Network News

X My Profile
View More Activity
The new Washington
Post Weather website
Jump to CWG's
Latest Full Forecast
Outside now? Radar, temps
and more: Weather Wall
Follow us on Twitter (@capitalweather) and become a fan on Facebook
Posted at 11:00 AM ET, 02/ 6/2009

Models Knew Monster Storm Was No Certainty

By Steve Tracton

Despite hype, 'Groundhogzilla' was never close to sure bet

* Weekend Warm-Up: Full Forecast | When Ants Come Marching In *

In his recent post Andrew Freedman justifiably criticizes the hyping last week and into the weekend by AccuWeather, a Pennsylvania-based private forecasting company, of a potential major east coast snowstorm this past Monday, Groundhog Day. The hype manifested itself in eye-catching headlines that did nothing to acknowledge the ever-present uncertainties in any weather forecast, especially one so complicated. It was a classic case of "cherry picking" the data that supported a significant storm and ignoring all else.

With the benefit of hindsight we know the certainty of a "monster storm" or "Groundhogzilla" -- both terms used in an AccuWeather news release -- was unjustifiable, seeing as the storm dumped a grand total of a dusting to a few inches on the mid-Atlantic and Northeast. But we knew this before the fact, too.

Keep reading for more on the disconnect between AccuWeather's forecast and what computer weather models were predicting...

At times toward the end of last week there was a consensus among the computer models meteorologists use to help predict the weather, which go by acronyms such as "GFS," "ECMWF" and "UKMET," that the AccuWeather forecast could come true. But, there were also times when these models fluctuated wildly, showing a big snowstorm on one run and virtually no snow at all on a run six or 12 hours later (most models are run every six or 12 hours). The fluctuations were enough to drive forecasters batty and send snow lovers on a roller coaster of emotions that alternated between periods of optimism (exaggerated by wishful thinking?) and pessimism (tempered by wishful thinking?).

AccuWeather placed its bets squarely on the big-snow horse, and now finds itself having to eat a very large slice of "humble pie." Importantly, though, I believe that by in large the blame here does not lie with AccuWeather's forecasters, who are no doubt aware and mindful of forecast uncertainties and the pitfalls of overconfidence and hype, but rather the policy dictates of management -- not unlike the relationship between some TV meteorologists and station management.

Nevertheless, it must be emphasized that up until late Saturday and early Sunday there was a very realistic possibility of AccuWeather hitting a home run. -- sometimes even blind squirrels find an acorn. Had the storm occurred as AccuWeather had predicted, you'd likely never hear the end of AccuWeather claims about how wonderfully skillful it had been -- rather than acknowledge the role of old fashion good luck.

The truth is that in a situation as complicated as this particular storm, the basic set of weather models mentioned above is totally inadequate to sample the much larger number of equally possible forecasts.

Models make predictions by, first, processing measurements of the initial state of the atmosphere -- current temperature, wind, pressure, humidity and other data collected across the country and around the world. Then, using mathematical formulas that model the physics and dynamics of the atmosphere, they predict how the atmosphere will evolve over time from its initial state. Equally plausible sets of uncertainties in the initial conditions and/or differences in the algorithms from model to model will produce an array of equally plausible forecasts. Differences between any two forecasts might be too small to be consequential, or large enough to be the difference between a blizzard, drenching rains or a sunny day.

This range of possibilities is an expression of the inevitable uncertainties in all weather forecasts. And, no forecast should be considered complete without conveying that uncertainty, either probabilistically or with some explicit measure of forecast confidence. It all goes back to those pesky butterflies of chaos theory, which inevitably will come back to haunt those who choose to ignore them.


Generating a more complete picture of the full range of possible forecasts is the basis for ensemble forecasting, where a single model is run up to 50 times, each time with a set of slightly different initial conditions (to account for possible inaccuracies in the measured initial conditions). Each of these runs is known as an ensemble member, and together the ensemble members provide a reasonably good portrayal of possible outcomes (e.g., storm track and intensity) and their relative likelihood. The more members that show a given scenario -- for example, a storm hugging the coast and producing significant snows, or instead a storm tracking too far off the coast to produce meaningful precipitation -- the greater the chance that scenario will pan out (at least in theory).

My appraisal of all available ensemble forecasts indicates that from the middle of last week, when the first signal of a realistic possibility for a significant snowstorm became apparent, there was never more than a 50/50 chance of it actually occurring. The chances markedly decreased to less than 10% from Sunday onward. Conversely, an offshore track of a relatively weak low-pressure area -- as turned out to be the case -- was always included in the array of possibilities, and became a 90% plus possibility by Sunday.

Bottom line: There was never a scientifically sound basis to hype the inevitability of a significant snowstorm with this weather system. To their credit CWG and most other weather providers never did.

Note: It was a different system that ended up burying localized areas with an unanticipated 8-12"+ of snow Tuesday night into Wednesday in parts of New Jersey and Pennsylvania, including the Philadelphia area.

By Steve Tracton  | February 6, 2009; 11:00 AM ET
Categories:  Tracton, Winter Storms  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   Del.icio.us   StumbleUpon   Technorati   Google Buzz   Previous: Forecast: Milder Air Moves In; 60 on Sunday?
Next: PM Update: Warmer, But not 'Warm' ... Yet

Comments

AccuWeather wasn't the only group hyping this up. CPC in Camp Springs was also forecasting a "monster" tracking from Mobile to the Delmarva.


That track would have produced more of a windy rain/mix mess [similar to the 1993 Superstorm] than the big snowstorm being hyped by AccuWeather. Towards the weekend, Camp Springs' predicted storm was becoming less and less intense on the charts. It eventually ended up more like the actual storm we got.

Posted by: Bombo47jea | February 6, 2009 12:47 PM | Report abuse

While this was a bad bust for accuweather and a number of other outlets, it pails in comparison with the march 2001 bust. With that storm, most forecasts were calling for 1-2 feet 3 days before the storm. we ended up with an inch or so. I am surprised accuweather seemed to forget about this 2001 bust and went ahead using language on their site that left little wiggle room for other scenarios (such as no snow).

Hopefully accuweather has now learned to use less definitive descriptions in their forecasts.

Posted by: jfva | February 6, 2009 2:56 PM | Report abuse

Bombo47, you're way off with the March 1993 Superstorm. In the metro area here, it produced, not a windy rain/mix, but what was, in effect, almost a winter hurricane.....an 11" blizzard and 60 MPH+ winds from the NE. The storm's warm sector, which crossed FL and then, as usual with winter storms, by-passed us to the east as it moved up the coast, had numerous severe squall lines and tornadoes.

Posted by: MMCarhelp | February 6, 2009 6:56 PM | Report abuse

The imagery invoked with the "butterflies of chaos...(coming) back to haunt those who choose to ignore them" almost makes up for our lack of snow.

Posted by: --sg | February 6, 2009 9:01 PM | Report abuse

The "Superstorm" of March 13, 1993 was different form any other east coast storm in ways never recorded and which you and I will likely never see again- REALLY!

I'll have a post on this on or around the storm's anniversary (non wonkish version).

Posted by: SteveT-CapitalWeatherGang | February 7, 2009 9:59 AM | Report abuse

In personal communications a high profile personality at AccuWeather notes that, just as as there are different model solutions, there is more than one voice at AccuWeather. Indeed, looking back you'll find that at least for this voice at AccuWeather unceratinty about snowstorm possibilities last week was at the forefront of his posts. AccuWeather does encourage eye catching headlines, but I'm told most (obviously not all) temper those in the contents of their stories

My article (and others), this source correctly claims, paints AccuWeather with a one size fits all brush. To this, I can only respond that by far the the most prominent face of AccuWeather in this case was that of Groundhogzilla and the one dominating the blogosphere. Also, I did specifically try to spread the "blame" from Accuweather professional forecasters to management policies (perhaps taken too far with Groundhogzilla (and past occurrences of excessive hype).

Posted by: SteveT-CapitalWeatherGang | February 9, 2009 9:04 AM | Report abuse

The comments to this entry are closed.

 
 
RSS Feed
Subscribe to The Post

© 2012 The Washington Post Company