Catastrophically warm predictions are more plausible than we thought

In an attempt to answer that question and evaluate the plausibility of a given model, EPFL scientists have developed a rating system and classified climate model outputs generated by the global climate community and included in the recent IPCC report. The EPFL climate scientists find that roughly a third of the models are not doing a good job at reproducing existing sea surface temperature data, a third of them are robust and are not particularly sensitive to carbon emissions, and the other third are also robust but predict a particularly hot future for the planet due to high sensitivity to carbon emissions. The results are published in Nature Communications.

"We show that the carbon sensitive models, the ones that predict much stronger heating than the most probable IPCC estimate, are plausible and should be taken seriously," says Athanasios (Thanos) Nenes, EPFL professor of the Laboratory of Atmospheric Processes and their Impacts, affiliate researcher at the Foundation for Research and Technology Hellas, and author of the study together with graduate student Lucile Ricard.

"In other words, the current measures to reduce carbon emissions, which are based on lower carbon sensitivity estimates, may not be enough to curb a catastrophically hot future," says Ricard.

Evaluating the plausibility of a climate model: big data analysis

Since the mid 1800's, the scientific community has been systematically observing the planet, measuring meteorological variables such as temperature, humidity, atmospheric pressure, wind, precipitation, ocean and ice status on Earth. Especially over the last few decades, with observational networks and the deployment of satellites, the amount of observational data is vast, and using this information to predict every aspect of the climate's future is a daunting task.

To evaluate a given climate model, the EPFL researchers developed a tool called "netCS" to cluster climate model outputs using machine learning, synthesizing their behavior by region and comparing the outcome with existing data. With the help of netCS, scientists can determine which climate simulations best reproduces observations in the most meaningful way -- and rank them accordingly.

"Our approach is an effective way to quickly evaluate a given climate model thanks to netCS's ability to sift through terabytes of data in one afternoon," notes Ricard. "Our model rating is a novel type of model evaluation, and highly complements those obtained from historical records, paleoclimate records and process understanding outlined in the 2021 IPCC AR6 assessment report."

Nenes, who is invited to participate in the IPCC AR7 scoping meeting to be held in Malaysia, is of Greek origin. He recalls giving a piano concert in Athens in the middle of the summer almost thirty years ago: "The temperatures back then peaked between 33 and 36 degrees Celsius and were considered to be amongst the highest temperatures of the year. I'll never forget how difficult it was to play the piano in that heat. Greece is now often plagued with summer temperatures above 40 degrees. Forest fires are commonplace, even invading cities, recently burning neighborhoods that I used to live in. And it will only get worse. The planet is literally burning. Temperatures worldwide are consecutively, year after year, breaking records with all of its consequences."

"Sometimes I feel that climate scientists are a bit like Cassandra of Greek mythology," concludes Nenes. "She was granted the power of prophecy, but was cursed so that no one would listen to her. But this inertia or lack of action should motivate not discourage us. We have to collectively wake up and really address climate change, because it may be accelerating much more than what we thought."

Other authors included in the study are Fabrizio Falasca from the Courant Institute of Mathematical Sciences at New York University, and Jacob Runge from the Technical University of Berlin. This research was supported by the European Union's Horizon 2020 research and innovation program under Marie Sk?odowska-Curie grant agreement No. 860100 (iMIRACLI), by the FORCeS project under the European Union's Horizon 2020 research program with grant agreement No. 821205, and by the CleanCloud project under the Horizon Europe research program with grant agreement No. 101137639.