Nation/World

Looking, quickly, for fingerprints of climate change in extreme weather events

When days of heavy rain in late May caused deadly river flooding in France and Germany, Geert Jan van Oldenborgh got to work.

Van Oldenborgh is not an emergency responder or a disaster manager, but a climate researcher with the Royal Netherlands Meteorological Institute. With several colleagues around the world, he took on the task of answering a question about the floods, one that arises these days whenever extreme weather occurs: Is climate change to blame?

For years, most meteorologists and climate scientists would answer that question with a disclaimer, one that was repeated so often it became like a mantra: It is not possible to attribute individual weather events like storms, heat waves or droughts to climate change.

But increasingly over the past decade, researchers have been trying to do just that, aided by better computer models, more weather data and, above all, improved understanding of the science of a changing climate.

Attribution studies, as they are called, can take many months, in large part because of the time needed to run computer models. Now scientists like van Oldenborgh, who is part of a group called World Weather Attribution, are trying to do such studies much more quickly, as close to the event as possible.

"Scientific teams are taking on the challenge of doing this kind of analysis rather rapidly," said Peter A. Stott, who leads the climate attribution group at the Met Office, Britain's weather agency.

The goal is to get sound scientific analysis to the public to help counter misinformation, deliberate or otherwise, about an event.

ADVERTISEMENT

"It's worthwhile to give the best scientific evidence at the time, rather than not saying anything and letting others say things that are not related to what really happened," said Friederike Otto, a researcher at the University of Oxford who works with van Oldenborgh.

[The world's clouds are shifting toward the poles, and they're getting taller]

In the case of the European floods, World Weather Attribution, which is coordinated by Climate Central, a climate-change research organization based in Princeton, New Jersey, released a report less than two weeks after the Seine and other rivers overtopped their banks. The group concluded that climate change had made the French flooding more likely, but could not draw a conclusion about the flooding in Germany.

"In the French case, we had five almost independent measures, and they all agreed," van Oldenborgh said. "With Germany, we only had two, and they disagreed."

Climate scientists have said for decades that global warming should lead to an increase in extreme weather like heat waves and droughts. Because more water evaporates from the oceans and warmer air holds more moisture, climate change should also lead to more intense and frequent storms.

Studies have shown that these effects are occurring on a broad scale. The National Climate Assessment, for instance, notes that heavy downpours have increased across most of the United States in the last 25 years.

But analyzing individual events is problematic, largely for two reasons: Weather is naturally variable, even without climate change, and global warming may be only one of several factors influencing a particular event. Since reliable data is required, studies are also less likely to be undertaken for events in countries that lack much data-collecting infrastructure, or where governments do not share data widely.

Stott, of the Met Office, was the lead researcher for an early attribution study, a 2004 paper in Nature that linked a deadly 2003 heat wave in Europe to human-caused increases in greenhouse gas concentrations in the atmosphere. Since then the pace of such studies has increased; last year, a publication of the American Meteorological Society, edited by Stott and others, had 32 studies of 28 events, covering all seven continents.

Not all of them found a link to climate change. A study of water shortages in southeastern Brazil during a dry period in late 2014 and early 2015, for instance, found that the shortages were most likely driven by increasing population and water use than by climate change.

But other analyses — of a 2013 heat wave in Argentina, extreme rainfall in the Cévennes Mountains in France in 2014 and an extremely hot spring in South Korea that year — found a connection.

David W. Titley, a professor of meteorology at Pennsylvania State University who was chairman of a National Academies committee that looked at developments in the field of climate-change attribution, said that at this point studies of heat waves and other extreme-temperature events appeared to produce the most reliable assessments.

Studies of extreme rainfall are considered less reliable in finding links to climate change, and studies of events like wildfires and severe thunderstorms even less reliable.

Still, Titley said, such studies are worth doing, as long as certain conditions are met.

"There are very legitimate reasons why people want to do this rapidly," he said. "But they need to state very clearly what the assumptions are, what the methods are, what the confidence level is," he said.

"This is still not like predicting what time is sunrise for New York City tomorrow," he added.

Attribution studies usually involve running climate models many times over. Because no model is a perfect representation of reality, varying them slightly for each run and averaging the results give scientists more confidence in their accuracy.

One set of runs simulates the climate as it actually is, incorporating the effects of greenhouse gas emissions, while the other set simulates the climate as if that human influence had never happened.

ADVERTISEMENT

Researchers then compare historical data, as well as data from the actual event, such as rainfall or temperatures, to the different model results to assess any climate-change impact. The analysis is usually given in terms of probabilities, or increased or reduced likelihood, that climate change had an effect.

Rather than running models after an event, researchers like van Oldenborgh and Otto shorten the process by using models that have already been run.

"The only way we can do this rapid attribution is by precooking everything that we can," van Oldenborgh said.

The process starts with emails among members of the group, usually followed by a Skype session to discuss whether a specific event is worthy of study. (One group member is in Australia, so arranging a conference call can be tricky.)

At least one member of the group also has to have time to do the work. "You have to put everything aside for a week" said van Oldenborgh, who did most of the work on the flood analysis because the other researchers were at a conference.

A crucial part of the task is to define the event — what happened and what meteorological variable is best to study it. In the case of the European study, the models the researchers used simulate rainfall, not flooding. So they used rainfall as the variable.

"We can't look at flooding, only extreme precipitation," Otto said. "Where did it rain, how much did it rain?"

But they also consulted with a hydrologist who understood how rainfall affects the river systems involved, van Oldenborgh said, "just to make sure we didn't do something stupid."

ADVERTISEMENT

Among the models they use is one that Otto's group at Oxford, the Environmental Change Institute, runs regularly, using the personal computers of a large number of volunteers, a project called climateprediction.net. While the model is a global one, regional results can be extracted and used for the rapid analysis.

Van Oldenborgh said that the European flooding analysis was a good illustration of the need for transparency, because while the researchers were confident in their findings for France, they acknowledged that they could not draw conclusions regarding Germany.

"We have to make sure we're open-minded enough to conclude that, although we've spent a lot of time on this, we can't conclude anything," he said.

When the initial analysis is completed and released to the public — Climate Central helps with the communication process — the work is still not over. Although the hope is that such studies will eventually become so routine that there is no need to publish the findings in a scientific journal, for now, at least, a research paper has to be written and submitted for peer review. So for the flooding analysis, that was another week's work for van Oldenborgh.

"The worst problem is that you're pretty tired and want to take a break and put your feet up," he said. "We haven't got a solution for that yet."

 
ADVERTISEMENT