As Hurricane Joaquin blew up from a tropical storm into a Category 4 hurricane over a few days this month, the U.S. National Hurricane Center warned residents of the Atlantic Coast that the tempest showed a high likelihood of slamming into the Eastern Seaboard, anywhere from the Carolinas to Cape Cod. As the storm dithered and lingered near the Bahamas, however, the warning cone drifted steadily to the right, eventually showing the hurricane on a track away from land and into the North Atlantic — the path that Joaquin actually took, days later.
U.S. authorities weigh the information supplied by a variety of computer models when choosing how to warn citizens about oncoming hurricanes or tropical storms. In the case of Joaquin, the various models offered widely differing predictions. Interestingly, the so-called "European model," developed by European Center for Medium-Range Weather Forecasts (ECMWF), accurately predicted Joaquin's eventual path from the beginning — while several U.S. models called the storm's track wrong for several days before changing to agree with the European model as new data became available. The anomaly set of a discussion among weather scientists about the differences between the models.
"It’s not the first time that the European model has led the pack," noted the New York Times (see: "Hurricane Joaquin Forecast: Why U.S. Weather Model Has Fallen Behind," by Nate Cohn. "It’s almost a repeat of what happened with Hurricane Sandy, but in reverse. Three years ago, the European model anticipated, far in advance, Sandy’s unusual 'left hook' into New Jersey. This time, the other models called for a left turn, and the European model dissented."
After the Sandy embarrassment, the Times reports, Congress gave the National Weather Service funding to upgrade the computers driving the Weather Service's "Global Forecast System" (GFS) — which appeared to pay off when the U.S. model more accurately predicted the Northeast blizzard of 2015 that missed New York, making a mockery of the city's preparations (but buried Boston, crippling that city for weeks). But even after the upgrade, meteorology professor Cliff Mass told the Times that the GFS is still "quite inferior" to its European counterpart.
But "individual high-profile events like Sandy and Joaquin can obscure the big picture," cautioned Weather Underground's Bob Henson (see: "Why Did the ECMWF Forecast Joaquin So Well?" by Bob Henson). The U.S. and European models "were virtually tied as the best-performing models for hurricane tracking when averaged over the period 2011-2014," Henson wrote, adding: "As a guiding rule, the best forecasts emerge from a consensus that include a wide variety of models, rather than by going solely with the ECMWF or any other single model. This is why the NHC track forecasts lean heavily toward multi-model consensus, as was the case last week."