Weather Blog

What happened to Seattle's snow?

What happened to Seattle's snow?
As snow raged to the north, east and south of Seattle on Feb. 23, 2011, @hollywoodheffne snapped this photo of a rather calm, partly sunny afternoon on Queen Anne Hill.

Cars literally buried in Mount Vernon. Winds gusting to 40 mph across the San Juans. Tacoma has snow shower after snow shower that melts when it's done, only to have one final blast of snow that created an icy mess on the streets.

Meanwhile, the sun was out in Seattle, as many who decided to stay home from work to dodge what would ultimately be a bare-and-dry network of roads unhappily noted, until they went out and tried to get home in the South Sound or Eastside in the late evening snowy mess.

Despite the crazy weather, the Seattle weather forecast community got a large area of the region's forecast right as far as actual predicted storm totals (granted, we gave a wide range in some spots)

It just so happened a bulk of the region's population live in the area where the snow forecast didn't pan out.

While many in the blank regions probably spent the day yesterday mumbling about the incompetence of the weather folks here, I'd respond that what happened yesterday was not a function of incompetence, but a great illustration of just how much of a challenge it is to forecast the weather here.

Think how complex the atmosphere is. Each molecule of air affects the ones next to it. It's an incredible ode to humanity that we have developed technology that can even begin to calculate the atmosphere and, more amazing still, forecast how the atmosphere will change -- even out several days -- with some degree of success. (Read more on how forecasting models work at my earlier blog entry: Forecast models: Just really fancy dart boards?)

But of course, it's not perfect. We have yet to develop enough computer power to compute each parcel of the atmosphere. And even if we did, we'd have to have to know what's going on at each parcel, and the perfect math to simulate what each parcel will do.

In the Northwest, it's even more daunting. For one, this region has a nearby ocean, an inland Sea, two mountain ranges, a rain forest and a desert, all within about 200 miles of each other. The interactions between those features can cause all sorts of complex ramifications. Computer models can try to account for it all, but it doesn't take into account small details in terrain that we have so much of. And small errors can be the difference between 18" in Mount Vernon and a dusting 30 miles away in Everett.

Ask a computer model scientist and nailing a snow squall within 20-30 miles would be considered a great success. Ask anyone in Seattle or Bremerton or Mountlake Terrace how the forecast went, and the response would not likely be positive.

And those have every right to be upset with the blown forecast. It is our duty to try to get the best forecast we can and when we predict 2-5" and something close to squat happens, we didn't do our job and we will learn from this. But I just wanted to get out there that it wasn't a totally busted forecast by any means -- the fact that we did reasonably get that a major snow storm happened in the Northwest within a few dozen miles with all the variables that were in play should be worth something, even If it didn’t go exactly to plan.

Better Laying Out The Odds

There are still lessons to be learned on our side as well about conveying uncertainty as well. I have a future blog in the hopper about weather forecaster's challenges in conveying the inherent uncertainty of a forecast. (It was one of the seminars I went to at the American Meteorological Society conference last month. Still haven't had time to write it up yet.)

We can to do better jobs in presenting just how much confidence we have in a forecast that can better help you weigh the risks of going about your daily routine or perhaps making changes. There's still some lessons to be learned. And from our side, there was a great article I read by blogger Paul Balcerak, who works in the local media web world but not in weather, about the audience's reaction to weather forecasts.

Another factor to consider when it comes to forecasting is simply the volume and the preponderance of information available to most people today.  People rarely get their information from one source credible or not.  That means that information we give out is combined with snippets from everyone from Radio DJ’s to talk around the latte stand.

Details like snow totals & timing get tossed about tweeted, and Facebooked.  And what are left out are often the nuances that we try so hard to get across.  Most people just want to know if it is going to snow.  Around here there is never a one word answer.

So as a forecaster you end up having to defend something that is not what you said. In effect one forecaster’s credibility is based on a perception developed from many sources.   On the other hand, we put what we believe in writing and will continue to take responsibility for it.  Regardless of perception on balance we do a pretty good job within the limitations of the technology and the incredibly complex topography of western Washington.

Yadda Yadda, why didn't Seattle get snow?

Caution: Weather geek speak ahead:

It all started so well: A cold low pressure system that had tapped into some very chilly air in the Gulf of Alaska was sweeping down from Vancouver Island. In the meantime, arctic air had pooled into British Columbia and was ready to be pulled out by the approaching low. (Cold air is more dense than warm air and thus has higher pressure, so when a low pressure center moves to our west and south, it will draw that arctic air in through the Fraser River Valley.)

So you've got your two main ingredients for snow: Moisture and cold air.

Models indicated as the low moved off the coast, southerly winds wrapping around the low would collide with those cold, north winds coming out of the Fraser River Valley. This is similar physics to what makes a Puget Sound Convergence Zone only the north component is coming from Canada, not the Strait of Juan de Fuca.

Previous events of this nature have shown that these areas of convergence have potential for great amounts of snow, and thus the very high level of concern. The only questions were: When would the low move south to veer the winds to begin the convergence, when would the arctic air start pouring in from BC, and when --and where -- would all the ingredients come together.

This is where some of the uncertainties were involved, and although there was general agreement in the models, it wasn't perfect.

I've put together a photo gallery of one model's predictions for snow from 4am Wednesday through 4 am Thursday. The gallery shows each model's run 12 hours apart starting Sunday night through Wednesday morning. The darker the colors, the greater the snow potential and the legend is on the right side (in inches)

But I'll put two here. This one was what the Monday morning model said:

That shows heavy snow up over Skagit and northern Snohomish County, less over Seattle. Ironically, this model run ended up doing quite well on the snow predictions.

But here is the model on Tuesday night -- right before the snow event began.

Note how it has major snow predicted for Seattle, and theoretically it would be the best model since it's the closest to the actual time of occurrence, it should have the best data to work with as far as what the storm was doing right before the event.

Our forecasts were tailored as such: The low was supposed to drop off the coast during the day Wednesday, first setting up a wide area of convergence up in Skagit/Snohomish Counties. As the low continued south and the arctic air increased, the convergence would increase. As the low continued to move south, the north wind would win the day and that big area of convergence snow would be pushed south into Seattle and beyond, likely during the commute or just after.

What really happened was slight differences from the original plan.

And those slight differences took away Seattle's snow.

As the low approached the coast and the area of convergence began as scheduled in Skagit County. Showers were possible to rotate around the low into Seattle during the morning -- the "1-2 slushy inch showery forecast" which verified for the far South Sound but never made it into the city.

Why? Because the low stalled a bit off the coast, delaying not only the showers getting into Seattle, but most importantly, getting the north winds going. That left Skagit County mainly stuck under the convergence longer. In addition, the flow wrapping around the south side of the low was due westerly -- which created a massive rain/snow shadow over Seattle and Kitsap County. A period of shadowing was expected in the morning, but it wasn't supposed to last long.

Instead, it lasted through the day. Finally, the low began its trek inland, but now a little late *and* a little north -- keeping that main area of convergence to the north of Seattle. After the low passed we finally did get a convergence zone to form -- over southern King and Pierce County that brought some accumulating snow to Tacoma and Kent -- snow that was probably originally destined for Seattle to Everett.

But between the snow shadow hanging on, the low going just a touch north and the convergence zone forming just a touch south, the Seattle to Everett and Kitsap corridor was left out of the big snow party.

So when all was said and done, some areas in Skagit County received 12-18" of snow. In nearby Whatcom County, there was hardly any in Bellingham and a few inches elsewhere. Snohomish County? A decent amount in Marysville but the usual snow belt of Lynnwood to Everett was left blank. Think about that variability. It's 55 miles between Bellingham and Everett, and you go from 0" to 12+" to 0". Where else in the country do you find that kind of variability?

The Northwest is lucky to have some of the brightest minds in the atmospheric sciences field who work here. The University of Washington has one of the top graduate Atmospheric Sciences programs in the world. UW Professor Dr. Cliff Mass is a well-renowned for his research and forecasting. The Seattle office of the National Weather Service has 250 years total of Seattle forecasting experience between their 18 staff members -- a good chunk of their current lead forecasters were there when I interned in 1993. I won't go too much into the media since I'm a part of it, but suffice to say you know there are quite a few TV weathercasters in town who have been here for several years. The people here know their stuff, but there will always be limitations on perfection.