January 18, 2020

Why Were the Snow Forecasts So Challenging During the Past Week?

This has been a hard week for meteorologists, with multiple threats of cold and snow.  Some aspects of the forecasts worked out, some did not.    For example, Seattle Public Schools cancelled classes, but the predicted snow did not show up.   I have gotten a number of emails asking:  why were the snow forecasts so good last February, but not so good this time around?


I will try to answer these questions in this blog.  A big message will be that my profession needs to work harder to communicate the uncertainty of our forecasts, and society needs to learn how to use it.

The successes

Well before the snow and cold hit, Northwest folks were warned.   Days ahead of time, meteorologists warned you that the weather pattern would shift, that cold and potentially snow would hit the lowlands, and that the mountains would get hit hard by huge amounts of snow.  Check my blogs of January 6 and January 7 if you need proof. The mountains did get massive snow, it did get colder, and the lowlands would get some snow.


The problems

But there were the problems.   Although we correctly forecast cooling, our models went too far, predicting temperatures 3-8F too cold.  The convergence zone snow on Sunday night was stronger and farther north than expected. And the snow band on Tuesday night was shifted 50-75 miles farther north than predicted, leaving Seattle pretty much snow free, while NW Washington had a major event.

Why did we seem to do so much better last year?

Last February represented a tour de force of snow prediction for our region.  But many of the events were relatively easy, since cold air was in place and the precipitation was associated with relatively large scale systems approaching the region (e.g., fronts, large low-pressure centers).   The large systems are easier to forecast and the precipitation areas were large (see satellite image from last February).

But that was not true of last week.   We had far more difficult to predict, smaller-scale features to contend with.  Last Sunday, a Puget Sound convergence zone was the key feature, in which small errors regarding the winds on the coast could greatly alter the position and intensity of the low-level convergence (between air coming in the Strait of Juan de Fuca and flow passing around the south side of the Olympics).   Not surprisingly, the excellent European Center model got the position/intensity better than the American models.

Then there was the unusual, small precipitation band that moved off the Pacific on Tuesday (see picture)

This small feature came barreling in, but the forecasts were off by 50-75 miles.  The Olympics (and Port Angeles) got hit hard--which was forecast--but the southern edge shifted north, leaving Seattle high and dry.  The more northern shift became evident during the early evening, but Seattle decided to cancel schools based on the earlier forecast.   

A communications failure mode we need to work on

Consider the Tuesday situation.   The models WERE showing lots of uncertainty--something I noted on my blog discussing the situation.    To illustrate, here is the ensemble of many model snow forecasts for that period for a location in Seattle (North Seattle near Magnusson Park).  HUGE variation in forecasts from very little to six inches.  Uncertainty was large.  We need to communicate this better.


And the fact that the high-resolution  models started backing off for Seattle snow after 6 PM was another sign of non-event over the city.  What Seattle Schools and others should have done was to hold on their decision to close the schools (or other events) until later in the evening, perhaps starting with a few hour delay.    The University of Washington waited to decide and keep the university going the next day.    Dealing with such uncertainty and securing the latest forecasts is one reason we created SnowWatch for city of Seattle.  I hope others will use it.

Lowland snow forecasting is always challenging in our region, since it involves the unusual occurrence of cold and precipitation at the same time.  But improving forecast models, coupled with the communication of uncertainty, should help society make better decisions.



32 comments:

  1. For what it's worth (nigh nada i'm afraid) from the point of view of a north Seattle Prof Mass blog reader (and reader of https://a.atmos.washington.edu/data/disc_report.html ), the forecast of this event seemed quite accurate indeed; particularly in terms of *when*. Damn near perfect timing.

    ReplyDelete
    Replies
    1. Very impressive indeed. Especially if you live in Port Angeles as I do. Well done!

      Delete
  2. I think most of the forecast discussion from the NWS on the uncertainty of the forecast was actually quite good, if people would read it. TV people are another story, however. Only a few of the TV weather announcers are good at communicating the subtleties. But even aside from that, when TV moves from the weather person to the general newscasters, it tends to become another thing altogether. Some stations were almost in 24-7 mode for what is and was fairly normal winter weather. Seems quite excessive to me. I think there is value to carefully reporting the hard-hit areas like Skykomish and Pt. Angeles, but they also need to keep it in perspective.

    The other thing is everyone, including government personnel and school districts, are quite gun-shy as a result of last February's snow fall and don't want to look like they were not being careful enough.

    ReplyDelete
  3. Weather here is not easy to predict. It's impossible for Cliff to provide a perfect, non-fallible forecast for the many areas his forecast covers. The same is true for schools and other institutions which have their own problems to deal with difficult weather events. We must follow nature and work with it as opposed to try to beat it. Let's use our time more constructively.

    ReplyDelete
  4. One thing that has really been off in the last week are wind forecasts, at least for Whatcom County. Currently, we are under a wind advisory ending at 10 am this morning for SE gusts up to 50 mph. So far, the biggest wind at the Bellingham Airport has been about 10 mph from the north. This is really a major miss. There are have been a number of similar big misses in the 10 days, but this is probably the worst. All of the models are way off.

    ReplyDelete
  5. I notice that Snowwatch seems to end at around Shoreline. Is there going to be any effort to extend it up to Bellingham?

    ReplyDelete
  6. There were parts of Seattle that *did* get snowed on ... after noon on Tuesday the 14th. Our alley was full of kids that were sledding/tobogganing. So the forecast and school closures helped avoid an afternoon mess.

    ReplyDelete
  7. Cliff,
    How do the large scale models such as Euro integrate with the high-resolution models such as WRF?

    ReplyDelete
  8. Nice! alright, since this snow event was not great (What i mean is, not like last year, if you know what I'm saying,) I'll hope we see a bigger event later this season, like February. We'll see!

    ReplyDelete
  9. Cliff, were the weather observation sailboats off the coast (mentioned in your January 9 post) of any assistance in forecasting our shifting weather this past week?

    ReplyDelete
  10. Was there really a forecast of anything close to the 18+ for port angeles?

    ReplyDelete
    Replies
    1. It was introduced to the forecast Tuesday morning as "Major Changes to forecast 8-12in" for Eastern Strait including PA. I suddenly sprung out of nowhere. The amounts were then downplayed slightly by the Winter Storm Warning that went into affect later that day. I got to witness the entire event as I worked in it all night. 18 inches to over 2ft was common. I dont think anyone expected this. Although I will say that Accuweathers winter cast gave probabilities of snow totals- listing the highest amount of 22 inches plus as roughly a 10 per cent chance. I thought- "that would be crazy"... And it was- lol. Absolutely incredible. Not as much snow as we had on the ground last Feb. But it all came at once.

      Delete
  11. If it shifted north, did the Vancouver area get hit hard?

    ReplyDelete
  12. Cliff,

    I'm learning more about the challenges of forecasting snow in our area. It's amazing to see what can be accomplished.

    Of course, I suppose individuals and groups of people have their own expectations. However, we seem to be moving toward increasingly reasonable public expectations given our resources and knowledge.

    We need to accept our capabilities and limitations in specific circumstances going forward. I can imagine our expectations will become more realistic over time. It's the only reasonable outcome as the other paths simply waste everyone's time and other resources.

    Personally, I don't have any complaints and choose to focus on appreciating what we can achieve now and hope to achieve in the future.

    ReplyDelete
  13. Cliff's point is a great one, about communicating more than the prediction, but also the certainty of their forecast. So Cliff, this seems common to your trade (or how a layperson hears it) -- most of the time you tell us rain, or sun or whatever with say "huge likliehood"..... but then enough times that it matters....we get a forecast without a good disclaimer of ....only a 50/50 chance of this happening........maybe you all do this somehow..."Super confidant" forecast versus "hard to tell"...and since you want a forecast, ok i'll supply it but don't hold me to it...

    ReplyDelete
  14. The one feature that was odd in the missed forecast was the cold air that never really made it here. We were warned about significantly cold temps that never really materialized. That really cold air could have changed things last week, but no dice.

    ReplyDelete
  15. Metro transit put chains on their buses when there was no snow on the roadway

    ReplyDelete
  16. NWS is doing it again for the Bellingham forecast for tonight, which is:

    "A 50 percent chance of rain. Cloudy, with a low around 41. South southeast wind 16 to 20 mph, with gusts as high as 28 mph."

    It is currently 36F at the airport with a 5 mph wind out of the north. Last measurable precip was at 10 am this morning. Of course, we did get a melt off early with temperatures in the 40's, but this just seems to be a fundamental miss. Looks like there is something going on over Vancouver Island. Hard to say if it gets here.

    ReplyDelete
  17. More unusual temperature swings in NW Bellingham today. The temperature rose from 32.4F at 10 minutes after 7AM to 46F at 10 minutes after 11AM. After peaking at 48.8F at 10 minutes to 5PM, the temperature fell to 35.6F at 35 minutes past 6PM. The rapid temperature increase in the morning was correlated to a sudden shift in wind direction from the NNW to the ESE and the rapid temperature decrease in the evening was correlated to a sudden shift in wind direction from the SE to the NW. Cold air remains trapped in the Fraser canyon and approaching low pressure systems can induce temporary shifts in pressure gradients from onshore to offshore such that areas adjacent to and north of the mouth of the Nooksack can be sporadically subject to cold, continental air from central British Columbia while areas south of the Chuckanuts bask in typical ~50F southeasterlies. Depending on the strength and location of the most intense pressure gradients, it's possible for areas of northern Whatcom County to remaining under offshore flow for considerably longer than their neighbors from Bellingham southward. For examnple, while locations from Ferndale southward reached the mid to upper 40s today, Lynden, to the northeast, remained in the mid 30s while Sumas, due south of Abbotsford, BC never rose above freezing.

    ReplyDelete
  18. Look, I think in general forecasts were only as good as the models, let’s face it, the GFS was absolutely horrible during the event, the cold never ever really got into the area, and a few news stations kept saying coldest air in 10 years to invade, very, very wrong, our news media blows things up, and overhyped the event, need to be realistic. I agree with other comments, look it was just normal, everyday winter weather. Definitely was not that cold, except for a few areas( PA, Stevens Pass, Skykomish, Bellingham) most of us did not get that much snow, twice I was under Winter Storm Warnings SE of Seattle ( Buckley) and barley got a flake of snow, yet the night I was under no advisory, I got 4-5 inches, so forecasts were not good, esp on the cold, and the media and some social media have got to stop overhyping events, even as the event was unfolding I heard news media saying coldest air in 10-15 years, and that simply was not true at all. Some areas that were under warnings just should not have been, and I agree, wind forecasts, the last week even here, have been bad. Been under high wind warnings that never happened either. Models need to be better, but I think the real problem was media crying wolf so to speak, when let’s face it, most areas were not that bad, and and hour of snow coverage for a few inches is ridiculous, it’s Winter!!! Yes the area up Hwy 2 was badly hit, as was parts of NW interior, but nothing that hasn’t happened before, and will happen again.

    ReplyDelete
    Replies
    1. Cliff Mass was the one who initially stated it would be the coldest air in years. The media just picked up on his erroneous prediction.

      Delete
    2. The forecast for extreme cold air was driven by all major weather models. This goes far beyond the scope of human prediction and are the best tools at our disposal. It wasnt just cliff, but the other forecasters interpreting the models that came to this conclusion. Rather remarkable to have such advance warning of an impending cold spell, even if it didn't arrive with the predicted severity. It is the very nature of where we live. If you are wise you will take the fair warning- prepare for the worse and hope for the best. It takes guts to be a weather forecaster- guess you gotta have thick skin

      Delete
    3. If you think the forecasts hinge on only his predictions, you are woefully mistaken

      Delete
  19. I think if you want usually always accurate (boring) forecasts go to San Diego.

    ReplyDelete
  20. Seems like the most likely help for forecasting is MORE DATA, maybe something along the lines like CoCoRaHS when volunteering interested individuals are reporting what's occurring at their specific locations throughout the country but in this case within our State. With the evolution of technology, hopefully in the not too distant future, real time reporting from individual reporting could be available to assist with the more complicated and elaborate forecasting models. More data can't be hurtful. And communication has to be worked on to prevent the over-hyping. Nobody said this was gonna be easy...

    ReplyDelete
  21. I live in Anacortes, in the highest elevation neighborhood on the island, at 400ft. We moved here 6 months ago from Ellensburg. Because of the some of the steep roads we wondered about snow. Some of our new neighbors said yes, it does sometimes snow here in the winter, but we don't usually get more than a couple of inches and it's gone quickly.

    Hah! We ended up getting more than a foot of snow over those several days last week ... at least 15 inches, probably more. Everyone was saying that as long as they've lived here, they've never seen this much snow! And, boy, were they jealous that we had a snow-blower!!! ;0)

    Luckily, the warmer temperatures on Saturday, coupled with windy conditions and a little rain, seems to have pretty much taken care of most of the snow. At least the streets & sidewalks are clear now (even in spots where no one shoveled or plowed). The plow never came into our neighborhood ... the kids in the neighborhood had a lot of fun sliding down the streets (we have two very steep hills in this neighborhood)!

    ReplyDelete
  22. Forecast is looking very routine for the rest of the month. So,there could be some copy paste of showers and +/- 50 degrees for highs. 40s at night. Mid 50s seems a bit warmish even for the wet side but it's not like we are looking down the barrel at 70.

    There is still time for another winter outbreak but once we get into February it's 50/50 in future retrospective if spring started this week or in March sometime.

    ReplyDelete
  23. Just add some metric for "certainty" to forecasts. Related to that, i am struck by how a forecast seems to have no "teeth" or consequence to the forecaster...vs say Vegas sports bets are very carefully thought out by the pros and thus are beyond quick guess and are instead predictions with big consequences.

    ReplyDelete
  24. Look, as I said earlier and agree, snow and cold are hard to forecast here, and I actually did not read cliffs earlier forecast of extreme cold, and I am not blaming Cliff or any other forecaster for that. All just a heads up, I have a degree in atmospheric science/meteorology, been doing this on the side for 40 years and teach a high school class, so I have a background, all major models DID NOT advertise a major cold snap, The European was actually much warmer, especially from about 3 days too the event, and had caught on, go back and look, but our TV people dont bother to tell you that, they just kept up the epic cold hype, which was simply not true, only point I am making, it was blown up by media and social media. I agree that human prediction can only go so far, and we need the models, but everyone I talk too, who knows or has a background in weather has said the same thing, our big problem was and is the media blowing it up, to be more than it was, dont think anyone can really argue with that, most recently last Feb was much worse in my opinion, def much colder, and yes some areas were hit, no doubt, but it is Winter,
    As far as being prepared, I think in the Winter months you should always be prepared, I agree, and I really think people pay attention to forecasts, which many dont, they just hear what might happen, The NWS did say many times, and I have gone back and looked, a lot of uncertainty, and yet everyone just runs with worst case. Never ever said a forecast was based on cliffs prediction, and I never thought that. The next two weeks look to be Ave to a bit above with rain, so for now, no worries or excitement.

    ReplyDelete
    Replies
    1. I agree with many of your points. The media did continue to overhype and did not properly deliver the probable scenarios and levels of certainty. I often wonder how the common person would handle this information under severely uncertain, yet possibly dangerous weather impending. I would have to say on the flip side that sometimes it may have saved lives. If it were only a 10 or 20 percent chance of a severe weather scenarios occurance, what would the general population do with that information? Probably go on that road trip. Would they make responsible preparations? Hopefully... even though there is a 90 percent chance things wont materialize and seeing that in previous events likely would associate it as a small threat, even though if it does materialize would have severe consequences. I think the winter storm and wind warning issued by NWS were warranted based off of the potential consequences if they event did materialize- Risk benefit analysis sort of thing. These warnings are often a que for media and people to take heed of the potential for the event. They are warnings of danger- or ar least the possibility thereof. I would imagine that many lives have been saved by advance warnings- even if the probabilities were somewhat low. Mike, not all of what is written here is directed toward you or your points which are very good. Just food for thought.

      Delete
  25. Seattle schools let out an hour early on Wednesdays, so a two hour delay (let alone a 'few hour' delay) would mean quite a short school day. The alternative is to close then use the Wed Jan 29 pre-arranged snow make-up day.

    I think it actually looks like a pretty optimal call given what they knew.

    The only way for it to be worse than a delay Wednesday is for the forecasts to be wrong AND for a new school closure event to happen during the following two weeks.

    ReplyDelete
  26. To be fair to Seattle Schools, they were already dealing with their scheduled one-hour early dismissal, so a late start (which would have been the right move any other day of the week) would have left them with three hours of classes, which split among six classes and other mandatory breaks would leave them with little reason to get everyone to school that day.

    ReplyDelete

Please make sure your comments are civil. Name calling and personal attacks are not appropriate.