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 10:00 AM ET, 03/ 3/2009

Post Mortem: Successful Forecast With Its Flaws

By Jason Samenow

* Full Forecast through the Weekend | Recap/Photos | Submit Your Photos | Chat Transcript | Deicing Tips | Winter Weather Driving Tips *

* View/Report Snow Totals - Map | Snow Totals from NWS *

* CWG on: Twitter & Facebook *

During the course of the storm and in its wake, there has been much debate about how well our forecast panned out. Our independent "Weather Checker" (i.e. our weather ombudsman), Jamie Yesnowitz, will weigh in with his own thoughts on this tomorrow. But I'll kick things off with a self-assessment reflecting on what I think went well and went wrong.

The short version of the assessment is that west of I-95 we overestimated the impact of the first half of the storm and slightly underestimated the impact of the second half of the storm (southeast of I-95 forecast we were close to being accurate throughout). But when all was said and done, our initial forecast and impact projections were spot on (pretty much everywhere) -- notwithstanding a modestly flawed mid-storm adjustment for areas west of I-95.

Keep reading for details about our successes and mistakes in forecasting the storm and vote for what grade you'd give the Capital Weather Gang's forecast...


A snowy scene at the Capitol yesterday. By Capital Weather Gang photographer Kevin Ambrose.

This storm snuck up on a lot of people who considered winter mostly over, especially in light of the fact March had arrived and we had received a meager 2 inches of snow to date. However, CWG's Matt Rogers first mentioned the threat of wintry weather a week ago today. We specifically noted the possibility of a coastal storm on Thursday, but cautioned there was a "decent" chance it would go out to sea. By Friday, the threat of a storm started to seem legitimate, so we issued a Snow Lover's Crystal Ball that spoke of the potential of two weekend waves of wintry weather -- though we indicated the second (but ultimately the more important) was a low probability.

Early Saturday morning (5 a.m., but post was updated at 11 a.m.) we started sounding the alarm bells noting a "[m]oderate risk of significant snow accumulation." Late Saturday morning, in a comprehensive assessment of storm possibilities , we went on the record with a preliminary accumulation estimate of 3-6" with more east -- which was pretty close to on target. As far as I know, no other local forecasting outlet in the area identified the threat as early as we did nor provided as early and accurate of an accumulation outlook. Although I give props to AccuWeather's Joe Bastardi who suggested a 6-12" storm was possible for D.C. north on Friday afternoon.


In our storm timeline, we indicated snow would begin between 1 and 4 p.m. and begin accumulating by early evening. In reality, snow didn't arrive until between 4 and 5 p.m. east and south, and until between 5 and 6 p.m. in the metro area. Also, the snow was mixed with sleet and rain in spots, which we didn't specifically call for. Most importantly, little meaningful accumulation occurred until after midnight along and west of I-95 even as areas 10 miles or more east of I-95 received several inches like we called it.

What happened was that the initial surge of moisture got eaten away by dry air coming in from the north. So even though radar showed precipitation surging through central Virginia Sunday afternoon, it hit a wall north and west Fredericksburg, Va. Then, even as the precipitation moved in, it was so light that some of the snowflakes forming in the cold air aloft melted near the surface (where temperatures were above freezing) causing a mix of rain/sleet and little accumulation. Had it been precipitating harder (like it was to the east), the air would've cooled via evaporation, we wouldn't have had rain, and the initial snow would've started sticking earlier. None of the models we use as a tool in forecasting predicted the dry air intrusion until it was practically happening. In the future, this is something we'll have to analyze more closely for an incoming storm of this sort.


Around 9:15 p.m., when the snow had temporarily stopped from the District and points west and not accumulated more than a sloppy half inch or so, we decided to lower accumulations for those areas. Because we had expected at least a couple inches of snow in these spots by then and the radar to the west and northwest was underwhelming, it seemed like the obvious decision. We didn't throw in the towel entirely though, noting the redevelopment of snow to the southwest as well as the likelihood of more snow late at night associated with an "energetic upper low." But we did drop accumulation estimates to 2-5" (from 4-8") in the immediate metro area and 1-3" north and west of the Beltway (from either 2-4" or 4-8"). That downward adjustment of 2-3 inches was unnecessary.

As it turned out, the "energetic upper low" (the same feature had produced thundersnow in Atlanta) dropped a couple more inches than even we expected. That made up for the lack of snow along and west of I-95 early on.

The lesson here is unclear. If we had stuck to our guns, our forecast would've been more successful. However, if for some reason the second phase of the storm had underperformed, the forecast would have been a horrendous bust.


Accumulation map issued by CWG before the storm. Compare with observed snow amounts.

If you look back at our post from 11 a.m. Sunday -- which basically served as an "executive summary" of what to expect -- it was on target save the second bullet ("snow to start between 1 and 4 p.m. this afternoon, and may quickly become heavy"). And the storm impact scales (StormCast, TravelCast, SchoolCast, and FedCast) provided in the preceding post verified nicely.


Overall, I'd give our performance a B (on an A-F scale): we were onto the storm early, got the big picture right but missed on some important details in the storm's evolution. We'll see if the Weather Checker agrees tomorrow...

By Jason Samenow  | March 3, 2009; 10:00 AM ET
Categories:  Capital Weather Gang  
Save & Share:  Send E-mail   Facebook   Twitter   Digg   Yahoo Buzz   StumbleUpon   Technorati   Google Buzz   Previous: Forecast: Freezing Today, Then Steadily Warmer
Next: PM Update: Some More Cold Before Moderation


I think you guys did a great job - I gave you an A. To those who gave you anything less (including yourselves) I say, stop splitting hairs!

So, what's with the inconsistent snow totals in this morning's Post? In the Metro section article, it states in text that the total at National was 5.6 inches. On the map it says 7.0 inches. On the graph showing the annual totals of the last nine years, it says so far National has had 11.5 inches which can't be right because in the text it says National had only gotten 2 inches before this storm. Neither 5.6 + 2.0 nor 7.0 + 2.0 add up to 11.5. I'm not nit-picking because I care about the exact amount of snow we got (if it were up to me, I'd just say the area generally got shin-high totals and call it a day!). But you gotta wonder - is anyone at the Post fact-checking?!

Posted by: paris7 | March 3, 2009 10:45 AM | Report abuse

The upper level low basically saved the forecast for DC and the western suburbs. What I found most interesting and frustrating about this storm was how the first surge of moisture hit a wall near DC, then slid east of town. I had also noticed that the surface low near the SC coast actually weakened during that time which did not help us. After watching the radar and satellite loops Sunday, I really thought the first surge of moisture would over-achieve, yet I only received a dusting of snow in Central Fairfax County. Later, the upper-level low later gave me 7" Monday morning, much more than I expected. My lesson learned is that nowcasting a storm can be as difficult as forecasting a storm.

Posted by: Kevin-CapitalWeatherGang | March 3, 2009 10:46 AM | Report abuse

For me, it's not crucial whether CWG hit the mark on snow totals. I don't have an expectation that you can foresee the future. My expectation is that you consider the information available to you, present it to your readers in a clear and thoughtful way, and maintain a high level of transparency. I'll add that the transparency is one of of things that really sets CWG apart from the vast majority of weather sources. I don't want to just be told that we're likely to get 4-8 inches of snow--I want to know why you made the forecast that you did, what are the factors that will affect the storm, how is the storm developing relative to predictions--all the technical background that goes into the forecast.

So with that said, I'd give CWG high marks for this storm. Your initial forecast fit the information you had in front of you. You adjusted your forecast as the situation changed. And you let us know why at each step of the process. So kudos on a good job all around.

Posted by: ElJocko | March 3, 2009 10:57 AM | Report abuse

Y'all did just fine. One thing I've learned living in the DC area on and off for the better part of 20 years (including the Veterans Day Storm of '87 my freshman year in college which totally drove me off the deep end in terms of being a snow nut) is that this area is MADDENINGLY difficult for forecasters to predict. But I suppose if y'all wanted an easier job, you could move to Boston... :-)

Seriously, bravo zulu and keep up the good work with this blog -- I thought I'd cured my weather addiction until I first saw it six months ago via the Nats/UnitedCasts...

Posted by: Juan-John | March 3, 2009 11:16 AM | Report abuse

BTW, how much props can you give AccuWeather? They're always pimping big snows, whether or not it happens. A stopped clock is right twice a day, too.

Posted by: ah___ | March 3, 2009 11:24 AM | Report abuse

This was one of the most unusual and at times frustrating and fascinating storm evolutions I have ever witnessed.

I can never remember radar returns dropping out as quickly as Sunday evening as heavy snow in sw Va. moved to within 40 miles of my location and died.

It is also very unusual for an upper level bundle of energy to overproduce as late Sunday night witnessed. I observed near blizzard conditions from 3-5 am Monday as 30-40 mph winds combined with heavy snow and 18 degree temps to create 3-5 ft. drifts. These drifts will be here for a while.

I believe the CWG performance overall was great, but there are some remaining mysteries about this storm that may never be completely understood. My personal forecast to many friends in my area at 7 am on Sunday was for an average total of 3-6 inches when the NWS was calling for 1-3. I feel very good about the 5 inches received. I don't talk much about the FACT that I didn't have a clue how we would arrive at that 5 inch total!!!

These always possible, totally unexpected and impossible to predict twists are what has always fascinated me about storms, whether a hurricane, thunderstorm or snowstorm.

Posted by: AugustaJim | March 3, 2009 11:54 AM | Report abuse

Kevin -- Totally agree with you on nowcasting being as difficult as forecasting. I'd go even further as to say that sometimes (not all the time) too much nowcasting too quickly can be dangerous. As quickly as something changes during a storm, like temperatures not dropping as quickly as anticipated or a surface low weakening -- things can change right back the other way just a few hours later. The challenge is to recognize when a real-time development in the storm is significant enough to justify deviating from the original forecast.

In this case it was a really hard call whether to knock down totals after the first part of the storm was so underwhelming, as the upper-level low could just as easily have weakened too much to give us the accumulations we ended up getting.

Posted by: Dan-CapitalWeatherGang | March 3, 2009 12:30 PM | Report abuse

i think you guys did a great job with this storm!

Posted by: madisondc | March 3, 2009 1:07 PM | Report abuse

I give CWG a B+...the final amounts were rather close to the original forecasts. A few details were overlooked.

My rant is reserved for the meteorological models and especially the TV weathercasters. The models kept jumping all over the place during the 72-hour period ending yesterday. It would help if the models stayed more consistent. I'd give the various models a C, possibly a C-minus.

The TV meteorologists got a collective D in my opinion. Many of them were downplaying the storm as early as Saturday evening. This continued through Sunday night, and there was some surprise Monday morning as the final burst moved through. A good rule of thumb when working an intense storm is to never say "done" until the upper-air system has moved through. Generally the upper-air low trails the surface low and will tend to drag out the precipitation, sometimes for 2 or 3 days after the surface low passes. Much of this late precipitation falls in the cold sector.

The Weather Channel gets a C, outperforming some of our local weathercasters, while my best grade, an A-minus, goes to the folks in Camp Springs and Sterling who do the extended forecast discussions. These pros had something big in the works off the Middle Atlantic coast up to a week in advance.

Posted by: Bombo47jea | March 3, 2009 1:10 PM | Report abuse

B+ on my scale, not sure I could have said it any better then El Jocko did above with his commentary, especially when he stated

"I want to know why you made the forecast that you did, what are the factors that will affect the storm, how is the storm developing relative to predictions--all the technical background that goes into the forecast."

That is always what interested me most about this site..and has kept me coming back. While I only maxed out at around 2 inches or so, the excitement of following the storm was there. Nice work guys!

Posted by: MikefromtheBlueRIdge | March 3, 2009 1:33 PM | Report abuse

I was proud of my good old Capital Weather team! For those of us that have been with you from the start, your coverage was by far the best over the years that I have been supporting you. Great job and may next winter be brutal!

PS - I will never fall for the Squirrel eating pumpkins = bad winter theory ever again.


Posted by: stinkerflat1 | March 3, 2009 4:46 PM | Report abuse

Strong B+ in my book. Well done.

I'll add that ElJocko expressed my interest in your site way better than I ever could.

I look forward to reading about spring thunderstorms soon.

Posted by: CapHillEast | March 3, 2009 7:30 PM | Report abuse

Its funny, the one thing I thought you guys missed on was identifying the threat in advance.

I can remember in storms past, you guys hinted sometimes strongly, up to a week in advance "somethings coming". Further, I know the SCLB can only make its appearance 48 hours before an event, but in past events, you guys were pretty much on top of it when the first SCLB came out. This time, as you note, your first SCLB pretty much dismissed the sunday main event.

To be fair, I dont think any site or service did a good job of seeing this thing in advance - even HM over at accuhype was not in full hype mode on this storm til weeks end.

Everything else about your forecast was good, and you all are still my #1 go to site. Keep up the good work!

Posted by: fishman1 | March 4, 2009 4:09 PM | Report abuse

The comments to this entry are closed.

RSS Feed
Subscribe to The Post

© 2012 The Washington Post Company