So what the hell happened with the GFS?

[Baroclinic Zone]

I mean, it's the 2nd best model, not the best. It's prone to more misses than the Euro. Not arguing that at all. But you guys make it sound like it's the 2007 version of the GFS, which was patently terrible. NCEP has come a looooooooong way with their last 2 upgrades, regardless of your perception of its reliability in SNE this winter.

Nah, I'm on Scott's side.  I just feel it's unreliable for us when it comes crunch time (T<3).  I most definitely look at it and use it.  Knowing it's biases and strengths help.  Since the GGEM upgrade a couple weeks back, I think it's performed better up here with less variability.  Not comparing it to the GFS or Euro, just to it's past history.

[CoastalWx]

A good example as using it as guidance was the Feb blizzard. While it was not 100% correct, it did have the idea of being more progressive than the euro. So, putting that into account...a dam good forecast could be made. The euro does have a nice bias of trying to hook back bombing lows a little too much.

 

And trust me, I would love for the euro to have a bad streak because everyone is praising it and it makes all the good stuff that NCEP does seem non-existent.

[CT Rain]

A good example as using it as guidance was the Feb blizzard. While it was not 100% correct, it did have the idea of being more progressive than the euro. So, putting that into account...a dam good forecast could be made. The euro does have a nice bias of trying to hook back bombing lows a little too much.

 

And trust me, I would love for the euro to have a bad streak because everyone is praising it and it makes all the good stuff that NCEP does seem non-existent.

 

Part of the problem is... and why this discussion has caught on like fire... is the GFS is readily accessible to everyone. We wind up getting these awful Facebook posts, Twitter posts and the like of accumulated snowfall products from a D6 forecast. Suddenly everyone has latched onto a solution that we all know will bust.

 

Having the Euro harder to access and certainly with less pretty maps like accumulated snowfall helps the public's perception IMO. 

[ORH_wxman]

I'll say this about the GFS, it performed very well in the medium range up here during the progressive January pattern. It was doing better than the Euro for stretches at D3-4...at least up here.

 

However, that said, the GFS is the biggest joke when it comes to short term in this area during E.C. cyclogenesis. That's why it gets a bad rap from many forecasters here. Its nice when its predicting 20F and sunny accurately or predicted the hemispheric heights almost as well as the Euro at D3, but when you really need it to come through in crunch time, its been terrible. Perhaps it has been nailing forecasts off to the southwest of here during significant storms, but that has been far from the case in New England. It has been awful in 3 major storms this season (Dec 26-27, Feb 8-9, Mar 7-8) inside of 48 hours in this region and that is not what the reputation has been built on either. It was awful for much of 2010-2011 as well with the big exception being Boxing Day where it performed excellent.  

 

Most models have strengths and weaknesses. The NAM is great at convection, but sucks at QPF in synoptic storms. Perhaps the GFS weakness is major coastal cyclogensis near New England. That's a crappy weakness to have when trying to be a reliable forecasting tool in the winter as a New England met...so its very easy to see where the hostility towards it comes from when talking specifically about this region.

[CoastalWx]

Part of the problem is... and why this discussion has caught on like fire... is the GFS is readily accessible to everyone. We wind up getting these awful Facebook posts, Twitter posts and the like of accumulated snowfall products from a D6 forecast. Suddenly everyone has latched onto a solution that we all know will bust.

 

Having the Euro harder to access and certainly with less pretty maps like accumulated snowfall helps the public's perception IMO. 

 

The media has even shown what the euro has day 6 which can bust just as hard. This is a terrible new shift in the weather segments...I don't know why this winter has been the test bed..but it has.

[CoastalWx]

I'll say this about the GFS, it performed very well in the medium range up here during the progressive January pattern. It was doing better than the Euro for stretches at D3-4...at least up here.

 

However, that said, the GFS is the biggest joke when it comes to short term in this area during E.C. cyclogenesis. That's why it gets a bad rap from many forecasters here. Its nice when its predicting 20F and sunny accurately or predicted the hemispheric heights almost as well as the Euro at D3, but when you really need it to come through in crunch time, its been terrible. Perhaps it has been nailing forecasts off to the southwest of here during significant storms, but that has been far from the case in New England. It has been awful in 3 major storms this season (Dec 26-27, Feb 8-9, Mar 7-8) inside of 48 hours in this region and that is not what the reputation has been built on either. It was awful for much of 2010-2011 as well with the big exception being Boxing Day where it performed excellent.  

 

Most models have strengths and weaknesses. The NAM is great at convection, but sucks at QPF in synoptic storms. Perhaps the GFS weakness is major coastal cyclogensis near New England. That's a crappy weakness to have when trying to be a reliable forecasting tool in the winter as a New England met...so its very easy to see where the hostility towards it comes from when talking specifically about this region.

 

It might not be statistically far off from the euro inside 3 days...but you understand what I'm about to say. Maybe it's off by a slight margin at H5..but that slight margin means the s/w spawning cyclogenesis is a bit weaker or further east...which then means low pressure is 40 miles or so further offshore than reality...and there lies the bust. So yes, these little nuances can have an impact on sensible wx here. Like we mentioned..it has done well in some longer range stuff..but we're just discussing it inside 48hrs or so.

[Baroclinic Zone]

I'll say this about the GFS, it performed very well in the medium range up here during the progressive January pattern. It was doing better than the Euro for stretches at D3-4...at least up here.

 

However, that said, the GFS is the biggest joke when it comes to short term in this area during E.C. cyclogenesis. That's why it gets a bad rap from many forecasters here. Its nice when its predicting 20F and sunny accurately or predicted the hemispheric heights almost as well as the Euro at D3, but when you really need it to come through in crunch time, its been terrible. Perhaps it has been nailing forecasts off to the southwest of here during significant storms, but that has been far from the case in New England. It has been awful in 3 major storms this season (Dec 26-27, Feb 8-9, Mar 7-8) inside of 48 hours in this region and that is not what the reputation has been built on either. It was awful for much of 2010-2011 as well with the big exception being Boxing Day where it performed excellent.  

 

Most models have strengths and weaknesses. The NAM is great at convection, but sucks at QPF in synoptic storms. Perhaps the GFS weakness is major coastal cyclogensis near New England. That's a crappy weakness to have when trying to be a reliable forecasting tool in the winter as a New England met...so its very easy to see where the hostility towards it comes from when talking specifically about this region.

I was going to mention this Will.  It was killing it during Jan when the flow was really fast off the PAC.

[Damage In Tolland]

Moyer. All your energy met buddies also talk about how bad of a model the GFS is globally and how poor it performs. It's pretty silly to try and defend it when your colleagues all admit its inferior. Winter has been alive and well in the east all of Feb and now all of March

[ORH_wxman]

It might not be statistically far off from the euro inside 3 days...but you understand what I'm about to say. Maybe it's off by a slight margin at H5..but that slight margin means the s/w spawning cyclogenesis is a bit weaker or further east...which then means low pressure is 40 miles or so further offshore than reality...and there lies the bust. So yes, these little nuances can have an impact on sensible wx here. Like we mentioned..it has done well in some longer range stuff..but we're just discussing it inside 48hrs or so.

 

 

There's no doubt its been a bad piece of NWP guidance here during larger storms. If that is its inherent weakness despite scoring well on hemispheric patterns, then that is a problem. The UKMET scores well too hemispherically, but it seems to have issues with EC cyclogensis as well. The model can score whatever it wants from a hemispheric perspective, but if it shows a consistent pattern of being a highly inferior piece of guidance during cyclogenesis here, then I'm not going to put much weight into it.

 

Maybe eventually it will stop its shenanigans inside of 48 hours during big storms here, but until it does, there's no reason to trust it.

[CT Rain]

The media has even shown what the euro has day 6 which can bust just as hard. This is a terrible new shift in the weather segments...I don't know why this winter has been the test bed..but it has.

 

Agreed.

 

The reason was Sandy. 

 

I have more people ask me outside of work about what the GFS or Euro is showing... it's unbelievable. Sandy was a turning point but unfortunately I don't think it improves communication of weather threat or uncertainty at all.

[powderfreak]

Most models have strengths and weaknesses. The NAM is great at convection, but sucks at QPF in synoptic storms. Perhaps the GFS weakness is major coastal cyclogensis near New England. That's a crappy weakness to have when trying to be a reliable forecasting tool in the winter as a New England met...so its very easy to see where the hostility towards it comes from when talking specifically about this region.

Bingo...when it matters (ie. big high impact event along the eastern seaboard) the GFS struggles. No way around that. No one cares about the CMC or UKMET, lol...but even I feel the CMC has been more reliable in east coast storms.

Hey but if its doing great forecasting partly cloudy skies in the Ohio Valley, that's sweet, but hey it also had several runs of warning snows to Montreal in the last storm when we came away with a sunny day. ECM never had a flake falling on any run and hit that spot on.

I always liked the GFS and used it all the time for mountain forecasts cause the data is easily available, but the storm failures this season has made me real nervous lately in trusting it.

[powderfreak]

It's been great in the Plains/Midwest with their last 3 storms. Again, just because it didn't perform the best in your little geographic area doesn't make it a bad model or worthy of derision.

That little geographic area from like Philly-NYC north and east though contains a helluva lot of people and commerce and has a much bigger impact than you seem to be alluding to. Nailing a forecast over a third of the country in Midwest farm land is one thing but big time misses lately in densely populated northeastern regions with east coast cyclogenesis, is another thing.

[ORH_wxman]

Bingo...when it matters (ie. big high impact event along the eastern seaboard) the GFS struggles. No way around that. No one cares about the CMC or UKMET, lol...but even I feel the CMC has been more reliable in east coast storms.

Hey but if its doing great forecasting partly cloudy skies in the Ohio Valley, that's sweet, but hey it also had several runs of warning snows to Montreal in the last storm when we came away with a sunny day. ECM never had a flake falling on any run and hit that spot on.

I always liked the GFS and used it all the time for mountain forecasts cause the data is easily available, but the storm failures this season has made me real nervous lately in trusting it.

 

 

Part of the problem is that in most major storms for our region, the upper air pattern gets complex with a lot of shortwave interactions. When the flow starts to get jumbled up and more complex with multiple shortwaves, vortmaxima, blocking, etc, the GFS seems to handle it so much worse and that is probably at least part of the issue of it failing when the chips are down for a big storm here.

 

It performed great when we had a fast flow in January without a lot of complex setups, but once we had storm opportunities with more blocking, it embarrassed itself inside of 48 hours. At least up in New England it did. I didn't follow it closely enough down in the Mid-Atlantic to make any claims about its performance during major cyclogenesis there.

[CoastalWx]

Agreed.

 

The reason was Sandy. 

 

I have more people ask me outside of work about what the GFS or Euro is showing... it's unbelievable. Sandy was a turning point but unfortunately I don't think it improves communication of weather threat or uncertainty at all.

 

Same here...."hey what's the euro or gfs showing...." I turn around and ask how the hell they know what the models are, and it's because of the OCMs showing it. That's great if you discuss it two days out, but 5-6 days out...that's a joke.

[CoastalWx]

Part of the problem is that in most major storms for our region, the upper air pattern gets complex with a lot of shortwave interactions. When the flow starts to get jumbled up and more complex with multiple shortwaves, vortmaxima, blocking, etc, the GFS seems to handle it so much worse and that is probably at least part of the issue of it failing when the chips are down for a big storm here.

 

It performed great when we had a fast flow in January without a lot of complex setups, but once we had storm opportunities with more blocking, it embarrassed itself inside of 48 hours. At least up in New England it did. I didn't follow it closely enough down in the Mid-Atlantic to make any claims about its performance during major cyclogenesis there.

 

Yeah exactly and that's all I meant by talking about its downfalls here. When the stakes have been high from all that was said above...it hasn't been stellar here in New England. Maybe it goes on a tear starting next week, but this is how it's been so far.

[Ellinwood]

Moyer. All your energy met buddies also talk about how bad of a model the GFS is globally and how poor it performs. It's pretty silly to try and defend it when your colleagues all admit its inferior. Winter has been alive and well in the east all of Feb and now all of March

Adam's been spot on thus far.

- An energy met

[ORH_wxman]

Adam's been spot on thus far.

- An energy met

 

 

I think he's directing his ire toward mostly points that a few weenies might spout in general about the GFS...but the GFS's struggles up here in coastal storms is very real and not some imagined concoction by weenies.

[am19psu]

Moyer. All your energy met buddies also talk about how bad of a model the GFS is globally and how poor it performs. It's pretty silly to try and defend it when your colleagues all admit its inferior. Winter has been alive and well in the east all of Feb and now all of March

Which ones?

[Cary]

I think he's directing his ire toward mostly points that a few weenies might spout in general about the GFS...but the GFS's struggles up here in coastal storms is very real and not some imagined concoction by weenies.

He admitted that it hasn't been good in NE, but that it doesn't mean it's performed similarly everywhere else.

[am19psu]

There's no doubt its been a bad piece of NWP guidance here during larger storms. If that is its inherent weakness despite scoring well on hemispheric patterns, then that is a problem. The UKMET scores well too hemispherically, but it seems to have issues with EC cyclogensis as well. The model can score whatever it wants from a hemispheric perspective, but if it shows a consistent pattern of being a highly inferior piece of guidance during cyclogenesis here, then I'm not going to put much weight into it.

 

Maybe eventually it will stop its shenanigans inside of 48 hours during big storms here, but until it does, there's no reason to trust it.

I mean, that's fine if you believe that in SNE. I'm telling you, it does fine in PHL and it basically the same as the Euro in the tropics for hurricane forecasting. And not just in the Atlantic.

[am19psu]

That little geographic area from like Philly-NYC north and east though contains a helluva lot of people and commerce and has a much bigger impact than you seem to be alluding to. Nailing a forecast over a third of the country in Midwest farm land is one thing but big time misses lately in densely populated northeastern regions with east coast cyclogenesis, is another thing.

The GFS has not been significantly worse than the Euro in PHL

[ORH_wxman]

He admitted that it hasn't been good in NE, but that it doesn't mean it's performed similarly everywhere else.

 

 

I don't think too many here have said that this has to be the case..maybe outside of a weenie or two. Its struggles here have been noted inside of 48 hours. The met discussion in here over the last month or so has very specifically stated how bad the GFS performs when the chips are down...i.e. large coastal cyclogenesis in the region.

[Damage In Tolland]

Which ones?

Lanza, Keggins, Gibbs to name a few

[am19psu]

It performed great when we had a fast flow in January without a lot of complex setups, but once we had storm opportunities with more blocking, it embarrassed itself inside of 48 hours. At least up in New England it did. I didn't follow it closely enough down in the Mid-Atlantic to make any claims about its performance during major cyclogenesis there.

Which is the whole problem with this thread. Just because it didn't perform well in YBY doesn't mean it doesn't perform well in, e.g. Chicago or Philadelphia.

[am19psu]

I don't think too many here have said that this has to be the case..maybe outside of a weenie or two. Its struggles here have been noted inside of 48 hours. The met discussion in here over the last month or so has very specifically stated how bad the GFS performs when the chips are down...i.e. large coastal cyclogenesis in the region.

Ask Kevin what he thinks

[am19psu]

Lanza, Keggins, Gibbs to name a few

I guarantee you Gibbs doesn't think that and I'd be shocked if Lanza thinks that. I don't know Keggins well enough to know what he thinks.

[Damage In Tolland]

I guarantee you Gibbs doesn't think that.

i KNOW he does

[ORH_wxman]

I mean, that's fine if you believe that in SNE. I'm telling you, it does fine in PHL and it basically the same as the Euro in the tropics for hurricane forecasting. And not just in the Atlantic.

 

Right, and that's great if we are forecasting hurricanes or Philly weather, but we aren't. It has a problem here with the short term in coastal systems and it isn't just confined to this year.

 

If you look in the pattern thread I made back in early February, you will see how we praised the GEFS for getting the idea more correct than the Euro ensembles for later that month at a 12-16 day lead time. The GFS bashing by me is not across the board on the model, it has its strengths and certainly has done great at times this year...but it has consistently failed in the short term during large winter storms for whatever reason.

[am19psu]

 The GFS bashing by me is not across the board on the model, it has its strengths and certainly has done great at times this year...but it has consistently failed in the short term during large winter storms for whatever reason.

Yeah, I know you know that, Will. The other weenies in this forum do not have the same beliefs.

[phlwx]

Lanza, Keggins, Gibbs to name a few

 

lol -- he just ripped on it on Facebook less than 48 hours ago.