How much can we trust computer models about this weekend's possible winter storm?

Don't bank on any of the snowfall totals computer models show this far out.
Publish date:

Computer models. Computer models. Computer models. 

You'll hear those words a lot over the next five days as a looming winter storm threatens to bring rain, snow and ice to parts of Minnesota Saturday and Sunday. 

But what the heck do meteorologists mean when they say "computer models," and how much can anyone trust what the models are saying when a storm is still five days away?

The models usually become more accurate as the storm gets closer, so take everything we show you in this story with a grain of salt. 

For example, the National Weather Service isn't going to start guaranteeing Minnesota will get slammed by a winter storm because the computer models they're using are tracking a storm that is still out over the middle of the Pacific Ocean. 

"I wouldn't give any credence to any model this far out," storm specialist Tim Purington told Bring Me The News. "You might have one event where the Euro is right on or one event where the GFS is right on. They vary. It's better to take a blend of all of them together and average it out." 

Euro? GFS? In layman's terms, please. 

In short, the National Weather Service and forecasters of all kinds use a bunch of different long- medium- and short-range computer models to predict what will happen. Among the models you'll hear forecasters talk about: 

  • European model (Euro)
  • American model (GFS)
  • Canadian model (GEM or CMC)
  • North American Model (NAM)

There are variations of each of the above models, but to keep this simple we'll leave them as they are. 

"The biggest thing for accuracy is always when the storm comes on shore for the very simple fact that we can get a lot more readings," says Purington, who in severe weather season serves as a storm chaser for FOX 9.  

"We can get temperature, dewpoint, barometric pressure, wind direction, all that stuff."  

Purington studies computer models but never gets overly confident about any particular storm until computer models start to come into agreement. 

"The biggest thing is agreement. You might have the Euro and the GFS show different tracks that are 200 miles apart and the storm is still four days out," says Purington.  

"They might not agree until the system is three days away or two days away or they might not agree until the storm is within 24 hours of hitting." 

Sign up for our BREAKING WEATHER newsletters

Another critical factor is determining how much liquid precipitation the storm will drop, which is especially important when forecasting how much snow will fall.

Each storm has it's own unique snow-to-liquid ratio, but an approximate norm is somewhere around 10 inches of snow for every inch of liquid. 

"It's a more accurate reading to take what the model shows as liquid precip. For example, if a model is printing out that we're supposed to get a half inch of actual precip, you can take that half inch and put in what your actual storm ratio is. Then you're able to get a more accurate reading of what the snowfall is going to be," says Purington. 

For example, take a look at one of the most recent runs from the Euro model. 

You can see in the map below that the Euro is projecting about an inch of liquid precipitation across the southern third of Minnesota between now and Sunday night. 

The Euro model's projected liquid precipitation. 

The Euro model's projected liquid precipitation. 

Now take a look at how many inches of snow the Euro is suggesting will fall this weekend. 

Screen Shot 2019-03-05 at 1.45.29 PM

The Euro is currently predicting 10-12 inches of snow in the Twin Cities.

Since the map above shows almost exactly an inch of liquid precipitation in the metro, we can determine that F5Weather, who provided the map, is currently expecting a snow-to-liquid ratio of about 10:1. 

Models can have big differences in snow, storm track

You've seen the Euro model above. Now let's take a look at what the American (GFS) and Canadian (GEM or CMC) models are saying. For these, we'll use screen shots from Tropical Tidbits, which is another great website anyone can use. 

The GFS shows parts of southwest and central Minnesota getting hammered with more than a foot of snow, and this is with a 10:1 ratio. 

Screen Shot 2019-03-05 at 7.13.19 PM

Here's the Canadian (GEM or CMC) model, which also shows a ton of snow falling in the same areas, also at a 10:1 ratio. 

Screen Shot 2019-03-05 at 2.00.45 PM

Again, nobody should buy into these projected snowfall totals this far in front of a storm, but it's notable that all three models agree that the storm will impact Minnesota. 

"Minnesota is going to be affected in one way or another, whether it's rain or snow," says Purington. "We have pretty good agreement that there's going to be something somewhere, and it's been consistent." 

The reason the Canadian and GFS models show less snow in southeast Minnesota is because it's anticipating rain at first before transitioning to snow. You can see it clearly in this GIF of the Canadian model. 


The GFS simulation reveals a similar rain to snow transition. 


The Euro keeps the rain south of the Minnesota border, hence the higher snowfall totals for the Twin Cities and southeast Minnesota compared to the Canadian and GFS models. 

These models are updated numerous times each day, and in the next couple of days the North American Model (NAM) will start to provide a look at the storm, which will only add to the variance or agreement in model trends. 

Meanwhile, March is just getting started and forecasters in Minnesota might be pretty busy over the next couple of weeks. 

"It looks like we've gone back into that stormy pattern where it's storm after storm after storm," says Purington. "We're basically on the storm train track again." 

If you're wondering, here's what the National Weather Service Twin Cities is currently saying about storm potential this weekend. 

Screen Shot 2019-03-05 at 7.10.06 PM

Next Up