How much warmer is the Earth’s surface likely to become if atmospheric carbon dioxide levels double from pre-industrial levels, as is likely to happen later in this century at current rates?

Output from a range of computer models accepted by the international climate community for years has estimated warming of 2 to 4.5 degrees C., or 3.6 to 8.1 degrees F.

Recently, powerful new models utilizing the latest supercomputer capabilities are giving a higher range, starting at 5 degrees C, or 9 degrees C.

These estimates have been produced by “at least eight of the next generation models produced by leading centers in the United States, the United Kingdom, Canada, and France,” according to a recent article in Science magazine

Superficially, the model results would seem to suggest that warming is happening faster and more dramatically than previously estimated.

At the moment, however, the higher projected temperatures are more puzzling than alarming, according to climate researchers at Lawrence Livermore National Laboratory (LLNL) and elsewhere.

The new models with higher readings are only the first in dozens of models expected to be released in the coming year as scientists prepare for the sixth report of the Intergovernmental Panel on Climate Change (IPCC) in 2021. IPCC is the authoritative international compilation of the state of climate science.

To Karl Taylor, who heads LLNL’s internationally known program for comparing and analyzing climate models, the higher temperature findings are “early” and “tentative”.

"We don't know whether results from other models that will come in later this year might lower or raise” estimated temperatures, he wrote in an email.

Still, the results have been a topic of conversation at climate conferences, such as one held last month in Spain in an ongoing effort to establish a consistent and transparent approach to comparing the performance of climate models from around the world.

That was the Coupled Model Intercomparison Project, forerunner to the coming ICPP report, organized and led in part by LLNL climate experts like Taylor.

Two other LLNL climate scientists, Ben Santer and Mark Zelinka, agree that the new model results published so far may not be representative in the longer run.

“It’s too early to…draw a definitive conclusion,” Santer said.

On the other hand, if the higher temperature projections were reinforced by future model results, “We will have to try to explain how and why.”

Zelinka noted that all models have the challenge of dealing with clouds, whose effects and dynamics have been a particular focus of his scientific interest.

Some clouds tend to warm the Earth and some tend to cool it. It is virtually impossible to model everything realistically, he said, from microscopically small droplets to weather fronts stretching hundreds of miles, constantly moving and changing.

Every model deals with clouds in its own way, which can help to explain some differences in output.

At NASA and at the National Center for Atmospheric Research in Colorado, scientists also wondered whether the challenge of modeling clouds might have skewed their new models to higher outputs, according to the Science magazine article.

In the larger climate community, there is agreement that new model results are too few and preliminary to reach any firm conclusions about the actual climate, but there is also a sense that the stakes are high and it will be important to pay close attention as further results emerge and old ones are refined

John Fyfe, a climate scientist from the Canadian climate modeling center in Victoria, summed up: "It’s a bit too early to get wound up. But maybe we have to face the reality in the future that’s more pessimistic than it was in the past.”