18z 12/24 model disco

[chris87]

let's not beat up HPC....it was a valid assumption to assume the differences between the intialization of the GFS and Euro was the reason for the 12z GFS track...unfortunately for that logic, the GFS is gaining a decent amount of support...at least suggesting a close to the coast track is possible...

unfortunately again for HPC, that logic may cost local NWS offices 12 hours of preparation because they have been told to discount the GFS instead of saying this may have played a role, but the possibility of a close to the coast track is higher than was shown in the 0z guidance

[yoda]

Honestly, I'm not sure. It seems to be a bit conservative in some areas, even using a 10:1 ratio, but it's the only site with model-generated snowfall graphics I'm aware of that covers the east coast.

It will be interesting to see the clown maps from Earl Barker in a lil while as well...

[Guest someguy]

My concern is the Euro has many more incidents of being owned by the GFS inside 72 hours, probably AS many as it has on the GFS outside 72....I still think this storm misses simply based on ridge/trough placement but I have always been skeptical of the Euro inside 72 hours, even though its verification is good we have to realize those are over every single run of the year, that doesn't mean it handles big systems well in that range.

DT strongly supports this view

. I have said in these many times that inside 84 and 72 hrs the GFS is pretty good and I can recall several storm

in recent years where the GFS caught last minute trends that the Euro did not

the 0z GFS is the run to watch and espcially the 0z GGEM with its somewhat progressive bias

[ORH_wxman]

euro did the same thing with the big runs.. a bit further south. storms last yr also did similar.. block?

The low cuts off and then gets brought east by the trough getting more progressive with time as that stuff from north interacts more. So you see a straight turn north initially, and then once it cuts off it drifts ENE. The same thing happens on the solutions further east but nobody noticed them because they were whiffs.

[dtk]

100% agree. Where is the data on which data points were bogus? Which winds were in error at what heights, which RAOBs? There's none of that. No specifics.

The 10am release said nothing about errors, it said specifically there were no signficant errors that would have impacted the 12z NAM. An hour later when the bomb is on the GFS it all started. So whatever was found passed the initial error checks.

Where is the error at 6h based on the water vapor and RUC etc. Why would the 18z NOGAPs pull back hundreds of miles. Why would the RGEM tighten up.

It makes no sense to me.

Don't get me wrong, there could be an issue (I'm skeptical). I just want to know what the issue is.

Btw, we have many mechanismns in place to toss or downweight 'suspect observations'

[Tiburon]

From PHL:

.DAYS TWO THROUGH SEVEN...SATURDAY THROUGH THURSDAY.

A COASTAL STORM MAY AFFECT OUR REGION DURING THE SUNDAY INTO

MONDAY TIME FRAME, ALTHOUGH THE THREAT IS NOT HIGH AT THIS TIME.

THERE STILL REMAINS SOME UNCERTAINTY REGARDING THE TIMING AND

TRACK, BUT THE GREATEST THREAT STILL WOULD BE TO EASTERN SECTIONS.

Very smart. See what 0Z has to say, and if the GFS continues this and the EURO ensembles shift west, then we wake up on Christmas morning to a Special Weather Statement or Winter Storm Watches. If it continues throughout Saturday, we awaken Sunday morning to Winter Storm Warnings (or more).

[Reale WX]

I have to go on air in Salisbury, MD in less than an hour with this! What a mess

Eww im sorry man, as of this morning I forecasted for them a few flurries/ brief periods of light snow Christmas afternoon not accumulating to anything, quit overnight and Sunday morning, then ? on Sunday haha. Gotta leave the possibility of being hit (more like grazed) with the coastal producing up to some moderate accums sun night. Cant put totals on it would not be smart.

[weatherweather]

It seems weird to me that if they knew there were errors with 12z gfs they couldnt do anything about it for 18z. They just have to let it run and call it useless?

[SnowGoose69]

The new UKMET is still a miss but as I've said already, the UKMET has as much weight with me now as a Brett Favre retiring promise.

[Amped]

Again errors in Nebraska and Kansas aren't a big deal.

It is the S/W over North Dakota that causes this and it looks legit on satellite. The problem would be if 18z initialized heights in the great lakes and NE too high.

[Tiburon]

...and as fast as it goes away, it sucks us all back in.

Merry Christmas folks.

[Kmlwx]

if the gfs is handling the southern wave right this is a believable solution.. the diff between a big hit on the coast and just offshore was always heavly tied to that. why the euro isnt seeing the same thing is truly puzzling in this range.

Agree with your post. This is pretty wild stuff on the models. I'm nowhere near saying I buy it but it's pretty weird having this all happen.

[Analog96]

Again errors in Nebraska and Kansas aren't a big deal.

It is the S/W over North Dakota that causes this and it looks legit on satellite. The problem would be if 18z initialized heights in the great lakes and NE too high.

Why in the hell doesn't HPC tell us then to make it more believeable and to give us an idea what to do with the forecast?

[Clinch Leatherwood]

My concern is the Euro has many more incidents of being owned by the GFS inside 72 hours, probably AS many as it has on the GFS outside 72....I still think this storm misses simply based on ridge/trough placement but I have always been skeptical of the Euro inside 72 hours, even though its verification is good we have to realize those are over every single run of the year, that doesn't mean it handles big systems well in that range.

There's been a lot of talk in the UK and Europe about recent poor performance in this pattern of the Euro. It's either about to restate dominance or have one of the biggest failures it has ever had.

[WhiteoutWX]

I understand everyones skepticism of the HPC tossing out the GFS in light of trends, but c'mon they arent going to lie to cover up a possibe flip-flop they would have to do after backing the Euro's storm, then dropping the storm, and then having to come back to that idea. Thats stupid. For now, until other non GFS based models (namely GGEM, Ukmet, and Euro) come around, this could all still be casued by some model error. 0z will be interesting for sure.

[jimmosk]

A background question from a relative newbie:

These were the errors that caused the 12Z GFS to be discarded:

12Z NAM RAOB RECAP...

BRW/70026 - 10142

GUM/91212 - 10142

NCC/78988 - 10142

YNN/78073 - 10142

ZED/71126 - MISSING TTAA/BB

ASY/70414 - 10145

KCR/78384 - PURGED ALL HGTS AND TEMPS

GYM/76256 - PURGED REPORT...TWIN REPORT

(10142 = Ground equipment failure; 10145 = Bad weather conditions, no launch)

My questions are, can someone tell us:

a) how important these radiosonde locations are for accurately modeling this situation,

b ) when and if replacement radiosondes would be sent up, and

c) which models other than the 12Z GFS would be affected by these errors?

[weatherweather]

Some of you are treating the HPC like this run already verified. They are the ones in the hotseat and doing the best they can. If this doesnt verify, they dont get any extra credit, and you have nothing to lose. So just chill out and let them do their thing

[A-L-E-X]

DT strongly supports this view

. I have said in these many times that inside 84 and 72 hrs the GFS is pretty good and I can recall several storm

in recent years where the GFS caught last minute trends that the Euro did not

the 0z GFS is the run to watch and espcially the 0z GGEM with its somewhat progressive bias

Yes, the 0z runs should apply the truth serum.... hopefully they keep this trend. Those runs are also within the 48 hr limit from when the storm is supposed to start, so that should add some more weight. We need to develop some sort of consensus here-- even with a schizoid pattern like this lol.

[Analog96]

Some of you are treating the HPC like this run already verified. They are the ones in the hotseat and doing the best they can. If this doesnt verify, they dont get any extra credit, and you have nothing to lose. So just chill out and let them do their thing

I didn't see anybody saying HPC is WRONG, but what we want to know is WHAT IS WRONG WITH THE MODELS EXACTLY? If that's too hard for them to answer, perhaps they shouldn't be working at HPC?

[rainstorm]

the snow here should begin tommor afternoon.

[A-L-E-X]

Some of you are treating the HPC like this run already verified. They are the ones in the hotseat and doing the best they can. If this doesnt verify, they dont get any extra credit, and you have nothing to lose. So just chill out and let them do their thing

Drawing a sports analogy here, its like a team going ahead in the top of the ninth on the road and claiming victory, forgetting that there's a bottom of the ninth to play.

And actually we arent anywhere close to the ninth inning yet!

[SnowGoose69]

Keep in mind, the lead time will still be fine if watches do not go up til tomorrow PM....the exception is DC/BWI/Delmarva...if those areas ultimately got hit then the lead time would be poor, but NYC/BOS would have 24-36 hours.

[winterymix]

Why in the hell doesn't HPC tell us then to make it more believeable and to give us an idea what to do with the forecast?

MODEL DIAGNOSTIC DISCUSSION

NWS HYDROMETEOROLOGICAL PREDICTION CENTER CAMP SPRINGS MD

143 PM EST FRI DEC 24 2010

VALID DEC 24/1200 UTC THRU DEC 28/0000 UTC

...THE RECOMMENDATION IS FOR 1/3 EACH NAM/ECMWF/UKMET.

...

JAMES

[weatherweather]

I didn't see anybody saying HPC is WRONG, but what we want to know is WHAT IS WRONG WITH THE MODELS EXACTLY? If that's too hard for them to answer, perhaps they shouldn't be working at HPC?

true that is a valid point, I was mainly referring to the non-mets in this thread

[yoda]

Hmmm... 987 mb

[Ian]

Agree with your post. This is pretty wild stuff on the models. I'm nowhere near saying I buy it but it's pretty weird having this all happen.

differences of a few hours make a huge impact on the end result. it would be nice to have a clearer picture but it seems the models can't read this pattern well. after all this time invested and the emotional swings it would be nice to get some snow lol.

[JERSEYSNOWROB]

I have to go on air in Salisbury, MD in less than an hour with this! What a mess

At this late stage in the game, you just have to tell it like it is. Be honest and let them know the possible scenerios, not just to cover your own butt, but to warn the public that such a scenerio does exist and to get them to keep a close eye on the situation.

[Clinch Leatherwood]

A background question from a relative newbie:

KCR/78384 - PURGED ALL HGTS AND TEMPS

GYM/76256 - PURGED REPORT...TWIN REPORT

(10142 = Ground equipment failure; 10145 = Bad weather conditions, no launch)

My questions are, can someone tell us:

a) how important these radiosonde locations are for accurately modeling this situation,

when and if replacement radiosondes would be sent up, and

c) which models other than the 12Z GFS would be affected by these errors?

KCR is in Iowa I think, deleting that for error would be significant given the current placement.

[Analog96]

And? Does that answer WHY?

[hurricaneschwartz]

WHAT AN AMAZING DAY AND WEEK!

I've never seen anything like this before...such a difference in models so close to a storm, AFTER a flip-flop of both EURO and GFS! Try attempting to make a forecast on TV on such an important weekend.

As for meteorological stuff, we get a special product at NBC10 in Philadelphia called eCast. This is a product from AER in Mass. that creates a MOS based on the EURO ensembles. For example, it gives the POP for every 12 hour period of 01, .10. and .50. A POP of 30 means that 30% of the 51 ensemble members had more than that amount of precip during the period in question. It has had a great track record over the past several years for us.

Curiously, it has had surprisingly low POPs for PHL this week. Even when the operational EURO was hammering us 3 days in a row, the POPs for .50"+ never got over 16%. Today's POPs are 57% for .01", 22% for .10", and 2% for .50"+. That was based on the 00z run.

The eCast maps from the 12z run show the storm track very close to the operational, and precip farther west (but still less than .25" at ACY, and less than .10 at PHL.

If we have the same issues with the 00z runs, I will not be able to sleep tonight (working the 10 and 11pm tomorrow).

My mind goes back to Jan, 2000, when everyone was so sure the storm would stay way offshore. But the radar echoes were way farther west than even the 12 hour forecasts were, so I was able to predict snow up here when other forecasts were still for "partly sunny" (still had no idea we'd get so much). This has taught me to never brag about a forecast before the event, and never say never until the event is over.

Glenn