TCS Daily

Conjecture vs. Science

By Patrick Michaels - March 30, 2005 12:00 AM

A new paper published earlier this month by Science magazine reconstructs the temperature history of the earth for the past 400 years using data gathered from 169 glaciers from around the world. However, problems with the methodology used in this research undermine its conclusions concerning the character of the global temperature record during the past several centuries and its usefulness in adjudicating between surface and satellite-based temperature trend differences. Again, I must wonder what is happening to the peer-review process at our major scientific journals.

The author, Dr. Johannes Oerlemans, relies on the historical accounts of changes in the length of 169 glaciers to develop a representative temperature history for the globe. However, there is no analysis as to whether the glaciers are distributed in such a way as to fairly reflect a "global" temperature history.

For instance, 154 of the glaciers are from locations north of latitude 35ºN. The remaining 15 are in the Southern Hemisphere -- five located on Mt. Kenya, two side-by-side in Papua New Guinea, two neighboring glaciers in New Zealand, and six glaciers in the Patagonia region of South America (five of which are tightly clustered together). At best, this represents only five locations in the entire Southern Hemisphere. And yet, in producing the reconstructed global average temperature, Oerlemans gives the combined temperature history of these 15 glaciers the same weight as the combined history of the 154 glaciers from the Northern Hemisphere (which are a weighted combination of smaller regions within the Northern Hemisphere). This procedure greatly biases the global temperature reconstruction to the few Southern Hemisphere sites.

This is problematical for two reasons.

First, the notion that five sites accurately reflect the temperature history of the Southern Hemisphere is not demonstrated. Performing such a demonstration is a rather easy proposition. Simply randomly select five long-standing and reliable weather stations and see if they how well they reproduce the Southern Hemisphere temperature history of the United Nations Intergovernmental Panel on Climate Change (IPCC).

Second, the glaciers from at least one of the sites, Mt. Kenya, are known to be strongly effected by climate aspects other than temperature. According to the recent work of Kaser et al., (2004), "Figure 1 [reproduced below] illustrates the strong recession trend of all glaciers in equatorial East Africa since the end of the 19th century. The dominant reasons for this strong recession in modern times are reduced precipitation and increased availability of shortwave radiation due to decreases in cloudiness." Later Kaser et al. (2004) add, "In contrast to the 'switch' in moisture conditions, there is no evidence of an abrupt change in air temperature...the glaciers of Mount Kenya and in the Rwenzori Mountains seem to have responded clearly to this change in moisture by retreating drastically and in spatially differential patterns."


Figure 1. Time series of glacier surface areas on Mt. Kenya (open circles), Mt. Kilimanjaro (closed circles), and in the Rwenzori Mountains (crosses). (Figure from Kaser et al., 2004).

Since it has been documented that the glaciers in equatorial East Africa, including those found on Mt. Kenya, have been receding in response to factors other than temperature changes, the records from the 5 glaciers on Mt Kenya should be excluded by Oerlemans as proxy indicators of the temperature during the recent past. Otherwise, they are producing a false signal.

While Oerlemans recognizes that factors such as precipitation and radiation can impact glacial extent, he too readily dismisses them as being insignificant for developing a long-term global record, contending instead that he considers these to be local changes that would be averaged out over larger spatial and temporal time scales. While this argument may be meritorious if the number of sites is large, it clearly is inadequate when only a few sites are involved, as is the case in the Southern Hemisphere where at least one, and perhaps even more, of the sites contain glaciers that respond strongly to changes other than temperatures. Given these limitations, it's clear that producing a temperature reconstruction that is representative of the historical conditions across the Southern Hemisphere cannot be done from Oerlemans' existing dataset. Emphasis should be placed on the temperature reconstruction of the mid-to-high latitudes of the Northern Hemisphere where a greater number of locations are represented by the existing data.

In doing so, the reconstructed temperature history would look significantly different than the one developed by Oerlemans' (non) global reconstruction. To illustrate this point I (coarsely) removed the Southern Hemisphere signal from Oerlemans global temperature reconstruction (Figure 2). This reconstruction looks quite different than the global reconstruction, especially during the latter half of the 20th century. The temperatures of the mid-to-high latitudes of the Northern Hemisphere bottom out in the mid-1800s (presumably the time when the glaciers were at their maximum extent during the Little Ice Age), and then rapidly warm to the mid-1900s. Thereafter, the temperatures cooled to about the 1980s and have begun to rise again to the present, not yet attaining the level of warming found earlier in the 20th century.

It is important to stress that even during the period of cooling -- from 1940 to 1980 -- that the glaciers continued to recede, indicating that there is a lag between surface air temperature and glacial response, and that even during the cooler 1970s, the temperature was still elevated enough above the Little Ice Age minimum to induce glacial retreat. This implies that even if temperatures were "rolled back" several decades through some global effort of atmospheric engineering and greenhouse policy, glaciers would likely continue to retreat. This point, along with the nature of the Northern Hemisphere temperature reconstruction, should not be swept under the rug by the author. Instead, together, they are the strongest findings of the paper.


Figure 2. Rough calculation of the glacier-reconstructed temperature history of the mid-to-high latitudes of the Northern Hemisphere. (Derived from Oerlemans, 2005).

The author also suggests that the glacial records from different altitudes can be used to shed some light on the discrepancy between surface temperature trends and temperature trends in the lower atmosphere derived from satellites and/or weather balloons. However, based upon the distribution of data available from glaciers, this claim is simply false. As such, this point should be removed from the paper. The reason that this claim cannot be tested with the available data is that the controversy involving the discrepancy in surface vs. lower atmospheric temperature trends mainly involves the time period of the past 25 years (since the start of the satellite microwave sounding record in 1979). Oerlemans notes that the number of glacial records drops off precipitously, making reliable comparisons during this period questionable. In fact, they are impossible. Most important, as noted by the National Research Council (2000), the primary region where the trends are different at the surface than those measured in the lower atmosphere is over the tropical oceans, regions without glaciers (see, for example, National Research Council, 2001; Douglass et al., 2004). In the mid-to-high northern latitudes, the surface temperature trends match quite closely those of the lower atmosphere. Thus, Oerlemans' finding that "The warming signals from glaciers at low and high elevations appear to be very similar" sheds no light on the satellite/surface temperature trend discrepancy. The glacier data is simply not of sufficient temporal or spatial quality to be robustly used to examine the differential temperature trends. Therefore, it should not be alluded to by the author, unless a strong statement about its lack of utility is included.

It is obvious that the conclusions of this research would be radically different had the author focused on those research areas that are best supported by his data. The results actually apply only to the mid-to-high latitudes of the Northern Hemisphere, not to the globe or the Southern Hemisphere. As such, these issues must be addressed before the paper should be accepted for publication. Publication in its current form would indicate that the editors of Science are more interested in conjecture than in firm scientific findings.


Douglass, D.H., et al., 2004. Disparity of tropospheric and surface temperature trends: New evidence. Geophysical Research Letters, 31, L13207, doi:10.1029/2004GL020212.

Kaser, G., et al., 2004. Modern glacier retreat on Kilimanjaro as evidence of climate change: Observations and fact. International Journal of Climatology, 24, 329-339.

National Research Council, 2000. Reconciling observations of global temperature change. National Academy Press, Washington, DC., 85pp.

Oerlemans, J., 2005. Extracting a climate signal from 169 glacier records. Sciencexpress, March 3, 2005.


1 Comment

Michaels is funded by the fuel industry
Ahem, I think our friend, the author, should disclose that he has received over $160,000 in funding from the fuel industry. At least, that's what I would do, if I was writing a stock article that spoke of future gains in MSFT, after I received a large payment from Microsoft.

It's an integrity thing. Just so we know why he chose to oppose thousands of scientists who all agree that global warming is not just a figment of our imagination.

If he doesn't post a disclaimer, then marching through every scientific article published on global warming, and systematically discrediting the findings, looks very foolish.

TCS Daily Archives