Arctic Sea-Ice: Another Hockey Stick?

By | October 31, 2007

style=”text-align: center” align=”center”>

style=”text-align: center” align=”center”>

style=”text-align: center” align=”center”> Arctic Sea-Ice: Another Hockey Stick?[1]

This figure, labeled as “Sea-ice Extent: Northern Hemisphere” was presented by Al Gore in the book version of his science (fiction) movie An Inconvenient Truth. But is this depiction of the Arctic sea ice extent over the course of the 20th century even close to reality?

Probably not.

Figure 1. Arctic sea-ice extent as depicted by Al Gore in An Inconvenient Truth. (Source: An Inconvenient Truth, p. 143)

It does, however, bear a lot of similarity to another (at-one-time-popular) depiction of an aspect of climate that was widely used to demonstrate just how unusual things had become under mankind’s stewardship—the paleo-reconstructed multi-century temperature history of the Northern Hemisphere, aka the “hockey stick” graph (Figure 2). That graph, when it first was published, had a many-centuries long “handle” with a single century uptick at the end, “the blade.” The hockey stick graphic made it look like, when left to its own devices, the natural variation of annual average temperature in the Northern Hemisphere during much of the past 1,000 years was very small (and temperatures themselves were trending downwards). Then mankind began pernicious economic activity, and temperatures took off like a rocket in the other direction (indicating a rapid warming).

Figure 2. The “hockey stick” reconstruction of the past 1,000 years of annual temperatures in the Northern Hemisphere as it was depicted in the Third Assessment Report (2001) of the Intergovernmental Panel on Climate Change (source: IPCC TAR, p. 3)

But the rigidness of the “hockey stick” didn’t hold up under scrutiny. Over time, refinements and improvements were made to the paleo datasets, more researchers got involved, the analysis methods changed, data handling techniques were updated, etc. with the net result being that the “handle” of the hockey stick, representing natural climate variability, is now a lot wigglier than it was first depicted (Figure 3). This doesn’t mean that humans haven’t had a significant impact on the earth’s climate in recent decades, but it does better place the impact in the context of a naturally variable climate.

Figure 3. In its latest Fourth Assessment Report (2007) the IPCC depicts the past 1,000 years of the Northern Hemisphere annual temperatures like this—many more reconstructions containing a lot more variance. It no longer looks much like a hockey stick (source, IPCC AR4, p. 467)

The saga of the “hockey stick” is an example of early research efforts into complex topics that produce a far too simplified result making the behavior of well-measured data (in this case, more recent data) look particularly unusual when compared against that of more sparse data (in this case, data from longer ago). Such results are often used to overemphasize the current human contribution to climate variability.

It looks like Gore has honed his own stick. His depiction of the Northern Hemisphere sea-ice extent (Figure 1) shows basically small annual variations, but no trend from about 1900 through about 1970, and then a large decrease in the period since. The decrease certainly looks pretty dramatic and gives the distinct impression (aided by Gore’s presentation) that human activities, which have been fingered in the recent Arctic sea-ice declines, are producing changes that are quite unusual, at least in the context of the last 100 years.

As new data and analyses come to light, it is looking less and less likely that the early-to-mid 20th century variations in Arctic sea ice were as low as indicated by Gore’s hockey stick.

For instance, as we have described in a recent World Climate Report article entitled “A million square miles of open water” there exist historic observations, as well as currently active research efforts, that strongly indicate that there was a large sea-ice extent decline from about the mid-1920s to the mid-1940s. Writing in 1953, arctic researcher Hans Ahlmann noted that “The extent of drift ice in Arctic waters has also diminished considerably in the last decades. According to information received in the U.S.S.R. in 1945, the area of drift ice in the Russian sector of the Arctic was reduced by no less than 1,000,000 square kilometers between 1924 and 1944.” And in a recent seminar at the National Snow and Ice Data Center (at the University of Colorado) researcher Andy Mahoney explained that “[Historic] time series of air temperature and the extents of pack ice, multiyear ice and landfast ice extents reveal three distinct periods of variability over the last 8 decades: a period of warm winters and decreasing summer and fall sea ice extent (period A), followed by a cool period of stable or slightly increasing extent (period B) before a period of year-round warm temperatures and ice loss (period C).” Yet there is no sign of such variation in sea-ice extent in the handle of Gore’s hockey stick from 1900 to 1970.

Gore wants to relate recent Arctic sea ice declines to the recent warm-up there. Seems reasonable. But since the record of Arctic temperatures not only shows a warm-up in recent decades, but also a similar in relative magnitude warm-up from the early years of the 20th century to about the mid-1940s (Figure 4) shouldn’t Gore have expected sea ice to respond in a similar fashion then as now and show a significant decline? Weren’t he suspicious when his Figure didn’t show one? And what about the period from the mid-1940s to the mid-1970s when Arctic average temperatures declined a healthy amount? Shouldn’t he have expected an increase in sea ice extent during that period? Where is that on his hockey stick graphic?

Figure 4. Arctic annual temperature history, 1900-2003 (source: ACIA, 2004).

It is not sufficient for Gore to hide behind the source of his Figure, which he lists as “Hadley Carter” (whoever that is, but we’ll bet it means “Hadley Center,” a British government entity known for the hyping of climate change). The closest thing we could find is this graphic from the Cryosphere Today site of the University of Illinois Polar Research Group. The dataset documentation file that accompanies the chart is full of caveats about how the dataset was put together including the warning “Please note that much of the pre-1953 data is either climatology or interpolated data and the user is cautioned to use this data with care.” Well, the incorporation of climatology (long-term averages) goes a long way towards explaining the lack of variation in early 20th century data. But nowhere in our paper copy of An Inconvenient Truth is any of this made clear. Instead, we just see a graph with little ice variability for 70 years, and then a steep drop off during the past 30.

And finally, a paper was published in 2004 (before An Inconvenient Truth was released) that discussed some Arctic ice data that wasn’t included in the dataset that underlies Gore’s image. This “new” set of old data sounds like the same Russian dataset discussed by Ahlmann and more recently Mahoney. When Johannessen and colleagues used the “hitherto under-recognized” Russian sea ice extent observations to create a long-term 20th century record of sea ice, they produced a Arctic sea-ice extent history that looked quite different from the Gore hockey stick version and, in fact, exhibited a much higher correspondence with the Arctic temperature history (as you might imagine). Figure 5 shows Johannessen’s Arctic sea-ice reconstruction (red line), together with the one likely used by Gore (green line), and the Arctic temperature history (black line).

Figure 5. Arctic sea ice extent reconstructed from the “under-recognized” Russian dataset (red), a version of the sea ice extent dataset likely used by Gore (green), and the history of Arctic annual temperatures (black). (source: Johannessen et al., 2004).

The authors note that the Russian sea ice observations do not encompass the entire Arctic (lacking about 23% of the total area, primarily that which lies along the coast of North America including the eastern Chukchi Sea, the Beaufort Sea and the Canadian Arctic straits and bays), and that the data are inadequate during World War II and the early post-war years. This probably explains the lack of correspondence between the Arctic sea-ice extent and falling temperatures during the decade of the 1940s, as well as why the Russian reconstruction doesn’t fall off as much in recent years (where a lot of sea ice loss was experienced off the northern coast of North America). But, despite these shortcomings, it is interesting to note that the Russian reconstruction includes a far greater degree of interdecadal variation, including a large decline from 1900 to the 1940s, a recovery from the 1940s into the late 1960s (quite possibly underestimated due to insufficient data during the early part of this period), and a then a subsequent decline to the present. The present decline has resulted in the absolute lowest sea ice extent area but it has not progressed at the absolute fastest rate—which occurred early in the 20th century.

Mahoney explained in his seminar, “[T]he Russian Arctic ice pack did not fully recover during [the mid century], suggesting that the early 20th Century warming…may have preconditioned the Arctic for greater change in recent decades.” In other words, human activity may be responsible for pushing Arctic sea ice to its lowest extent in the past 100 years or so, but we had quite a bit of help from Mother Nature.

Gore often brags about getting classified records of Arctic sea ice observations taken by the U.S. Navy during the cold war released for scientific analysis that apparently shows what he wants—a recent decline in Arctic sea ice. But he turns his back on extant datasets that tell a different, less alarming story—that there has been a large degree of variation in Arctic sea-ice extent over the course of the 20th century, much of which was fuelled by non-human induced climate variations.

Gore’s hockey stick, like the one the came before his, is simply wrong in that it underestimates the behavioral complexities of the real world and paints a false picture as to the relative magnitude of the human contribution to date.

Chalk this up to another in the growing list of “errors” that lie within An Inconvenient Truth.


Ahlmann, H. W., 1953. “Glacier Variations and Climatic Fluctuations”. Series Three, Bowman Lecture Series, The American Geographical Society, George Grady Press, New York, available from

Arctic Climate Impact Assessment, 2004. Cambridge University Press. Cambridge U.K. (

Gore, A., 2006. An Inconvenient Truth, Rodale, pp. 327.

Johannessen, O.M., et al., 2004. Arctic climate change: observed and modelled temperature and sea-ice variability. Tellus, 56A, 328-341.