18z 12/24 model disco

[chris87]

You could reduce the flights out of an airport like ATL by 75%, and you'd still get a ascent/descent sounding from aircraft every 5 minutes or so....which is more then sufficient.....

all three points are trumped up a bit....

good luck keeping the peace

Huge point for two reasons:

1. Christmas eve flight volume normally.

2. Thousands of flights were again cancelled into and out of europe including flights into the US. We are getting a lower sampling of data than we would normally get. Bad timing really.

3. Thousands of flights are being altered today. Delta has already cancelled about 500 out of Atlanta on XMas for weather reasons which means those planes were held elsewhere and weren't in the air potentially today.

In a nutshell IMO a bad combination of events. Models choked on some bad data at 12z, and we have a bundle of flights cancelled across europe into major US hubs, and domestic flights being scrubbed for weather. Ouch.

[CTWeatherFreak]

Huge point for two reasons:

1. Christmas eve flight volume normally.

2. Thousands of flights were again cancelled into and out of europe including flights into the US. We are getting a lower sampling of data than we would normally get. Bad timing really.

3. Thousands of flights are being altered today. Delta has already cancelled about 500 out of Atlanta on XMas for weather reasons which means those planes were held elsewhere and weren't in the air potentially today.

In a nutshell IMO a bad combination of events. Models choked on some bad data at 12z, and we have a bundle of flights cancelled across europe into major US hubs, and domestic flights being scrubbed for weather. Ouch.

Talk about a crazy feedback loop!!!

[Clinch Leatherwood]

Ask and you shall receive. Here is the 500mb analysis as of 6:40pm, 20min before 00z. Hope this helps! FWIW, I think the 18z GFS is handling the two features nicely.

Only major issure aside of the dual vortices instead of one tight s/w in TX is up nw North Dakota, too fast with the trailing s/w. Sure there will be changes at 0z init but as you note in the MS Valley it looks okay.

[usedtobe]

Even without the data assimilation problems, this would be a tough call on how to weight the models as the Euro is the best model and has some support. The ruc seems to be going to a solution that might not support it but....We at teh Capital Eather Gang raised out POPS for an inch or snow or more at DCA to 40 to 50 percent and the pops for 4" or more to 25 since this system looks like it will have a really strong precipitation gradient. I still don't trust the GFS but ....I'm not good enough to completely gainsay before seeing another run. Anyway, it's certainly one of the toughest forecasts I can recall since I didn't work the Jan 200o event. I do remember blowing a forecast on a late phasing system like this one by going for heavy snow and then waking up the next morning to sun. Such systems are the worst to deal with.

[Clinch Leatherwood]

You could reduce the flights out of an airport like ATL by 75%, and you'd still get a ascent/descent sounding from aircraft every 5 minutes or so....which is more then sufficient.....

all three points are trumped up a bit....

good luck keeping the peace

True it should be partially mitigated but keep in mind the Euro is probably on about day 7-10 of grossly reduced input from aircraft. It's all bad timing, freakish events really that in the post mortem could have come into play either for bad euro performance or the US models inability to adjust at all at 18z.

[ORH_wxman]

Even without the data assimilation problems, this would be a tough call on how to weight the models as the Euro is the best model and has some support. The ruc seems to be going to a solution that might not support it but....We at teh Capital Eather Gang raised out POPS for an inch or snow or more at DCA to 40 to 50 percent and the pops for 4" or more to 25 since this system looks like it will have a really strong precipitation gradient. I still don't trust the GFS but ....I'm not good enough to completely gainsay before seeing another run. Anyway, it's certainly one of the toughest forecasts I can recall since I didn't work the Jan 200o event. I do remember blowing a forecast on a late phasing system like this one by going for heavy snow and then waking up the next morning to sun. Such systems are the worst to deal with.

Good post Wes. Its very difficult to say how much weight to throw around each solution. The RGEM (which I assume would be free of data assimilation issues) started amping up pretty good at 18z, so I am starting to give a little weight to a further west solution.

That said, you said it well WRT Euro...its not like its by itself. It has support from the other foreign models (GGEM/UKMET) on a further east solution. I personally haven't strayed too far from the ECMWF ensemble mean which has hardly moved more than 50 miles the past 4 or 5 runs. Its been consistently near the benchmark (for up this way) with some minor oscillations. If I had to make a forecast this minute, I'd probably keep it near there with a minor tug to the west.

But this could easily change in a much larger fashion. Volatile setup.

[Kaner587]

True it should be partially mitigated but keep in mind the Euro is probably on about day 7-10 of grossly reduced input from aircraft. It's all bad timing, freakish events really that in the post mortem could have come into play either for bad euro performance or the US models inability to adjust at all at 18z.

after all this is done with, storm or no storm, I highly doubt lack of data from flights over the CONUS is responsible for this "error"

[CT Rain]

Good post Wes. Its very difficult to say how much weight to throw around each solution. The RGEM (which I assume would be free of data assimilation issues) started amping up pretty good at 18z, so I am starting to give a little weight to a further west solution.

That said, you said it well WRT Euro...its not like its by itself. It has support from the other foreign models (GGEM/UKMET) on a further east solution. I personally haven't strayed too far from the ECMWF ensemble mean which has hardly moved more than 50 miles the past 4 or 5 runs. Its been consistently near the benchmark (for up this way) with some minor oscillations. If I had to make a forecast this minute, I'd probably keep it near there with a minor tug to the west.

But this could easily change in a much larger fashion. Volatile setup.

I think this is the way to go right now. 9 out of 10 times I'll take GGEM/UK/Euro (op and ensemble) consensus over the GFS.

[Eskimo Joe]

Not speaking for Eskimo Joe here but its probably using RUC data like the SPC Mesoanalysis.

I will always defer to you, after you do have the red tag.

[Chagrin Falls]

Even without the data assimilation problems, this would be a tough call on how to weight the models as the Euro is the best model and has some support. The ruc seems to be going to a solution that might not support it but....We at teh Capital Eather Gang raised out POPS for an inch or snow or more at DCA to 40 to 50 percent and the pops for 4" or more to 25 since this system looks like it will have a really strong precipitation gradient. I still don't trust the GFS but ....I'm not good enough to completely gainsay before seeing another run. Anyway, it's certainly one of the toughest forecasts I can recall since I didn't work the Jan 200o event. I do remember blowing a forecast on a late phasing system like this one by going for heavy snow and then waking up the next morning to sun. Such systems are the worst to deal with.

Wes,

Could the extremes between operational runs and various ensemble members be caused by the change in the long-wave pattern over the hemisphere? Before this storm we were in a consistent upper-air pattern for about a month, after the storm we are in a fast zonal flow with an eventual upper level ridge developing in the east. I seem to remember the individual runs really having a tough time when the long-wave pattern changes.

Also, seems like a hypothetical super-ensemble has been fairly consistent. If one was to blend the NAEFS & ECMWF ensembles all of the extremes over the last few days have been "blended out". I think it is good the HPC has rejected the GFS runs, I'm quite skeptical.

The GFS did not handle the northern stream features well in the upper Mid-West. As the long-wave pattern transitions further I can't see it doing better.

[donsutherland1]

Even without the data assimilation problems, this would be a tough call on how to weight the models as the Euro is the best model and has some support. The ruc seems to be going to a solution that might not support it but....We at teh Capital Eather Gang raised out POPS for an inch or snow or more at DCA to 40 to 50 percent and the pops for 4" or more to 25 since this system looks like it will have a really strong precipitation gradient. I still don't trust the GFS but ....I'm not good enough to completely gainsay before seeing another run. Anyway, it's certainly one of the toughest forecasts I can recall since I didn't work the Jan 200o event. I do remember blowing a forecast on a late phasing system like this one by going for heavy snow and then waking up the next morning to sun. Such systems are the worst to deal with.

Great points, Wes. I do like the idea of somewhat raising the POPs for an 1" or more of snow and 4" or more for DCA. If the 0z runs come in anything close to the 18z runs while the foreign models are much farther to the east, the stage will be set for one set or the other to have an incredible short-term bust. I still like something in between the last night's 0z GFS and today's 6z GFS until a consensus begins to form, but we'll see what happens.

[yoda]

Wes,

Could the extremes between operational runs and various ensemble members be caused by the change in the long-wave pattern over the hemisphere? Before this storm we were in a consistent upper-air pattern for about a month, after the storm we are in a fast zonal flow with an eventual upper level ridge developing in the east. I seem to remember the individual runs really having a tough time when the long-wave pattern changes.

Also, seems like a hypothetical super-ensemble has been fairly consistent. If one was to blend the NAEFS & ECMWF ensembles all of the extremes over the last few days have been "blended out". I think it is good the HPC has rejected the GFS runs, I'm quite skeptical.

And the NAM? Just so I get it correct, you believe that its good the HPC rejected its own American models from 12z on?

[thunderman]

Not speaking for Eskimo Joe here but its probably using RUC data like the SPC Mesoanalysis.

I can confirm that this is correct

[chris87]

certainly possible, but not all together likely to have played a large role....it's hard to really pin down the impact of missing a certain percentage of data you normally have...remote sensing instruments are much more important than aircraft data, and in most cases, the data is redundant...so you could completely take our aircraft data and satellite data would still provide the needed observations...

here's a good example...

link

go to slide 38 and you can see sensitivity analysis for each observation platform....the 3 most important observations to the ECMWF are AMSU (microwave sounders) and AIRS and IASI (IR sounders)

aircraft data is just not that important anymore

True it should be partially mitigated but keep in mind the Euro is probably on about day 7-10 of grossly reduced input from aircraft. It's all bad timing, freakish events really that in the post mortem could have come into play either for bad euro performance or the US models inability to adjust at all at 18z.

[eduggs]

The strikingly small GFS ensemble spread seems fishy to me (both 12z and 18z). A few of the members should cluster with or near the offshore consensus of Euro/GGEM/UK and their respective ensemble means. The fact the the GFS ensemble spread is sooo small compared to the inter-model spread is a huge red flag IMO.

[ORH_wxman]

And the NAM? Just so I get it correct, you believe that its good the HPC rejected its own American models from 12z on?

I'm pretty sure HPC gave some weight to the 12z NAM.

[usedtobe]

Wes,

Could the extremes between operational runs and various ensemble members be caused by the change in the long-wave pattern over the hemisphere? Before this storm we were in a consistent upper-air pattern for about a month, after the storm we are in a fast zonal flow with an eventual upper level ridge developing in the east. I seem to remember the individual runs really having a tough time when the long-wave pattern changes.

Also, seems like a hypothetical super-ensemble has been fairly consistent. If one was to blend the NAEFS & ECMWF ensembles all of the extremes over the last few days have been "blended out". I think it is good the HPC has rejected the GFS runs, I'm quite skeptical.

The GFS did not handle the northern stream features well in the upper Mid-West. As the long-wave pattern transitions further I can't see it doing better.

Certainly rapidly changing regimes are tougher to forecast but these kinds of storms that have really late phasing because of the ridge position upstream being so far east probably are the worst. For dc to get much, the upper low has to close off south of us strong enough to get a cold conveyor belt going before the storm gts past our latitude and the low has to develop far enough west to get us in the deformation zone or we get zilch since we don't have the warm advection out ahead of the storm like we have when the ridge is a littel farther west allowing for a better shortwave ridge to develop. Farther north for Will, it's not quite as critical and for guys iaroudn teh Cape, they should get plenty of precip.

[Clinch Leatherwood]

after all this is done with, storm or no storm, I highly doubt lack of data from flights over the CONUS is responsible for this "error"

That's not what I said.

What I said was it's possible one of the reasons we didn't smooth out as much as normally would have been the case is that 18z there was likely less flight data than on a normal day. It might be .01%, I don't know.

At 6h for the 18z NAM and GFS they do both now have their problems. The NAM appears too strong and too west with the vortmax west of Missouri, and the GFS didn't seem to notice the energy NW of ND and missed the elongated nature of the s/w in texas. In the last couple of hours these differences have emerged. Normal 6 hour changes? Maybe, but we'll know in an hour.

[JERSEYSNOWROB]

I think this is the way to go right now. 9 out of 10 times I'll take GGEM/UK/Euro (op and ensemble) consensus over the GFS.

Even within 72 hours? I thought the GFS was better at this timeframe than the Euro, which usually does better over 72 hours?

[yoda]

I'm pretty sure HPC gave some weight to the 12z NAM.

Some, but not a lot it appears

THIS APPROACH DISREGARDS THE

SUBSTANTIALLY DEEPER AND WESTWARD SHIFT OF THE 12Z GFS REGARDING

THE POWERFUL LOW TRACKING UP THE EASTERN SEABOARD...AND TO A

LESSER EXTENT THE 12Z NAM WHICH LIES NEAR THE FAST EDGE OF THE

GUIDANCE WITH THE DEVELOPING LOW.

http://www.hpc.ncep.noaa.gov/discussions/pmdhmd.html

[ORH_wxman]

Even within 72 hours? I thought the GFS was better at this timeframe than the Euro, which usually does better over 72 hours?

Well I certainly wouldn't pick the GFS over the ECMWF if I was forecasting qpf 72 hours out (at least normally)....which is really what a lot of these forecasts boil down to whether they are viewed as busts or hits

[Kaner587]

That's not what I said.

What I said was it's possible one of the reasons we didn't smooth out as much as normally would have been the case is that 18z there was likely less flight data than on a normal day. It might be .01%, I don't know.

At 6h for the 18z NAM and GFS they do both now have their problems. The NAM appears too strong and too west with the vortmax west of Missouri, and the GFS didn't seem to notice the energy NW of ND and missed the elongated nature of the s/w in texas. In the last couple of hours these differences have emerged. Normal 6 hour changes? Maybe, but we'll know in an hour.

I just think you're "looking" for the supposed errors that the GFS has when in reality it would be more helpful to compare it to all major models. If the GFS is too fast with the northern stream feature but the others are too slow then the result is probably a compromise. Granted you clearly have more expertise in this than I do, but in all your posts it seems as if you're grasping at straws trying to use lack of aircraft data as a potential reason when every met who has responded has said its probably not enough of a factor to make a visible difference. I'm not saying you're wrong or that your analysis is off but include at least one other model like the EURO or GGEM in these analyses I suspect it'd be a lot more helpful...

[JERSEYSNOWROB]

Well I certainly wouldn't pick the GFS over the ECMWF if I was forecasting qpf 72 hours out (at least normally)....which is really what a lot of these forecasts boil down to whether they are viewed as busts or hits

[/quot

interesting. thanks ORH. .

[Srain]

I just think you're "looking" for the supposed errors that the GFS has when in reality it would be more helpful to compare it to all major models. If the GFS is too fast with the northern stream feature but the others are too slow then the result is probably a compromise. Granted you clearly have more expertise in this than I do, but in all your posts it seems as if you're grasping at straws trying to use lack of aircraft data as a potential reason when every met who has responded has said its probably not enough of a factor to make a visible difference. I'm not saying you're wrong or that your analysis is off but include at least one other model like the EURO or GGEM in these analysis I suspect it'd be a lot more helpful...

The TX S/W is doing quite well and is nicely wrapped and perhaps a touch W of guidance, fwiw as well.

[Eskimo Joe]

The TX S/W is doing quite well and is nicely wrapped and perhaps a touch W of guidance, fwiw as well.

I've been watching that like a hawk and then comparing it to the water vapor/radar loops. I really think a plowable snowfall is possible for the I-95 corridor.

[usedtobe]

Well I certainly wouldn't pick the GFS over the ECMWF if I was forecasting qpf 72 hours out (at least normally)....which is really what a lot of these forecasts boil down to whether they are viewed as busts or hits

The euro is best even during the 1st 36 hrs. The sref mean for modest amounts is usually also pretty good which leads to a dilemma as the two best verifying systems are somewhat at odds. But like you, I think the euro is the more likely solution but we're just playing the odds and sometimes the guy drawing to a flush beats the guy holding 3 of kind. Time to go watch a movie.

[JERSEYSNOWROB]

The euro is best even during the 1st 36 hrs. The sref mean for modest amounts is usually also pretty good which leads to a dilemma as the two best verifying systems are somewhat at odds. But like you, I think the euro is the more likely solution but we're just playing the odds and sometimes the guy drawing to a flush beats the guy holding 3 of kind. Time to go watch a movie.

24 hrs of a Christmas Story has started on TBS.

[AlYourWxPal]

I think when I go on at 11 in Salibury, I will post 2 snowfall maps with the different scenarios. One further east and one further west. That is the best way I think I should do it at this point in time unless the GFS goes east at 0Z.

Anyone else have other ideas?! They are welcome!

[ORH_wxman]

The euro is best even during the 1st 36 hrs. The sref mean for modest amounts is usually also pretty good which leads to a dilemma as the two best verifying systems are somewhat at odds. But like you, I think the euro is the more likely solution but we're just playing the odds and sometimes the guy drawing to a flush beats the guy holding 3 of kind. Time to go watch a movie.

Agreed Wes, I'm not exactly comfortable having to pick between the SREF and ECMWF...maybe a compromise is in order. At least we can do that in meteorology to some degree unlike poker, lol.

You try and play the percentages plus a bit of experience and hope for the best.

[yoda]

I saw this in the PHL/NYC thread... thought it might help some in here.

Here she comes. I think the 12Z computer models underestimated the strength of this storm yesterday and how sharp the trough is.