Snowstorm forecast bust spotlights need to better communicate uncertainty
Published Date: 2/14/2024
Source: axios.com

The storm that brought New York City the most snow in more than two years was also one of the biggest forecast busts in recent memory for the highly populated I-95 corridor.

Why it matters: The miss, with computer models shifting significantly on all aspects of the event within 24 hours of the first flakes flying, calls into question scientists' abilities to anticipate extreme weather events.


The big picture: When some of the country's biggest cities are involved, the economic consequences of a forecast miss can be significant.

  • For example, from Sunday into the middle of Monday, the forecast for Boston called for up to a foot of snow, with some areas potentially picking up more than that.
  • That unraveled Monday evening, as computer models showed the heavy snow sliding further and further southward. The city's total? Less than an inch.
  • Meanwhile in New York City, a forecast of an inch of slush turned into a bona fide snowstorm over the same timeframe, with 1 to 2 inches per hour falling on Tuesday morning.

The intrigue: The southerly swerve in computer model projections for the storm's path and intensity was an outlier in two ways.

  • First, most forecasts become more confident over time as model projections align. This did the opposite.
  • Also, in recent years there has been a tendency for computer models to project Northeastern snow storms moving farther north as they near the Mid-Atlantic and Northeast; this one retreated south.
  • The last-minute shift had practical consequences: Boston Public Schools, for example, ended up closing Tuesday for only a trace of snow.

Our thought bubble: "I'll bet most Boston-area parents aren't happy their kids missed school Tuesday, but really, it's just disappointing going from forecasts calling for a foot of snow to the dusting we saw today," Axios Boston's Steph Solis said.

Between the lines: Experts tell Axios that this storm illustrates the need for better ways of communicating uncertainty, as well as making the full range of dicey projections clearer to forecasters.

  • "The models were unusually sensitive to small changes in the initial conditions, and we wound up with wildly different forecasts run to run, model to model," Ryan Hanrahan, chief meteorologist at NBC Connecticut, told Axios.
  • "I think what people don't understand is not every forecast has the same level of certainty. Some forecasts are far more uncertain than other forecasts," he said.

Of note: Climate projections show future snow events in southern New England will occur in marginally cold environments similar to what occurred Tuesday, when the rain-snow line and elevation played a central role in the outcome.

  • This could make future snowfall harder to predict than it already is in this previously colder region.

What they're saying: "We've known for a long time that the best way to convey information is through a probabilistic framework," which shows the odds of particular outcomes, Hanrahan said.

  • "This event shows us that the tools that we use to come up with those probabilistic forecasts are inadequate," he said.
  • He'd like to see the National Weather Service invest more in tools that show the full range of possible outcomes.

Yes, but: Tuesday's forecast failure in many locations does not mean that it is time to doubt all weather and climate projections.

  • "We clearly still have room for improvement in numerical weather prediction modeling, but the poor overall performance for one event should not overshadow the clear progress being made in overall model forecast accuracy," said Steve Bowen, chief science officer at Gallagher Re.

What's next: Bowen anticipates upcoming improvements in forecast accuracy and communication as more AI technologies are incorporated into weather forecasting.

  • This is particularly the case, he said, with extreme events that may be tied to climate change.

Go deeper:

In photos: Heavy snow hits New York City, Southern New England