Jump to content
Snow?
Local
Radar
Cold?
IGNORED

Gfs And Different Input


johnholmes

Recommended Posts

Posted
  • Location: Lincoln, Lincolnshire
  • Weather Preferences: Sunshine, convective precipitation, snow, thunderstorms, "episodic" months.
  • Location: Lincoln, Lincolnshire

It is worth noting that prior to the easterly blast of late Dec '05, the 06Z repeatedly went for the easterly, and the 12Z repeatedly went for a milder solution before gradually falling into line with the 06Z runs.

Link to comment
Share on other sites

Posted
  • Location: Co Dublin, Ireland
  • Location: Co Dublin, Ireland

Hi everyone, I know this is a GFS help thread but I didnt want to open yet another topic on the GFS so I hope you dont mind.

Basically I dont believe the GFS should be available beyond 168hrs to the public simply because of the amount of ridiculous nonsense it comes out with beyond this range on a daily basis. It is, in every sense of the word, useless beyond this range. It begs the question though, why make it available at this range? Whats to gain? Absolutley nothing is the best answer for that I think. I mean, if you've had a few drinks, and want a few laughs, a great place to start would be FI on the GFS (not refering specifically to last nights insane 18z).

At closer range I find it very useful and the detail is magnificent, however the 9 other runs associated with each operational run is bizzarre simply because logic should be that the operational run is the average of all the runs, currently it clearly isnt. What this means as far as I can assertain is that (and ive seen this) as close as 42 hours you can have a cold or mild outlier. That would be my main concern when looking at a single run. Hope I havnt gone off the beaten track of this thread but I think its a genuinely important point :lol:

Edited by Darkman
Link to comment
Share on other sites

Posted
  • Location: Beccles, Suffolk.
  • Weather Preferences: Thunder, snow, heat, sunshine...
  • Location: Beccles, Suffolk.

But you can't blame the GFS just because some people are stupid?? :lol:

Link to comment
Share on other sites

Posted
  • Location: Aviemore
  • Location: Aviemore
Hi everyone, I know this is a GFS help thread but I didnt want to open yet another topic on the GFS so I hope you dont mind.

Basically I dont believe the GFS should be available beyond 168hrs to the public simply because of the amount of ridiculous nonsense it comes out with beyond this range on a daily basis. It is, in every sense of the word, useless beyond this range. It begs the question though, why make it available at this range? Whats to gain? Absolutley nothing is the best answer for that I think. I mean, if you've had a few drinks, and want a few laughs, a great place to start would be FI on the GFS (not refering specifically to last nights insane 18z).

At closer range I find it very useful and the detail is magnificent, however the 9 other runs associated with each operational run is bizzarre simply because logic should be that the operational run is the average of all the runs, currently it clearly isnt. What this means as far as I can assertain is that (and ive seen this) as close as 42 hours you can have a cold or mild outlier. That would be my main concern when looking at a single run. Hope I havnt gone off the beaten track of this thread but I think its a genuinely important point :lol:

I posted this into the winter model thread earlier, which explains a bit about the use of the GFS beyond 7 days:

There is some concern about NCEP model variability from run to run, particularly at extended ranges beyond 7-8 days. The American Meteorological Society Statement adopted on January 14, 2001, titled "Statement on Seasonal to Interannual Climate Prediction" (Bull. Amer. Met. Soc., 82, 701), states, in part, the limit of predictability for day-to-day weather changes for deterministic forecasts is "in the range of 1-2 weeks." NCEP agrees with this statement and notes that for current state-of-the-art models, the limit of "useful skill" is about 7-8 days. Beyond that range, deterministic numerical weather predictions are characterized by little or no skill and suffer from large run to run variability. However, ensemble or probability forecasts may offer improved predictive skill.

Another AMS statement adopted January 13, 2002, titled "Enhancing Weather Information with Probability Forecasts" (Bull. Amer. Met. Soc., 83, 450), praises the value of probability forecasts and endorses ensemble model forecasts as a means for improving skill levels in the extended ranges while quantifying the level of certainty (uncertainty) in any forecast. NCEP agrees with this statement.

NCEP is working to making these ensemble forecasts products more readily available. In the meantime, we recommend using the GFS, or any global forecast system, with care beyond day 7 and recognizing the inherent variability in long range deterministic forecasts. Also recognize that probability forecasts, while providing much more flexibility and value to the user, require the user be trained to properly understand and use the new product suites based on ensemble model output. Some relevant training material can be found at http://www.hpc.ncep.noaa.gov/ensembletraining/

Wrt your point about having outliers at close range, I think you may be misunderstanding how the ensemble system works. It isn't simply that the control run is the mean of the runs. The control run is the run that uses the raw data input - untouched (aside from quality control). The other ensemble runs have the data adjusted - 5 negatively and 5 positively. This is because no matter how much data you put in and how much qc you apply to it, you can never have a perfect dataset to start. So by changing the data slightly, you can show a range of likely possibilities based on slightly different starting positions.

Invevitably this will mean that even at closer range, some of these runs will diverge - as they have different data within them.

Link to comment
Share on other sites

Posted
  • Location: Co Dublin, Ireland
  • Location: Co Dublin, Ireland
I posted this into the winter model thread earlier, which explains a bit about the use of the GFS beyond 7 days:

Wrt your point about having outliers at close range, I think you may be misunderstanding how the ensemble system works. It isn't simply that the control run is the mean of the runs. The control run is the run that uses the raw data input - untouched (aside from quality control). The other ensemble runs have the data adjusted - 5 negatively and 5 positively. This is because no matter how much data you put in and how much qc you apply to it, you can never have a perfect dataset to start. So by changing the data slightly, you can show a range of likely possibilities based on slightly different starting positions.

Invevitably this will mean that even at closer range, some of these runs will diverge - as they have different data within them.

Cheers Paul, I read the post in the model thread about the variability and it pretty much confirms everything the more experienced forcasters are saying on the forum. The GFS carries a health warning, like every model beyond a week in advance and my take on it now is that, that far ahead is best left untouched. I suggested that the control run would be better being the mean run not that it is the mean run. A valuable insight you have pointed out there about the raw data, certainly worth keeping in mind and perhaps even more significantly the different data within each run. Thanks again. :lol:

Edited by Darkman
Link to comment
Share on other sites

Posted
  • Location: Barnet, North London
  • Location: Barnet, North London

I would also add, it's worth hoping that one day the models may be pretty accurate out to T252 and beyond.

That's what the scientists (and their 'puters) are striving for. If you don't publish any info beyond say T96, how will you ever expand on your model accuracy?

Did I say that right?!! :lol:

Smich

Link to comment
Share on other sites

  • 8 months later...
Posted
  • Location: just south of Doncaster, Sth Yorks
  • Location: just south of Doncaster, Sth Yorks

tks for that gust will have a look when I've got some spare time to try and digest what it shows.

Welcome to Net Wx.

A favour please?

Can you put your nearest town in your Avatar.

many tks

John

Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
×
×
  • Create New...