CWG January 15 two week outlook

[usedtobe]

Here's my outlook.  I wrote most of it without seeing today's Euro and then having Angela edit the teext to include it. The above normal snow chances was based on the 5-day ens mean pattern and how they looked to me. 

 

 

http://www.washingtonpost.com/blogs/capital-weather-gang/wp/2015/01/15/two-week-outlook-seasonal-then-a-temperature-plunge-and-increased-odds-of-snow/

[H2O]

Nice write up, Wes. All we can do is hope one of these pattern storms verifies

[usedtobe]

Nice write up, Wes. All we can do is hope one of these pattern storms verifies

Thanks.  Trouble is a good pattern doesn't always get you a snowstorm.  We also need luck. Plus, a forecast good pattern may end up verifying as a not so good pattern.  

[Bob Chill]

Thanks.  Trouble is a good pattern doesn't always get you a snowstorm.  We also need luck. Plus, a forecast good pattern may end up verifying as a not so good pattern.  

 

So true. If we pulled composites on all the non-slam dunk storms (and there's plenty of them), they wouldn't look so hot. We can also pull up plenty of 2 weeks periods that look good on paper but brown on the grass. 

 

Great write up. I wish we could fast forward one week and have some clarity for the last 10 days of the month. I think we score some accum snow. How we do it might not be same how we think today. lol

[usedtobe]

So true. If we pulled composites on all the non-slam dunk storms (and there's plenty of them), they wouldn't look so hot. We can also pull up plenty of 2 weeks periods that look good on paper but brown on the grass. 

 

Great write up. I wish we could fast forward one week and have some clarity for the last 10 days of the month. I think we score some accum snow. How we do it might not be same how we think today. lol

 

I'm not very optimistic about the 234 hour euro storm but think once we get beyond ti we have a pretty decent chance if we keep the big ridge to our west. 

[Bob Chill]

I'm not very optimistic about the 234 hour euro storm but think once we get beyond ti we have a pretty decent chance if we keep the big ridge to our west. 

 

At least we're moving towards something with some real promise and it starts happening well before the magical land of unicorns d10+. 

 

Ensemble guidance is going all out frigid for d11-15+ now. I hope I'm wrong but from what I saw on the euro ensemble run, it doesn't look like a particularly wet pattern in the MA. Am I wrong?

[gymengineer]

So true. If we pulled composites on all the non-slam dunk storms (and there's plenty of them), they wouldn't look so hot. We can also pull up plenty of 2 weeks periods that look good on paper but brown on the grass. 

 

Great write up. I wish we could fast forward one week and have some clarity for the last 10 days of the month. I think we score some accum snow. How we do it might not be same how we think today. lol

You've mentioned last year's little lead time on events and how that might be a better template than the long lead times we can get for -AO/-NAO patterns... I just went through the CWG archives for last winter, and we definitely had short lead times for at least the 1/2-3 and 1/21 events. On the morning of 1/1, the expectation was for just snow showers for the event starting the next evening. On the morning of 1/19, the word "dusting" was still used in the forecast for 1/21. 

[usedtobe]

At least we're moving towards something with some real promise and it starts happening well before the magical land of unicorns d10+. 

 

Ensemble guidance is going all out frigid for d11-15+ now. I hope I'm wrong but from what I saw on the euro ensemble run, it doesn't look like a particularly wet pattern in the MA. Am I wrong?

It's not but we don't nee it real wet to get accumulating snow.  All we need is a strong clipper

[WinterWxLuvr]

It's not but we don't nee it real wet to get accumulating snow. All we need is a strong clipper

It sure looks like the gfs ens want to bring a pretty strong one through next week. Just hope we end up in the right spot.

[PDIII]

The scatter plot is great. It really tells the story. I also love how you simplify things... Like "follow the lines on the map to see where the air is coming from". Your ability to explain stuff to the layman is a true talent. You base a lot of this on the ensemble mean forecasts. I think I have an idea as to why you use the ensembles in the long range as opposed to other guidance.. I guess it has something to do with resolution and large scale/ long wave length simulation. Could you explain technically why the ensembles are better in the long range?

[Scarlet Pimpernel]

Could you explain technically why the ensembles are better in the long range?

 

If I may, I'll make a half-hearted attempt to try and explain this.  Anyone else, do feel free to add more and/or correct what I say that may not be quite accurate.  First, there are several peer reviewed papers on ensemble forecasting, I think the first ones go back to the 1980s (authors include Kalnay, Tracton, Toth, and others).  My understanding is that in the medium range (>~ 3 or 4 days out), the ensemble mean tends to out-perform any single ensemble member and the higher resolution deterministic run.  Essentially the ensemble members are a bunch of runs from perturbed initial conditions (you can perturb the various initial fields, slightly tweak the physics, etc.).  So it's an attempt to estimate the possible range of solutions given the uncertainty in the initial conditions.  The mean and spread, basically, give one a very good indication of the overall forecast variables (such as 500 mb height) and the level of uncertainty.  It sort of smoothes out, if you will, the more "jittery" individual ensemble members.  Another thing one can glean from the ensembles is clusters of forecast solutions...e.g., if the spaghetti plots of height or storm tracks have a preference toward a particular solution.  If you have a wide range of such solutions with or several different possible clusters, that makes things more difficult to assess as you can perhaps imagine.  So, looking at the ensemble mean flow for d+10, say, rather than an individual member evolution or the high-res deterministic, gives one a lot more insight to what the true conditions might be (the mean tends to be more "stable", so to speak).  This can be extended to taking the 5-day mean fields around a particular date at medium-longer ranges.

 

There's more to the "math" than what I'm saying here, and to be honest I kind of have forgotten some of the precise reasons as to "why", but hope this long-winded description gives at least some insight.

[PDIII]

If I may, I'll make a half-hearted attempt to try and explain this. Anyone else, do feel free to add more and/or correct what I say that may not be quite accurate. First, there are several peer reviewed papers on ensemble forecasting, I think the first ones go back to the 1980s (authors include Kalnay, Tracton, Toth, and others). My understanding is that in the medium range (>~ 3 or 4 days out), the ensemble mean tends to out-perform any single ensemble member and the higher resolution deterministic run. Essentially the ensemble members are a bunch of runs from perturbed initial conditions (you can perturb the various initial fields, slightly tweak the physics, etc.). So it's an attempt to estimate the possible range of solutions given the uncertainty in the initial conditions. The mean and spread, basically, give one a very good indication of the overall forecast variables (such as 500 mb height) and the level of uncertainty. It sort of smoothes out, if you will, the more "jittery" individual ensemble members. Another thing one can glean from the ensembles is clusters of forecast solutions...e.g., if the spaghetti plots of height or storm tracks have a preference toward a particular solution. If you have a wide range of such solutions with or several different possible clusters, that makes things more difficult to assess as you can perhaps imagine. So, looking at the ensemble mean flow for d+10, say, rather than an individual member evolution or the high-res deterministic, gives one a lot more insight to what the true conditions might be (the mean tends to be more "stable", so to speak). This can be extended to taking the 5-day mean fields around a particular date at medium-longer ranges.

There's more to the "math" than what I'm saying here, and to be honest I kind of have forgotten some of the precise reasons as to "why", but hope this long-winded description gives at least some insight.

Awesome. Perfect explanation.

[Scarlet Pimpernel]

Awesome. Perfect explanation.

 

Glad it helped!  As I said, I'm sure there are others who can probably add a more accurate description, but hope I've got the correct general idea here.  In one of those published papers, I believe, they proved why the ensemble mean performs better in the medium range (but not in the short range).

[usedtobe]

The scatter plot is great. It really tells the story. I also love how you simplify things... Like "follow the lines on the map to see where the air is coming from". Your ability to explain stuff to the layman is a true talent. You base a lot of this on the ensemble mean forecasts. I think I have an idea as to why you use the ensembles in the long range as opposed to other guidance.. I guess it has something to do with resolution and large scale/ long wave length simulation. Could you explain technically why the ensembles are better in the long range?

In the longer past about 7 days I don' rely much on the daily forecasts by the models as they can jump around.   Instead I tend to look at 5-day mean patterns of the ensemble mean. I do that for two reasons.  One, is that a 5 day mean pattern teds to show the longer wavelength in the atmosphere which are the easier ones to forecast thoough even those are suspect at times.  The reason for using ensembles rather than one model is the mean is made up of a number of forecasts that are averaged but each of those forecasts start with slightly different initial conditions.  We do that because we know there are going to be errors in any analysis and slight differrences will grow.   By averaging the forecasts you hope to eliminate those times when the higher resolution model is an outlier and its solution has been impacted negatively by not having adequate initial conditions.  Years ago,  I saw a talk buy someone from the ECMWF showing that the esnemble mean usually beats the ECMWF in the longer time ranges.  I don't remember what hour the ensemble mean starts winning but by day 10, I'm pretty sure it does. 

[kurtstack]

Looks like the operational models may be backing off that western ridge we were banking on to deliver cold air, ugh. I imagine it could come back since we've seen the ops have 0 skill this year in the 7-10 day range.

[Ian]

Looks like the operational models may be backing off that western ridge we were banking on to deliver cold air, ugh. I imagine it could come back since we've seen the ops have 0 skill this year in the 7-10 day range.

Ensembles have a good cold signal by the end of the month.. tho that's like pretend land I suppose. 

[kurtstack]

Ensembles have a good cold signal by the end of the month.. tho that's like pretend land I suppose. 

It's far more enjoyable in pretend land this year:

[usedtobe]

Regardless of the fantasy storms which are usually wrong, I liked today's runs of the ensembles better than yesterday's.  We still never really get a negative NAO but should have enough cold to give us a chance or two by the end of the month.

[Bob Chill]

Regardless of the fantasy storms which are usually wrong, I liked today's runs of the ensembles better than yesterday's. We still never really get a negative NAO but should have enough cold to give us a chance or two by the end of the month.

We're still not out of the game wed-thurs. Obviously not a high prob event but h5 isn't far from decent. Gfs shows multiple little vorts zipping along in succession.

Agree about ensembles last night. Very active look. Especially the euro. Op runs are going to drive everyone nuts over the next week. Fast/active ns flow is going to present a multifamily yardsale of different evolutions

[usedtobe]

An update on the pattern from my Thursday's CWG outlook. Below I've attached the GEFS superensemble 5-day mean pattern. CPC takes that mean and finds the 10 closest matches since around 1950. Nine of the 10 analogs had snowstorms for DCA within 3 days of the centered mean. the snowstorm amounts in inches were:

 

1.8, 6.8, 1.9, 3.7,2.6,3.6,3.4,2.6 and 2.8.

 

That's as many snow analogs as I've ever seen on the superensembles.  Doesn't mean we get snow but the pattern is not the total loss that some seem to be saying.

 

The February 1996 storm had a stripe of more than 10 inches south of the city and Andrews Air Force base recorded 13". Today's mean maps pretty much support Thursday's CWG outlook. Looks like it will get cold for the end of the month and that we'll be threatened by northern stream systems. Even the 1996 storm was a northern stream shortwave that dug far enough to produce a coastal low.  The 500h below is not a bad look.

 

 

 

 

 

[gymengineer]

An update on the pattern from my Thursday's CWG outlook. Below I've attached the GEFS superensemble 5-day mean pattern. CPC takes that mean and finds the 10 closest matches since around 1950. Nine of the 10 analogs had snowstorms for DCA within 3 days of the centered mean. the snowstorm amounts in inches were:

 

1.8, 6.8, 1.9, 3.7,2.6,3.6,3.4,2.6 and 2.8.

 

That's as many snow analogs as I've ever seen on the superensembles.  Doesn't mean we get snow but the pattern is not the total loss that some seem to be saying.

 

The February 1996 storm had a stripe of more than 10 inches south of the city and Andrews Air Force base recorded 13". Today's mean maps pretty much support Thursday's CWG outlook. Looks like it will get cold for the end of the month and that we'll be threatened by northern stream systems. Even the 1996 storm was a northern stream shortwave that dug far enough to produce a coastal low.  The 500h below is not a bad look.

 

 

And the models correspondingly had a tough time with this storm... forecast was for flurries to a dusting as of the previous afternoon, 2/15. The timing of realizing what was about to happen was fairly similar to 1/24/00-- 10 pm and 11 pm newscasts led with the impending snow, but amounts were still hedged somewhat lower. The NWS had to keep on upping the snow totals during the next morning. 

[Ian]

Active stream pattern ahead. Hard to resolve but hard to imagine we totally fail given peak climo and cold air around. I'd still wager avg or above for Jan and probably the same for Feb.

[HighStakes]

The 1/17/2000 date was a couple days before the 3-5 inch event that kicked  off that 10 day great period. The Friday system on today's 12z GFS reminded me of that storm.

[WxUSAF]

Active stream pattern ahead. Hard to resolve but hard to imagine we totally fail given peak climo and cold air around. I'd still wager avg or above for Jan and probably the same for Feb.

Agreed.  Active pattern, even if it looks a little more northern stream dominated than it did a few days ago.  Hard to imagine we totally strike out before the end of the month.  I think we get at least 1 more widespread snow event.  I have no ideas whatsoever about February.  El Nino seems to have been stillborn. 

[Bob Chill]

Active stream pattern ahead. Hard to resolve but hard to imagine we totally fail given peak climo and cold air around. I'd still wager avg or above for Jan and probably the same for Feb.

Looks like 2-3 shortwaves move through from wed-sat. I think we have a decent chance of at least one of them amplifying enough to give us a minor to moderate event.

[Ian]

Agreed. Active pattern, even if it looks a little more northern stream dominated than it did a few days ago. Hard to imagine we totally strike out before the end of the month. I think we get at least 1 more widespread snow event. I have no ideas whatsoever about February. El Nino seems to have been stillborn.

I don't know much about Feb either tho I think this phase lags into early month at least.. Even if coming out of it. Our avg is puny enough it's not too bold to think.

[Kay]

Looks like it will get cold for the end of the month and that we'll be threatened by northern stream systems. 

 

Right? Even as a weenie and perpetual noob I can see this. I don't get the cliff jumping.. 

[usedtobe]

Today's  12Z GEFS ensmean keeps the nice look that the D+11 had last night through Feb 2.  At the longer time ranges I never worry about individual storms on either the GFS or operational Euro until they get within 6 days or so and even then it's just to watch them as the models handling of shortwaves isn't that good. The pattern isn't an ideal one as the Atlantic still is not great but the big ridge in Alaska and cold look to the upper flow coupled with a positive PNA certainly gives us chances for clipper type lows to get us and for one of those rare cases when the trough amplifies enough to actually form a coastal low like we had Feb 15, 1996.  A clipper look is the more likely type hit.  7 of the 9 analogs from last night had at least 2 inches of snow and even the two that didn't quite make 2 inches had 1.8 and 1.9. 

[AlaskaETC]

DT is issuing a "rebuttal" soon...

[feloniousq]

DT is issuing a "rebuttal" soon...

Yea was about to say...if DT hates the article it means Wes is on point.