This article was published in Scientific American’s former blog network and reflects the views of the author, not necessarily those of Scientific American
It is difficult to focus on hurricane warnings right now, when Oklahoma is reeling from some of the worst tornadoes ever recorded. But the storms do raise questions about the abilities of U.S. scientists to predict severe weather, and the answers are not clear.
Just last week the National Oceanic and Atmospheric Administration released an internal study that judged how well its National Weather Service (NWS) did in predicting Hurricane Sandy. The report came to a curious conclusion: “The National Weather Service provided accurate forecasts for Sandy, giving people early awareness of the significant storm churning toward the mid-Atlantic and Northeast…. Forecasters performed well predicting the track of this extremely large and complex storm.”
But the weather community knows full well that the so-called European model for medium-range forecasts predicted that Sandy would “turn left” from the Atlantic Ocean toward the mid-Atlantic coast—a highly unusual path—while all the American models had the hurricane drifting northeast off to sea. It took several days before U.S. models began to show the same track that the European model had been indicating all along. The new head of the NWS, Louis Uccellini, recently acknowledged that the lag was a “miss” and told the Reuters news service that the European model is “the number one model, there’s no question about that.”
On supporting science journalism
If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.
NOAA’s report may be political cover, or it may be an attempt to change the subject, because it goes on to say that its review did reveal that the NWS can do better in getting people to heed its storm warnings. That challenge is a sticky one, though, involving complex social and psychological factors.
Perhaps the NWS itself knows it can improve hurricane forecasts. Last week it announced it would use $25 million recently appropriated by Congress to upgrade some of it supercomputers. U.S. hurricane models have a resolution of 25 kilometers, yet the European model has a resolution of 16 kilometers. Tighter resolution could allow forecasters to better assess how a hurricane is growing and moving, given that the inner core of such a storm is often in the range of 80 kilometers across. Predicting how high a coastal hurricane’s storm surge might be—often the cause of the greatest damage—is even harder than predicting its path, however, and might require different advances.
Technology alone is not the answer, however. “A new computer is really good, but you also need the people to use it,” says Chris Davis, a senior scientist at the National Center for Atmospheric Research (NCAR) in Boulder, Colo. “It takes a lot of work to use that better resolution well.” Researchers must properly recalibrate models to exploit the new resolution, and must learn how to interpret the results.
Although Davis thinks improved resolution is helpful, he maintains that better coordination among researchers could provide greater gains. Right now the NWS uses its exclusive models, university-based researchers like those at NCAR use their own, and private weather services use their own as well. If, instead, the three research groups integrated their work, forecasts could improve regardless of the technology used. “There has to be a better way to entrain all the people in the research community,” Davis says. “I would argue that the European Center for Medium-Range Weather Forecasting does that better.”
Greater cooperation also begs the question of whether U.S. models have to be the best in the world, which is what most NWS supporters say. Not necessarily, Davis notes. “We would want a model that on average is as good as the European model, but it doesn’t have to be better.” Actually, forecasts would be more likely to improve “if the U.S. model was equally as good but made different kinds of errors.”
All models have strengths and weaknesses, he explains, but if an improved U.S. model essentially duplicated the European model, it would not provide much new insight. If new U.S. models had different traits, “then we’d get much more insight when we combined the U.S. and European models,” which could provide earlier and more accurate forecasts. Davis suggests that NWS use funding not just to upgrade computers but to develop new forecasting techniques that might lead to models with unique attributes.
Infrared image of Hurricane Sandy courtesy of U.S. Navy