In 2003, a deadly heat wave struck Europe that would usher in a new era of climate science. In July and August alone, temperatures upward of 115 °F claimed nearly 70,000 lives. However, while average global temperatures have increased at a steady clip since the mid-20th century, strong heat waves had been documented from time to time before then. For climate scientists, that meant that attributing the heat wave to global warming would be next to impossible.
So when a team of British researchers used environmental data and model simulations to establish a statistical link between climate change and the heat wave, they got attention.
Though they couldn’t prove that global warming had “caused” the scorcher, the scientists did assert that warming from human emissions had doubled the risk of extreme weather events. Published in Nature, their first-of-its-kind study launched the new field of “attribution science,” which uses observations and models to tease apart the factors that lead to extreme climatic events.
In the years since, better models and more data have helped climate scientists get much better at predicting extreme weather. But how confidently can scientists attribute these extreme weather events to anthropogenic climate change? Will they ever be able to definitively say that our emissions caused a specific drought, tornado or heat wave?
We put these questions to three experts who use environmental data and modeling techniques to study extreme weather and global climate change.
To be clear, scientists can and do assert that anthropogenic climate change has wide-ranging global effects, from ice caps melting and sea level rise to increased precipitation. “Many lines of evidence demonstrate that human activities, especially emissions of greenhouse gases, are primarily responsible for recent observed climate change,” reads a federal climate change report published in draft form in January, and publicized by the New York Times last week.
Thanks to advances in supercomputing and pooling hundreds of climate models developed by researchers across the world, they are also more statistically confident than ever in saying that intense storms, droughts and record-breaking heat waves are occurring with increased frequency because of humans. “Ten years ago we wouldn’t have been able to do so,” says Ken Kunkel, a climate scientist at North Carolina State University who also works with the National Oceanic and Atmospheric Administration.
But teasing apart individual weather events is harder. The planet’s history is dotted with unexpected, prolonged heat waves and sudden damaging storms far before humans began pumping out greenhouse gases. “The big challenge is that these kind of extreme events have always happened,” says Kunkel, whose work focuses on heavy storms that cause considerable damage in the U.S. But, he says, “Can you say, ‘This event was caused by global warming? No.'”
The difficulty of isolating a culprit behind extreme weather is a similar to the diagnostic challenge that medical doctors face, says Noah Diffenbaugh, an earth system scientist at Stanford University. Just because one patient recovers from cancer after taking a particular drug, for instance, isn’t enough evidence for doctors to widely prescribe that substance as a cancer cure. Instead, the drug needs to go through hundreds of replicated experiments on multiple populations before doctors are confident enough that it works.
In both medicine and climate science, “the default position is the null hypothesis: that every event occurred by chance," Diffenbaugh says. "We have a very high burden of proof to reject that null hypothesis."
But unlike in medicine, when it comes to Earth, we don’t have the ability to do clinical trials on hundreds or thousands of similar planets to overturn that null hypothesis. We only have one planet, and one timeline. So scientists have had to get creative in finding ways to observe other possible realities.
To conduct planetary experiments—the equivalent of clinical trials in medicine—they use computer models that mimic the variables on Earth, and turn the knobs. “With model simulations, you essentially have large populations you can look at,” Diffenbaugh says. “That’s where the models come in, they allow us to have more Earths to look at.”
A climate model works by dividing the Earth’s atmosphere and surface into a grid, like the the lines of latitude and longitude on a globe. “The model has to break up space into chunks,” says Adam Schlosser, a senior research scientist at the Center for Global Change Science. The smaller the chunks, the more precise the model will be.
These climate models work well when it comes to capturing large-scale patterns. They "are quite good at simulating the global-scale temperature,” Diffenbaugh says. But extreme weather events are more challenging, because they’re rare, localized and brought about by a swirling mixture of environmental factors. Currently, most climate models operate at a fairly coarse scale due to limitations of super computing power, Schlosser says.
This is part of the reason that modeling extreme events like heat waves is easier than modeling, say, individual storms or tornadoes. Heat waves happen over huge geographic regions that coarse models can easily capture. “When you see news about tornado hunters, they’re looking at weather events that are the size of a small town. A climate model can’t get down to that resolution,” Schlosser says.
Not yet, at least. Computers are getting faster, and climate scientists are figuring out ways to crunch more data to strengthen their predictive abilities. “We analyze every variable that we could possibly get our hands on,” Schlosser says. Still, challenges remain when it comes to building enough evidence to make claims of increased probability. As Diffenbaugh puts it: “Science is highly conservative.”
The increasing and sometimes alarming frequency of floods, droughts, heat waves and heavy storms may have a silver lining: They provide troves of data for researchers to plug into their models. In other words, they’re making the connections between the occurrence of localized extreme events and anthropogenic climate change more clear.
Things you hear the meteorologist mention on the nighly news—wind speed, pressure fronts, temperature, humidity, instability in the atmosphere—are all ingredients in the cookbook of extreme weather.
“We can use those telltale signs as a recipe—anytime you see these ingredients come together you’re going to be in an environment for a storm,” Schlosser says. “Those are the sorts of things we’ve been using and they’ve been successful in making a nice leap in our confidence in model concensus in where all this is going in the future.”
Diffenbaugh agrees. When it comes to predicting specific weather events, “we’ve moved really rapidly from saying ‘we don’t do that’ as our public stance, to some bold pioneers trying to do it, to now a number of groups working hard.”
As the recent climate report shows, researchers now have greater confidence when they make assertions about the role of anthropogenic climate change in increasing extreme weather events. “The consensus is getting stronger and stronger,” Schlosser says. “It doesn’t really matter which direction it goes, we just want to be confident about it.”
Yet the challenges of teasing out the causes of something as complex as weather also illustrates the ways in which climate change is unlike any other field of science. “It would be nice to have 100 Earths, so you could turn the knobs and increase this or decrease that and see what happens,” Kunkel says. “We don’t have that. We’re living our experiment.”
He pauses, and adds: “unfortunately.”