Lead forecast times before our 12"+ snowstorms

[gymengineer]

Clearly forecasting has gotten much better with the improvements in models, so the lead times in forecasting major (widespread 12"+ for east, west, or all parts of the area) snows have gotten longer even within the past decade. Here's a summary of lead times based on memory and a bunch of other sources for the storms since 1990:

 

3/93- Early watch out more than 30 hours from onset; Friday morning talk, 6-12" was the general range thrown around; Friday PM- blizzard warnings raised with >1' amounts on the higher end entering every forecast; snow starts near midnight.

 

1/96- Early signs of the storm showed up by 1/2 in the medium range models. The potential for a major east coast snowstorm was flagged by Thursday, 1/4. Overnight runs suggested a suppressed storm, so during the day Friday, talk was of a VA bulls-eye. Then, the infamous ETA run came out Friday evening, spurring Bob Ryan's: "This is likely to be a top-3 snowstorm for DC" on the 11 pm newscast Friday night. By Saturday morning, every forecast was for that historic snow. The snow started Saturday late evening in the region. 

 

1/00- Monday morning, 1/24, forecasts were still focused on Wednesday, 1/26 as the day to watch for a possible snowstorm. By the 4 and 5 pm newscasts, local TV mets (Doug Hill, for example) were noticing the heavy precip in central SC and promised to keep an eye on the low even though we weren't forecast to get more than flurries. LWX issued a WWA for the SE MD counties for the afternoon package. The WSWarning came out just before the 10 pm FOX 5 newscast. Snow started just a few hours later. 

 

2/03- Help me fill this one in? I was out of town for the lead-up to it, but knew by Thursday (2/13) that there would be a large storm for DC, and by Friday evening-- right before the first wave started, it was clear this was going to be a historic storm starting early Sunday AM. 

 

2/06- For a storm that really got going Saturday overnight for the immediate area, by Friday mid-day there was confidence for a major snowstorm. But, like so many other storms when forecasters look at QPF outputs verbatim, the heaviest band was forecast as of Friday night for south and east of DC, with less heading north and west. The forecast amounts increased Saturday morning for the metro area, only to have some forecasters pull back dramatically Saturday evening as the transition to accumulating snow was slower than anticipated in some areas. (This thread that MN Transplant re-located captures the angst: http://www.easternuswx.com/bb/index.php?/topic/83822-vadcmdwv-feb-1112-mecs-thread/)

 

12/09- This is the storm I have the most questions about--- What was it about this storm's set-up that the model consensus could not nail down a historic event until within 24 hours? Given the classic DC major storm look in the ensemble means well ahead of time, I would think the models would have locked in well ahead of time like for the subsequent 2/5-6/10 storm.

 

30 hours before onset is our current expectation for locking down the "historic" label for a storm, but it took until 12-18 hours before onset for the lock in this storm. Look at this article from CWG as of Thursday, 12/17 at 6:30 pm. The NWS had recently issued a WSWatch (just a bit over a day before the onset of the storm), and here was the CWG's take:

 

Accumulation Totals: Uncertainty about snow totals still exists, but the potential is growing for a major event. Here's our current thinking on accumulation chances (though as any D.C. area veteran knows, with more than 24 hours until the storm's arrival, there's still time for these numbers to shift in either direction)...

15%: Less than 1"
15%: 1-3"
20%: 3-6"
30%: 6-12" (most likely as of now)

20%: 12"-24" 

 

2/5-6/10: Don't need much comment about this one--- huge success in medium-range forecasting with the potential identified a week ahead of time and the model consensus of a major storm coming days ahead-- even as the Tuesday night-Wednesday morning storm was still ongoing. The HPC all the way back from 1/31: 

VERY HEAVY PRECIPITATION EVENT SEEMS LIKELY FOR THE MID ATLANTIC REGION SATURDAY VERY POSSIBLY INTO SUNDAY SPREADING NORTH AND NORTHEASTWARD. A SIGNIFICANT WINTER WEATHER THREAT IS POSSIBLE. HIGHER CONFIDENCE IN THE STRONG ONSHORE WIND CONSENSUS FOR A PROLONGED PERIOD OF TIME ALONG THE SOUTHEAST AND MID ATLANTIC SEABOARD WITH BEACH EROSION/COASTAL FLOODING FRIDAY INTO SUNDAY.

 

2/9-19:10: Even as snow was winding down on 2/6, the next event started to look more and more significant. From CWG on Sunday, 2/7, almost 48 hours before the arrival of the storm:

 

The next storm: Precipitation, primarily snow as it looks now but maybe mixed with sleet at the start, is on track to arrive Tuesday afternoon or evening. Snow, or briefly a snow/sleet mix, may be on the light side at first before probably becoming moderate to heavy overnight into Wednesday.

At least 4 inches is looking like a decent bet, with 8 inches or more a possibility. Here are our accumulation probabilities, which are likely to change as the storm gets closer:

15%: Less than 2"
25%: 2-5"
35%: 5-8"

25%: 8"+ 

 

2/12-13/14: We all still remember this progression...

 

 

 

So, what stands out most to me is 12/09, firmly in our new era of really good modeling. This was a storm in an amazing pattern with blocking. What caused various models to pulse between having it and losing it and then keeping it southeast focused before finally converging?

[gymengineer]

And looking back at the 12/18-19/09 remembrance thread, Wes was excited for the prospect of a major snowstorm for our area well ahead of the consensus time citing the 72-hr ensemble mean and analogs like 2/83 showing up a week ahead.

From the easternwx threads, it looks like models *were* converging on a historic storm on the 0Z 12/17 (Thursday) runs.... I guess many were just a bit gun-shy to fully go with it.  

[anotherman]

Interesting read, thanks for posting.

[HighStakes]

Nice post. Thanks.

[WxUSAF]

My memory is that 12/09 was well-modeled from a week out.  I remember us talking about it for at least a week on Eastern and telling people about it several days before I would normally notify friends/family/coworkers. 

[gymengineer]

My memory is that 12/09 was well-modeled from a week out.  I remember us talking about it for at least a week on Eastern and telling people about it several days before I would normally notify friends/family/coworkers. 

It was a classic case of looking great a week out and then the GFS, etc., losing the storm the week-of, only to bring it back in the shorter range. The period of most interest to me is Wednesday through Thursday (12/16-12/17). A typical storm of that magnitude would have seen a watch issued much earlier than Thursday PM and the CWG, for example, identifying the historic aspect of the storm certainly by 30 hours before onset. 

[mitchnick]

My memory is that 12/09 was well-modeled from a week out.  I remember us talking about it for at least a week on Eastern and telling people about it several days before I would normally notify friends/family/coworkers. 

hmmmm....not a week

Monday was a minor threat as I recall with only the JMA and JB calling for it to come up the coast

things slowly improved and by Wednesday 18Z, the NAM started to go crazy and the other models followed (at least the way I remember it)

[mitchnick]

I found this pic on my conputer

it's the 120 hr. 12Z GFS on 2/1/10

everyone remembers the computers nailing this storm from far out, but this suggests otherwise

 

[gymengineer]

I found this pic on my conputer

it's the 120 hr. 12Z GFS on 2/1/10

everyone remembers the computers nailing this storm from far out, but this suggests otherwise

Maybe it was more that the "the board" had learned the lesson from 12/09-- when Wes gets really excited by a pattern and its associated primary analogs (again 2/83 in this event), it's ok to get totally excited even if the GFS isn't showing it verbatim yet

[MDstorm]

Having followed many winter storms through the years, some of the good mets have taught me that patterns and analogues help identify threats at longer distances. Then models variably latch on to the threats, but are prone to "wobble" as small changes aloft can lead to more significant surface changes. Thinking about following threats from a distance reminds me of a guy named Gary Gray. Some of the older timers here will remember him from Millennium Weather ( 1990's-2005). He would break down in vivid detail every model run dealing with a threat from the time it was identified till the time it ended. Back then, the Euro was only run once day. However, there was still plenty of model data to keep him busy. Reading his play by play of the models, even for the big storms, revealed enough differences in the details to keep things interesting up to game time. I see the same things with more recent storms-- hence the board tension with every model run. Models may be statistically better , but with actual sensible weather forecasting, experience and even instinct are invaluable ( even for the big storms ).

MDstorm

[gymengineer]

Having followed many winter storms through the years, some of the good mets have taught me that patterns and analogues help identify threats at longer distances. Then models variably latch on to the threats, but are prone to "wobble" as small changes aloft can lead to more significant surface changes. Thinking about following threats from a distance reminds me of a guy named Gary Gray. Some of the older timers here will remember him from Millennium Weather ( 1990's-2005). He would break down in vivid detail every model run dealing with a threat from the time it was identified till the time it ended. Back then, the Euro was only run once day. However, there was still plenty of model data to keep him busy. Reading his play by play of the models, even for the big storms, revealed enough differences in the details to keep things interesting up to game time. I see the same things with more recent storms-- hence the board tension with every model run. Models may be statistically better , but with actual sensible weather forecasting, experience and even instinct are invaluable ( even for the big storms ).

MDstorm

Gary Gray's posts leading into the 1/96 blizzard still make me smile.  He was the internet guru for model interpretations back in the day...

[HighStakes]

hmmmm....not a week

Monday was a minor threat as I recall with only the JMA and JB calling for it to come up the coast

things slowly improved and by Wednesday 18Z, the NAM started to go crazy and the other models followed (at least the way I remember it)

I think you're right. I remember watching a JB Monday morning and he was convinced a storm was coming up the coast despite no model support and he explained why. Not the first time he called for something like that lol. GFS had nothing on it's 12z run following the JB's video. I'm pretty sure the Euro showed nothing as well early in the week. There were still a lot of questions as to snow amounts and how far north the low would go as late as Friday morning.

[HighStakes]

Having followed many winter storms through the years, some of the good mets have taught me that patterns and analogues help identify threats at longer distances. Then models variably latch on to the threats, but are prone to "wobble" as small changes aloft can lead to more significant surface changes. Thinking about following threats from a distance reminds me of a guy named Gary Gray. Some of the older timers here will remember him from Millennium Weather ( 1990's-2005). He would break down in vivid detail every model run dealing with a threat from the time it was identified till the time it ended. Back then, the Euro was only run once day. However, there was still plenty of model data to keep him busy. Reading his play by play of the models, even for the big storms, revealed enough differences in the details to keep things interesting up to game time. I see the same things with more recent storms-- hence the board tension with every model run. Models may be statistically better , but with actual sensible weather forecasting, experience and even instinct are invaluable ( even for the big storms ).

MDstorm

His blogs leading up to PDII were classic. Basically gave a detailed breakdown of every model run as you said. He used to say he was short on time but still would have these incredibly long posts.

[midatlanticweather]

As much as you are trying to gain from this, I gotta tell ya, these are great memories.

[MDstorm]

His blogs leading up to PDII were classic. Basically gave a detailed breakdown of every model run as you said. He used to say he was short on time but still would have these incredibly long posts.

Right you are. I also remember how he would be posting a model run and invariably a new piece of data would be "rolling in" during his post. He was a true one man show. His model discussions really highlighted the run to run variability that existed.

MDstorm

[jacindc]

Right you are. I also remember how he would be posting a model run and invariably a new piece of data would be "rolling in" during his post. He was a true one man show. His model discussions really highlighted the run to run variability that existed.

MDstorm

Unfortunately the Internet Archive only got this from February 2003 (actually, late January). At least it's an example of Gary's work. I followed him religiously.

https://web.archive.org/web/20030201095329/http://www.millenniumweather.com/winter/cast2.html

You can click the bar at the top to look at other dates.

[Deck Pic]

Great thread.  My hazy recollection of PD2, was that it was pretty well forecast, but with the typical incremental bumps up.  I believe early in the week, there were indications that a can't miss massive complex producing prodigious amounts of precipitation was going to run into a cold dome.The modeled QPF outputs were monstrous.I think there were questions about precip type, and also suppression.  I think by 2/12-13 we had an idea of the evolution.  That this tongue was going to produce a frontrunning rain to snow but miss central VA, and then the motherlode would start around midnight.  The ETA was producing insane amounts of precip at that point.  I was at a bar in Dupont the night of the 14th and it was raining and not particularly lightly either, but the models were locked in and nothing to worry about.  I slept in Old Town and when i woke up on Saturday morning there was snow on the ground, but only patchy on the grass and nothing special.  On the news Saturday evening, Bob Ryan looked like the country had just been attacked as he went 18-24".  Topper also went similarly. Maybe 16-24"?  Doug Hill obstinately went 8-14" and was overly defensive about it.  That there would be mixing and that the setup wasn't conducive to our biggest snows.  I stayed up all night Saturday night north of Dupont, and there was no "entry period".  5 minutes after the snow started, we were absolutely ripping.   So I believe it was well forecast early on.  Of course the 40N jog wasn't well forecast at all.

[high risk]

    For PD2, I have a number of the Eta QPF maps saved.   Here is the total precip from the 00z Friday run

 

              That's pretty decent lead time.    Here is the same from 12z Saturday:

 

[Deck Pic]

great maps...question..I don't understand the text at the bottom...If the 1st map is the 0z Friday night run, how can you have 66 hour precip 18 hours later?

 

EDIT - Is it is blend of some sort of 48-66 hour precip?...hmmm....can't figure it out

[Eskimo Joe]

Clearly forecasting has gotten much better with the improvements in models, so the lead times in forecasting major (widespread 12"+ for east, west, or all parts of the area) snows have gotten longer even within the past decade. Here's a summary of lead times based on memory and a bunch of other sources for the storms since 1990:

 

3/93- Early watch out more than 30 hours from onset; Friday morning talk, 6-12" was the general range thrown around; Friday PM- blizzard warnings raised with >1' amounts on the higher end entering every forecast; snow starts near midnight.

 

1/96- Early signs of the storm showed up by 1/2 in the medium range models. The potential for a major east coast snowstorm was flagged by Thursday, 1/4. Overnight runs suggested a suppressed storm, so during the day Friday, talk was of a VA bulls-eye. Then, the infamous ETA run came out Friday evening, spurring Bob Ryan's: "This is likely to be a top-3 snowstorm for DC" on the 11 pm newscast Friday night. By Saturday morning, every forecast was for that historic snow. The snow started Saturday late evening in the region. 

 

1/00- Monday morning, 1/24, forecasts were still focused on Wednesday, 1/26 as the day to watch for a possible snowstorm. By the 4 and 5 pm newscasts, local TV mets (Doug Hill, for example) were noticing the heavy precip in central SC and promised to keep an eye on the low even though we weren't forecast to get more than flurries. LWX issued a WWA for the SE MD counties for the afternoon package. The WSWarning came out just before the 10 pm FOX 5 newscast. Snow started just a few hours later. 

 

2/03- Help me fill this one in? I was out of town for the lead-up to it, but knew by Thursday (2/13) that there would be a large storm for DC, and by Friday evening-- right before the first wave started, it was clear this was going to be a historic storm starting early Sunday AM. 

 

2/06- For a storm that really got going Saturday overnight for the immediate area, by Friday mid-day there was confidence for a major snowstorm. But, like so many other storms when forecasters look at QPF outputs verbatim, the heaviest band was forecast as of Friday night for south and east of DC, with less heading north and west. The forecast amounts increased Saturday morning for the metro area, only to have some forecasters pull back dramatically Saturday evening as the transition to accumulating snow was slower than anticipated in some areas. (This thread that MN Transplant re-located captures the angst: http://www.easternuswx.com/bb/index.php?/topic/83822-vadcmdwv-feb-1112-mecs-thread/)

 

12/09- This is the storm I have the most questions about--- What was it about this storm's set-up that the model consensus could not nail down a historic event until within 24 hours? Given the classic DC major storm look in the ensemble means well ahead of time, I would think the models would have locked in well ahead of time like for the subsequent 2/5-6/10 storm.

 

30 hours before onset is our current expectation for locking down the "historic" label for a storm, but it took until 12-18 hours before onset for the lock in this storm. Look at this article from CWG as of Thursday, 12/17 at 6:30 pm. The NWS had recently issued a WSWatch (just a bit over a day before the onset of the storm), and here was the CWG's take:

 

Accumulation Totals: Uncertainty about snow totals still exists, but the potential is growing for a major event. Here's our current thinking on accumulation chances (though as any D.C. area veteran knows, with more than 24 hours until the storm's arrival, there's still time for these numbers to shift in either direction)...

15%: Less than 1"

15%: 1-3"

20%: 3-6"

30%: 6-12" (most likely as of now)

20%: 12"-24" 

 

2/5-6/10: Don't need much comment about this one--- huge success in medium-range forecasting with the potential identified a week ahead of time and the model consensus of a major storm coming days ahead-- even as the Tuesday night-Wednesday morning storm was still ongoing. The HPC all the way back from 1/31: 

VERY HEAVY PRECIPITATION EVENT SEEMS LIKELY FOR THE MID ATLANTIC REGION SATURDAY VERY POSSIBLY INTO SUNDAY SPREADING NORTH AND NORTHEASTWARD. A SIGNIFICANT WINTER WEATHER THREAT IS POSSIBLE. HIGHER CONFIDENCE IN THE STRONG ONSHORE WIND CONSENSUS FOR A PROLONGED PERIOD OF TIME ALONG THE SOUTHEAST AND MID ATLANTIC SEABOARD WITH BEACH EROSION/COASTAL FLOODING FRIDAY INTO SUNDAY.

 

2/9-19:10: Even as snow was winding down on 2/6, the next event started to look more and more significant. From CWG on Sunday, 2/7, almost 48 hours before the arrival of the storm:

 

The next storm: Precipitation, primarily snow as it looks now but maybe mixed with sleet at the start, is on track to arrive Tuesday afternoon or evening. Snow, or briefly a snow/sleet mix, may be on the light side at first before probably becoming moderate to heavy overnight into Wednesday.

At least 4 inches is looking like a decent bet, with 8 inches or more a possibility. Here are our accumulation probabilities, which are likely to change as the storm gets closer:

15%: Less than 2"

25%: 2-5"

35%: 5-8"

25%: 8"+ 

 

2/12-13/14: We all still remember this progression...

 

 

 

So, what stands out most to me is 12/09, firmly in our new era of really good modeling. This was a storm in an amazing pattern with blocking. What caused various models to pulse between having it and losing it and then keeping it southeast focused before finally converging?

 

RE:

 

1.)  Anybody have a .gif or link to that run?  I have never seen that before.

 

2.)  I remember there was considerable amount of discussion about the climatology of big (Read:  10" + snowstorms) in December across the Mid Atlantic.  Also, there was herculean blocking downstream and some skeptics were prudent to think the storm would not come together as it thankfully did.  

[Symblized]

Unfortunately the Internet Archive only got this from February 2003 (actually, late January). At least it's an example of Gary's work. I followed him religiously.

https://web.archive.org/web/20030201095329/http://www.millenniumweather.com/winter/cast2.html

You can click the bar at the top to look at other dates.

 

Speaking of Gary's model-watching and forecasts, there's also this in regards to 1/96.

 

http://wow.americanwx.com/gg/

[high risk]

great maps...question..I don't understand the text at the bottom...If the 1st map is the 0z Friday night run, how can you have 66 hour precip 18 hours later?

 

EDIT - Is it is blend of some sort of 48-66 hour precip?...hmmm....can't figure it out

 

    good question.    it was some sort of strange way that the buckets were kept in the gempak files when trying to sum odd (i.e. 66 in this case) time periods.   To get a 66 (and don't ask me why I chose 66), I had to sum a 48 total and then an 18.  So the first map is the 48 hour total valid at 00z on the 16th plus the 18 hour total valid at 18z on the 16th.

[RodneyS]

1/00- Monday morning, 1/24, forecasts were still focused on Wednesday, 1/26 as the day to watch for a possible snowstorm. By the 4 and 5 pm newscasts, local TV mets (Doug Hill, for example) were noticing the heavy precip in central SC and promised to keep an eye on the low even though we weren't forecast to get more than flurries. LWX issued a WWA for the SE MD counties for the afternoon package. The WSWarning came out just before the 10 pm FOX 5 newscast. Snow started just a few hours later. 

First, great post!  Second, the above was one of the lesser moments in DC forecasting history.  I distinctly remember hearing on WTOP around 7 PM on the evening of Monday, January 24th, that there could be "flurries" on Tuesday.  As I was getting ready for bed Monday evening, Sue Palka led the Channel 5 news with breaking news about a major snowstorm heading our way. However, the Office of Personnel Management did not take the updated forecast seriously, with the Federal Government opening on time on Tuesday.  Big mistake.  I lucked out, as my bus never arrived that morning.  So, I trudged back home through the snow and discovered that the Government had just closed.  The final snow totals were 9.3 inches at DCA and 10.3 inches at IAD. Probably the all-time best unanticipated snow day in the history of DC (bear in mind that the Veterans Day snowstorm of November 11, 1987 occurred on a Government holiday), and the Government remained shuttered on Wednesday as well. 

[Farmboy]

How early was the Great Blizzard of March 4, 2001, spotted?

[Deck Pic]

    good question.    it was some sort of strange way that the buckets were kept in the gempak files when trying to sum odd (i.e. 66 in this case) time periods.   To get a 66 (and don't ask me why I chose 66), I had to sum a 48 total and then an 18.  So the first map is the 48 hour total valid at 00z on the 16th plus the 18 hour total valid at 18z on the 16th.

 

Thanks.  Does that mean that on the 1st panel for instance, that the initialization time was 2/14/03 0z?  

[WinterWxLuvr]

Question about March 1993 since it is the most phenomenal weather event of my lifetime IMO.

I know the models were nothing comparable to now, and I know my first indication that there was going to be a storm came from a radio forecast on Wed evening for a storm on Saturday. My question is how far in advance, and with how much agreement did the models of the day see that storm? Also, given those lead up conditions, how far in advance would today's models see that storm?

Thanks in advance for any replies.

[gymengineer]

Question about March 1993 since it is the most phenomenal weather event of my lifetime IMO.

I know the models were nothing comparable to now, and I know my first indication that there was going to be a storm came from a radio forecast on Wed evening for a storm on Saturday. My question is how far in advance, and with how much agreement did the models of the day see that storm? Also, given those lead up conditions, how far in advance would today's models see that storm?

Thanks in advance for any replies.

At 12Z on 3/13/93, the actual low was 971 mb near Savannah, GA. 

From 5 days out valid for 12Z 3/13, the Euro had a 995 mb low near New Jersey. The UKMET had a 991 mb low near Cape Cod, and the MRF had a 993 mb low just off the VA capes.

From 4 days out valid for 12Z 3/13, the Euro had a 990 low right off the VA capes. The UKMET had a 994 low east of Ocean City, MD, and the MRF had a 996 low near the FL/GA line.

From 3 days out valid for 12Z 3/13, the models all caught on to the storm forming later. The Euro was a 1007 low just west of Key West, FL. The UKMET was a 983 low near Chattanooga, TN. The MRF had a 996 low just west of Athens, GA. 

 

At 12Z on 3/14/93, the actual low was 964 mb near Portland, ME.

From 5 days out valid for 12Z 3/14, the Euro had the low at 979 mb near the southern tip of Nova Scotia. The UKMET had a 977 mb low over northern Nova Scotia. The MRF had a 983 mb low near Chincoteague, VA. 

From 4 days out valid for 12Z 3/14, the Euro had a 993 mb low hundreds of miles east of NJ out in the Atlantic (progressing from its location in the above 3 day prog west of Key West). The UKMET had a 966 mb low over Lake Ontario (progressing from its location in the above 3 day prog from TN). The MRF had a 975 mb low near Baltimore (progressing from its location in the above 3 day prog from GA).

From 3 days out valid for 12Z 3/14, the Euro had low moving from central AL to Lake Erie reaching 968 mb. The UKMET had low moving from the coast of GA to well east of NJ reaching 975 mb. The MRF had a low moving from near Augusta, GA to northern ME reaching 974 mb. 

 

So, the MRF performed the best, but there was a very strong signal for a major event on all three models.  

[gymengineer]

If anyone is curious about the forecasting of the March 1993 event, I really suggest purchasing/locating "Forecasting the 12-14 March 1993 Superstorm" by Uccellini, Kocin, et al. This was published in the Bulletin of the American Meteorological Society. 

 

A really interesting short-term forecasting dilemma was that at 24-hrs out, the NGM and ETA both shifted the low further inland to central VA (compared to their 48-hr solution) and were showing a pressure in the mid-950's mb. Following that output verbatim would have meant shifting the rain line further west and substantially reducing the snow forecast for the big cities. The AVN kept a 960 mb low near the St. Mary's/Charles County line, which was much closer to what verified. Forecasters followed the AVN guidance because of verification results showing the AVN doing better than the regional models, and kept heavy snow into the cities. 

[high risk]

Thanks.  Does that mean that on the 1st panel for instance, that the initialization time was 2/14/03 0z?  

 

       correct!

[high risk]

First, great post!  Second, the above was one of the lesser moments in DC forecasting history.  I distinctly remember hearing on WTOP around 7 PM on the evening of Monday, January 24th, that there could be "flurries" on Tuesday.  As I was getting ready for bed Monday evening, Sue Palka led the Channel 5 news with breaking news about a major snowstorm heading our way. However, the Office of Personnel Management did not take the updated forecast seriously, with the Federal Government opening on time on Tuesday.  Big mistake.  I lucked out, as my bus never arrived that morning.  So, I trudged back home through the snow and discovered that the Government had just closed.  The final snow totals were 9.3 inches at DCA and 10.3 inches at IAD. Probably the all-time best unanticipated snow day in the history of DC (bear in mind that the Veterans Day snowstorm of November 11, 1987 occurred on a Government holiday), and the Government remained shuttered on Wednesday as well. 

 

        The models certainly weren't good that day, but there were some hints in the trends.    This was the 00z 1/24 Eta precip forecast     

 

   Now here is the same 12 hours later:

 

    it's still not a direct hit, but the westward trend was alarming a lot of us that morning at the World Weather Building.   Yes, you had only few hundredths shown for DC, but the 1" QPF line was only 75 miles to the southeast and had jumped 75 miles to the northwest in 12 hours.    Satellite and radar trends through midday only made us more curious.   I remember seeing the 18z GFS show the 0.25" line going through the local area, and I knew we were in for at least a decent event.   The 00z Eta sealed the deal.    As I recall, LWX was a bit gunshy on being more aggressive with an earlier call for more snow, as they had busted on calling for a 2-4" event 2 days previous.

 

      I also recall the morning of the 24th that the Peachtree City raob winds at upper levels were rejected by the models because they were so far off the first guesses.    I think it was something like 62 m/s at 250 mb where the first guesses were in the 30 range.   They were playing catchup to the strength of the upper jet.   Also, the 12z Tampa raob didn't launch or had bad data or something.    The initialization of the upper wind field over the southeast that morning didn't help us at all.