Solar Raditaion 1000 Years

10.2. Several models of the ice-sheet response to July solar radiation changes in the Northern Hemisphere predict that ice sheets should have begun growing several thousand years ago.

100,000, 41,000, and 22,000 years in response to changes in solar radiation (chapter 4). Given the evidence for this long-term relationship, why not try to link the two in a mathematical formula?

For the "target signal" in their exercise, the Imbries used the marine record of oxygen isotopes, which provided an estimate of the size of the ice sheets through the last several hundred thousand years (chapter 4). However, even though the same orbital cycles are present in both the solar radiation and the ice-volume signals, linking the two is not as easy as it might sound. The main problem is that the 100,000-year signal is much stronger in the ice-volume changes than it is in the solar radiation changes, and the reason for this mismatch is not well understood.

The result of the modeling exercise is shown in figure 10.2. When the Imbries published these results, they noted that their model matched the past behavior of ice sheets well enough to conclude that natural orbital variations should carry Earth's climate into the next glaciation within the next few millennia (ignoring current and future human impacts on greenhouse gases). But looking over their results from a fresh perspective, I realized that their model had revealed something they had ignored. Their simulation showed that the current interglaciation should have peaked about 6,000 to 5,000 years ago and that the climate system should have been slowly drifting back into a new glaciation since then. The Imbries would have had valid reasons for ignoring this part of their model's simulation, including complications from temperature overprints on the ice-volume target signal they were trying to match.

Probably the most important reason for ignoring the model simulation of the last 5,000 years would have been the simple fact that no glaciation has developed. This would have made it appear that the model simulation failed during this interval. But now I was looking at this 20-year-old paper from a different viewpoint: Could the model simulation for the last 5,000 years actually be correct? Did this seeming "failure" result from humans having overridden the natural behavior of the climate system by introducing greenhouse gases and stopping the very glaciation the model had simulated? From this perspective, the Imbrie model provided me with an additional line of support for the idea that a new glaciation is actually overdue.

The Imbries' results are a direct outgrowth of Milankovitch's original ice-age theory. Long ago he suggested that ice sheets are so sluggish in their response that changes in their size lag an average of about 5,000 years behind the changes in summer radiation that cause them to grow or melt. The last maximum in summer radiation at the latitudes of the ice sheets occurred 10,000 years ago, after which the solar radiation levels have continually fallen. Allowing for the 5,000-year lag that Milankovitch invoked, the ice sheets should have begun growing about 5,000 years ago, just when the Imbrie model suggests.

I also reexamined simulations from other climate models that had been developed in the last two decades to simulate ice-sheet trends in the past and then project the trends into the future using known changes in Earth's orbit. Some (but not all) of these models had also predicted that at least a small amount of ice should have started forming sometime within the last several thousand years. And yet the scientific community had ignored the basic discrepancy between the model simulations of renewed glaciation and the fact that no ice sheets have formed.

Piecing this information together, I came up with the scenario in figure 10.3. Natural climate was warm 8,000 years ago because of strong solar radiation in summer and high natural levels of greenhouse gases. Since that time, the summer solar radiation level has been falling in response to natural (orbital) changes and causing a natural cooling. Within the last few thousand years, this cooling would

Years Ago

10.3. A natural cooling trend in the Northern Hemisphere should have passed the threshold for initiating a new glaciation several thousand years ago, but greenhouse gases added by humans kept climate warm enough to avoid the start of a new ice age.

Years Ago

10.3. A natural cooling trend in the Northern Hemisphere should have passed the threshold for initiating a new glaciation several thousand years ago, but greenhouse gases added by humans kept climate warm enough to avoid the start of a new ice age.

have reached the threshold at which glaciation became possible, but humans had begun adding greenhouse gases to the atmosphere in amounts sufficient to keep climate warm enough to avoid glaciation. Overall, polar climates did cool during the last 5,000 years because of the solar radiation changes, but ice sheets failed to form because of the human additions of greenhouse gases. During the last two centuries, industrial-era emissions have driven greenhouse gases farther above the temperatures at which glaciation is possible.

I knew that this hypothesis—that an ice sheet of some size should now exist in northeastern Canada—would be provocative and controversial, and I turned to a long-time friend, John Kutzbach, and his colleague Steve Vavrus to try to test the idea a different way. We ran an experiment using the same kind of model that Kutzbach had relied on to test his orbital-monsoon hypothesis (chapter 5), a model that reproduces in three dimensions Earth's climatic response to any kind of nudge away from its present state. The present-day climate is the baseline state in the model, but it can be altered for such experiments.

Our experiment was designed to answer two simple questions: How much cooler would Earth be today if no greenhouse gases of human origin existed in the atmosphere? And, more specifically, would it be cold enough to support a new ice sheet in Canada, or anywhere else for that matter? In our experiment, we made just one change in the baseline state, removing all human-generated greenhouse gases from the model's atmosphere. Out came the CO2 and methane generated by humans prior to the industrial era (based on the analysis described in chapters 8 and 9), and out came the additional gases generated during the industrial era, including not just CO2 and methane but also chlorofluorocarbons (CFCs) (from refrigerants) and nitrogen gases from fertilizers and other sources.

The resulting model simulation produced a global-mean cooling of just under 2°C. This is a very large number: on average, Earth was only a little more than 5°C colder during the last glacial maximum. Slightly under half of the simulated 2°C cooling in our experiment was caused by removing the preindustrial gases and a little over half by removing the industrial-era gases. At higher latitudes of both hemispheres, the amount of cooling was larger, especially in the winter season when model-simulated changes in snow and sea-ice extent amplify the cooling. The largest cooling over any continent was centered on northern Hudson Bay in east-central North America, with a mean-annual decrease of 3—4°C and a winter cooling of 5-7°C (fig. 10.4).

Such a deep cooling would seemingly be favorable for the growth of ice sheets, although the kind of model used in this experiment can give us only a partial answer to whether or not ice actually grew. The model simulates changes in temperature, precipitation (rain or snow), and sea-ice extent and thickness across the entire surface of the planet, as well as properties like wind strength and atmospheric pressure in the overlying atmosphere. But simulating so many variables in so many regions comes at a cost. The model is so complex and expensive to run that only a few decades of Earth's climatic response can be simulated in any one experiment. As a result, very slow processes that develop over many millennia are beyond the scope of this kind of model, including the gradual buildup and melt-back of ice sheets.

Still, the model gives an indication of whether or not ice is likely to have formed in any particular region. It simulates the thickness of snow cover (and sea ice) throughout the year, balancing the accumulation of snow and ice in autumn and winter against the melting in spring and summer. The critical issue is whether or not any snow cover or sea ice persists through the entire summer. If it does, the next winter's snow can add to the thickness, and then the next and the next. In time these snowfields will coalesce and turn to solid ice, marking the beginning of glaciation.

The model results showed just one region on Earth entering this state of "glacial inception"—Baffin Island, the location of Williams's modeling study. Today, Baffin Island is snow-free for an average of 1 to 2 months each summer, both in the model baseline simulation and in reality. But for the simulation in which we removed human-generated greenhouse gases, snow now persisted year-round in a few areas of higher terrain along the high spine of Baffin Island (fig. 10.4). The

10.4. When greenhouse gases of human origin are removed from the atmosphere in a climate-model run, North America becomes much cooler in winter, and snow persists year-round on Baffin Island and for 11 months across high terrain in Labrador.

simulation also showed that that a second area just east of Hudson Bay moved much closer to a state of incipient glaciation. In the Labrador region of eastern Canada, a large plateaulike feature rises to an elevation of 600 meters (2,000 feet). In the baseline run, this plateau had been snow-free for 2 or more months each summer, but in the run with lower greenhouse gases it was snow-free only during the month of August.

The fact that these two regions, and only these two, were in or close to a state of incipient glaciation is highly suggestive if you look back at the map of the melting North American ice sheet shown in figure 4.1. Baffin Island was the last area from which the great ice sheet melted, and it is the first region the model simulates as poised to enter a new glacial cycle. The Labrador Plateau was the next-to-last region to lose its glacial ice cover, and the model simulation shows it as the next area likely to return to a glacial state. In addition, the model results with lower greenhouse-gas levels show sea ice lasting one month longer in parts of Hudson Bay and disappearing for just the month of September, again conditions very close to those needed for glaciation. This simulation provided confirmation for the hypothesis that at least a small part of northeastern Canada would be in a state of incipient glaciation were it not for humans and our early greenhouse-gas emissions.

Still, I was initially disappointed with the results from this simulation. I told my colleagues that I had hoped we might hit a home run with this experiment, with clear indications of incipient glaciation across a broad area, but the localized evidence of a few regions with snow persisting through summer felt more like beating out an infield single.

Yet it was worth keeping in mind that several important processes were absent from the model we used for this experiment. At high northern latitudes, climatically driven changes in vegetation provide an important positive feedback that can amplify the size of initial changes in climate. One important change is linked to shifts in the boundary between the northern limit of conifer forests and the southern limit of treeless tundra. When cooling occurs along this boundary, the forest retreats southward and is replaced by tundra. In such areas, the dark-green, heat-absorbing forest canopy is replaced by bright white, snow-covered tundra (mostly grass and low-lying shrubs). The advancing tundra reflects more solar radiation than the forest it replaces, and so the area cools by an additional amount. This added cooling in turn favors longer persistence of a thicker snow pack into and through the summer months. Our first experiment did not include this positive feedback. But results from other experiments indicate that the large size of the cooling shown in figure 10.4 would push tundra south of its present-day limits, further cool local climate, and increase the extent and persistence of snow and sea-ice cover. Future runs that include these and other feedbacks will likely strengthen our conclusion that at least a small glaciation is overdue in northeastern Canada.

The question of how big an ice sheet might exist in Canada today if nature were still in control is difficult to answer. Baffin Island is the most likely place to have been glaciated, and it is similar in size and dimensions to the combined size of California and Oregon, spanning a similar 10° range of latitude. The high plateau in Labrador that moved close to a glaciated state in the model experiment is about the same size as New England and New York State combined. If future experiments show that both areas would have begun to accumulate ice, the extent of that ice sheet would be much smaller than the enormous glacial-maximum ice sheets, and yet larger than the size of the present-day ice sheet on Greenland. Small and yet big, depending on how you look at it. And in any case, ice was present on this continent and slowly growing toward the south and west.

This reinterpretation of the basic course of climate change during the last several millennia sheds a revealing light on one aspect of the current policy debate about global warming. The confirmation in the 1970s of Milankovitch's theory that gradual orbital variations controlled the growth and decay of ice sheets was one of the great success stories of climate science (chapter 4). The 1980 paper by the Imbries extended these findings, showing that the next glaciation was "imminent," that is, due within no more than a millennium or two. Unfortunately, a few scientists at the time used these results to jump to an erroneous short-term conclusion: they inferred that the brief leveling out and slight downturn in global temperature under way during the 1960s and 1970s might be the onset of a new ice age.

By the 1980s, decades of direct measurements of rising CO2 concentrations in the atmosphere had convinced most climate scientists that the rapid CO2 increase would be a much larger factor in climate change in the immediate future than the much slower orbital changes. As a result, concerns about near-term warming, rather than longer-term cooling, came to dominate deliberations that by then were beginning to receive unprecedented attention from the media. At times, spokespersons for environmental organizations made exaggerated, alarmist statements about the possible harmful effects of future warming.

The 1990s saw a backlash against these alarmist statements, with doubts expressed that the scientific community had a good grip on the complexities of this issue. The switch from predictions of a future glaciation to warnings of a future heat wave was (and still is) cited as an example of scientific incompetence.

The results summarized here bring a wider perspective to this issue. Most scientists who drew on the newly confirmed orbital theory in the 1970s to infer that a glaciation was imminent were clearly doing so in the long-term context of the slow orbital cycles. They saw a new glaciation as being "imminent" in the sense of being due within a few centuries or millennia, not tomorrow. The reinterpretation in this chapter suggests that most of those scientists were actually understating the case, rather than being alarmists. The findings described here indicate that Earth should have undergone a large natural cooling during the last several thousand years, and that at least a small glaciation would have begun several millennia ago had it not been for greenhouse-gas releases from early human activities. The next glaciation is not "imminent"; it is overdue.

On the other hand, the few scientists who jumped to the premature conclusion that the minor climatic cooling during the 1960s and 1970s was a harbinger of glaciation deserve the criticism they have received. They were simply wrong.

Was this article helpful?

0 0

Post a comment