Earth’s climate may warm considerably more than expected in response to a doubling of atmospheric carbon dioxide, a new study of a broad range of climate models hints. The reason, the scientists say, is that simulations that now show only a moderate amount of warming don’t accurately depict the amount of cloud formation in the lower atmosphere, thus cooling the climate more than real-world data suggest will actually occur. If true, warming of the planet will fall toward the high end of the range offered in every expert climate assessment of the past 3 decades.
Carbon dioxide is a so-called greenhouse gas: The more of it there is in the atmosphere, the more heat is trapped there and the higher the global average temperature climbs. Scientists have long debated how sensitive Earth’s climate is to this planet-warming trace gas. Specifically, they ask, how much will worldwide temperatures rise if the level of CO2 becomes double that seen in the era before human activity began spewing the gas into the atmosphere?
Current models and a range of observations suggest that Earth will warm somewhere between 1.5° and 4.5°C once carbon dioxide levels are twice the preindustrial concentration of about 280 parts per million and the climate system adjusts, says Steven Sherwood, an atmospheric scientist at the University of New South Wales in Sydney, Australia. That’s a wide range, he notes—a range that hasn’t narrowed since the first computer simulations of climate debuted in the 1970s. Broad analyses have hinted that a model’s climate sensitivity depends, in large part, on how the model estimates cloud formation at low altitude, he adds. If a simulation produces generous amounts of low-level clouds, more sunlight is reflected back into space, and Earth, on the whole, is cooler than it would have been without the clouds.
In an attempt to narrow the range of climate sensitivity, Sherwood and his colleagues analyzed the results from 43 different climate models. Specifically, they looked at how the simulations represented mixing in the lowest few kilometers of the atmosphere, where many clouds form, as climate gets warmer. Then, they compared model results with data gathered worldwide.
The team found that on the whole, the global climate models with low climate sensitivity—all 15 of those in which global average temperature rose less than 3°C for each doubling of CO2—produced far too many low-altitude clouds. “These [low-sensitivity] models are doing it all wrong,” Sherwood says. On the whole, he and his colleagues say, increased convection in the lowest portion of the atmosphere will tend to dry out the air there, making cloud formation less likely. That, in turn, suggests that the low-sensitivity models shouldn’t be trusted, and that Earth will most likely warm more than 3°C for each doubling of CO2, the researchers report in today’s issue of Nature.
The team’s results suggest that about half the variation in climate sensitivity is explained by differences in how the models depict mixing in the lower atmosphere, climate scientists Hideo Shiogama and Tomoo Ogura of the National Institute for Environmental Studies in Tsukuba, Japan, comment in the same issue of Nature. The rest of the variation can’t yet be explained, they note, but important factors could include how the models simulate overall changes in the amounts of sea ice or of high-level clouds.
Because the team’s analysis of the climate models focuses on the processes incorporated into those simulations, “it’s pretty credible,” says Gavin Schmidt, a climate scientist at NASA's Goddard Institute for Space Studies in New York City, who also was not involved in the new work. However, he notes, previous studies have indicated that climate sensitivity below 3°C can’t be ruled out. Therefore, the new analysis simply makes low values of climate sensitivity less likely but not altogether out of the question.
“This new study is just one bit of information, but I believe it pushes the likely climate sensitivity closer to where it’s always been, up around 3°C,” Schmidt adds. “It’s always difficult to predict the future, we’re always limited by what we don’t know.”