As Hurricane Joaquin gained wind over the Atlantic today, meteorologists scrambled to figure out where it’s headed next, leaving the rest of us scratching our heads. Will this storm currently punishing the Bahamas turn west and make landfall along the East Coast, or will it turn east and leave us alone? Both of these forecasts aired today. The American Global Forecasting System (GFS) predicted the storm would make landfall on the U.S. while the European Center for Medium-Range Weather Forecast (ECMWF) expected Joaquin to skip the East Coast entirely. Meanwhile, the hurricane grew from Category 3 to Category 4 on the Saffir-Simpson scale, with winds clocked at around 125 mph.
The primary cause of the discrepancy in the predictions of Joaquin’s path was a trough of low pressure developing northeast of the storm. The American model predicted that the trough would draw the storm into the eastern coastline of the U.S. The European model, on the other hand, predicted that hurricane will interact with the low-pressure system at a later point in time, enabling Joaquin to escape and curve out to the sea.
American and European models have been at odds before. In 2012, the American GFS predicted Hurricane Sandy would fizzle over the ocean. The European model anticipated a more disturbing scenario: the storm would turn west and strike the East Coast. Sadly, the Europeans were right. Sandy struck the Eastern Seaboard, causing at least 233 deaths and $75 billion in damage.
Overall, the consensus among meteorologists and other scientists is that the European model is better overall in forecasting weather. “When you’re talking specifically about hurricane tracks, if you look at the last three years, the American and European models perform virtually the same,” says Jeff Masters, co-founder and director of meteorology at The Weather Underground. “But when you look more generally at weather forecasts around the whole world, such as cold fronts and rain showers, the European model is clearly superior.” There are also a few high profile cases, such as Sandy, where the European group made more accurate predictions than the American GFS. The main differences between the two approaches lie in data simulation, computing power and the underlying physics used by each.
To run these weather track models, scientists start by gathering information about the atmosphere from various sources, including ships, balloons and satellites. Then, using simulations, the data points are packed onto a grid, essentially creating a three-dimensional description of the atmosphere. “The [Europeans] are better at this because their simulation system is better, and they use more data than we do,” says Cliff Mass, a professor of atmospheric sciences at the University of Washington.
The European system also draws on more computing power, which enables the model to run on a finer grid, allowing higher resolution and better forecasts. Another important component is the underlying physics used to determine the differing physical properties of clouds. Here again, the European centers do a better job.
In addition to running the models at the highest resolution possible to obtain the most accurate forecasts, scientists also conduct something called ensemble models, which allow them to determine the accuracy and consistency of their prediction. To do this, they run simulations at slightly lower resolutions under different hypothetical conditions to determine where the uncertainties are in the forecasts. The European system runs 50 of these cycles, whereas the American GFS only runs 20. On Wednesday (September 30), only 10 of the 50 European ensembles predicted the hurricane hitting the U.S., while 10 of the 20 American ones predicted the same.
The superiority of ECMWF is largely due to Europe’s willingness to invest in software and spend on computers. “The Europeans have simply thrown more people and more money at it,” Masters says. “So they get better results.” In comparison, the U.S. has remained largely reactive—efforts are now being made now to upgrade computing power and improve data simulation systems, but our accuracy still lags behind. Back in 2006, after the devastation following Hurricane Katrina the National Science Board called for an increase of $300 million in hurricane research funding per year. According to Masters, the budget hasn’t changed much since.
According to the latest reports, the forecasts of Joaquin’s path are beginning to converge toward the European predictions.