We all know those people.
Is that a Budweiser? Cool. I only drink craft beer myself.
You’re seeing a doctor for your sore knee? You should really go see my guy. In New York. The best.
Oh, you guys are worried about the hurricane because of TV? Well, I just checked the latest run of the Euro. It’ll miss us.
Alright, that last one might be specific to those of us who spend a lot of time on social media during hurricane season.
We’re only a few years removed from an era when only meteorologists had access to the mathematical weather predictions produced around the globe. Then it would filter down to ordinary folks after the pros decided what we should know.
Today anyone can access a kaleidoscopic array of maps and charts and data on sites like tropicaltidbits.com, or the overwhelming Mike’s Weather Page (spaghettimodels.com), which made Oldsmar resident and amateur weather enthusiast Mike Boyland a mild internet weather celebrity.
On Reddit recently, a user discussing the accuracy of American versus European forecasting models during Hurricane Dorian explained that they ride hard for UKMET, the numerical weather prediction model operated by the United Kingdom METeorological Agency, but that the “Euro and GFS are trash.” They’d dismissed the wildly powerful, expensive and supercomputer-powered high technology of the European Centre for Medium-Range Weather Forecasts and America’s National Oceanic and Atmospheric Administration in the same tone one would clown an NBA player’s brick jump shot or a chicken sandwich that didn’t stack up to Popeyes. Another user noted the “strong showing” of the Canadian model in the sort of weather model horse race, but everyone seemed to ignore that person.
Even if you haven’t encountered the so-called “model wars” online, you may have heard the meteorologist on your local news refer to “the Euro” or “European model” and wondered what they were talking about. We talked to some experts to break it down, ask if one model is really better than the other and find out what they think of weather enthusiasts picking a favorite.
First of all, what is a model?
A model is a set of mathematical equations solved to predict possible weather outcomes. In 2019 that means extremely complex equations crunched by a very powerful computer. A lot of physics is involved. There are dozens of operational models from around the world that are used by forecasters, though two that we hear about most often are the European model operated by the European Centre for Medium-Range Weather Forecasts and the American model, known as the Global Forecasting System, or GFS, operated by the U.S. government. Today’s models are a gigantic improvement over even just 20 years ago. These models are used to forecast all kinds of global weather, not just hurricanes. A model does not equal a forecast.
What’s the difference between the American and European models?
They’re different software run on different computers, the specific details of which would make most of our heads hurt, but there are some basic differences.
“The Euro is run only twice a day because it takes more computing power. The GFS is run four times a day. So we have it more often,” said Andy Johnson, former WTVT-Ch. 13 on-air meteorologist and current certified meteorological consultant. “In America, we have free access to all the data that National Oceanic and Atmospheric Administration offers. The European community charges up to $250,000 if you want everything they offer. So they have more money to spend on research and development.”
The difference also has to do with “initialization,” Johnson said, which is determining the current weather conditions used as a starting point to feed into the models. “You think we know exactly what the weather is, but we really don’t, because large portions of the earth are covered by water. We have ships and buoys out there, but they’re spaced far apart. ... So we have to use satellite or radar, which gives us good info, but isn’t as accurate as having a weather station there. We have to make assumptions.”
The same models are run repeatedly with different variables to account for errors. The outcome can be called an ensemble. An ensemble made of varying models’ ensembles can be used to smooth out the rough edges in a forecast. Sometimes on TV, an ensemble of varying runs of the same model or a bunch of different models’ forecasts, illustrated as a bunch of noodley lines, is called a “spaghetti model.”
So who wins the model war, Europe or the USA? Is the Euro actually better?
Marshall Shepherd, director of the University of Georgia’s Atmospheric Sciences Program and host of the Weather Channel’s Weather Geeks wrote in Forbes recently that “I don’t get it. If you follow the meteorological community or weather enthusiasts on Twitter, there is a level of hyperventilation (and vitriol) when it comes to the Euro versus American debate.”
“The European model is statistically superior but it is not a huge difference,” Shepherd told the Tampa Bay Times. “Both models have their strengths and weaknesses. Remember, the Euro was also one of the models that had Dorian initially going into Florida. It falters at times, too. Overall, my sense is that people think we are dealing with a sports car versus a horse and carriage. Actually, both are solid sports cars as models go.”
“There are times every model busts, and truthfully, the Euro didn’t do very well with Dorian, at least not until the last three or four days,” said ABC Action News chief meteorologist Denis Phillips, who often mentions “the Euro” in his Facebook updates.
Phillips thinks armchair forecasters go on about the Euro always being best, “because historically it is, and people like a winner. I’ve heard the Euro called the King. Some are calling the UKMET the Queen. And for some reason, the American model GFS is closer to the court jester. It has a long way to go to compete with the Euro in overall accuracy. There are a ton of reasons this is true, and probably the biggest reason is money.”
The idea that the Global Forecasting System is inferior probably entered the public consciousness after Hurricane Sandy. Early on, the American model predicted Sandy would float out harmlessly to sea, while the Euro accurately predicted its crash into the Northeast. The Global Forecasting System did accurately predict Sandy’s landfall once it got closer, and with time to spare, but the media took notice of the early error.
However, the idea that the European model is always better is a myth.
Can the American model get better?
One thing casual weather enthusiasts might not realize is the biggest upgrade to the Global Forecasting System’s “engine” in four decades became operational in June. It’s called GFS-FV3, which stands for “Finite-Volume Cubed-Sphere dynamical core,” and it’s a big deal for National Oceanic and Atmospheric Administration, developed over years by more than 100 scientists, modelers, programmers and technicians from around the country and at the Geophysical Fluid Dynamics Laboratory in Princeton. It’s too early to say how this huge upgrade of the Global Forecasting System is performing this hurricane season, but the National Oceanic and Atmospheric Administration ran it alongside the old system for several years and found it to be far more accurate. Below is an image of the old system versus the upgraded forecasts for several previous storms.
So which model should I go by?
None of them! No one should be basing their hurricane preparedness off data from a single model. Models are tools used by the experts to create a forecast and the official forecast from the U.S.’s National Hurricane Center is the “gold standard,” said Vijay Tallapragada, chief of the modeling and data assimilation branch at the National Oceanic and Atmospheric Administration’s Environmental Modeling Center.
The team that creates those Hurricane Center forecasts looks at not only the Global Forecasting System, but also the Euro, the Canadian model, the U.S. Navy’s model, the Japanese model and others, and they look at them with all sorts of variables that create ensembles. The European Centre for Medium-Range Weather Forecasts and the National Oceanic and Atmospheric Administration are also highly cooperative.
“There are a select few who understand the strengths and weaknesses,” Tallapragada said. “Based on their expertise they’ll make a consensus approach for a forecast.”
There is an important human factor. And that’s what your local on-air meteorologist goes by.
“The NHC has the top experts in the world putting this together every day. Pretty much every station’s policy is to go by the official NHC forecast,” Johnson said. “They might customize it for their viewers in their area, but they’re not changing it.”
Is this whole model argument silly?
Yes and no. Sharanya J. Majumdar, a professor and associate dean at the University of Miami’s department of atmospheric science, said that while the model wars “are largely a drama created by people outside the respective agencies, they may have served a positive purpose in alerting the National Oceanic and Atmospheric Administration and the U.S. Congress to the continuous need to accelerate the national forecasting capability via improved funding, infrastructure and organization.” The National Oceanic and Atmospheric Administration says its upgrade will be the basis for many future improvements.