Top scientists have ruled out best-case global warming scenarios

A major new study of the relationship between carbon dioxide and global warming lowers the odds on worst-case climate change scenarios while also ruling out the most optimistic estimates nations have been counting on as they attempt to implement the Paris agreement.

A group of 25 leading scientists now conclude that catastrophic warming is almost inevitable if emissions continue at their current rate, even if there’s less reason to anticipate a totally uninhabitable Earth in coming centuries.

The research, published Wednesday in the journal Reviews of Geophysics, narrows the answer to a question that’s as old as climate science itself: How much would the planet warm if humanity doubled the amount of carbon dioxide in the atmosphere? That number, known as “equilibrium climate sensitivity,” is typically expressed as a range. The scientists behind this new study have narrowed the climate-sensitivity window to between 2.6 and 3.9 degrees Celsius.

That’s smaller than the current range accepted by the United Nations-backed Intergovernmental Panel on Climate Change, which has for almost a decade used a spread between 1.5 and 4.5 degrees Celsius — a reading of climate sensitivity that has changed little since the first major U.S. climate science assessment in 1979.

Improving these estimates is “sort of the holy grail of climate science,” said Zeke Hausfather, director of climate and energy at the Breakthrough Institute and one of the study’s authors.

Climate sensitivity is one of the most iconic numbers in climate science, but it’s not necessarily intuitive. The range isn’t a projection; it’s more like a speed limit that influences projections. “It informs all the other things — like 2100 warming projections, for example-that depend on the sensitivity of our models, and our scenarios,” Hausfather said.

What gave the authors confidence is that three independent lines of evidence-the modern temperature record, geological evidence, and the latest Earth systems models-all agreed on the same answer.