Saint Patrick's Day Snow Event II

[mitchnick]

wouldn't be funny if it came north and we had to worry about about rain

no

[Spanks45]

NAM thinks the dewpoints are in the mid teens tomorrow afternoon...

[snowfan]

Sfc temps are below freezing by 26.

[T. August]

Hr 26 on 00z already has precip north of Baltimore, where on the 18z it just got to DC.

[mattie g]

Another idea could be to place 2 smaller min/max map graphics below an enlarged "most likely" one

This is actually a pretty good idee-er.

[Ltrain]

2m line is right along 95 at hour 25. Light precip moving in.

[Interstate]

Surface low is also ever so slightly north of 18z

 

Looks south to me

[T. August]

I think this came quite a bit north... At least the precipitation did.

[PhineasC]

Horrible weenie analysis as usual.

[snowfan]

Nice run. Heaviest still south, but most are near inch an hour rates by 4z. Nice band set up at 29.

[BTRWx]

2m line is right along 95 at hour 25. Light precip moving in.

It looks like mostly snow at the onset, if not all

[bricktamland]

Looks south to me

 

Yeah, I spoke too soon. Was north at hr 20 or so, now it looks to be south

[T. August]

Yeah, I spoke too soon. Was north at hr 20 or so, now it looks to be south

As did I. At hr 26 it appeared north, but after that it went south (get it lol?).

[EasternUSWX]

South of last run. 

[Interstate]

As did I. At hr 26 it appeared north, but after that it went south (get it lol?).

 

Was never north... even at panel 12... it was south of the 18z

[T. August]

Was never north... even at panel 12... it was south of the 18z

I said "appeared". I was basing it on the precipitation field.

[yoda]

South of last run. 

 

If you are up by M/D line yes... down by DC and south its fine

[WxMan1]

They're using SREF members, which in this case are an outlier and often are, as the core starting point for a major "groundbreaking" product.....it is standard govt stuff.

 

I can tell you that one of the improvements we hope to make at WPC is to limit the percent weight of the SREF members.  Right now, WPC's automated PWPF product (probability of winter precipitation forecast) is a 33 multi-member ensemble, with the "mode" being the WPC winter weather forecaster's deterministic snowfall (and ice) forecast based on the preferred model blend(s) over a particular region. Now, all 21 SREF members are included, along with 7 operational runs, and 5 GEFS members.  That's a 33 member system, HOWEVER, 64% of which (21/33) are based on the SREF -- 7 NMB members (which go into the NAM), 7 NMM members (GFS), and 7 EM/WRF members (RAP).  The problem is, when the NCEP models, particularly the NAM and SREF, are not the preferred runs for a particular winter storm, the "spread" among the mode (which again is the winter weather desk's deterministic forecast) is going to widen toward that SREF output.  

 

What we want to do going into next season is to not make the PWPF so "SREF heavy".  In doing so, we're going to incorporate some (perhaps half) of the ECMWF ensemble members.  One may be asking the question right now, "how come you (WPC) haven't done that already?"  Well, it's a matter of resolution.  Right now, we are getting the EC members at a full degree resolution, or 111 km. That's a pretty course resolution -- the kind that will make a NW flow clipper event bleed east of the Appalachians into the DC area. Why are we getting the ECMWF members at such a low resolution currently?  I think the main reason is computational -- being able to get the data and process it in our system in a timely manner.  However with the new supercomputer (part of the Sandy suppliment), that's going to change, as we hope to have 0.5 degree resolution ECMWF ensemble data by next season (closer to 54.5 km). 

 

So, we are getting there folks, gradually.  In the meantime, keep in mind a couple of things, at least from our (WPC's) perspective:

 

1) Those google-earth background probability graphics you see (i.e. our "automated" PWPF) for the time being will remain 64% SREF weighted.  Translation: when the SREF is not preferred -- especially when the other operational NCEP runs are not preferred (NAM and GFS), you will get a fairly large (too large) distribution with those snowfall and ice probabilities. 

 

2) WPC in the meantime still has the ability to manually edit the "final" 10, 40, and 70 percent probabilities for the 4, 8, and 12" snowfall probabilities, as well as for the 0.25" of ice. So, when we see such a huge disparity with the WPC's manual winter weather output vs. the automated PWPF, it is because we did not buy into the SREF output (and possibly NAM and/or GFS as well).  A great example of this is with the 03/03 event: when the WPC winter weather forecaster had the highest snowfall axis along the mid Atlantic region (our area), while the SREF was pointing toward NYC and SNE.  So, in this case, the automated PWPF looked AWFUL, with higher probs of 4, 6, and even 8 inches stretching from SNE all the way down into central VA.  Again, we'll get there...

[Bob Chill]

Close to .35 on dc by 2am.

[snowfan]

This run looks fine to me. Take some off the top and call it 3-6".

[T. August]

looks like the last 4 posters

 

2 in Pennsylvania and 2 near the PA border dont like the run

This is a Mid-Atlantic forum, not a "DC forum".

 

DC and south do well this run. The northern cutoff is probably too sharp as Bob Chill pointed out earlier.

[Bob Chill]

Nearly .25 falls between 2-5am. Solid

[snowdude]

I know not many people care about SBY, but I have been impressed with the SREFS here in Salisbury for predicted snowfall with this storm. They haven't been the best this year but this has been the highest totals it's shown all winter for here. 

[yoda]

Looks like .60 DCA... .75 line is in S Fairfax

[Huffwx]

I can tell you that one of the improvements we hope to make at WPC is to limit the percent weight of the SREF members.  Right now, WPC's automated PWPF product (probability of winter precipitation forecast) is a 33 multi-member ensemble, with the "mode" being the WPC winter weather forecaster's deterministic snowfall (and ice) forecast based on the preferred model blend(s) over a particular region. Now, all 21 SREF members are included, along with 7 operational runs, and 5 GEFS members.  That's a 33 member system, HOWEVER, 64% of which (21/33) are based on the SREF -- 7 NMB members (which go into the NAM), 7 NMM members (GFS), and 7 EM/WRF members (RAP).  The problem is, when the NCEP models, particularly the NAM and SREF, are not the preferred runs for a particular winter storm, the "spread" among the mode (which again is the winter weather desk's deterministic forecast) is going to widen toward that SREF output.  

 

What we want to do going into next season is to not make the PWPF so "SREF heavy".  In doing so, we're going to incorporate some (perhaps half) of the ECMWF ensemble members.  One may be asking the question right now, "how come you (WPC) haven't done that already?"  Well, it's a matter of resolution.  Right now, we are getting the EC members at a full degree resolution, or 111 km. That's a pretty course resolution -- the kind that will make a NW flow clipper event bleed east of the Appalachians into the DC area. Why are we getting the ECMWF members at such a low resolution currently?  I think the main reason is computational -- being able to get the data and process it in our system in a timely manner.  However with the new supercomputer (part of the Sandy suppliment), that's going to change, as we hope to have 0.5 degree resolution ECMWF ensemble data by next season (closer to 54.5 km). 

 

So, we are getting there folks, gradually.  In the meantime, keep in mind a couple of things, at least from our (WPC's) perspective:

 

1) Those google-earth background probability graphics you see (i.e. our "automated" PWPF) for the time being will remain 64% SREF weighted.  Translation: when the SREF is not preferred -- especially when the other operational NCEP runs are not preferred (NAM and GFS), you will get a fairly large (too large) distribution with those snowfall and ice probabilities. 

 

2) WPC in the meantime still has the ability to manually edit the "final" 10, 40, and 70 percent probabilities for the 4, 8, and 12" snowfall probabilities, as well as for the 0.25" of ice. So, when we see such a huge disparity with the WPC's manual winter weather output vs. the automated PWPF, it is because we did not buy into the SREF output (and possibly NAM and/or GFS as well).  A great example of this is with the 03/03 event: when the WPC winter weather forecaster had the highest snowfall axis along the mid Atlantic region (our area), while the SREF was pointing toward NYC and SNE.  So, in this case, the automated PWPF looked AWFUL, with higher probs of 4, 6, and even 8 inches stretching from SNE all the way down into central VA.  Again, we'll get there...

 

Thanks for the information-- wonderful. 

[Deck Pic]

I can tell you that one of the improvements we hope to make at WPC is to limit the percent weight of the SREF members.  Right now, WPC's automated PWPF product (probability of winter precipitation forecast) is a 33 multi-member ensemble, with the "mode" being the WPC winter weather forecaster's deterministic snowfall (and ice) forecast based on the preferred model blend(s) over a particular region. Now, all 21 SREF members are included, along with 7 operational runs, and 5 GEFS members.  That's a 33 member system, HOWEVER, 64% of which (21/33) are based on the SREF -- 7 NMB members (which go into the NAM), 7 NMM members (GFS), and 7 EM/WRF members (RAP).  The problem is, when the NCEP models, particularly the NAM and SREF, are not the preferred runs for a particular winter storm, the "spread" among the mode (which again is the winter weather desk's deterministic forecast) is going to widen toward that SREF output.  

 

What we want to do going into next season is to not make the PWPF so "SREF heavy".  In doing so, we're going to incorporate some (perhaps half) of the ECMWF ensemble members.  One may be asking the question right now, "how come you (WPC) haven't done that already?"  Well, it's a matter of resolution.  Right now, we are getting the EC members at a full degree resolution, or 111 km. That's a pretty course resolution -- the kind that will make a NW flow clipper event bleed east of the Appalachians into the DC area. Why are we getting the ECMWF members at such a low resolution currently?  I think the main reason is computational -- being able to get the data and process it in our system in a timely manner.  However with the new supercomputer (part of the Sandy suppliment), that's going to change, as we hope to have 0.5 degree resolution ECMWF ensemble data by next season (closer to 54.5 km). 

 

So, we are getting there folks, gradually.  In the meantime, keep in mind a couple of things, at least from our (WPC's) perspective:

 

1) Those google-earth background probability graphics you see (i.e. our "automated" PWPF) for the time being will remain 64% SREF weighted.  Translation: when the SREF is not preferred -- especially when the other operational NCEP runs are not preferred (NAM and GFS), you will get a fairly large (too large) distribution with those snowfall and ice probabilities. 

 

2) WPC in the meantime still has the ability to manually edit the "final" 10, 40, and 70 percent probabilities for the 4, 8, and 12" snowfall probabilities, as well as for the 0.25" of ice. So, when we see such a huge disparity with the WPC's manual winter weather output vs. the automated PWPF, it is because we did not buy into the SREF output (and possibly NAM and/or GFS as well).  A great example of this is with the 03/03 event: when the WPC winter weather forecaster had the highest snowfall axis along the mid Atlantic region (our area), while the SREF was pointing toward NYC and SNE.  So, in this case, the automated PWPF looked AWFUL, with higher probs of 4, 6, and even 8 inches stretching from SNE all the way down into central VA.  Again, we'll get there...

 

This sounds great and thanks for the detailed summary

[StudentOfClimatology]

I was basing it on the precipitation field.

Why would you ever do that? Look at the friggin pressure field. The contradictory pbp is annoying as hell.

[ers-wxman1]

Another idea could be to place 2 smaller min/max map graphics below an enlarged "most likely" one

That is what we do now. Plus the highlighting of the most likely

[mitchnick]

by 12Z Monday, ,53" at BWI and .7" at DCA

[BTRWx]

I can tell you that one of the improvements we hope to make at WPC is to limit the percent weight of the SREF members.  Right now, WPC's automated PWPF product (probability of winter precipitation forecast) is a 33 multi-member ensemble, with the "mode" being the WPC winter weather forecaster's deterministic snowfall (and ice) forecast based on the preferred model blend(s) over a particular region. Now, all 21 SREF members are included, along with 7 operational runs, and 5 GEFS members.  That's a 33 member system, HOWEVER, 64% of which (21/33) are based on the SREF -- 7 NMB members (which go into the NAM), 7 NMM members (GFS), and 7 EM/WRF members (RAP).  The problem is, when the NCEP models, particularly the NAM and SREF, are not the preferred runs for a particular winter storm, the "spread" among the mode (which again is the winter weather desk's deterministic forecast) is going to widen toward that SREF output.  

 

What we want to do going into next season is to not make the PWPF so "SREF heavy".  In doing so, we're going to incorporate some (perhaps half) of the ECMWF ensemble members.  One may be asking the question right now, "how come you (WPC) haven't done that already?"  Well, it's a matter of resolution.  Right now, we are getting the EC members at a full degree resolution, or 111 km. That's a pretty course resolution -- the kind that will make a NW flow clipper event bleed east of the Appalachians into the DC area. Why are we getting the ECMWF members at such a low resolution currently?  I think the main reason is computational -- being able to get the data and process it in our system in a timely manner.  However with the new supercomputer (part of the Sandy suppliment), that's going to change, as we hope to have 0.5 degree resolution ECMWF ensemble data by next season (closer to 54.5 km). 

 

So, we are getting there folks, gradually.  In the meantime, keep in mind a couple of things, at least from our (WPC's) perspective:

 

1) Those google-earth background probability graphics you see (i.e. our "automated" PWPF) for the time being will remain 64% SREF weighted.  Translation: when the SREF is not preferred -- especially when the other operational NCEP runs are not preferred (NAM and GFS), you will get a fairly large (too large) distribution with those snowfall and ice probabilities. 

 

2) WPC in the meantime still has the ability to manually edit the "final" 10, 40, and 70 percent probabilities for the 4, 8, and 12" snowfall probabilities, as well as for the 0.25" of ice. So, when we see such a huge disparity with the WPC's manual winter weather output vs. the automated PWPF, it is because we did not buy into the SREF output (and possibly NAM and/or GFS as well).  A great example of this is with the 03/03 event: when the WPC winter weather forecaster had the highest snowfall axis along the mid Atlantic region (our area), while the SREF was pointing toward NYC and SNE.  So, in this case, the automated PWPF looked AWFUL, with higher probs of 4, 6, and even 8 inches stretching from SNE all the way down into central VA.  Again, we'll get there...

thanks for the info, good stuff