Could Global Warming Be Worse Than You Think?

From the comments to the Krugman post on Exxon and global warming, this appears to be a topic that could use more discussion. Here's a place to start. This was posted at Scientific American's blog two days ago. It's somewhat long, so here's a very condensed version followed by a longer version in the continuation page. The principle being invoked here is the same as for monetary policy in the face of model uncertainty, choose the policy that is robust across models and avoids the chance of a catastrophic outcome. As the San Francisco Fed states, "A policy can be made "robust" to model uncertainty by designing it to perform well on average across all of the available fully specified models ... (McCallum 1988)":

SciAm Observations: One of the questions that came up in the earlier global warming thread was whether climate models have been tested against historical data. ... Climatologists who think global warming is serious and human-driven actually agree with skeptics who say that models have not been adequately tested. But whereas the skeptics think that the models overstate the threat, the mainstream researchers think they could understate it...

Now, what should we make of all this? ... To me, the main lesson of worst-case scenarios is that uncertainty cuts both ways. Skeptics often invoke uncertainty as a reason to defer action because global warming may not be as bad as the headline predictions. But uncertainty equally well means that the outcome could be even worse. Our response should be neither complacency nor panic, but risk-management -- exactly what we do when we buy insurance or strap on seat belts. As David Wasdell of the Meridian Programme said at a workshop I went to last weekend, the scenarios are alarming but not alarmist.

Here's the longer version:

SciAm Observations: One of the questions that came up in the earlier global warming thread was whether climate models have been tested against historical data. As I collect my thoughts on this issue, I wanted to share with you one observation. Climatologists who think global warming is serious and human-driven actually agree with skeptics who say that models have not been adequately tested. But whereas the skeptics think that the models overstate the threat, the mainstream researchers think they could understate it.

Their concern stems from one simple fact: the projected increase in temperature over the coming decades takes us out of the range encountered in the natural ice-age cycle of recent geologic history. It could "imply changes that constitute practically a different planet," climate scientist Jim Hansen told the Washington Post in January, and neither climate models nor humanity is up to it.

Consider Antarctic temperature data from the Vostok ice core for the past 420,000 years:

On this graph, time runs from right to left and 0 represents the current Antarctic temperature. The maximum of 3.2 degrees Celsius was achieved about 130,000 years ago. A rule of thumb is that the temperature variation at high latitudes is about twice the global average, so the maximum corresponds to about 1.6 degrees of global warming. That is, if anything, at the low end of predictions for this century. It is not far off what even most skeptics think we're experiencing.

When the world reaches that point, it will presumably start to look something like it did 130,000 years ago. Back then, Greenland was truly green and sea level was six meters higher than it is today. The case for a recurrence has been articulated by Hansen (among others) in a variety of articles, including one in our March 2004 issue. Last month, a batch of Science papers reiterated earlier suggestions that we already be seeing the early stages of this melting; see my colleague Dave Biello's news story and the discussion at RealClimate. Such melting is not usually included in models, which predict only a gradual sea-level rise due to thermal expansion.

If the warming continues past 1.6 degrees, we have to look even further back in Earth's history to find an analogous situation. Last year Hansen and his colleagues pointed to the middle of the Pliocene epoch, 3 million years ago, when the temperature was 3 degrees warmer and sea level 25 meters higher than today. And we may be plumbing even deeper into history. The 2001 report of the Intergovernmental Panel on Climate Change estimated that it has been at least 20 million years since CO2 levels were so high.

One of the largest, fastest episodes of climate change in the geologic record is the Paleocene-Eocene Thermal Maximum, or PETM, 55 million years ago. Within about 30,000 years, the globe warmed by 5 to 10 degrees Celsius, causing a huge disruption of plant and animal life. The warming coincided with a sharp change in the isotopic composition of carbon in the ocean and atmosphere, suggesting that it was driven by a sudden release of greenhouse gases. A leading candidate is methane stored in marine sediments, which may have been destabilized by an existing warming trend or by a more abrupt event such as a volcanic eruption. The total release of carbon was comparable to what humanity would achieve if it burned all the world's fossil fuels. The gaseous pulse flipped Earth's climate into a new state that lasted for 70,000 or so years before long-term negative feedback mechanisms reasserted themselves.

Scott Wing of the Smithsonian Institution and his colleagues wrote in Science last November: "The PETM provides an important analog to present-day anthropogenic global warming, because the two episodes are inferred to have similar rates and magnitudes of carbon release and climate change." Jim Zachos of the University of California at Santa Cruz has described the event as an example of the tipping points that may await humanity if it continues to push the climate beyond its present range of relative stability.

One recent paper attempts to test climate models by looking back 84 million to 100 million years ago, during the Cretaceous period. The lead author, Karen Bice of the Woods Hole Oceanographic Institution, presented the results at the American Association for the Advancement of Science meeting in February (as discussed in The Economist and Science). The team published its paper last week, and it is worth looking at for an appreciation of the uncertainties involved in paleoclimate data and how scientists wrestle with them.

The researchers took core samples of sediments off the coast of Suriname and analyzed the composition of fossilized plankton, whose oxygen isotope and magnitude-to-calcium ratios reflect the temperature at the time of their formation, and whose carbon isotope ratios reflect the concentration of carbon dioxide in the air. Interpreting these ratios involves a number of assumptions about, for example, the composition of the ancient seawater and the effects of pH on the isotope enrichment process. So the team considered a range of assumptions and was able to put a floor of 33 degrees on the temperature during the Cretaceous (5 degrees warmer than the same region is today) and about 600 parts per million on the CO2 level (roughly one and a half today's concentration). The warming could have been as much as 14 degrees, but even a conservative interpretation of the data is worrisome enough.

According to most current models, doubling the present concentration of CO2 raises the temperature by only 3 degrees. In the specific model that Bice's team applied, the 5-degree rise would require 2500 ppm of CO2, which is above the likely range of values for the Cretaceous. In short, CO2 seems to pack a bigger punch than expected, perhaps because the warming becomes self-reinforcing. The authors conclude that either the data are wrong, the models are wrong, or some other input to the models (such as the assumed methane level) is wrong...

Some scientists are even seeing parallels between present trends and the granddaddy of geologic catastrophes, the Permian-Triassic mass extinction 251 million years ago -- the worst extinction in Earth history, when 70 percent of land species and 90 percent of marine ones died out. Researchers debate the causes, but opinion now leans in favor of massive volcanic eruptions in Siberia, which vented CO2 and other gases. The resulting greenhouse warming unleashed other noxious consequences, including the release from the deep oceans not only of methane but also, as Lee Kump of Pennsylvania State University and his colleagues argued last year, of hydrogen sulfide. These gases amplified the greenhouse, pushing up temperatures by 10 to 30 degrees.

Such apocalypses are thankfully rare, so researchers think they involve multiple insults acting in concert. But paleontologist Peter Ward of the University of Washington worries that humanity might be able recreate some of the conditions if it pushed CO2 levels above 1000 ppm. He wrote in Discover magazine in August 1998:

The Permian extinction is now shaping up as an entirely new type of mass extinction. It had nothing to do with extraterrestrial causes, yet it happened far faster than typical extinctions triggered by internal changes to Earth's climate and chemistry. And if our hypothesis is correct, it raises some very disturbing implications about our current situation. We humans are producing carbon dioxide at a prodigious rate, and many climatologists believe that we are already raising temperatures and altering weather patterns. Are we walking down the same path that killed off so much life 250 million years ago--not from carbon dioxide liberated from the oceans but from carbon dioxide liberated by our cars and industry?

Now, what should we make of all this? You don't have to go so far as to predict another Permian-Triassic extinction to realize that humanity is playing trapeze without a safety net. To me, the main lesson of worst-case scenarios is that uncertainty cuts both ways. Skeptics often invoke uncertainty as a reason to defer action because global warming may not be as bad as the headline predictions. But uncertainty equally well means that the outcome could be even worse. Our response should be neither complacency nor panic, but risk-management -- exactly what we do when we buy insurance or strap on seat belts. As David Wasdell of the Meridian Programme said at a workshop I went to last weekend, the scenarios are alarming but not alarmist.