Climate change is likely to be worse than many computer models have projected, according to a new analysis.

The work, published yesterday in Science, finds evidence that Earth's climate is more sensitive to the amount of carbon dioxide in the atmosphere than some earlier studies had suggested.

If the new results are correct, that means warming will come on faster, and be more intense, than many current predictions. Moreover, the impacts of that warming, including sea level rise, drought, floods and other extreme weather, could hit earlier and harder than many models project, said study co-author John Fasullo, a climate scientist at the National Center for Atmospheric Research.

"Temperatures are likely to go up to the high side of current projections, as is [atmospheric] water vapor," he said. "To the extent those environmental impacts influence events like [Superstorm] Sandy, expect the impacts to be on the high side."

For scientists like Fasullo and co-author Kevin Trenberth, head of NCAR's climate analysis section, determining the climate's precise sensitivity to the CO2 accumulating in the atmosphere has been an unusually tough task.

Researchers normally define the problem, known as equilibrium climate sensitivity, as the warming that would result if the amount of CO2 in the atmosphere rises to twice the preindustrial level.

Probing the behavior of clouds
A groundbreaking 1979 climate report from the National Academy of Sciences put the likely warming from that doubling of CO2 at 1.8 to 8.1 degrees Fahrenheit. But over the next 28 years, scientists were able to narrow that range only slightly. The estimate listed in 2007 report of the Intergovernmental Panel on Climate Change is 3.6 to 8.1 degrees F, with a best guess of 5.4 degrees.

That uncertainty is represented in the latest crop of global climate models, which assume a climate sensitivity of anywhere from about 3 to 8 degrees F.

What the new study does is attempt to change the angle of attack on the sensitivity problem. It tries to turn a major factor in the uncertainty in climate sensitivity estimates -- the behavior of clouds -- into a strength.

Clouds can act to heat or cool the atmosphere, depending on their location and structure. But scientists have had a hard time representing cloud behavior in models, in part because there is little high-quality data on cloud behavior.

To get around the problem, Fasullo and Trenberth decided to examine how well 16 global climate models reproduce recent satellite observations of relative humidity in the tropics and subtropics, a quantity that is directly related to cloud formation.

Relative humidity is the measure of the actual amount of water vapor in a given location's atmosphere, compared with the maximum it could hold. When relative humidity is high enough, it triggers cloud formation.

"The trick here is we're not using any trends, [not] looking at trends in dry zones," Fasullo said. "Instead, we're just trying to ask the question, do the models resolve dry zones? We're trying to get at the processes that drive sensitivity, instead of observing the actual changes themselves."

When the scientists compared the output of climate models with a decade of satellite measurements of relative humidity, they found that the models that best reproduced observed conditions were built on the premise that climate sensitivity is relatively high -- 7 degrees F or more.

The models that were least accurate also had the lowest climate sensitivity baked in, the scientists said.

The results are promising, Fasullo said, but he also sounded a note of caution.

A way to judge the models
"This problem has been around for 30 years, and very little progress has actually been made on it in 30 years," he said. "In no way would I be under the delusion that this solves the problem."

But experts who were not involved in the research said it is a promising new approach to an issue that has sometimes seemed intractable.

Karen Shell, an atmospheric scientist at Oregon State University, called the new findings "one piece of the climate sensitivity puzzle."

"Knowing that a modeled relative humidity is incorrect does not directly translate into the necessary model improvements, and models might not correctly simulate the dependence of cloud properties on relative humidity," she wrote in an essay accompanying the research, which was also published in Science.

Even models that correctly capture cloud behavior may fail to fully account for other climate feedbacks from factors like changing snow and sea ice cover, atmospheric water vapor content, and temperature. Yet even with those caveats, Shell praised the approach laid out in the new research as a "simple diagnostic ... [and] an encouraging step that links observations to climate sensitivity."

Similarly, Reto Knutti, a climate scientist at the Swiss Federal Institute of Technology in Zurich, called the findings "impressive."

"For years we've been trying to find relationships between quantities we can measure and climate sensitivity in order to narrow the plausible range for climate sensitivity. We have largely failed in doing so," Knutti said.

"This new paper is one of the few cases where there is a clear relationship that offers hope to constrain the models, or to identify the better ones based on observations. And it's particularly nice because the authors offer a plausible physical mechanism to explain the results, not just a correlation."

Reprinted from Climatewire with permission from Environment & Energy Publishing, LLC., 202-628-6500