In June 2021, a jet stream charged with heat and chaotic energy from a nearby cyclone stalled over the Pacific Northwest. The mass of trapped air baked the already hot landscape below to a record 49.6°C. More than 1000 people died from heat exposure.
Scientists quickly began working to figure out how much of the blame for the heat wave could be laid to global warming. But the heat was so unusual, the weather so weird, that it broke their methods. “It challenged our techniques, our climate models, and our statistical analysis methods,” says Michael Wehner, a climate scientist at Lawrence Berkeley National Laboratory who participates in the World Weather Attribution (WWA) initiative. WWA ultimately issued a statement, finding the heat wave was “virtually impossible” without global warming. But, Wehner admits, that statement masked plenty of doubts. “We sort of kludged it.”
Next time, he and fellow modelers expect to do better.
Critics complain that extreme event attribution, as the effort is known, overemphasizes public communications and underestimates uncertainty. But new approaches promise to increase the field’s rigor and more precisely capture the relationship between climate change and extreme weather—even for events so extreme that there is no historical record for comparison.
In one existing method, researchers use climate models to simulate decades of climate history under current conditions and in preindustrial times, before warming set in. They tally up weather events that are similar to the extreme at hand in the two simulations and compare the frequencies to see whether warming increased the odds of the event occurring. In a second method, they shorten the timescale and run a model several hundred times under current and preindustrial conditions to see how often the extreme event would occur in each world. In a third method, researchers force models to re-create the actual atmospheric conditions recorded in the few years leading up to an event, then see how they unfold in counterfactual worlds with less warming. Although this “storyline” approach doesn’t show how much more likely global warming made the extreme event, it does offer clues to how much worse it was because of warming.
By highlighting how the slow process of climate change can affect dangerous weather, these techniques have “revolutionized our ability to communicate the findings of climate science to the public,” says Wim Thiery, a climate scientist at the Free University of Brussels. The results are increasingly being considered as evidence in lawsuits seeking damages for fossil fuel emissions and in the push for wealthy countries to pay for their role in worsening global warming. And the studies have led to new insights, Wehner says, for instance by showing that warming can increase hurricane flooding even more than expected and exposing how poorly models handle local causes of heat waves.
But these efforts rely on climate models with known biases that often fail to capture the detailed, local weather phenomena behind a heat wave or a flood. As a result, attribution studies that conclude, for example, that an event was 10.9 times more likely because of warming risk a false precision, says Markus Donat, a climate scientist at the Barcelona Supercomputing Center. “I don’t think such a quantification factor means anything.”
Because of such criticisms, WWA no longer relies on a single climate model, says Friederike Otto, a climatologist at Imperial College London; instead, it uses a suite of models to compare extreme events in preindustrial and greenhouse climates. The group has also standardized how it defines and categorizes an extreme event, a choice that’s critical when hunting for and tallying up similar events in the models.
Yet these methods still falter with weather so extreme that it appears unprecedented, such as the Pacific Northwest heat wave. Such weather doesn’t necessarily signal a new climate state; weather is so chaotic that even centuries of records can miss what’s possible, says Laura Suarez-Gutierrez, a climate scientist who studies heat extremes at the Max Planck Institute for Meteorology. “Our range of events we can look at in the real world are so limited.”
A method called “ensemble boosting” described last month at the annual meeting of the European Geosciences Union (EGU) could help modelers dissect such singular events. A team led by Erich Fischer, a climate modeler at ETH Zürich, has run a climate model many times over, looking for heat extremes similar to the 2021 heat wave. In a few instances, the model produced a heat wave resembling the real event, though not quite as extreme. The team took those outliers and rewound the model, perturbing the atmospheric conditions a few weeks before the heat wave to introduce more chaos and see whether they could produce events more like the real one. “We’re trying to push the model to the most extreme state,” Fischer says. In future research, he says, modelers could create such an ensemble of artificial scenarios for both current and preindustrial atmospheres. Then they could analyze these outliers and say whether global warming made the heat wave worse.
Other techniques forgo models entirely. Davide Faranda, a climate scientist at the University of Paris-Saclay, compares recent records of sea level air pressure to records from 1950–79, when warming was only getting started. He says these pressure analogs reflect large-scale air flows that can drive extreme events such as the Pacific Northwest heat wave. The method ensures that comparisons are like-to-like, unlike the model techniques, which might sometimes hinge on comparisons of regional rainfall, which can have diverse causes—a thunderstorm or a hurricane, for example. And it identifies when a weather event is truly new, resulting from chaotic pressure flows unseen in past records—which makes it statistically impossible to say whether climate change made the event more likely. In a preprint posted in February, Faranda and colleagues applied the method to maps of pressure leading up to a series of 2021 extremes, including flooding in Germany. They found the rainfall from similar extreme events has increased in modern times, a sign that warming made the flooding worse.
Finally, one group of researchers, led by Nicholas Leach, a graduate student at the University of Oxford, is betting that detailed weather forecasting models can give a more precise picture of global warming’s role in weather than coarser climate models. They simulated the 2021 Pacific Northwest heat wave with the world-leading model of the European Centre for Medium-Range Weather Forecasts. After lowering carbon dioxide levels and removing human-driven heat from the ocean, the model reproduced a similar heat wave, although the highs were 2°C or so cooler than in the actual event, the team reported last month at the EGU meeting. The model’s many runs for each scenario will allow them to assess whether climate change made the heat wave more likely. But WWA’s statement that the event was “virtually impossible” without warming was likely an overreach, Leach says. “We don’t find that in our model.”
Such advances are encouraging government agencies to move into attribution studies. The National Oceanic and Atmospheric Administration, for example, has funded a pilot study on rapid attribution using its operational forecast model and climate models. The effort will seek to tease out the influence not only of global warming, but also of large-scale weather patterns like El Niño. Thiery can see a day coming where weather forecasters might incorporate an attribution like Leach’s into their everyday work. “You can imagine a weather forecaster saying this heat wave is coming and it’s 2°C warmer because of climate change.”
Source: Science Mag