Glaciers on Kilimanjaro
    Gore error list                                                                      Cool It, British                        Cool It, American        

In his film, Al Gore talks about the rise in atmospheric CO2 levels: "It just keeps going up. It is relentless. And now we´re beginning to see the impact in the real world. - This is Mount Kilimanjaro more than 30 years ago - and more recently (2000). And a friend of mine just came back from Kilimanjaro with a picture he took a couple of months ago. . . Within a decade there will be no more snows of Kilimanjaro."

The earliest estimate of the glacier surface area on Mount Kilimanjaro refers to the year 1880, when it made out about 20 km². Since then, it has declined continuously; in 2000 the area measured 2.52 km² and in 2007 1.85 km². From 1912 to now, the decline in the horizontal parts of the glaciers has been nearly linear (i.e. the same size of area disappeared each year). Details are given in the paper by Kaser et al. (2004) and Cullen et al. (2006), which are cited by Lomborg, and in Thompson et al (2002) and Thompson et al (2009)..

According to Kaser et al. and Cullen et al., glaciers around the summit have horizontal surfaces which end abruptly in vertical walls to the side facing the summit and the central crater. The position of these vertical walls is gradually shifted by `ablation´ of the ice. This does not imply that the ice melts away - the temperature of the ice never surpasses ÷ 1.6° C - but instead the ice disappears by sublimation, i.e. direct transition from solid phase to vapour phase. According to Kaser et al., once a vertical wall has receded by some distance, it cannot advance again, because snow does not deposit on a vertical wall. This means that the walls are inherently unstable - they can only recede, never advance. The implication is that once these walls formed in the past, the glacier was doomed to dwindle. In addition, however, there are also sloping glacier surfaces some of which are more nearly in balance between addition of snow and removal by sublimation.

Kaser et al. indicate that there is practically no melting of ice on the horizontal surfaces. This contrasts sharply with Thompson et al. who found considerable surface melting. In their 2009 paper, they found that half of the ice loss is due to surface melting, and only half to shrinking from the sides. In one of the glaciers, the glacier surface had lowered 3 m in just 7 years. The top ice layers originating from the period after 1952 have melted away. Ice cores show that similar melting has not occurred during the preceding centuries.

Each year there are two rainy/snowy seasons and two dry seasons. During the former, snow is deposited on the glaciers. During the latter, a relative lack of clouds means that sunlight has direct access to the glacier surface, and the incoming radiation supplies the energy for sublimation without any involvement of sensible heat. An instructive explanation of this is found in
this link. The most crucial parameter in this process is the surface albedo - the whiter the glacier surface, the more sunlight is rejected and the less is available for the sublimation process. Frequent snowfall gives a white surface and hence little sublimation. Careful studies of the mass balance of a sloping glacier on Kilimanjaro has shown that the rate of sublimation is affected five times more strongly by a 20 % decrease in snowfall than by a 1° C increase in air temperature. This means that changes in both precipitation and temperature could in principle cause glacier retreat, but changes in precipitation, if they occur, will have the greater impact. In addition, there could be effects of increased deposition of dust and soot (from forest fires) on the glacier surface (this would reduce the albedo).

Information on climate changes in the region is scanty and contradictory. Precipitation has changed markedly at some points of time. This is known espcially from variation in water level in Lake Victoria which is not far away. Precipitation was low in the first part of the 17th century, but increased from 1850 onwards, and there were excessive rains during the years 1876-1878. After that, precipitation dropped suddenly to low levels from 1890 onwards. These levels then increased again to rather high levels after 1960. According to Thompson et al. (2009) there are no overall general trends in precipitation in East Africa, whereas there are rising trends in some aspects of temperatures. This paper states
(p. 3994) that "drying has not occurred over tropical eastern Africa."

If the extent of the ice cap is mainly determined by precipitation, then it is conceivable that it had its maximal size just around 1880 when the first recordings were made. The marked decline in ice cover from 1880 to 1912 could be explained by the time sequence of record amounts of precipitation followed by a period with much less precipitation. The continued decline during the next decades could also be explained by continued low precipitation. But for the years after 1960, the situation is different. During
the mainpart of the 20th century, recordings from weather stations in the region indicate no significant changes in average temperature or average rainfall.  See for instance rainfall data in Rohde & Virji (1976): Monthly weather review 104: 307-315. But in the latter part of the 20th century, precipitation has increased. Røhr & Killingtveidt (2003): Hydrological Sciences 48(1): 65-77 indicate that some of the highest weather stations on the south slope of Kilimanjaro show a significant change in rainfall after 1970-1980. This is most clear in the highest station (above 1500 m), where annual rainfall during 1984-1999 was 42% higher than during 1954-1983. Thus, if glacier retreat was governed by precipitation, it should have slowed down or stopped after 1984. In addition, after c. 1990 there have been clear surface temperature increases in east Africa (link). Thompson et al. (2009) write: "A 25-year temperature and precipitation history recorded in the Amboseli Basin, a few kilometers from the northern base of Mount Kilimanjaro, reveals a warming rend in both maximum and minimum temperatures and large interannual variability in precipitation but no long-term trend."Therefore, the situation now is that precipitation has changed so as to  favour ice retention, whereas temperature has changed so as to favour ice melting. This suggests that temperature rather than precipitation plays a role for what we are seeing now.

Judged from the situation in Lake Victoria, the period 1800-1850 had very low precipitation, maybe slightly lower than during 1900-1960. If the extent of the ice cap were governed solely by precipitation, it should have been very small before 1850. But this seems not to have been the case. There has been more ice after 1700 than before that year (judged from the fact that the small glacier inside the crater has existed only for 300 years, according to the ice core study).
However, a series of facts indicate that there was a 300 year long period of extreme drought around 4,000 years BP.
In the ice cores studied by Thompson et al., this period appears as a marked dust layer. Some of the ice cores are no older than this, indicating that where they were drilled, there was no ice 4,000 years BP. That is, the ice cap then was even smaller than now, at least to the side facing the crater. On the other hand, the fact that the largest ice field continues further back, right down to the end of the last glaciation, indicates that even during a period much drier than now, this ice cap persisted. This suggests that low precipitation can not, or not fully, explain that the ice cap is disappearing now. Furthermore, the upper layers of the northern ice field contain evidence of surface melting that does not appear anywhere else in the ice core representing an 11,700 year long history (Thompson et al. 2009). Thus, what is happening now is unprecedented.

Precipitation depends on conditions in the Indian Ocean. In certain years, the usual distribution of temperatures in this ocean (warmest to the east) is reverted (warmest to the west). In such years, precipitation in east Africa is higher than usual. There is information that this climate pattern prevailed much more often during the period 1820 - 1880 than before and after this period (
link). This is based on proxy data such as historical accounts of the water level of Lake Victoria, and of winds and currents in the Indian Ocean. It is believed that precipitation was higher during that period, and that it decreased suddenly by about 20 % in 1880. This could explain that the glaciers on east African mountains, especially Kilimanjaro, receded from 1880 onwards.

Is the changed temperature distribution in the Indian Ocean a reflection of the ongoing climate change? In an attempt to answer this question, Mote and Kaser (link) write: "It is possible . . that there is an indirect connection between the accumulation of greenhouse gases and Kilimanjaro´s disappearing ice. There is strong evidence of an association over the past 200 years or so between Indian Ocean surface temperatures and the . . precipitation patterns that either feed or starve the ice on Kilimanjaro. . Any contribution of rising greenhouse gases to this circulation pattern necessarily emerged only in the last few decades; hence it is responsible for at most a fraction of the recent decline in ice and a much smaller fraction of the total decline."

As explained in
this link, partially based on this reference, air temperatures in the mid troposphere in the altitude of the tip of Mount Kilimanjaro (c. 5,000 m) are fairly uniform all the way around the equator. These temperatures have risen by an average of 0.2°C per decade during the period 1960-1979 and afterwards, although the rise is not seen in every  part of this  period.  With a  fairly uniform trend around the globe, it is  tempting  to assume that the simultaneous shrinking of glaciers everywhere in the tropics is caused by one single factor, the air temperature at this altitude. However, those that favour the hypothesis that changes on Kilimanjaro are not due to rising temperatures, present other data for the mid troposphere that do not show a rising temperature (link). The data treatment of the satellite-based temperature measurements is complicated, and different interpretations may have led to the different estimates of temperature trends.

In conclusion, it must be said that the issue is complicated and not settled. There remain two explanations of what is happening. These need not to be competing alternatives, but may supplement each other. It is likely that changes in precipitation are the main cause for glacier retreat in the decades after 1880, whereas surface melting due to rising temperatures seems to explain about half of the glacier decline happening today.
What is described here refers only to Mount Kilimanjaro. Here the glaciers are situated so high above the mean 0° isothermal surface that melting due to heat transfer plays a small role. Other tropical glaciers situated at lower altitudes and in more wet climates are more affected by changes in temperature, especially
glaciers in the Andes (Favier et al. (2005): J. geophysical research 109:D18105, link).

A list of informative papers on the subject from the tropical glaciology group in Austria and Switzerland is available at
this web site.