The most advanced and comprehensive analysis of climate sensitivity undertaken has revealed with more confidence than ever how sensitive the Earth’s climate is to carbon dioxide.
Climate warming depends on the changing composition of the atmosphere, and the “climate sensitivity” represents the dominant indicator of change, defined as the increase of global air temperature resulting from a doubling of atmospheric CO2.
For more than 40 years, the estimated likely range of the global equilibrium temperature response to a doubling of atmospheric carbon dioxide, compared to preindustrial levels, has stubbornly remained between 2.7-8.1˚Fahrenheit.
This new research, revealed in a 165-page, peer-reviewed journal article, finds that the true climate sensitivity is unlikely to be in the lowest part of the 2.7-8.1˚F range. The analysis indicates that if atmospheric carbon dioxide levels double from their pre-industrial levels and are maintained, the world would be likely to experience eventual warming from 4.1-8.1˚F. There would be less than a 5 percent chance of staying below 3.6˚F and a 6-18 percent chance of exceeding 8.1˚F.
The research, written over four years, was commissioned by the World Climate Research Programme (WCRP) and includes Lawrence Livermore National Laboratory (LLNL) authors. It appears in the July 22 edition of the American Geophysical Union’s Reviews of Geophysics.
With the Earth’s temperature already at around 2.2˚F above pre-industrial levels, if greenhouse gas emission trajectories continue unabated, the world can expect to see a doubling of carbon dioxide in the next 60-80 years.
Baseline estimates show that there is a 66 percent chance that climate sensitivity is between 4.7 and 7.0˚F – which cuts the uncertainty of the 40-year-old estimate in half.
“It has been a constant goal of climate science since 1979 to reduce the uncertainty in estimates of climate sensitivity,” LLNL climate scientist and co-author Stephen Klein said. “After reviewing all of the evidence and performing new calculations, our paper does exactly that. Our estimates differ from past estimates mainly by raising the lower bound of climate sensitivity.”
The research was made possible by bringing together an international team of researchers from a wide range of climate disciplines. Using temperature records since the industrial revolution, paleoclimate records to estimate prehistoric temperatures, satellite observations and detailed models that examine the physics of interactions within the climate system, the team was able to combine more independent evidence than any previous study to get its results.
This evidence was combined in a statistically rigorous way, allowing the team to find where the results overlapped and converge on the best estimate of climate sensitivity. The team found that, with new multiple scientific analyses, the various lines of evidence corroborated one another, leading to more confidence in the result. The likely 4.1-8.1˚F range accounts cautiously for alternative statistical modeling assumptions (to allow for “unknown unknowns”), with a more straightforward calculation yielding an even narrower 4.7-7.0˚F likely range.
“Previously, we could not really estimate how much more confident we can be if different lines of evidence such as the historical period and palaeoclimate point at the same sensitivity range. In this paper, we have for the first time managed to combine carefully estimated lines of evidence,” noted co-author Gabi Hegerl from the University of Edinburgh.
“An important part of the process was to ensure that the lines of evidence were more or less independent,” said lead author Steven Sherwood, a University of New South Wales chief investigator with the ARC Centre of Excellence for Climate Extremes. “You can think of it as the mathematical version of trying to determine if a rumor you hear separately from two people could have sprung from the same source; or if one of two eyewitnesses to a crime has been influenced by hearing the story of the other one.”
The researchers then went another step further and identified the conditions that would be required for the climate sensitivity to lie outside the most likely range.
The researchers show that low climate sensitivities previously thought to be plausible, around 2.7-3.6˚F, could only occur if there were multiple unexpected and unconnected errors in the data analysis (for example, unexpected cloud behavior and patterns of long-term ocean warming), underlying their judgment that these low values are now extremely unlikely.
A different set of circumstances make it unlikely that global temperatures would rise more than 8.1˚F for a doubling of carbon dioxide from pre-industrial times, although these higher temperature responses are still more likely than very low sensitivities.
LLNL’s contribution included dealing with climate feedbacks in general and cloud feedbacks (which are the greatest uncertainty) in detail. Klein led a group of scientists, including LLNL’s climate scientist and co-author Mark Zelinka, in assessing all of the evidence from satellite observations, global climate models, large-eddy simulations and theory to produce a new estimate of climate sensitivity. This estimate was then combined with the estimates of climate sensitivity from the two other areas of the historical warming since the late 1800s and paleoclimate.
The three-year-long research process initiated by the WCRP with double-checking at every step, a detailed examination of the physical processes and an understanding of the conditions required for the estimate has finally consolidated a more robust result on the 40-year problem.
“These findings are a testament to the importance of cross-disciplinary research along with slow, careful science, and perfectly highlight how international cooperation can unpick our most vexing problems,” said co-author Eelco Rohling from the Australian National University.
“If international policymakers can find the same focus and spirit of cooperation as these researchers then it will give us hope that we can forestall the worst of global warming.”
The Department of Energy Office of Science, Office of Biological and Environmental Research, funded the LLNL portion of the work.