Communicating a Snow Forecast
The Pacific Northwest is a tough place to forecast the weather for in general, but nothing is tougher here than forecasting snow. This was particularly true this past January, with two events that wreaked havoc across some locales while leaving others unscathed. The first—an arctic front and associated Puget Sound Convergence Zone—occurred on 1/12–13 and brought blizzard conditions to Bellingham while dropping over 6 inches of snow to parts of Snohomish County. The second—a persistent band of snow that dropped 2 feet of snow over Port Angeles—occurred on late 1/14–15 and was not even forecast until 12 hours prior to the start of the event. In both events, south Seattle saw much less snow than north Seattle; Sea-Tac only received 0.7 inches of snow for the month, while the Seattle NWS forecast office near Magnuson Park saw 3.8 inches.
But as tough as forecasting snow is—communicating a snow forecast, particularly in an area with as many micro-climates and local terrain features as our own— is even harder. And unfortunately, social media allows eye-catching but incorrect forecasts (such as a blizzard from a cherry-picked model run) spread like wildfire, giving people the impression that “Snowmageddon 2020” is imminent and that they need to head to their nearest PCC to stock up on fresh fruit and vegetables!
Models hinted at the potential for a cold and snowy mid-January more than a week before our events. However, these models (particularly the American model) showed much colder temperatures than the ones that verified. The American model showed highs staying in the teens and lows dropping to the single digits, and while the long-range forecasts spit out by weather apps recognized this as an outlier, they were still calling for mid-January to be the coldest period since at least early February 2014.
As the event approached, the models (as expected) backed off on the cold, but there was a tremendous amount of uncertainty in snow forecasts due to (1) temperatures being on the fringe for snow, especially from Seattle southward and (2) uncertainty in the location and amount of precipitation. Meteorologists can interpret and visualize this uncertainty using “ensembles,” which is a suite of many slightly different forecasts from a given model. Ensembles will play a bigger and bigger role in forecasting as computers become faster and faster—they take massive computational resources but are more exact and better convey uncertainty than just a single model run.
Communicating uncertainty to the public is not an easy task—a consumer would much rather hear how much rain they are going to have to deal with on their evening commute than hear that there’s an “80% chance of rain” while they are driving home. But a forecast that doesn’t communicate uncertainty is a very dangerous one indeed—one only needs to go back to the infamous “Ides of October” 2016 windstorm forecast bust to see how focusing on the worst-case-scenario can come back to bite you! In recent years, the National Weather Service and media outlets here in the Pacific Northwest have pivoted their forecasts to talk in terms probabilities and multiple, potential outcomes rather than a single, in-depth forecast to align their forecasts more closely with the probabilistic nature of forecasting as a whole while still providing easy-to-digest information to the public.
So, here’s my challenge to you, my fellow Leschi News readers! Next time you hear about snow in the forecast through a weather app, weather model, or social media, see if you can find out the amount of confidence is in a potential event and what the reasonable range of outcomes are. In a city that can be crippled by an inch of snow, it’s important to be prepared for any flake that might fall!
Charlie Phillips, a Madrona resident, received his B.S. in atmospheric sciences from the University of Washington and works in Portland as a meteorologist. Check out his weather website at to charlie.weathertogether.net.