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, 02/15/2008

Recap: A Fair Forecast, With Flaws

By Capital Weather Gang

By A. Camden Walker and Dan Stillman

Logan Ice Accumulation on Tuesday night
Capital Weather Gang meteorologist A. Camden Walker captures Tuesday evening's deserted, ice-coated Logan Circle.

Sometimes Mother Nature has an uncanny sense of timing, whether it's severe storms on the Fourth of July, a soaking rain or significant snow on a holiday weekend, or most recently this past week's tricky wintry mix on the day of the Potomac Primary. Getting the forecast right is always the goal, but the stakes are even higher when active weather coincides with a circled day on the calendar.

Looking back, our forecasting of what turned into Tuesday's icy mess was fair, but not without flaws.

Our Monday morning forecast was the first to mention a "slight chance" of a wintry mix on Tuesday afternoon for areas north and west of town.

It wasn't until Monday evening at 8:30 p.m. that we warned of an "increasing chance for some iciness in the northern and western suburbs" Tuesday afternoon. By 10 p.m. we were predicting "a wintry mix of sleet, freezing rain and rain" to develop Tuesday afternoon, and that the Tuesday evening commute "could be a bit nasty, especially north and west of DC."

Come early Tuesday morning, our forecast headline cautioned, "Today's biggest threat: Light afternoon ice," and we identified the morning as the best time to vote. We predicted that "some slick spots are definitely possible on sidewalks, bridges, overpasses, and less traveled streets" during the afternoon and early evening for all parts of the metro area, not just north and west. And we noted that "forecast confidence, on a scale from 1 to 10, is about 6.5" due to the complexity of the storm.

This line from a washingtonpost.com story published Tuesday night -- "Officials across the region were surprised by a sheet of freezing rain that blanketed the region just after 3 p.m. " -- was rather perplexing considering that most forecast outlets were highlighting by Tuesday morning the potential for freezing rain later in the day.

While we caught up pretty well with the forecast through early evening, we misjudged (along with most other forecast outlets) how deep into the night temperatures would stay at or below freezing, especially north and west of town where temperatures didn't reach the 32-degree mark until Wednesday morning. It wasn't until around 7 p.m. Tuesday night, when the National Weather Service upgraded its Winter Weather Advisory for the metro area to an Ice Storm Warning, that we alerted "temperatures may remain subfreezing through the early morning hours, especially in the north and west suburbs." Later, our 11 p.m. update was mostly accurate in nailing down the timing of temperature rise.

The main culprit for the delay in rising temperatures?... the phenomenon known as cold air damming, which is when cold air at the surface along the East Coast gets trapped against the Appalachian Mountains. Tuesday afternoon and night, the D.C. area was on the edge of this difficult-to-dislodge wedge of cold air as warmer, moist and less-dense air streamed in from the south, riding up and over the more dense cold air.

We factored the tendency for models to underestimate the magnitude and duration of cold air damming into our forecast, but obviously not quite enough. A wind from the north that was more persistent than expected also helped keep the cold air in place. The difference of just a few degrees resulted in a more treacherous Tuesday evening commute and icier overnight than we expected for some parts of the metro area.

That's our assessment. What's yours?

By Capital Weather Gang  | February 15, 2008; 11:00 AM ET
Categories:  Recaps, Winter Storms  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   Del.icio.us   StumbleUpon   Technorati   Google Buzz   Previous: Forecast: Flip Flopping Through President's Day
Next: CommuteCast: Mild with Clouds Returning

Comments

Because of your forecast I bailed from my office, here at NOAA in Silver Spring, at 1:30 on Tuesday. Along with half the other people in the office. So I was home in McLean well before the ice hit.

Posted by: wiredog | February 15, 2008 11:32 AM | Report abuse

wiredog: Glad our forecast was useful. Thanks for the feedback.

Posted by: Jason, Capital Weather Gang | February 15, 2008 11:39 AM | Report abuse

I also paid attention to your forecast and voted in the morning. Didn't leave work until 7PM, but I live close by and the ride home was fine.

Posted by: Hemisphire | February 15, 2008 11:52 AM | Report abuse

Yes thanks for the head's up, I too left a little early to go vote, and in the time between when I got home safely, it was literally less than two hours later that the street in front of my house was impassable. We live on a hill, and not one of my neighbors could go up or down that hill after 6:30 pm. Didn't stop school from happening the next morning though, but I'll let go of that rant now. Between the poor TV forecasts(remember "any brief iciness north and west will quickly turn to rain" Bob Ryan?) and the VDOT mixing bowl debacle there were plenty of poor performances from those in charge the other day. We have clearly learned nothing from 9-11 about effective real-time communication and decisive action in an emergency...surprise, surprise. This time it was only the weather.

CapWx was on-it though. There's a reason I turn to this site ahead of all the others!

Posted by: Curtis | February 15, 2008 12:01 PM | Report abuse

I think everyone missed badly with the latest storm on a number of fronts. Missing the extent of the CAD and associated ice on Tuesday (as noted), busting on the temperature on Wednesday (predicted highs were in the upper 40s and we were 10 degrees short), then missing on the snow forecast Wednesday night (even CWG said 70% chance of at least a dusting, as I recall).

Glad to see the self-assessment in here, however. It lends a whole lot of credibility to what you're doing, and shows that when it comes to winter weather, we all still have much to learn.

Posted by: Jamie Y (Potomac) | February 15, 2008 12:35 PM | Report abuse

The errors in the forecast were minimal compared to the inability of VDOT to adapt to the variance from forecast to reality. Based in part on your predictions I voted at 11 AM; I left Herndon at 5:30 headed for the Arlington campus of George Mason for my 7:20 class. Wiehle Avenue was an ice rink, as was Old Dominion Drive (I purposely avoided the Toll Road) and by and large people were driving appropriately for the conditions (surprise!). I arrived at GMU at 7:30 only to be told the Professor couldn't get to class because of the weather so I got back in traffic for the drive back to Herndon. All told I was on the road for four hours and only saw one salt truck the entire time, inbound on Old Dominion Drive in McLean.

Posted by: Dave Richardson | February 15, 2008 12:43 PM | Report abuse

any time we have precip-type issues, the forecast will be diffcult. The models will never get the surface exactly right. Experience, and collaboration help, but ultimately when dealing with precip-type/changeover, we will always bust somewhere...overall I think we did well...I know the public wants us to be as deterministic as possible, but I think it is always prudent to emphasize the uncertainty when it is there

Posted by: Matthew Ross, Capital Weather Gang | February 15, 2008 12:49 PM | Report abuse

The real problem is that forecasters never seem to LEARN from underestimating the CAD east of the Appalachians. Those forecasting mistakes will happen again and again and again.....just like they have for decades. Forecasters, so often, believe the models that that show the warm front north of us. At a couple of thousand feet up, that is true....but at the surface, though, that usually doesn't happen. An "S" shaped front......usually stationairy......curves down over the mountains from Western Pennsylvania south and then curves back north again on or near the coast. That is why Norfolk and Ocean City are much warmer than us when there is CAD.

I don't want to sound excessively negative here, but we have heard this time and time again, even from the CapitalWeather staff...."forecasters DO learn from this....you're just being too hard on them". Well, no, they don't. If they DID, the same errors wouldn't be repeated over and over. But history proves that what said was correct.

Posted by: Mike | February 15, 2008 1:07 PM | Report abuse

Mike: Not every cold air damming situation is the same. You can learn from one event and the next event is a different beast entirely. Yes--the models usually overdo the amount of surface warming in these events (any competent forecaster knows that), but the magnitude they vary by differs each time. For one event a model could be 10 degrees too warm, and the next 2 degrees too warm. When a few degrees matter, this is a big deal. You're understating the complexity of forecasting these variable events. There's no simple formula of learning from one event and then you'll get it right forever into the future.

Posted by: Jason, Capital Weather Gang | February 15, 2008 1:24 PM | Report abuse

Forecasting has, however, improved tremendously over the years.

29 years ago(!) when I was in 8th grade we had the Washington's Birthday storm. On the 6 O'Clock news forecasters called for a couple inches. By 11 they were calling for 6 inches.

We got (IIRC) 2 feet. By morning.

Posted by: wiredog | February 15, 2008 1:41 PM | Report abuse

~sg/anyone else,

Where would I go for marine forecast? My boss is going tuna fishing, and needs a link.

Posted by: Kalorama Park | February 15, 2008 1:45 PM | Report abuse

This is the most impressive thing the Cap Weather gang does - critically evaluate their forecasts and open their evaluation to all. It is why I have been a follower of their advice since long before they moved to WaPo. Keep up the good work.

Posted by: David | February 15, 2008 1:49 PM | Report abuse

OK, Jason. I'm not trying to be arrogant, I respect the CapWeather staff, and I know that no two systems are exactly the same. But there is a reason for what I said. I worked in and around NWS for over 30 years before retiring. I've watched many hundreds of these systems move over us during the years, and usually it is more or less the same pattern. A long-wave trough develops west of us, strong SW flow aloft from the Gulf develops, and cold high-pressure sticks in N of us over New England or southern Canada. A surface low forms W of the mountains and moves NE, accentuating the SW flow aloft. It drags the surface warm front north with it W of the mountains, up and over the trapped cold air E of the mountains, and sometimes up along the coast as well. Meanwhile, the Piedmont region never sees the warm air at the surface.....the surface warm front is depressed sometimes as far S as Atlanta. The surface front assumes a classic "S shape" and remains nearly stationairy. It is a situation that the computer models simply cannot forecast with much precision. It is just too-small a scale event in the boundary layer......and forecasters simply have to KNOW that and learn from that. Some do......too many don't.

These systems can be very interesting if the dew points in the warm air are high enough. I've seen it happen in winter, on the satellite shots when you actually had CB tops and severe thunderstorms with the surface warm front in western PA at the same time we were in the trapped cold air getting sleet.

In some cases (again, usually), the system occudes over or to the west of us as the system passes and we go directly from a NE to a NW flow...in others, depending on the vorticity, the energy from the first low west of the ridges is transferred SE to the classic baroclinic zone off the VA/NC coast.....and a new low forms and takes over, disspating the coastal warm front. Then steady precipitation starts and moves NE...that is what happened with out first snowstorm last December.

Posted by: Mike | February 15, 2008 1:52 PM | Report abuse

Mike-- Very nice summary... you are a knowledgeable poster and we appreciate your input

Posted by: Jason, Capital Weather Gang | February 15, 2008 2:02 PM | Report abuse

This is an excellent debate and very informative!

Mike,

You seem very knowldegeable on CAD and make very good points..

Posted by: HEELS | February 15, 2008 2:12 PM | Report abuse

Mike -- Appreciate the discussion, and your analysis is spot-on for a significant percentage of storms that come through here. But there is also a significant percentage that do not fit perfectly into your description. There are plenty of times, given a setup somewhat similar to Tuesday's event, that a strong enough surface wind out of the southeast or east manages to nudge temperatures over the freezing mark for much of the metro area fairly early on in the event.

Behind the scenes, I'll tell you that on Monday evening we identified what we felt would be the determining factor in this storm. I'll quote directly from an email exchange amongst the team that evening:

"Key question is whether 18Z GFS is right with winds kind of light (not much above 5 mph) and mainly from the east, which would have trouble pushing out cold air at the surface. Or if there's any validity to 12Z ETA, which is more southeasterly with winds and at closer to 10 mph, which would more easily scour the cold air out."

(For those who are curious, GFS and ETA are the names of the two main models used to forecast U.S. weather. 18Z is 1 p.m., 12Z is 7 a.m.)

In the end, we correctly leaned more toward the model predicting lighter winds. We just didn't lean enough. Only time will tell if we can carry any lessons over to future storms that may be similar in nature, but probably not exactly the same.

Posted by: Dan, Capital Weather Gang | February 15, 2008 2:20 PM | Report abuse

PS: Agree with Jason -- appreciate your clearly knowledgeable input and am glad to have you part of the discussion here.

Posted by: Dan, Capital Weather Gang | February 15, 2008 2:25 PM | Report abuse

This was indeed a difficult forecast and the CapitalWeather team did better than most with it's predictions. Being completely objective however your forecast from 11:40 a.m. on Tuesday was as follows:

"By mid afternoon, a very light mix of snow, sleet, freezing rain and rain could develop in the southern suburbs. This precipitation should move into the rest of the area by late afternoon and evening hours. At that time, temperatures should be at or above freezing inside the beltway -- so major travel issues are not anticipated. However, north and west of central Fairfax and central Montgomery counties, below freezing temperatures may cause some slick spots particularly on bridges, overpasses, ramps and sidewalks."

Clearly, even a few hours before the event materialized you didn't anticipate the hazardous conditions that would develop in DC and the inner suburbs. This doesn't mean we should let the local highway crews off the hook. Being responsible for public safety; they should always err on the side of caution.

Posted by: 2BFair | February 15, 2008 3:10 PM | Report abuse

The key to this event IMHO was the stronger than usual intensity of the cold air mass, supported by the circulation around a very strong low pressure area to the northeast, over eastern Canada.

Posted by: Steve, Capital Weathe Gang | February 15, 2008 5:04 PM | Report abuse

Kalorama Park: For marine forecasts, go to http://www.erh.noaa.gov/lwx/marine.htm and click on the segment corresponding best to the area needed.

Posted by: ~sg | February 15, 2008 5:35 PM | Report abuse

I love this website! Keep up the good work.

Posted by: Susanna | February 17, 2008 7:18 AM | Report abuse

The comments to this entry are closed.

 
 
RSS Feed
Subscribe to The Post

© 2012 The Washington Post Company