in the sun on a day when the air temperature is quite cold. Glaciers experience the same thing.) Oerlemans addresses this by using a simple linear model that relates the glacier length to temperature, with adjustments for the glacier geometry and the local annual precipitation for each glacier. It should be noted that a lot of work was required to do these calculations, much of it presumably by Oerlemans’s student L. Klok. Many of the details are not given in the paper due to the short space provided by Science, but all the information most will want is in the online Supplemental Data on the Science website. (If you want more, see the paper by Klok and Oerlemans in The Holocene.)

Second, Oerlemans’s reconstruction doesn’t say anything about the ongoing debate of whether the “Medieval warm period” was as warm as today. Certainly there is evidence that some glaciers were as small or smaller than they are today at some locations, around 1000 years ago. However, the extent to which the “Medieval warm period” was a pervasive, essentially synchronous retreat of glaciers worldwide (as is happening now) is still open to question (see e.g. Bradley et al., 2003).

Finally, Oerlemans’s work doesn’t address whether or not the worldwide glacier retreat is part of a “natural” phenomenon. Indeed, the fact that glaciers were generally more advanced in the 19th century than they are today is exactly what gave rise to the term Little Ice Age (coined by a newspaper reporter in California, writing about F.E. Matthes work on glaciers in the Sierra Nevada). Again though, the evidence that the Little Ice Age advances were as synchronous worldwide as the current glacier retreats are today is sketchy.

In any case, what Oerlemans’s paper does very well is to demonstrate (one more time) what we already knew: global temperatures have risen more than 0.5 degrees C in the last century (up to 1990 — we don’t yet have a compilation of the latest data). As Oerlemans points out, the only way for this to be substantially in error is if there has been worldwide decreases in summertime cloudiness (by 30% or so!), or in winter precipitation (by 25%!). There is no evidence for either of these changes occurring, and if there were, it would be a remarkable discovery in and of itself.

67 Responses to “Worldwide glacier retreat”

  1. 51
    Tim says:

    Re #49. There is a lot of “underlying climate variability”. One of the best known is ENSO [El-Nino Southern Oscillation] which has a timescale of ~5 years but there are other aspects of climate variability that operate on other timescales both longer and shorter. Unfortunately the observational record is too short [and too polluted by the global warming signal] for these to be known about exhaustively. There are ‘theoretical’ modes of variability with periods of order 20 years [ie the timescale for certain types of wave activity to propagate across ocean basins].

    When you look at climate model output from “control” runs [ie with constant radiative forcing] you can see these sorts of variability for example in multi-decadal cycles of weakening and strengthening thermohaline circulation in the Atlantic. How accurately these relate to the ‘true’ internal variability of the world’s climate is hard to say because in many instances the observations are not there to compare against.

    Another example would be “the Little Ice Age” which it has been suggested was a result of internal climatic variability producing regional cooling over Europe [rather than overall global cooling]. This could have been due to a weakening of the thermohaline circulation. Globally tremperatures could have been quite stable which would require regional temperatures to be anomalously high elsewhere [perhaps the South Atlantic].

    [Response: Note however that many (most) people attribute the Little Ice Age at least in part to solar forcing, not internal climate variabilty. There’s a nice recent paper relevant to this: Shindell et al (2001).–eric]

  2. 52

    I’ve noticed that the global growth rate in CO2 peaks in the second year of moderate-strong EL Ninos (82-83, 86-87, 97-98).

    I’ve also noticed that average dewpoints reach high levels during the second year of El Ninos in the Midwest.

    I remember hearing that stronger El Ninos can be expected as global warming heats up. Do other’s agree that El Ninos are likely to get stronger with more global warming?

    [Response: No offense intended Pat — and this is no comment on what you wrote — but we were supposed to be talking about glaciers! I appreciate that everyone has useful things to say, but this isn’t the place. I don’t have the time to respond to all these off-the-subject comments. Nor do we at RealClimate feel any obligation to have space perennially available for bringing up new subjects. If you think there is a worthy topic we ought to do a post on, please let us know by writing to the contrib at realclimate address, and we will take it under advisement. Since evidently people have run out of things to say that relate to this post, I will phase out the comments space shortly. Thanks! — eric]

  3. 53
    Jeff Norman says:

    Again thank you. This is quite an eye opener.
    How can you tell the difference between climate changes due to variability and climate changes due to forcing? For example, how do you differentiate between the the 97/98 el Nino variability and the climate forcings superimposed over top?

    [Response: This is a rather short answer to a very very good question, but the simple answer is: this is really hard! Indeed, it is what much of the research that many of us do is devoted to, but it is not at all an easy problem. To use my bicyle analogy again, suppose I fall over and try to sue the driver of the car. The defense will undoubtedly point to my inherent wobbiliness while riding bicycles. The only way to resolve this unambiguously is to observe bicyles in the absence of cars — rather difficult to do in most cities. Similarly, we really would like to be able to observe climate in the absense of (changing) forcing. Essentially impossible. We can get close with either a) comptuer simulations or b) looking at paleoclimate data during time periods in earth history for which we have independent evidence of unchanging forcing. Hard to do — I would say, has not really been done very well yet.–eric]

  4. 54
    Joseph O'Sullivan says:

    The receding glaciers are not the only changes that show a warming climate in these regions. The changes in the mountain ecosystems confirm the conclusions that the climate is warming that scientists have drawn from the glacier data. I don’t know if the ecological changes could be used as a climate proxy.
    Where the glaciers are melting, ecological changes that demonstrate a warming climate are also occurring. The record of the ecological changes is not as extensive globally as glacial records but there are efforts to monitor ecological changes. The Global Observation Research Initiative in Alpine environments (GLORIA) is an organization attempting to conduct this monitoring.
    The website is:

    It cites a many scientific papers that have a lot of info on these issues. A very informative one was: “The Changing Face of the Alpine World” by L. Kullman in the Global Change Newsletter No.57 March2004. Itâ��s on the web at:

    For the non-scientist or non-ecologist mountain regions are divided into ecological zones that are divided by climate. From the highest altitude and coldest to the lowest altitude and warmest the general pattern is ice/bare rock to mosses/lichens to grasses/brush to trees/forests. As the climate is warming the ecological zones are moving up the mountain and in some cases the colder ecological zone are disappearing.

    Thanks Eric for recommending the article in Science, “All Downhill From Here?” Science Vol303 12March2004. It’s on the web at:
    It was a great starting point to do some independent research on how the receding glaciers were affecting the local ecosystem.

    Thanks also for recommending the Freely et al 2004 article on the effect of CO2 on seawater chemistry and the ecosystem. It’s on the web on a non-subscription site at:
    A less technical summary in the popular press is at:
    I have read some about the change of the chemistry of the oceans caused by anthropogenic CO2 and I wondered about how it could affect the marine ecosystem. It was great to read something about this. The short comment about scientists raising alarms was also very instructive.

    Finally thanks to Lynn Vincentnathan for the cite for the paper in Nature. The abstract is at:
    Lynn Vincentnathan basically got it right, if climate change disrupts the ocean current system that brings nutrients up from deep waters that the plankton in surface waters need, the plankton productivity would drop. The study’s statement that this effect could last centuries was something that grabbed my attention.

  5. 55
    Joseph O'Sullivan says:

    A correction of my earlier comment, the study in Nature Lynn Vincentnathan commented about, “Decline of the marine ecosystem caused by a reduction in the Atlantic overturning circulation” abstract can be found on the web at:

  6. 56
    John Dodds says:

    Re #36.
    Isn’t the latent heat of vaporization released when you go from steam to water at 100 C, not from water vapor to water?
    I thought that water vapor in air is already water so there is no big heat resevoir to melt all that ice as claimed.

  7. 57
    John Dodds says:

    Re the UNATTRIBUTED responses to #38.
    My point was that in the past it has gotten warmer, and glacier melt was due to all the listed causes- be they apples or fish. In the previous warming (130,000 years ago), they melted without the added benefit of anthropogenic CO2.

    My comment on the shape of the hockey stick being represented in the Solar MAGNETIC FLUX, (but now that you identify it, it is also evident in the solar (energy) vaiations,) was intended to point out that not all of the hockey stick is due to anthropogenic CO2 (which seems to be a common perception).

    It is quite evident to me, that the Earth Temp and the Solar Mag flux and the Solar energy variations,
    –bottom ~1820, rise to the mid 1800s
    –fall back to retest the bottom in 1905 (with the Solar fluxes staying higher & the Earth temp retesting to the 1820 low -probably indicating a time LAG on earth?)
    — Rise to the 1940s top, and fall back to the 1970s bottom
    — Rise to the 1998 top (with the Earth temp rising MUCH more than the solar (presumably due to CO2- which apparently resulted in your “recent…” comment), and then begin to fall again in 1999, 2000 etc.

    To me this is a hockey stick starting in 1820, on the SUN and also on the earth.
    It also says to me that MOST of the hockey stick is solar induced, altho there is NO doubt in my mind that anthropogenic CO2 has added to the 1970-98 temp rise.

    BUT if I go back to the last 15,000 years of Greenland (ice core) temperatures, that the current temp is still well below the peaks induced by the sun (without the use of anthropogenic CO2) about 8000, 7000 and 3500 years ago.
    It also raises questions with me as to how does the ever increasing CO2 forcing account for the temp drops in 1999 or 2000 etc if the CO2 forcing is so so so much more dominant that the solar forcings?

    My bottom line, Glaciers are melting, the earth is warming, there is more CO2 inducing some warming, we are not yet warmer than we were several times in the past 15,000 years.
    I do not yet believe that the computer simulation models are accurate enough to model the systems, and I think that there is probably an under estimate of solar energy (and especially the fluctuations) AND maybe even solar mag flux induced warming, and an overestimate of CO2 effects.
    I also think that the modelling is deficient by not including the supposedly “small” Milankovitch effects AND their multiplying effects on solar forcings and the associated water vapor feedbacks. (If they are so small then WHY in the ice core histories, have they been responsible for the “sudden” changes in temperature (eg 2-4 degrees over 10s to 100s of years) ?

  8. 58
    John Dodds says:

    Re: #38, Gavin’s comments on melting Glaciers in Greenland.
    Yes I agree, there is divergence of ice near the base of glaciers. BUT why did the ice core happen to stop ~123,000 years ago a few thousand years after the peak of the last warming? and why did the ice start accumulating again a few thousand years after the Vostok temp (Global Cooling?, since Greenland temp is not available) dropped by 2-3 degrees? – just coincidence??
    Did the second GISP core stop at a similar time?
    Why would GISP not include more than one ice age cycle like Vostok? (other than the obvious that there is less ice in the north)
    AND there are NOW trees in Alaska and Russia at latitudes higher than in the bottom part of Greenland still covered in ice, but I would expect lower Greenland to be warmer due to the sea & warmer ocean currents.
    AND due to the Earth Precession, the teardrop shape of the north ice sheet rotates around the north pole in about 20,000 years, sequentially exposing some places with more ice at lower latitudes than others. (eg Chicago was covered to 40+ degrees latitude but much of western Russia (@60 degrees lat) was not, at the previous ice peak).

    With the current orbital precession minimizing solar input over North America, I would expect it to be colder there, BUT in 5000 years (or 1/4 of a precession cycle) the teardrop will have moved exposing Greenland to less precession induced cold- hence opening up the possibility of more faster warming and ice/glacier melting in lower Greenland. Added to this in 5000-10,000 years we will again be approaching the eccentricity of 130,000 years ago, further reducing the overall size of the northern ice sheets. and the tilt will have also reduced resulting in cooler summers & warmer winters (less ice), not to mention the higher CO2 levels.

    I do not think I would so cavalierly reject the idea that glaciers will melt to expose significant amounts of land in Southern Greenland, but It will not happen in the next few hundred years.

    Yes the northern ice sheet has existed for 3 million years, but before that it was intermittent, and we know that it moves around due to sea/land, weather and orbital influences. Likewise the southern ice sheet has existed for 30 million years BUT it was intermittent for 12-15 million of those 30. (ref: Zachos 2001)
    (You realy do have a tendency to exaggerate your arguments with the use of “all”, just like I do!!)

    Sincerely, I really do thank you for taking the time to respond to all of us. You are providing a useful service. (even if in my view a little biased!!) I only hope we also expose you to other alternatives.

    [Response: Don’t underestimate the power of coincidence! GRIP and GISP2 did hit bedrock at about the same depth, but given their proximity, that’s maybe not surprising. Note that no-one knows exactly how old the base ice was because it had been deformed and mixed up. NGRIP goes back further into Stage 5e. Talking to ice core people, they think that even older ice might be found further north still, but that is a little speculative. The trees found under the Greenland ice are Pliocene in age – supporting the idea that Greenland has not been substantially de-glaciated since then. Greenland’s ice is there partly because it’s there. That is if the glacier disappeared, it’s unclear whether it would reform. Both the albedo and the altitude effects help Greenland stay (relatively) stable even though similar latitudes do not have glaciers. Another help is that the topography helps keep the ice trapped – Greenland is like a bowl with mountains around the side). This was not true for the Laurentide or Fenno-Scandanavian ice sheets, which therefore come and go more easily. -gavin]

    [Those unattributed responses above were mine; my apologies. In any case, if I may further clarify this: Greenland MAY have been somewhat or even substantially smaller than today at the last interglacial warm period (see Cuffey and Marshall, 2000, Nature). But the age of the ice at bedrock has very little bearing on this. The oldest ice in Antarctica is probably not much more than 1 million years, but the age of the ice sheet is at a minimum several million. Ice sheets and glaciers are like slow rivers. The age of the oldest water in the Mississippi river is of order decades, yet the river has obviously been there for much much longer. -eric]

  9. 59

    Re #36.
    > Isn’t the latent heat of vaporization released when you go from steam to water at 100 C, not from water vapor to water? I thought that water vapor in air is already water so there is no big heat resevoir to melt all that ice as claimed.

    Snowmelt with high humidity is more rapid than with low humidity, other things held constant. The same is true for thawing of ice on lakes and rivers.

  10. 60
    John Dodds says:

    Thanks for the apology & Cuffey Marshall ref in #58.
    I think you are helping my Greenland melting case.
    If the C & M Abstract IS actually correct, then of the 6-7m sea level rise that Greenland ice is now capable of, if 4-5.5m came from Greenland during the last warming then 60-80% of the ice would have melted. And as Gavin pointed out (logically) less ice would melt further north & higher up.
    I see this as further evidence that it was probable that the GISP ice core site in lower Greenland could have been uncovered or at least have had at least 60-80% less ice. Which might explain (along with the mashing & melting at the base of glaciers) why there is NOT 3 or 4 ice age cycles of ice there.
    Now given that 60-80% of the current 3000m of ice is one hellofa lot AND that ice ages are apparently cyclical (and caused by Milankovitch & solar cycles?) and we are approaching similar to eemian time orbital conditions in 10-20K years, then I again say do not be surprised if a hellofa lot of Greenland ice melts (even without CO2 influences).

    The point being that I believe that we have a tendency to think that our current & recent conditions have always been there. Apparently such major changes as (some/much?) bare land in central Greenland would not be so unexpected. AND no amount of Kyoto CO2 intervention will stop it.

    Now, care to comment on my (admittedly off topic) observation about the hockey stick (apparent cause of recent glacier melting) is actually evident in solar and solar mag flux data going back to the mid 1800s?
    One other observation, CO2 Warming theory can NOT explain why ice ages occur, hence leaving the admittedly not perfect Milankovitch theroy (which to me includes Solar variations) in place. and Past ice age data shows multi degree temp rises & drops within 100s of years. If this is valid, then WHY do you (all) keep insisting that orbital influences which currently would be a positive multiplier on solar and would also increase the water vapor feedback, are so small? I wish you would/could revisit this assumption in your models or maybe make it a discussion topic?
    Thanks & keep up the effort.

    [Response: A couple of quick comments. I think you are confusing the timescales of things. Sure, in the long term, climate change will happen with our without human intervention, and if I were around 20 kyr from now I wouldn’t be suprised to see Greenland melted away, CO2 or no. With respect to our own lifetimes though, Milankovitch is simply not important – orbital changes are swamped by other forcing. You are also conflating Milankovitch orbital changes and solar irradiance changes. Milankovich theory has nothing to do with irradience changes — only the seasonally varying distance of Earth to the sun, and orientation of the rotation axis with respect to the sun. Now, the argument against the sun-drives-the-hockey-stick has a least two basic underpinnings: 1) the solar changes don’t competing with the CO2 changes in magnitude of equivalent forcing; 2) the solar changes don’t go in the right direction (that is, they are not as highly correlated with the temperature changes as are the GHG changes). We’ve had some other posts and I’ll track them down and put the links here. We probably ought to do a longer post on this topic too, and will when time permits — eric.]

  11. 61
    Thomas Mölg says:

    Glacier recession on Kilimanjaro

    As one member of the group who is currently doing research on glacier-climate interactions on Kilimanjaro (joint project between Massachusetts and Innsbruck universities), I would like to make some comments on Kilimanjaro glacier recession. They may hopefully be useful for colleagues as well as for non-scientists who are in contact with glaciology and climatology. – And maybe a contribution to the RealClimate ‘tropical glaciers’ backgrounder proposed in comment no. 8.?

    Up to date we have published three studies that address the issue. Unfortunately, the third one has got little attention until now, although it is the one which is most supported by in-situ data.

    STUDY 1, link
    Kaser, G., D.R. Hardy, T. Mölg, R.S. Bradley, and T.M. Hyera (2004): Modern glacier retreat on Kilimanjaro as evidence of climate change: Observations and facts. – In: International Journal of Climatology, vol. 24, no. 3, pp. 329-229.

    In this study we review a variety of papers, ranging from the first observations of Kilimanjaro glaciers by Hans Meyer in the 1880s, to 20th century satellite data of tropospheric temperature. This happens with the intent to develop a working hypothesis for our research. Based on all these studies, a late 19th century moisture drop is by far the most likely event that has initiated the retreat of glaciers on Kilimanjaro. A subsequent drier climate was the main driver for maintaining this retreat. As correctly mentioned in comment no. 8, it cannot be ruled out that this local climate change driving glacier retreat (drying) is connected to the large-scale change of our atmosphere, as we suggest on pages 336 and 337 of the paper. We intend to explore during the next three years (official duration of current project) if such a connection does exist. – – Hence, we certainly don’t deny general global warming at all. Unfortunately, climate skeptic groups have misused mainly this study (but also the others below) to argue against the global warming issue. All we aim at is to explore glacier recession on Kilimanjaro in its full complexity.

    STUDY 2, link
    Mölg, T., D.R. Hardy, and G. Kaser (2003): Solar radiation-maintained glacier recession on Kilimanjaro drawn from combined ice-radiation geometry modeling. – In: Journal of Geophysical Research, vol. 108, no. D23, 4731, doi:10.1029/2003JD003546.

    Although submitted after the paper above, it was published three months earlier due to the fast JGR review process. – – This is the first study that builds on the concept from Kaser et al. (2004). It shows with a simple ice cap model that the lateral retreat of the vertical ice walls (that form the margin of the ice bodies on the summit plateau) is controlled by the (spatial distribution of) energy from solar radiation. Solar radiation receipt at the surface is much more tied to moisture-related parameters (especially cloudiness) than to air temperature. The existence of sharp features like the ice walls further excludes a significant impact of air temperature on the glacier. This is nicely explained in the paper of Kraus (1972, cited in Kaser et al., 2004) from a physical viewpoint. – – On page 4 of our paper we note that we pursue an “exploratory approach”, thus we are aware of the fact that this is not a ‘classical’-scheme study. Nonetheless, it revealed the basic process forcing ice wall retreat, and helped us to design the new measurements (see below).

    STUDY 3, link

    Mölg, T. and D.R. Hardy (2004): Ablation and associated energy balance of a horizontal glacier surface on Kilimanjaro. – Journal of Geophysical Research, vol. 109, D16104, doi:10.1029/2003JD003546.

    The first study that uses an extensive in-situ data set (recorded by the Massachusetts automatic weather station) for application to an energy balance model. This model has a standard design as used by glaciologists to quantify the impact of weather/climate on glaciers. Only such models give a more precise picture of the complex energy exchange processes taking place between atmosphere and glacier surface (cf. comment no. 15). – – Applied to Kilimanjaro’s Northern Icefield, the model demonstrates how important precipitation is for energy exchange and therefore mass loss at the glacier surface. The key behind is that albedo (portion of incoming solar radiation reflected at the surface) controls the entire energy balance, and albedo depends on snowfall amount and frequency mainly. The sensitivity study in the paper shows that a moderate reduction in precipitation (-20%) would lead to much stronger mass loss than an increase in air temperature of 1°C. – – This control by albedo is the case for most glaciers all over the world. Glaciers that are located in the vicinity of the 0°-level (a first, rough approximation for dividing between rain and snowfall zone) thus strongly suffer from air temperature increases as the rainfall zone is pushed towards higher elevations (and mean glacier albedo decreases). But what is special about the Kilimanjaro glaciers? These glaciers are located at such high altitude (> 5,000m; measured annual mean temperature = -7°C) that any realistic increase in air temperature will not push the rain zone as high as the glaciers are located. The impact of local air temperature on these glaciers therefore remains small – which is well visible in the energy balance model results.
    As said above, to get precise insight into the impact of climate on glaciers it is inevitable to know the surface energy balance of a glacier. Everything else includes a certain bit of speculation. Thus, I encourage everyone who is interested in the Kilimanjaro glaciers to download this paper, especially as it has got little attention compared to the two other papers.

    Some specific comments to the discussions on this site, as well as some general remarks:

    Kilimanjaro glaciers and local deforestation
    Regarding the importance of moisture for the Kilimanjaro glaciers (as outlined above), it is likely that deforestation on the Kilimanjaro slopes (cf. comment no. 4) contributes to the changes (precip. deficit) in the summit area. Several studies have shown that a decrease in vegetation initiates several processes (e.g., increase in moisture divergence) which promote a decline in precipitation. Deforestation might therefore be a direct human impact on glacier retreat (we are about to test this hypothesis… ). – – An interesting study on the link between vegetation changes and climate is the following: Pitman, A. J.; G. T.Narisma, R. A. Pielke Sr., and N. J.Holbrook (2004): Impact of land cover change on the climate of southwest Western Australia. â?? In: Journal of Geophysical Research, vol. 109, D18109, doi:10.1029/2003JD004347.

    Temperature change in the tropics
    It is no doubt that air temperature has also risen in the tropics. We haven’t said that air temperature isn’t rising in the tropics, as suggested by comment no. 8. I guess this refers to our statement “Temperature increases in the tropics on the surface and in the troposphere have been little in recent decades compared with the global trend” in our Kaser et al. (2004) paper? – – which does not imply that air temp. did not increase (and may be unfortunate regarding more recent studies like Fu et al.). However, East Africa seems to be a region which has experienced scattered trends in temperature evolution over the 20th century. The study of Hay et al. (2002, full citation in our paper STUDY 1) found no temp. trend (since 1911) in records of East African highland stations, and other authors (Kingâ??uyu, S.M., L.A. Ogallo, and E.K. Anyamba (2000): Recent trends of minimum and maximum surface temperatures over Eastern Africa. â?? In: Journal of Climate, vol. 13, no. 16, pp. 2876-2886) find different trends in East African stations. This leaves of course the question if such trends measured at lower elevations have also occurred at the high altitude of the Kilimanjaro glaciers….. The automatic weather stations (see below) may be helpful in finding answers (Doug Hardy is currently analyzing this). Hence the magnitude of temp. change in the middle troposphere of East Africa seems unclear (Hense et al., 1988, Meteorol. Atmosph. Phys., 38, 215â??227, even detected a cooling from radio sonde data of Nairobi….), while in South America many researchers have documented a rise in air temp. also at the altitudinal level of tropical glaciers. E.g., Ray Bradley and his group have done work on this.

    Automatic weather stations (AWSs) on Kilimanjaro – two new stations
    Three AWSs are currently collecting data on/near glaciers. The longest recorded is provided by the Massachusetts AWS on the Northern Icefield (summit plateau, since 2000). Two new AWSs have been installed recently (02/2005) and are operated by Innsbruck University; one in front of a vertical ice cliff (to better validate STUDY 2), the other on a steep southern slope glacier. More information on the stations is available on our homepages, including monthly updated weather data on the Massachusetts site (links at the end of posting). These measurements will help to characterize current Kilimanjaro summit climate as well as extending the energy balance modeling. Regarding palaeoclimate, Lonnie Thompson of Ohio University has drilled ice cores (in 2000) to reconstruct Holocene climate variability (paper).

    Tropical glaciers and moisture
    The importance of moisture for the glaciers on Kilimanjaro is no surprise. This importance has been as well demonstrated for other tropical glaciers, worldwide. The obvious background for this is that seasonality in tropical climate is mainly caused by the annual cycle in moisture-related parameters (precipitation, clouds, …), while monthly air temp. fluctuates little over the year. Hence, as correctly noted in comment no. 16, the relation with air temperature is quite different compared to outertropical glaciers. Hans Oerlemans of course knows this, which may explain why he has focused on outertropical glaciers in his recent analysis. – – The strong sensitivity of tropical glaciers to moisture-related climate parameters has been recognized decades ago! – mainly by the studies of Stefan Hastenrath (1970s, 80s), and also in the earlier papers of Georg Kaser (long before I joined his group…). More recently, the French group of Patrick Wagnon has done valuable energy balance studies on South American glaciers. However, we feel it is only the most recent years that tropical glaciers have come into broader interest.

    Following all my comments above, using Kilimanjaro glacier recession as an icon of global warming is – in our opinion – simply a bad choice. There are many better examples experiencing a direct impact of the recent warming. However, Kilimanjaro is an icon of climate change, and with all its complexities represents quite nicely the problems that scientists have in understanding climate change.

    We had lots of media contact since we have started this research several years ago. In many cases this was frustrating. Since Kilimanjaro is one of the famous symbols of our earth, many people seem to believe they also understand what is happening with these glaciers. However, non-experts like politicians and journalists should be careful in making quick comments about the causes of glacier retreat on Kilimanjaro, as all their ‘stories’ usually attract broad interest. We have also made good experience with scientific journalists and are therefore willing to provide any information – before arbitrary assumptions are reported to the public. Comparing two pictures of Kilimanjaro – a 1993 photo after an immediate snowfall event, and a 2000 photo of the dry season without any seasonal snow cover (link) – was presented for quite a while as evidence how fast global warming forces glacier recession, and therefore used to totally mislead the public. Only since a short time ago it has been added that snow is involved in the 2000 photo, and thus a comparison in terms of glacier retreat does not make sense (cf. editor’s note in linked page). This is only one example of how Kilimanjaro glaciers are ‘abused’ and the exciting scientific questions behind are ignored and oversimplified….

    Sorry, this posting became rather long, but hopefully appropriate as so many different Kilimanjaro stories without scientific background are circulating these days…… – – Thomas.

    the links:
    Tropical Glaciology Group, Univ. of Innsbruck, Austria:
    Climate System Research Center, Univ. of Massachusetts, USA: :

    [Response: Thanks for all that useful info, no apology for length needed – William]

  12. 62

    […] in Verbindung gesetzt. Hier sind einige Beispiele mehr aus dem IPCC TAR. Realclimate hat ebenfalls einen Artikel zum Thema globaler Gletscherrückgang […]

  13. 63

    […] this, and a slight cooling in parts of Antarctica (now abating), most of the world’s glaciers have been receding as part of an inter-decadal trend, ice loss has accelerated in Greenland (resulting in a large net […]

  14. 64

    […] this, and a slight cooling in parts of Antarctica (now abating), most of the world’s glaciers have been receding as part of an inter-decadal trend, ice loss has accelerated in Greenland (resulting in a large net […]

  15. 65

    […] Despite this, and a slight cooling in parts of Antarctica, most of the world’s glaciers have been receding as part of an inter-decadal trend, ice loss has accelerated in Greenland (resulting in a large net […]

  16. 66

    […] Despite this, and a slight cooling in parts of Antarctica, most of the world’s glaciers have been receding as part of an inter-decadal trend, ice loss has accelerated in Greenland (resulting in a large net […]

  17. 67

    […] gráfica claramente muestra que los datos extraidos de los avances y retrocesos de los glaciares concuerdan bastante con los datos de temperaturas medidos y con la reconstrucción de Mann de las te…(la famosa curva de hockey que tanto critican los […]