In the immediate aftermath of a weather disaster, people like to wonder if climate change contributed to the severity of the disaster. The problem is that any sensible, serious answer to that question takes time – so much time that public attention has shifted before we get an answer.
To a group of climate scientists, that unfortunate problem sounded more like a challenge. With a good plan and setup, the team thought it could quickly run the necessary climate model simulations and spew out some basic results. By comparing a virtual world where humans have not increased greenhouse gas concentrations to those we live in, the models can be used to see if there is any change in weather patterns associated with the latest disaster.
At the end of May, the team had the opportunity to test drive the system. Weather associated with a persistent low-pressure system poured rain on France and Germany. Three days of continuous heavy rain, following a wet spring, caused flooding on the Seine and Loing rivers upstream from Paris. In Paris, the water level in the Seine rose more than 6 meters (more than 20 feet), prompting the evacuation of art from the Louvre’s basement levels. Thousands of people were also evacuated in other areas.
In Germany, the damage was due to flash flooding due to heavy rainfall caused by severe thunderstorms on May 29 and 30. In the town of Gundelsheim, for example, more than 12 centimeters (nearly 5 inches) of rain fell in less than a day. Radar estimates picked up a peak speed of about 9 centimeters (3.5 in) per hour near Braunsbach, where cars washed away.
The researchers used a cache of simulations previously run with various climate models, including a run with free CPU time from the volunteers of the weather@home network. In each case, multiple simulations of worlds with and without climate change were combined to calculate statistical probabilities of extreme rainfall events – three-day rainfall in northeastern France and one-day rainfall in southwestern Germany.
Unlike the months or years it took in the past, the team had results less than a week ready to share with the media. A week later they had the paper, which the authors shared with Ars. It has since been submitted for open peer review in the journal Hydrology and Earth System Scienceswhere it appeared on Wednesday.
In France, the timing of the flood was quite unusual. The highest historic flood levels in Paris have all come in winter rather than late spring. As for three-day rain in the spring, it was only expected to occur once every few centuries – although rainfall in other parts of the area was only about 50 years. These rare events occurred about twice as often in the greenhouse gas-induced global warming simulations. In other words, a 100-year rainfall is now about 6 percent heavier than before.
For southwest Germany, the investigation did not reveal much. Available data shows that extreme one-day rainfall at this time of year has actually become less common over the past 50 years. Because the thunderstorms responsible depend on small-scale processes, most models have simulated it poorly. The model that performed best in the thunderstorms simulated a increase in this kind of rainfall. The researchers would have to take a closer look to figure out what to make of that.
So was climate change partly responsible for the severity of the flooding in France and Germany a few weeks ago? According to this incredibly fast analysis, the answer for France is “probably” and the answer for Germany is “we don’t know”. There may be further research into this — and the details of this study may even evolve somewhat during the peer review process — but a scientific answer was available before the ground had even dried. You can expect that more often in the future.
Will be assessed on Hydrology and Earth System Sciences2016. DOI: 10.5194/hess-2016-308 (About DOIs).