EURO model side talk

[OSUmetstud]

Not sure how much to trust it, but isn't the DGEX doing just that? Last run I thought I saw a complete whiff?

it's using the GFS physics though. It's not exactly the NAM extrapolated.

[Clinch Leatherwood]

Its nice to see it be near the Euro at 84 hours but given how much the NAM is very frequently poor at that range I'm not sure how much we can believe it.

Think about that though, if it's frequently poor at this range is it good that it's in decent agreement with the euro?

500mb verifications will be very interesting with and without the amdar data outage we are seeing right now. As flights begin to resume I'd expect things (this run) start to solidify.

[NortheastPAWx]

I haven't seen the individual Euro ensembles... just the mean. Cisco down at HPC overnight said a number of individual members did support the operational. I'm going on their word. And all I really meant by that is I lean heavily towards the Euro on an 84 hour forecast at 500mb than I would the NAM at 84 hours.

It was posted here by another met that 37/50 members were a miss.

[thegreatdr]

He would have either been referring to the 12z EC and Canadian members or 00z GEFS members. The 00z Canadian and 00z ECMWF individual members are not available until 10-11z, or 5-6 am EST. I know I've recently posted this in other threads, but not sure if I have in this one yet.

DR

I haven't seen the individual Euro ensembles... just the mean. Cisco down at HPC overnight said a number of individual members did support the operational. I'm going on their word. And all I really meant by that is I lean heavily towards the Euro on an 84 hour forecast at 500mb than I would the NAM at 84 hours.

[ohleary]

Think about that though, if it's frequently poor at this range is it good that it's in decent agreement with the euro?

500mb verifications will be very interesting with and without the amdar data outage we are seeing right now. As flights begin to resume I'd expect things (this run) start to solidify.

Good first point, I don't know how to take all the contradiction about the Euro being great in that range, the NAM being lousy and them both looking similar...but the Euro still looks great.

I don't know how much of an influence this amdar thing is, first off, all the globals likely use it to some extent, and secondly the lost data wouldn't be from the area where this system is taking shape. If the dearth of data was out of the CONUS west coast, maybe more of an impact?

[thegreatdr]

Personally, I wouldn't go with a solution until it had the support of about 25% of the global ensemble members, but every forecaster has a different comfort level with risk.

DR

However more than half (in fact 75%) of its ensembles were a miss. That's ugly.

[Clinch Leatherwood]

Good first point, I don't know how to take all the contradiction about the Euro being great in that range, the NAM being lousy and them both looking similar...but the Euro still looks great.

I don't know how much of an influence this amdar thing is, first off, all the globals likely use it to some extent, and secondly the lost data wouldn't be from the area where this system is taking shape. If the dearth of data was out of the CONUS west coast, maybe more of an impact?

ohleary but some of the data is. Those euro planes fly into airports all over the eastern 1/2 of the country and those flights aren't taking place. So there are thousands of sampling points per day being lost just here.

The EC inputs about 45,000 datapoints a day, the CMC about 100. I think the US models are only in the upper hundreds/low thousands. I could be wrong on this, but that's what I found last night. The EC really built out that network with all the airlines over there.

[pazzo83]

ohleary but some of the data is. Those euro planes fly into airports all over the eastern 1/2 of the country and those flights aren't taking place. So there are thousands of sampling points per day being lost just here.

The EC inputs about 45,000 datapoints a day, the CMC about 100. I think the US models are only in the upper hundreds/low thousands. I could be wrong on this, but that's what I found last night. The EC really built out that network with all the airlines over there.

A lot of flights out of Europe resumed yesterday. Further, the only big airports to be significantly impacted were Heathrow, de Gaulle, and Frankfurt.

[tcutter]

ohleary but some of the data is. Those euro planes fly into airports all over the eastern 1/2 of the country and those flights aren't taking place. So there are thousands of sampling points per day being lost just here.

The EC inputs about 45,000 datapoints a day, the CMC about 100. I think the US models are only in the upper hundreds/low thousands. I could be wrong on this, but that's what I found last night. The EC really built out that network with all the airlines over there.

is this really true?? seems like a huge disparity of data input......

[WBRE Wx]

A lot of flights out of Europe resumed yesterday. Further, the only big airports to be significantly impacted were Heathrow, de Gaulle, and Frankfurt.

Lol. That's like saying Newark, Chicago and Atlanta! But you're right it is getting much better today... I still think tomorrow's 12z all around will lock onto a consensus. Whether it's because the Euro folds because of better data resuming or the others figure it out remains to be seen... but hey, this is why we all do this, right?

[Organizing Low]

ohleary but some of the data is. Those euro planes fly into airports all over the eastern 1/2 of the country and those flights aren't taking place. So there are thousands of sampling points per day being lost just here.

The EC inputs about 45,000 datapoints a day, the CMC about 100. I think the US models are only in the upper hundreds/low thousands. I could be wrong on this, but that's what I found last night. The EC really built out that network with all the airlines over there.

im not saying you are not right

but that seems almost unbelievable.

if that is the case, i will stop looking at all the other models.

EDIT on further read, i think you are talking of how much each model contributes to the network, with all models using the entire network of data points?

[vinylfreak89]

People are still focusing on the wrong thing here, its the energy in the gulf that the Euro keeps more compace and amplifies that allows this thing to become a monster. It is very similar to all other models with the trough axis and even timing now. Its that the euro keys on the stj energy and really amps up the low in the gulf allowing a sharper turn and bomb it up the coast. Other models are keying on the northern stream and washing out the stj energy. This has the storm take a wider turn with not as much digging. Its the developments in the gulf that make the difference and unfortunately in that regard the NAM is still closer to the GFS/UKMET/GGEM then it is the euro.

excelent point... the euro also is less strung out overall in the northern branch than most other models... the question is why is it keying so much on the southern branch feature? I'm going to re-examine some things in the 00Z euro.

[NortheastPAWx]

Without seeing them I don't know what that means... A miss for who? Everyone? New England? Raleigh? DC? Philly? Hazleton?

All I'm saying is there is some support. Not unanimous, but some. There are a few GFS Ensembles supporting the western track too. Not unanimous by any stretch, but some.

edit - meant to say "majority" not unanimous

I thought someone here said 37/50, not sure, but here's the non-enthusiasm:

Adding some fuel to the fire- the 00Z Euro ensembles are MUCH less impressed- kind of a moderate mainly coastal event-the Op is a big outlier. Sorry, this increases my unease that we are all (including myself for a while) getting way too worked up about what may end up being a nice event for some, and a complete bust for others.

[IsentropicLift]

If you take a look at the HPC discussion...it mentions that the higher res of the euro might be adding to its solution. I was a tad suprised to see them weigh the euro so heavily considering it seems to be an outlier. I've learned my lesson the hard way...when in doubt, trust the euro. The GFS seems to be about as usefull as an umbrella in a hurricane.

[Feb]

I thought someone here said 37/50, not sure, but here's the non-enthusiasm:

I never saw it but i did see a post that said their were 4 or five members west of the OP Euro.

[thegreatdr]

With the ECMWF, you might have a point. The EC is occasionally too phased with systems from 120 hours onward (sometimes too retrogressive or not progressive enough as well). If the next run or two trend eastward or weaker, it would be worth paying attention to. Besides, it's rare nowadays for any kind of slowing trend to occur for days without some later quickening trend occurring once the system is within the short range. The EC has done this enough to make me suspicious of its solutions beyond 108 hours.

DR

Aren't we at the range where notoriously the usual suspects lose these storms only to bring them back later?

[North Balti Zen]

Hey all. A plea from a lurker. Can folks that are not red-tagged or not giving model info take the chatter to another thread? I think that is in keeping with the request from Stormtracker et al. Thanks.

The analysis on the model runs today is crucial for a lot of us making travel plans, and being able to read the analysis without what people "think" based on nothing more than gut would be awesome.

[ohleary]

ohleary but some of the data is. Those euro planes fly into airports all over the eastern 1/2 of the country and those flights aren't taking place. So there are thousands of sampling points per day being lost just here.

The EC inputs about 45,000 datapoints a day, the CMC about 100. I think the US models are only in the upper hundreds/low thousands. I could be wrong on this, but that's what I found last night. The EC really built out that network with all the airlines over there.

From the graphs it looks like NCEP puts an average of about 38,000 amdar reports in the NAM daily, and about 49,000 in the GFS. Not sure how many pass QC though. Plus there are many other aircraft reports, tamdar, acars, camdar, pireps. Also, I believe a majority of the data over the oceans is from satellites so both still have that. NCEP model dump stats:

http://www.nco.ncep....wprod/realtime/

Today's 12Z GFS dumped 13,302 amdar reports, for example:

http://www.nco.ncep....x.summary.shtml

Sorry about the OT, but thought these links would be interesting to many.

[chris87]

this statement could not be more false....each GFS and ECMWF basically simulate the same number of obs...it is how they are assimilated which is the difference in the two models...

we're talking about millions of observations each cycle

is this really true?? seems like a huge disparity of data input......

[usedtobe]

With the ECMWF, you might have a point. The EC is occasionally too phased with systems from 120 hours onward (sometimes too retrogressive or not progressive enough as well). If the next run or two trend eastward or weaker, it would be worth paying attention to. Besides, it's rare nowadays for any kind of slowing trend to occur for days without some later quickening trend occurring once the system is within the short range. The EC has done this enough to make me suspicious of its solutions beyond 108 hours.

DR

Wasn't there an e-mail at HPC saying that the euro has even been something overdeveloping lows in europe? That's at least the rumor I've heard.

[NortheastPAWx]

I never saw it but i did see a post that said their were 4 or five members west of the OP Euro.

I'd be more concerned that the majority of members are a whiff, but to each his own I suppose.

[Clinch Leatherwood]

From the graphs it looks like NCEP puts an average of about 38,000 amdar reports in the NAM daily, and about 49,000 in the GFS. Not sure how many pass QC though. Plus there are many other aircraft reports, tamdar, acars, camdar, pireps. Also, I believe a majority of the data over the oceans is from satellites so both still have that. NCEP model dump stats:

http://www.nco.ncep....wprod/realtime/

Today's 12Z GFS dumped 13,302 amdar reports, for example:

http://www.nco.ncep....x.summary.shtml

Sorry about the OT, but thought these links would be interesting to many.

Excellent thanks. it's relevant to the topic, not sure it's OT at all. It's one aspect of dynamic modeling we rarely talk about and in reality it's one of the fastest growing sampling methods.

[Clinch Leatherwood]

this is statement could not be more false....each GFS and ECMWF basically simulate the same number of obs...it is how they are assimilated which is the difference in the two models...

we're talking about millions of observations each cycle

Right, and if the data doesn't exist it's not existing for all at the same time to be assimilated. What I don't know is how much sharing goes on between the EC and NCEP.

Time for the red taggers to take over as they get the data earlier than most of us in terms of the GFS.

[thegreatdr]

I'll have to look that up Wes, and I could ask Steve once he's back from Europe, because he's been looking at the weather over there on the forecast guidance on a regular basis. I know OPC was concerned about this with the new GFS over the open Atlantic ocean, which I noticed when it tried to strengthen Igor to a 920 hPa extratropical cyclone. That did not verify within 30 mb. As Bruce might say, holy smokes!

DR

Wasn't there an e-mail at HPC saying that the euro has even been something overdeveloping lows in europe? That's at least the rumor I've heard.

[Ed Lizard]

it's using the GFS physics though. It's not exactly the NAM extrapolated.

Typo. I think. DGEX, IIRC, uses NAM physics but initializes off the GFS. So its early hour solutions should look a lot like the GFS, and then diverge as different physics and finer grid scale resolution take over.

If I'm wrong again, be gentle.

[thegreatdr]

With the last Northeast "miss", a simple compromise of the GFS and ECMWF would have worked out wonderfully for most days, even through day 7. There have been numerous occasions other than the last miss where this has worked out fine since the GFS upgrade. From a pressure perspective, the GFS won during that event at longer ranges since it had a larger field of low pressure, and a stronger solution, 6-7 days into the future. And the overall pressure distribution is important because the wind grids used by NWS field offices are winds derived from the HPC pressures that far out into the future. Even though the surface low may not be all that important this time of the year for precipitation production (fronts aloft, the trowal, and the position of the warm conveyor belt circulation are more important), since we live at the surface, which has wind, there is importance placed on the overall pressure pattern. With this current forecast storm, it wasn't more than a couple days ago that the ECMWF had a 1040 high in southeast Canada. Even if its low position near the East coast is right, its winds will likely be overdone, per the most recent guidance (should it verify). =)

DR

If you take a look at the HPC discussion...it mentions that the higher res of the euro might be adding to its solution. I was a tad suprised to see them weigh the euro so heavily considering it seems to be an outlier. I've learned my lesson the hard way...when in doubt, trust the euro. The GFS seems to be about as usefull as an umbrella in a hurricane.

[dtk]

Right, and if the data doesn't exist it's not existing for all at the same time to be assimilated. What I don't know is how much sharing goes on between the EC and NCEP.

Time for the red taggers to take over as they get the data earlier than most of us in terms of the GFS.

In terms of observations, there is a general agreement between almost all of the centers (through the WMO) to share almost everything. Now, there are always potential issues with data transmission, or getting certain pipelines to work correctly with new observing systems....but overall, the same observations are/should be available by all the major operational centers for usage.

BTW, in terms of actual impact on reducing forecast error, aircraft obs (i.e. amdar, etc.) is pretty low on the list (I'll see if I can find the actual figure from collaborators at NASA GMAO for their system).

[Clinch Leatherwood]

In terms of observations, there is a general agreement between almost all of the centers (through the WMO) to share almost everything. Now, there are always potential issues with data transmission, or getting certain pipelines to work correctly with new observing systems....but overall, the same observations are/should be available by all the major operational centers for usage.

BTW, in terms of actual impact on reducing forecast error, aircraft obs (i.e. amdar, etc.) is pretty low on the list (I'll see if I can find the actual figure from collaborators at NASA GMAO for their system).

I posted elsewhere a link to some analysis of the impact. It wasn't huge, but it was definitely positive.

[DJ Amenzo]

This is an incredible thread. Intelligent discussion between mets and very useful information about data ingestion. Thanks guys. Its a refreshing break in the middle of my long day.