NWS to adopt new snow forecasting techniques

Since we’ve discussed recent heavy snowfall, this seemed like a good story to cover.

via Eurekalert: Better snowfall forecasting

National Weather Service adopts U of Utah powder prediction method

IMAGE: Jim Steenburgh, professor and chair of atmospheric sciences at the University of Utah, skis through powder snow in Days Fork in the Wasatch Range backcountry near Alta, Utah. Steenburgh and…

Click here for more information.

SALT LAKE CITY, Feb. 22, 2010 – University of Utah scientists developed an easier way for meteorologists to predict snowfall amounts and density – fluffy powder or wet cement. The method has been adopted by the National Weather Service for use throughout Utah – and could be adjusted for use anywhere.

Based on a study of 457 winter storms during eight years at 9,644 feet in the Wasatch Range at Utah’s Alta Ski Area, the researchers determined that forecasters could predict snowfall density – known as snow-to-liquid ratio (SLR) – most accurately using only two variables: temperatures and wind speeds at mountain crest level.

The American Meteorological Society is publishing the study in the February issue of its journal Weather and Forecasting.

“We’ve developed a formula that predicts the water content of snow as a function of temperature and wind speed,” says the study’s senior author, Jim Steenburgh, professor and chair of atmospheric sciences at the University of Utah.

“This is about improving snowfall amount forecasts – how much snow is going to fall,” says Steenburgh. “As a nice side benefit for the ski community, this will tell you whether you’re going to get powder or concrete when it snows. We are working on incorporating this into the UtahSkiWeather.com website” run by the university.

The new method “is also helpful to avalanche forecasters,” says the study’s first author, Trevor Alcott, a doctoral student in atmospheric sciences. “We’re forecasting snow density, which is related to the stability of freshly fallen snow.”

A Better Handle on Snowfall, Skiing and Avalanche Conditions

The National Weather Service (NWS) in Salt Lake City has used the method since November, says Randy Graham, the science operations officer.

“Forecasters really like it because it gives us a more realistic depiction of how snow density will vary across the Wasatch Range and with elevation,” he says. “Instead of anticipating a singular density of snow or fluffiness of the snow over the Wasatch, Trevor’s and Jim’s tool has allowed us to have different snowfall densities in our forecasts for different areas based on forecasts of [crest-level] temperature and wind.”

“We’ve always had some insight into the difference between a real powder day versus a really wet snowfall event,” Graham adds. “What this tool has enabled us to do is to better differentiate how dense the snow is going to be over an area with really complex terrain – the state in general, but in particular the Wasatch Range.”

Bruce Tremper, director of the Utah Avalanche Center, isn’t familiar with the new method, but says predicting “new snow density is a very important factor in avalanche forecasting. If low-density snow falls first – light powdery snow – then heavy, wetter snow falls on top, it instantly creates a slab of ‘upside-down snow’ as we sometimes call it. These slabs can easily be triggered by people.”

Resorts “really care about the water equivalent of the snow,” Graham says. “It’s really important to them. Powder is better. And it’s important for them to know what kind of avalanche [prevention] work they’re going to have to do.”

Alcott, an NWS intern, extended the technique so it can be used throughout Utah, and says the agency’s Elko, Nev., office may use the method to improve forecasts. It could be extended to other regions by making local snow measurements in different locations and using them to devise predictive formulas for snow density.

Graham says the method “is a really good example of taking a complex problem, boiling it down to the most important variables to describe the problem, and then coming up with a technique that can be applied in operational forecasting.”

The study was funded by the National Weather Service, its parent agency, the National Oceanic and Atmospheric Administration, and the National Science Foundation.

Flakey Forecasting

Steenburgh says that to accurately predict snowfall amounts, “getting the snow density right is critical. To forecast snowfall amounts, you need to know how much water is going to fall and how dense the snow is going to be.”

Meteorologists predict how much water a storm will produce and translate that to snowfall based on predicted snowfall density, which is the snow-to-liquid ratio (SLR) – the ratio of the depth of new snowfall to the depth of water from melting that snow. SLR reflects how powdery or wet and heavy the snow will be.

“The best way to think of it is how much does an inch of water translate to in terms of inches of snowfall? So a snow-to-liquid ratio of 5-to-1 means 5 inches of snow for every inch of water, or a water content of 20 percent,” says Steenburgh.

Higher SLRs mean the snow is more powdery. Typical Utah SLRs are:

  • Heavy, wet Utah snow has an SLR around 7 (an SLR such as 7-to-1 is commonly referred to only by the numerator), with a water content of 14 percent.
  • Average Utah snow has an SLR of 14, or 7 percent water content. Steenburgh says “that is still pretty dry, especially when you compare it with coastal ski areas” with SLRs around 9 or 10.
  • Very dry, light snow has an SLR of 25. That’s the same as 4 percent water content. Anything above SLR 25 is extremely dry, fluffy snow known as “wild snow.”

Steenburgh says the driest snows ever recorded had SLRs of 100 in Japan and Colorado. Alcott says the record high 24-hour SLR at Alta – known for its powder – is 50.

Learning to Predict Powder

To devise their method, Alcott and Steenburgh studied the relationship between measured snow density or SLR and various recorded atmospheric measurements at a single site at Alta, named the Collins Snow Study Plot.

Steenburgh says he and Alcott chose to study that site “because Alta gets a ton of snow [almost 43 feet annually]. You get as many samples in Alta in one year as you get in Salt Lake City in 10 years.” In other words, Alta provided numerous snowstorms that could be analyzed and used to develop a formula for predicting snow density.

Alta snow safety crews measure snow depth at the Collins site twice daily. Precipitation measurements are made automatically each hour.

Alcott and Steenburgh analyzed temperatures, wind speeds and other factors such as relative humidity for 457 “snow events” or storms at Alta during November through April of 1999 through 2007.

The depth of new snow was divided by the depth of water measured by a rain gauge to determine actual snow density and see what variables best correlated with it.

The study showed that only two variables – crest-level wind speeds and temperatures – were most critical in predicting snow densities. In fact, for all the storms studied during 1999-2007, those two variables alone explained 57 percent of the variance in snow density. And for large, wet storms, crest-level wind speed and temperature explained 73 percent of the variance in the snow density or SLR.

That means that much of the storm-to-storm difference in whether new snow is powdery or wet can be predicted by the new technique.

“It’s the KISS method – keep it simple, stupid,” Steenburgh says. “How much can we strip down the number of variables analyzed and get a good result?”

He says the new technique “does a good job of predicting how the snow density changes from storm to storm, and it does especially well for the larger storms.”

Alcott says the Weather Service’s previous method was less accurate because it tried to predict snow density based on surface temperature at the forecast location – a method developed in the Great Plains – rather than what the study showed was more accurate: temperatures and wind speeds above mountaintops where snow is forming.

Secrets of the Snows

In analyzing Alta snow conditions as they developed their formula for predicting snow density, the researchers discovered some interesting aspects of Alta snow:

  • The fluffiest snow tends to occur when a storm contains less than 0.8 inches of water in 24 hours, when crest-level wind speeds are 18 to 26 mph and when temperatures are 0 to 10 degrees Fahrenheit, with snow heavier at either colder or warmer temperatures due to the type of ice crystal formed at different temperatures.
  • Snowfall density can vary radically from day to day. For example, during Jan. 3-12, 2005, it ranged from heavy, wet snow with a snow-to-liquid ratio of 5.2, to “wild” powder with and SLR of 35.1.
  • Snow densities at Alta have the widest range in February, from a wet SLR of 3.6 to fluffy powder at 35.1.
  • The most extreme powder – “wild snow” with snow-to-liquid ratios of 25 or more – peaks in mid-winter. Of 26 wild snow events during the eight-season study period, 24 occurred in December, January and February, with none in April.
  • Extremely wet snow, with SLRs less than 7, occurred in 28 of the 457 storms during the 1999-2007 study period, or 6.1 percent of the storms.
###
Advertisements

43 thoughts on “NWS to adopt new snow forecasting techniques

  1. Don’t know if our weather service is using this new snow forecasting method, but we could be in for up to 1/2″ of global warming tomorrow afternoon in San Antonio.
    Also 1″- 2″ north of the city, and 4″ further north in the hill country.

  2. Two variables? That’s nothing!
    The climate change models only use one variable, CO2, and look how well they work. Everything that happens is consistant with the models!! ;o)
    Seriously, I’d like to see a followup post sometime reporting how well the method works in various areas of this country and other parts of the world. Does the formula hold up at sea-level as well as it does at the mountain site where they gathered data?
    Neat-o post!

  3. Randy Graham, quoted in this article, used to be the SOO at the NWS office in my area and is one sharp fellow. We use a tool here in the Great Lakes called the Bufkit, developed by the Buffalo NWS, that works quite well. It uses the maximum temperature in a vertical profile to forecast SLR and snow amounts. We obviously don’t have mountain tops.

  4. Better results with far less variables? Are they trying to put computer modelers out of work?
    Not that such would be a bad thing, with regards to certain organizations and individuals…

  5. Only two variables: Temperature and Wind Speed???? ….what about CO2 level? I thought that was THE critical factor in predicting everything.

  6. 2-3″ of global warming predicted tomorrow north of Houston in Montgomery County. Won’t believe it till I see it though. Is Gore flying through IAH?

  7. “Dave L (14:46:00) :
    Only two variables: Temperature and Wind Speed???? ….what about CO2 level? I thought that was THE critical factor in predicting everything”
    Co2 causes higher temperature, supercharging the air with moisture, leading to higher snowfall. So he can leave out the CO2. The scientific basis is robust and has been more heavily scutinized by over 2500 scientists than any other in the history of mankind.
    The tricky part is telling the supercharged warm air that makes it snow from the anthropogenically heated air that makes it thaw (and icebergs melt).

  8. I don’t know much about snow, but it seems curious that humidity isn’t a factor on snow fall depths.

  9. Snow forecasts more than half a day ahead of the storm or even 5 minutes before it can be a useless prediction because either they change the forecasted amount too much or the actual snow totals are a bit different.
    Sometimes they get it fairly close, but usually those forecasts are made while the snow is falling or just before. But right now it’s apparent Texas gets socked with snow again and IceAgeNow is saying another foot for Flagstaff (which had roofs collapsing because of all the snow they already got)
    Mega snowstorms especially are about the most difficult of any type of weather to forecast.

  10. Some friends of mine who live around Chi Town have been telling me horror stories about shovelling up to six feet of “partly cloudy” out of their driveways, to get to work some days.
    Yes definitely we need an improvement in snow forecasting.

  11. Joe Romm on climateprogress predicted a permanent dustbowl in the southwest. he claimed his forcast used NOAA. Today we have a major snow event in NM and going into Texas. Why are the polluted forecasts so far off?

  12. We were supposed to get 4 inches of snow today. It certainly snowed for about 6 hours, but what little of it that deposited in the moning was thawed by mid afternoon. The UK Met Office could do with a better method of prediction too!

  13. Nursing a pulled muscle all winter-(Thanks to my Springer Spaniel, a Squirrel,
    and a longish leash back in November.) So all this talk of power makes me
    a little-depressed (since I have a season pass at the local ski area)…

  14. There is a lot more moisture to work with in this storm in TX than the last one. But this one is much faster moving. A lot of weather models predicted it as early as 7 days ago. Dallas could break its all-time snow record tomorrow. Waco seems certain to break its record if the conservative estimates hold out.

  15. Here’s something that I just don’t get: With second-to-second radar and satellite weather tracking, weather stations in every city and town throughout the world, meteorologist on every TV station, experts behind the scenes analyzing up to the minute data and using high tech computer models, accounting for every factor that can be tracked, they still can’t tell me with 50% accuracy if it’s going to precipitate in the next several hours or even predict the temperature within a few degrees… YET, a smaller group of scientists, only armed with CO2 forcasts, tree rings, and ancient coral, can tell me with 100% certainty that the planet will warm by 0.7oC and that will cause all sorts of horrendous things like drought, rising sea levels, tornadoes, hurricanes..and the list goes on. Is anyone really buying this? I mean really…look around you folks. Think about it.

  16. Dave L. (16:16:39)
    “they still can’t tell me with 50% accuracy if it’s going to precipitate in the next several hours or even predict the temperature within a few degrees…”
    You must have some really bad meteorologists where you live. 🙂
    I agree, it does seem a little silly to say with certainty the earth has warmed by .8 degree C when the thermometers are only accurate to within .5 degrees C and reported to the nearest whole degree F. They are calibrated only once per year or if a forecaster thinks the thermometer is off by more than 3 degrees F. That is far greater than the supposed anomaly, therefore, the change is statistically insignificant.


  17. Dave L. (16:16:39) :
    Here’s something that I just don’t get: With second-to-second radar and satellite weather tracking, weather stations in every city and town throughout the world, meteorologist on every TV station, experts behind the scenes analyzing up to the minute data and using high tech computer models, accounting for every factor that can be tracked, they still can’t tell me with 50% accuracy if it’s going to precipitate in the next several hours or even predict the temperature within a few degrees…

    And that’s because … upper-level conditions are based on spatially-sparse balloon launches (radiosondes) every -wait for it- 12 hours!!!
    The ‘input’ to all those super-computers, the direction, temperature, the humidity, the pressure at height, yes, updated TWELVE hours apart from a sparse network of NWS facilities ….
    Now, there are a few NWS faclities that have launched balloons at reduced intervals, such as OUN (near Oklahoma City) did during their last snow event, to get better ‘soundings’ and a feel for the makeup (moisture content) and temperature profile and air movements (Jet, 500 mb flow etc.) for some nearer-term forecasting.
    .
    .

  18. At this stage does it really require a new model to tell them that when it snows at 30F it will be heavier and wetter than when it snows at 10F or 0F. Any moron who ever took a shovel to his driveway after a snowfall could tell them that.

  19. Dave L. (16:16:39) :
    Yeah the difference is with the amount of time between now and the climate forecasts for 2030 is they have lots of time to make adjustments in steps, Homogenize station data together and fill in areas with make believe numbers so the climate forecasts, can be verified.
    With weather it happens so fast they cannot adjust things fast enough to get it right the next morning, the devil is in the details. Weather is very detailed and hard to forecast but one big average number drawn from many stations can be fudged a little at a time so the decline is hidden very well.
    Like Jim Hansen says “2010 WILL be the hottest year on record,”(by the time he gets done with it.)

  20. Dave L
    “they still can’t tell me with 50% accuracy if it’s going to precipitate in the next several hours”
    Dave, I don’t know where you live, but our forecasts are 100% accurate here.
    “50% chance of rain” 😉

  21. This is a fantastic study! Combine this with the 30% reduction in wind speeds nation wide, which impact wind turbine output and reliability, and it helps explain reductions in snow cover EVEN IF TEMPERATURES REMAIN CONSTANT OR DROP. Thus you can have cooling and reductions in snowfall if the wind speed drops more than temperature does, because you’ll still get the same water content, but the snow will be denser.

  22. Sophomore: I didn’t ask for the difference between climate and weather, that would be sophomoric…I asked why they can’t predict weather with any tolerable certainty but they feel they can predict climate with absolute certainty. Let me try and clarify a few points: weather is observable right now…with accurate and timely measurements, and the predictions are only required for a few hours or days…and they fail most times than not. Climate study contains hundreds of years of poor-quality data (proxy data, many factors are simply not known like the humidity, barometric pressure, sunspots, wind speed, etc. for that same time span) and make projections decades into the future. And we think the climate models stand a chance????? Not a chance. And we see it every day…back in 1998 nobody was saying the winter of 2009/2010 was going to be record breaking snowfall. They said we’d be on the steep temperature upswing of the hockeystick. Just admit it…they were as wrong then as the weather guy was this morning when he said it was going to rain this evening…and it snowed.
    But thanks for that frightening website designed to indoctrinate helpless children with bad science. For example, they state “Climate can change too, but in the past it has taken a very long time to do so”. I believe there is evidence that climate can change within a few decades, even without human intervention.

  23. I didn’t ask for the difference between climate and weather, that would be sophomoric…I asked why they can’t predict weather with any tolerable certainty but they feel they can predict climate with absolute certainty.
    If you understood the difference, you wouldn’t be asking that question.

  24. ” Sophomore (19:13:45) :
    I didn’t ask for the difference between climate and weather, that would be sophomoric…I asked why they can’t predict weather with any tolerable certainty but they feel they can predict climate with absolute certainty.
    If you understood the difference, you wouldn’t be asking that question.”
    The main difference is when you project something out a 100 years none of your contemporaries will be around to say “you blew it!”
    So far – dying polar bears – nope.
    ice free arctic – nope
    catastrophic ocean rise – nope.
    increased hurricanes – nope.
    I can’t keep up with all the things that didn’t happen – or were just not related –
    http://www.numberwatch.co.uk/warmlist.htm

  25. Interesting. I wonder what wind speed weilds its influence. I’d guess that it’s related to the rate of vertical lifting.
    Crest level wind speed? That may limit its usefulness outside of mountainous regions. However, if the important component is uplift rate then that makes it generally useful.
    BTW, there’s a very nice snow crystal morphology diagram at http://www.its.caltech.edu/~atomic/snowcrystals/primer/morphologydiagram.jpg and really awesome stuff at http://www.its.caltech.edu/~atomic/snowcrystals/primer/primer.htm

  26. We recently have learned that politicians have acquired the powers to control the weather and our climate.
    The Mayer of Moscow last year promised his City a snow free winter and prepared a big budged for cloud seeding measures.
    http://www.federaljack.com/?p=12228
    It has been all in vain and the reputation of Moscow’s Mayer is severely dented because yesterday Moscow broke it’s all time snowfall record from 1966!
    http://news.bbc.co.uk/2/hi/europe/8529506.stm
    This opens up a new reality!
    If a politician makes a promise about the climate or the weather, expect the opposite of his or hers prediction in extrema.
    Other examples are Global Warming, Sea Level Rise, Glacier- and Ice Cap Melting, Hurricane frequency and intensity, Droughts, extinctions, you name it.
    Let all those politicians be warned!
    Stay away from predictions and promises regarding weather and climate!
    It’s bad for your career!

  27. He was a good professor and an expert in synoptic/mountain forecasting.
    A supporter of AGW, but that’s a must if you’re a department chair.

  28. R. de Haan (20:07:51) :
    All of which underlines the point made before: Weather modification is a disaster looking for a place to happen, a fools errand.
    The weather is the irresistable force, bound & determined to deliver what it wants where it want to dump it.
    Of course it ended in catastrophe. Anyone attempting to stand in front of a freight train is going to get squished like a bug.
    Just do the math: How much energy reaches the Earth every day from the Sun? And man is supposed to intervene with how much power to alter the atmospheric motion of the Earth acting in unison?

  29. Funny they never thought to ask the folks who *make* snow for ski resorts.
    Back in the ’60s, we’d adjust the nozzles (thereby controlling the air velocity) of the snow guns according to the temperature to get wet base or fluffy powder — whichever we needed for a particular slope.
    But then, we were only college kids working over Christmas break — what’d we know?

  30. c james (14:13:25) : Bufkit
    I believe “Bufkit” was developed to help predict Lake-Effect Snow as the wind direction variation is important as to whether an area is missed or gets dumped on. As it worked well it was adopted beyond Buffalo. Sort of like Buffalo Wings, I think.

  31. As the spatial coverage of a forecast increases so does its accuracy. A statement, say, “A 30% chance of rain.” put out by Seattle that covers from the Pacific Coast to the crest of the Cascades with an approaching low off the Ocean is a good bet – it doesn’t have to rain over the entire region, just somewhere in an area from Oregon to B.C. and from sea level to 14,000 feet.

  32. “Sophomore (16:49:30) :
    You can start here: http://tinyurl.com/cliweath
    “Greenhouse gases in the atmosphere behave much like the glass panes in a greenhouse.””
    Sophomore, read Gerlich and Tscheuschner, they explain why it’s not working that way.

  33. Nice to see a simple model that works. However, wind direction also plays a major part in coastal areas and can be as important as wind speed. I hope the UKMet Office adopts this model as current forecasts for snow amounts are very poor.

  34. @ John F. Hultquist (21:28:25)
    Sort of like Buffalo Wings, I think.
    But… but… buffalo’s don’t have wings.
    More on topic, I wonder if they use the Bufkit for forecasting snowfall in my neck of the woods. Living in Michigan’s UP we get a lot of lake effect snow and its amazing how localized it can be. Throw in the fact that we’ve got enough geography around here to get some orographic lift and it adds yet another factor to who gets snow and who doesn’t.
    @ Sophomore (16:49:30)
    Coming from a place of humility is awesome.
    You might try that sometime. And for the record, green houses in cold climates require heaters to keep them warm enough to prevent the plants from dying.
    Charlie K

  35. Charlie K (04:46:01)
    Yes, all the NWS offices in the Great Lakes now use the Bufkit and not just for lake effect either. It is used for all types of winter events as well as severe thunderstorm events.

  36. Charlie K (04:46:01) : But… but… buffalo’s don’t have wings.
    Of course they don’t, we’ve eaten them all… 😉
    So with 50 ish to 75 ish percent predicted by these two variables, was there any hint what the other variables were? Or is it just unknowns?
    I’d expect something like: Temperature, “mixing” where wind is a proxy, and total water percentage. Once you have the heat flow (mass of water vs temp) and the mixing (air at temp as percentage of total mass and any physical effect from rate of mixing / shear ) I start running low on ideas…
    Has anyone asked a Chemical Engineer what parameters drive “snow” formation in commercial operations?

Comments are closed.