Sunday, March 4, 2012

Snow Threat Evaporates

I just took a look at the latest model runs---both the UW and NWS models--and I had quite a shock.   The lowland snows suggested in the previous model runs were GONE.  To illustrate, here is the 24-h snowfall for the period ending 4 PM Monday.  Lots of snow in the mountains, but virtually nothing in the lowlands.

 So what is going on?  This is really an interesting case and reveals some dilemmas in how we forecast and communicate our forecasts.

As I noted in my earlier blog, getting much snow during a frontal passage, as will happen tomorrow morning, is very, very unusual.  It is going against all my experience and other forecasters I talked to about.    But the models "thought" it was physically reasonable.   Things would have to be just right...cold air moving in aloft, enough precipitation intensity to drive the snow level to the surface, etc.   It appears that relatively subtle changes in frontal intensity, precipitation timing, and other features caused the model to back off on the snow.

There is a real lesson in this event, a lesson discussed at the NW weather workshop yesterday.  We really need to get away from providing single value deterministic forecasts that can change in time sharply like this, but rather move to probabilistic predictions based on ensembles, providing the odds of snow and other parameters.

By the way, much of the rest of the forecast has not shifted that much, but because the snow forecast was so excruciatingly sensitive, minor variations in the solutions made a huge difference.

The latest forecasts indicate that the precipitation should reach us around 4 AM and be over around 9 AM.  There is substantial uncertainty in the snow forecast...We will know soon....

21 comments:

AndrewM said...

Another 'miss' on the forecast was the heavy afternoon winds. It was gusting strongly on Bainbridge from 3-6PM.

patrick said...

Gotta be honest, I really do believe the WRF is handling this situation VERY POORLY and this wouldn't be the first time, we'll have to wait and see what we wake up to tomorrow morning

Jeff said...

Good grief, what a swing. Accuweather is way more accurate, I'll stick with them.

wymanbr said...

Perhaps we should move away from forecasting and move on to now casting!

Zathras said...

When I was growing up in Lake Stevens we got a foot of snow toward the end of the second week in March one year. The upper trough that moves through Monday night is just the sort of setup that ought to give some snow showers to areas north and east of Seattle--especially places like Lake Stevens.

The 00z WRFGFS has some precip up in that are Monday night--but doesn't have snow at low elevations despite the very cold air aloft.

It will be interesting to see the 12z run handle the cold upper trough Monday night. There ought to be some snow showers around by that time. The front itself was always a long shot--statistically you probably don't want to have snow in the forecast too often when the 850mb temps is only forecast to be -2C. But Monday night it will be -8C and -38C at 500mb--convective perhaps which is good!

julie said...

Two years ago in second weekendvof March I was traveling across state westbound and a snow and ice event created a pileup on Snoquslmie. So a little cluster of Subaru types took our time with heavy snow over Blewett, Stevens and down to Monroe. It was exciting but ultimately wearing driving. Regular six hour drive turned into nine. Some weather had been predicted, but not extent,

Steve said...

Cliff: Many years ago one of the TV stations did a series of focus groups about how to present weather. One thing they found was that the population did not understand percentages-- like 30% chance of rain. Some thought that meant part of the area would get rain, or it would rain 1/3rd of the time, etc. So I don't see how ensemble forecasts will work. Forecasts on "broadcast media" have become less than useless.

Erik said...

Is it common practice to do sensitivity analysis to find which updates caused such a change? You could use old model run inputs with a subset of updated inputs to understand what effects each type of updated input had? I assume this is part of the tweaking you need to do to get a model working well? Would be curious how much adjusting is done after these types of events and how to avoid going overboard to make the model accurate for what might be an outlier?

Buddy said...

I just drove over Snoqualmie Pass (8am) and it was raining the entire time. Boy, Seattle barely got missed this time. Oh so very close lol.

Ferdi said...

Can't say I've been too impressed with recent forecasts. The dirty ridge this weekend was a case in point. Rained here off and on all day on Sunday with a heavy overcast. I thought the snow forecast was a bit forced because it didn't seem like the air mass was going to be cold enough given the warmth of the air it was replacing. But I know only too well, coming from a family of meteorologists, how hard it is to make these judgment calls. I totally agree with you that probabilistic forecasts make much more sense. A synopsis of what might happen would obviously be more useful than a wrong forecast.

Hal said...

Convergence zone snow right now in Lynnwood.

abbi said...

Maybe not so far off....snow is coming down fast out here in the snoqualmie valley. About an inch on the hill in less than the past hour. Started out with hail which has turned into large snowflakes

Unknown said...

looking out my window in snohomish, those are might big, fluffy, white rain drops.

sequimteeth said...

Monday at 2PM in Sequim it was very sunny, but windy. At about that same time my wife was getting the ferry at Kingston and reported cold and blowing snow/rain. March Madness.

coolio700 said...

Snowing hard right now in the south mill creek area. Sticking to the roads but only about 1/4 to a 1/2 has fallin

Unknown said...

Seems like ever since the coastal radar came on line the prediction accuracy has decreased. Do the modeling algorythms need to be adjusted?

Ferdi said...

I spoke to soon. There is an absolutely wild squall line moving down from the north, black as night. Orcas Island is disappearing in snow. This should be a good show. 5:30PM

soaringbrain said...

Well, its snowing at sea level at Point Lawrence Orcas Island (Eastside) right now! Big wet flakes! Awesome!

soaringbrain said...

Its snowing on Orcas Island right now.

cornbread said...

Snow in Bellingham at about 200' ASL.

Unknown said...

I'm a teacher in Duvall. It began snowing around 1:30 pm and accumulated about an inch and a half of weird "snow-ball" type flakes that stuck on my car at 3:30 pm as I drove to sunny Redmond where I live. Huh. Well, at least it snowed somewhere.