March 14, 2012

Why is it so hard to forecast lowland snow?

Meteorologists can get windstorm forecasts right nearly all of the time.  We hit the record high temperatures in July 2009 nearly dead on.  We can give you a skillful forecast of a marine push in the summer.

But snow forecasts, although improving, are often in error.

Why?   Well, there are reasons....a lot of reasons.

(1) You've got to get the amount of precipitation right--really right-- or you have a very bad snow forecast.  On most days, the forecasts hardly talk about the amount of precipitation, just the probability of precipitation.   Most people don't care whether .1 inch or .4 inch of precipitation falls...they would hardly notice it.  But a difference like that would be HUGE for a snow forecast.  Around here there is typically a 10 to 1 ratio of snow versus precipitation in the form of liquid water.  Thus, .1 inch of precipitation would be .1 inch of rain or, if cold enough, 1 inch of snow.  (Precipitation is always reported in terms of liquid water--so snow has to melted before measuring it).  So the difference between .1 and .4 inches of precipitation, which you would be oblivious to of if it were rain, would be the difference between 1 and 4 inches of snow.  That you would notice.

Getting the amount of precipitation exactly right is very hard, much harder than getting the probability of precipitation correct, thus making the snow forecast difficult.  I can tell you why in a future blog.

(2)  But it is worse than that.  The majority of time our temperatures are marginal for snow...we are right on the edge of rain or snow.  The reason...all the warm water (e.g., the Pacific and the Sound) close by.    On most days a few degree error would not even be noticed (like the difference between 58 and 61F), but that is a big error on marginal snow days.  It is very hard to reduce the error under a few degrees.

(3)  But it is even worse than that!   The temperature of the air is affected greatly by the evaporation and melting of precipitation.  And that is dependent on the intensity of precipitation and the relative humidity of the air below!   Remember:  virtually all of our precipitation starts as snow.  Usually it melts before it reaches the ground, but if the intensity is great enough, the evaporation and melting of the snow can cool the air below so that the freezing level (0C) and snow level (about 1000 ft below the freezing level where all the snow does melt) progressively head towards the surface.  So we need to get the intensity of precipitation right, the initial freezing level right, the relative humidity of the air below the precipitating clouds get the message.  And this is hard.

(4)  And there is more.  To get the depth of the snow right, you also need to know the characteristics of the snow--is it dense or light?  That depends on crystal type and temperatures, which can cause the ratio to vary from 5 to 1 for wet snow to 20 to 1 for dry stuff.

(5)  There is another problem.  Snow is so infrequent and takes such unusual conditions to occur that meteorologists have less practice and experience that for more typical weather.  Why unusual?  Well, it is easy to be warm and wet around here (warm being in the 40s and 50s) for air coming off the Pacific,  pretty easy to be cold and dry (when we have high pressure over or to the east of us).   But it takes a lot of doing for us to be cold and wet.  The set up has to be just right.  Meteorologists are like anyone else:  you gain skill with practice.   Only old timers have lots of cases in their heads...and even that is not good enough since the modeling and observation systems have changed so much.

(6)  Still think it is easy to forecast snow?   Well, local meteorologists have more problems to deal with: a multitude of local effects.  The potential for snow is generally much less near local water bodies, particularly near sea level.  Chances increase greatly up on hills...even 200-300 ft can make a world of difference.   We have gaps that allow cold air to move through the mountains (like the Fraser River gap allowing cold air into NW Washington).  Convergence Zones (like the Puget Sound convergence zone) produce areas of enhanced precipitation and sometimes snow.   And their are dozens of local effects I haven't mentioned.

You get the picture.  Snow prediction is hard, real hard, particularly around here.   Forecasting winter snow back east,  say in  North Dakota, is a cake walk---you KNOW is will be cold enough.  Huge simplification.

So unless you are a Subaru advertising executive....have a little sympathy for the poor NW weathermen when a snow forecast goes wrong.  And keep in mind, a lot more technology is coming:  ensemble-based snow prediction, better use of the new dual-polarization radars, and more.  Meteorologists will make progress....and then we can attack even harder problems, freezing rain and shallow fog.


  1. As you said, there are even more issues. One of which you alluded to was precipitation intensity. As you have noted in the past, pretty much all precipitation in our area forms as snow, and then melts to rain as it falls below the melting level. But this doesn't happen instantly, so you can see snow with above freezing temperatures/dew points. Typically this snow doesn't accumulate on the ground, especially during the day. But if the precipitation is hard enough, it can exceed the melting rate of the ground and so you get some accumulation. Convergence zones and convection are notorious for this, and are extremely difficult to forecast.

  2. Good article Cliff. Lots of snow just east of Mt. Vernon yesterday. Several inches. Also lots of snow on Cypress and Orcas Islands above about 700ft elevation.

    Not sure I agree with your idea that winter in the NW is over by the end of February. March is often colder and nastier than February. It seems a month for cold upper level lows and troughs. This month is following that pattern.

  3. Two points to add:
    1. Variability in snow form has tremendous influence too. Graupel requires more heat from the surface to melt than a delicate doily-like snowflake. PSCZ-generated and springtime convective squalls commonly start off with this precip type locally influencing the surface temps leading up to eventual snow & accumulation. Convection is notoriously hard to pinpoint, even in the plains the best we can really do is draw a box of relative-confidence around the feature... here we deal with the same thing... just colder and it produces our snow.
    2. Communication method: our nuanced and extremely diverse local micro-climates are constantly changing in time. Weather features that generate our high-impact weather are hardly static in nature and do not operate or conform to 6hr forecast intervals. The timeline evolution of a storm event for a ground observer in Everett will often be very different from an observer in Bothell, and still further different in Issaquah, etc. 2-3min news segments that are digested by the public at large cannot cover the detail needed to highlight the 'true' forecast around here. Even IF we get better at pinning down snow, the delivery of such info, and manner it is received must be both unified, and more effectively disseminated to be oriented to accommodate the observer's point of view.

    More to your original point, this is especially the case here in the Pac NW where during high-impact events the differences are tattooed on the ground.

  4. Nice explanation Cliff. Well done. Still, I'm waiting for you to post on the incredibly dreary weather we've having. I don't know what an atmospheric scientist would say, but to me and every single person I know, it is dreadful. And this from dyed in the wool northwesterners.

  5. "Meteorologists can get windstorm forecasts right nearly all of the time. We hit the record high temperatures in July 2009 nearly dead on. We can give you a skillful forecast of a marine push in the summer."

    If this is true, why is the NOAA forecast so often inaccurate? When the 12 hour forecast can miss mark often it leads me to believe NOAA does not really think it is too important to provide quality services instead of token forecast to appear they are earning their taxpayer funded endeavors. If NOAA can not do better than provide very general computer generated forecast then maybe NOAA should step aside. Let another entity (private sector?)that values putting out a quality product take over instead of just providing "sleeping at the wheel" NOAA forecast.

  6. Where is that first picture taken at?

  7. The picture is University Way in the Seattle U-District, looking north from NE 45 St.

  8. One of the long range 10 day outlooks (Weather Underground)has snow next week and wind chills of 19 degrees... and its not even April Fools Day!

  9. [url=""]http://www.nnvl.noaa...ageoftheday.php[/url]

    Check the image for the 12th.

  10. @Lance - looks to be University Way NE. Somewhere in the 4300 block or so would be my guess.

    -Duvall John

  11. Weather forecasting has come one heckuva long way in my 60 years of life, Cliff. You are correct. Snowfall is tough to forecast in Western Washington state. Heck, it is even tough to forecast in Yakima and Wenatchee because of the Cascade mountain rain shadow.

    You guys and gals have got it more together than many give you credit for.

    Jeez, when I lived in Yakima in the 50s and very early 60s, the forecasters would predict snow, and there would be nary a flake.

    The temp was right but the precip. was lacking.

    Seattle is the opposite much of the time.

    The general forecasts are GREATLY improved. Everywhere.

  12. I'm amazed that Tillamook and Newport, Oregon got significant (6 inches plus) accumulating snow Monday, and we in Seattle didn't -- because it was too warm.

    So I'm learning to appreciate with how odd the snow patterns can be around here. (I'm from Maryland)

  13. Having grown up in the Pacific Northwest, snow events have always felt serendipitous to me. I remember those mornings as a child listening to the list of school closures on the radio and what a joy it was on those rare days when school was closed, the snow stuck, and all the neighborhood kids went to the hilly streets to lug our sleds and saucers up and slide down all day. Because it is difficult to forecast, it always feels somewhat special. :)

  14. Cliff, if I change my opinion re: the impacts of mankind on aggregate global warming, will you promise to cut back on the rain this week?!?!?!?!?

    Any updates re: the snowpack? I am sure with the last month of snow, we are definitely above normal for this time of year. What is the likelihood the ski resorts will remain open late this year given current models?

    Thanks for all of the updates!


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

Is Mid-June Getting Warmer or Colder?

 As I will demonstrate below, this past week has been unusually cool around the region. But that leads to another question.... is mid-June g...