Model output discussion - 7th January onwards

[Guest Delete Me]

GFS needs to be ignored until the federal government shutdown ends, NWS don't have the manpower to accurately model currently.

Lovely ECM run this morning, time to get excited? 

[Day 10]

14 hours ago, ghoneym said:

Regarding GFS output, I would have a read of this. Its still relevant. The government shutdown is affecting it, no doubt in that. https://www.washingtonpost.com/weather/2019/01/07/national-weather-service-is-open-your-forecast-is-worse-because-shutdown/?noredirect=on&utm_term=.c917a9b7b0ee

I would take the GFS and the FV3 with a very large rich of salt and on to compare long term trends, and even at that I would say don't trust it.

Thanks ghoneym.

I have pulled out this bit for peeps to read, maybe the GFS should be taken with a huge dose of salt at present.

 

But in the meantime, the current Global Forecast System — or the GFS — the United States' premier weather model, is running poorly, and there’s no one on duty to fix it.

“There was a dropout in the scores for all of the systems” on Dec. 25, Saha said of the scoring system used to rank how the forecast models are performing. “All of the models recovered, except for the GFS, which is still running at the bottom of the pack.” Not only does that mean the day-to-day weather forecast is worse, she said, it is also a national security risk.

Saha thinks it has to do with the data format. The model brings in data from all over the world, from dozens of different countries that are now standardizing the format to adhere to new regulations. The Environmental Modeling Center was working to adjust for the new formats when the shutdown started. Saha said that even though the Weather Service is getting the data, the GFS doesn’t recognize the format, so it can’t use it. And a model forecast is only as good as its input data.

[ArHu3]

10 minutes ago, Leo97t said:

Omg that is one of the largest differences between 240hr runs I've ever seen. Suspect ecm is gonna have egg on the face. When there is cross-section model disagreement. Always pick the less favourable solution - it's almost always right

GFS operational is a warmish outlier within the ensemble, now we have to wait for the ECMWF's ensemble to see where it lies

[Chris.R]

 I think it is indisputable that it is having some effect. Take this quote from the article linked above. 

Suru Saha, a union steward at the Environmental Modeling Center in College Park, Md., said the main impact has been on the National Weather Service’s new global forecast model, which was scheduled to go live in February but will surely be delayed because of the shutdown.

But in the meantime, the current Global Forecast System — or the GFS — the United States' premier weather model, is running poorly, and there’s no one on duty to fix it.

 

Edited January 15, 2019 by Chris.R

[O'Maille80]

13 minutes ago, Leo97t said:

Omg that is one of the largest differences between 240hr runs I've ever seen. Suspect ecm is gonna have egg on the face. When there is cross-section model disagreement. Always pick the less favourable solution - it's almost always right

Surely should go with what the majority of models go for? Seems counter intuitive to go for the odd one out?

[Guest Delete Me]

GFS is like any other modelling system, its only as good as the data being fed into it......and there's no readable data for it to output.

[Guest Delete Me]

Just now, O'Maille80 said:

Surely should go with what the majority of models go for? Seems counter intuitive to go for the odd one out?

We all wish that even when it runs at 100%

[Ali1977]

ECM mean at 216 , Op is supported strongly I’d say. 850s on mean are also much colder than the Op which is great.

Edited January 15, 2019 by Ali1977

[nick sussex]

6 minutes ago, karlos1983 said:

It will be a warm outlier in terms of 850’s I’d say. Synoptically probably a bit eager as well. Just my opinion But what a run it was! 

Staggered the gfs is still not budging! Look forward to hearing ncep thoughts when Nick reports back.

They haven’t said much yet in terms of which model they prefer .  It’s still 6 days out but they definitely will because the track will effect the longevity of the snow and amounts . It’s still quite early there so we’d expect an update later on. 

A bellwether though is how far inland the storm tracks , this will correspond to the downstream jet track. The more out at sea the flatter the downstream solution .

 

[bluearmy]

OMG .... do you guys really think that the NOAA people at cpc would be using gfs output in their charts if they weren’t confident about its reliability ????

Surely, its just nuts to think otherwise ........

anyway, the ec mean is supportive of the op pattern until days 9/10 when the spread becomes wide to our west and north and the NH profile begins to look like it could head in a few v different directions re tpv behaviour and where we may see blocking establishing at a decent latitude 

[karlos1983]

7 minutes ago, Ali1977 said:

ECM mean at 216 , Op is supported strongly I’d say. 850s on mean are also much colder than the Op which is great.

I stand corrected 

No mild outlier. So must be well supported 

Edited January 15, 2019 by karlos1983

[MattStoke]

It’s amusing that people think the GFS output is being affected, as if it isn’t like this every time our normal west to east flow is disrupted. It’s performance is always poor in this situation.

Edited January 15, 2019 by MattStoke

[Day 10]

5 minutes ago, bluearmy said:

OMG .... do you guys really think that the NOAA people at cpc would be using gfs output in their charts if they weren’t confident about its reliability ????

Surely, its just nuts to think otherwise ........

anyway, the ec mean is supportive of the op pattern until days 9/10 when the spread becomes wide to our west and north and the NH profile begins to look like it could head in a few v different directions re tpv behaviour and where we may see blocking establishing at a decent latitude 

Have you fully read the article, do you know that for sure? Why would they make something like that up?

It's a very unusual situation over there and the huge discrepancy between the GFS and others is maybe the reason why.

[nick sussex]

The best solution is the ECM at day 7 then just a touch flatter after that .

The reason for this is to get the low centre further east and se. You don’t want the low dropping too far west and want the direction of energy se .

If offered I’d take the ECM mean position at day 9 rather than the op.

[Nick F]

13 minutes ago, Chris.R said:

 I brought this up yesterday. There’s definitely  something not right with the GFS. With all the chaos going on with the shutdown there’s bound to be a significant effect. 

I think we need to draw the line under the whole ‘government shutdown affecting GFS’ thing, as @snowking points out, such an idea is unfounded and the Washington Post article is based on the view of one NOAA employee who is perhaps disgruntled with being on furlough.

GFS may perhaps be in a spell of poor performance related to the trop - strat interactions as the easterly winds in the lower strat at high latitudes start to imprint on the top of the troposphere. Given the EC has apparently better parameterization for upper layers, would favour its handling of flow changes from the strat down. 

Anyway, 00z EC det. For London well within the ensemble spread envelope, if a little on the mild side later in the run.

 

[Johnp]

10 minutes ago, bluearmy said:

OMG .... do you guys really think that the NOAA people at cpc would be using gfs output in their charts if they weren’t confident about its reliability ????

Surely, its just nuts to think otherwise ........

anyway, the ec mean is supportive of the op pattern until days 9/10 when the spread becomes wide to our west and north and the NH profile begins to look like it could head in a few v different directions re tpv behaviour and where we may see blocking establishing at a decent latitude 

Why would the Washington post fabricate a story on weather models?

Hardly a sexy topic is it?

[bluearmy]

4 minutes ago, Johnp said:

Why would the Washington post fabricate a story on weather models?

Hardly a sexy topic is it?

It’s called politics 

like I said, I would believe what people are actually doing with the model rather than what that are saying about it - actions speak louder etc etc 

anyway- a big change in the ec mean at day 10 reflects the mean ridge being much stronger just to our nw rather than mid Atlantic .... need to see the clusters to understand if there are two distinct camps or a general no idea on where to place the mean ridge from mid Atlantic across to scandi 

[nick sussex]

As far as I know the input data into the model is automatic . Someone doesn’t sit there inputting observations from say aircraft and balloon data .

It used to be that some Stateside used to call the GFS GIGO. Garbage in garbage out! As some believed its issues weren’t just the model but the data that went into it. I’m dubious of that theory.

 

 

[Day 10]

4 minutes ago, Nick F said:

I think we need to draw the line under the whole ‘government shutdown affecting GFS’ thing, as @snowking points out, such an idea is unfounded and the Washington Post article is based on the view of one NOAA employee who is perhaps disgruntled with being on furlough. 

GFS may perhaps be in a spell of poor performance related to the trop - strat interactions as the easterly winds in the lower strat at high latitudes start to imprint on the top of the troposphere.

 

Ok Nick fair enough, let's just hope the GFS is just being GFS = wrong

[Yarmy]

I don't know if the GFS is affected by the shutdown or not; there are conflicting reports, as we see above. What I do know is that its verification stats are currently atrocious, as bad as I've seen them:

Trailing the GEM and barely ahead of the NAVGEM. Its new sibling is faring better, but still way behind the Euros.

 

 

[Met4Cast]

EPS mean shifts the high more towards Greenland/NW of the UK rather than sitting it in the mid-Atlantic like previous runs

Extended EPS much the same though more watered down re: Euro troughing, clusters will be more telling to see just how much support the EC Op has

[smhouston]

Find the whole "GFS isn't get data which is why it's rubbish" laughable. People werent claiming this few days ago when it started coming in to line with the other models. As soon as it goes on a different tangent then out comes the conspiracy theories to somehow justify its wrong rather than picking up on a trend that things may not be as cold and rosey as it may seem.

Not surprising an employee would come out with it when they're not getting paid. Threat to national security? Hmmm

As for incompatible data formats, yeah maybe true, but a person will not be converting the data by hand, there will be software routines doing so

 

Anyway, disappointing to still see the GFS not swing back, and even more so the UKMO taking a slight step away from cold. Hopefully not a trend? As ever though, FI is T96. No model handles cold scenarios well. Could argue ECM usually is too agressive with cold and usually waters it down nearer the time, purely based on what seen in the past

 

[Blessed Weather]

OK folks. I think it's time to move on from the "is the GFS impacted by Gov shutdown" debate. The simple fact is we don't know based on media reports and quotes from one employee (a union rep) and I can't find ANYTHING on the NOAA/NCEP website to suggest there's a problem. If anyone can find official info please post it, otherwise let's draw a line for now. Thanks.

Edited January 15, 2019 by Blessed Weather

[matty40s]

12 minutes ago, smhouston said:

Find the whole "GFS isn't get data which is why it's rubbish" laughable. People werent claiming this few days ago when it started coming in to line with the other models...

 

Pray tell us which day this was.....☺

[mulzy]

Ext EPS broadly similar to last night though less intense heights to the north-west than what was showing last night.  Though not as good as last night, a very good ensemble mean.

Though still a risk, the position of the mean trough (over the UK rather than to the west of the UK) has been trending east which is what we want to see.