Jump to content
  • Member Statistics

    17,588
    Total Members
    7,904
    Most Online
    LopezElliana
    Newest Member
    LopezElliana
    Joined

Saint Patrick's Day Snow Event II


stormtracker

Recommended Posts

  • Replies 827
  • Created
  • Last Reply

Thanks Chris. You know a lot more about this stuff than me. In my simpleton view I've always thought that the same things that make the nam do well with fine details early in the run make it become really inaccurate later on. Is this a fair assessment for a hobbyist?

 

Well, first I'll say you shouldn't be so hard on yourself! :)

 

But whatever this input is worth, I'd have to think constantly putting in BCs that are from a 6-hour older model (and a different type of model, though that's probably integrated well enough) would eventually deteriorate the solution even if the NAM's physics/thermodynamics are good.  The (6 hour older) GFS, of course, will also have more error over time as its forecasts are used as BCs, which I'd guess adds to that.  If this makes any sense.

 

Not sure if that totally explains the really crazy ULL and surface development it's showing for the current situation, since that wave is now well in the NAM's grid range.  That might be from something totally different as far as I know.

Link to comment
Share on other sites

    yes, that's basically it in a nutshell.    But the key is that, because we are running the NAM before the GFS, the 6-hr old GFS has to supply the boundary conditions.    In a perfect world, the GFS would run first, and then the NAM would use it for boundaries.

doesn't seem very hard to make it perfect...run the GFS first

Link to comment
Share on other sites

    yes, that's basically it in a nutshell.    But the key is that, because we are running the NAM before the GFS, the 6-hr old GFS has to supply the boundary conditions.    In a perfect world, the GFS would run first, and then the NAM would use it for boundaries.

Very good point, concerning using a 6-h old model.  I know this in some way has to do with the supercomputer usage, and when you can schedule in various models (and that schedule is already very tight).  But it would be interesting to see what would happen if the 00Z GFS is used on the 00Z NAM; of course, that would mean the NAM and GFS would come out around the same time, assuming they'd be run almost concurrently.

Link to comment
Share on other sites

Thanks Chris. You know a lot more about this stuff than me. In my simpleton view I've always thought that the same things that make the nam do well with fine details early in the run make it become really inaccurate later on. Is this a fair assessment for a hobbyist?

That's a difficult question -- I think in a case like this the fact that it is a regional model hurts it more than resolution -- then you can get into a scenario where certain parameterizations may not be valid at higher resolutions and then eventually you can get into scales where we use discrete convection.

The ecmwf is a great model bc of the resources that are dedicated to its development. There are issues within our system that preclude us getting on level ground (not really something I'll get into now). Not sure who mentioned it but first two things I'd do would is limit the forecast period for the NAM to 48 hrs then eliminate the 6/18 cycle for a more robust (higher res) global ensemble system (also I'd rework the SREFs -- no ensemble system should have as extreme of a spread as we often see there on short lead times).

In general, I think we focus too much on going to higher resolutions without fully understanding how that affects accuracy -- most people adept at NWP would agree that future improvements in NWP has less to do with higher resolutions and more to do with better DA and improvements in parameterizations

Link to comment
Share on other sites

That's a difficult question -- I think in a case like this the fact that it is a regional model hurts it more than resolution -- then you can get into a scenario where certain parameterizations may not be valid at higher resolutions and then eventually you can get into scales where we use discrete convection.

The ecmwf is a great model bc of the resources that are dedicated to its development. There are issues within our system that preclude us getting on level ground (not really something I'll get into now). Not sure who mentioned it but first two things I'd do would is limit the forecast period for the NAM to 48 hrs then eliminate the 6/18 cycle for a more robust (higher res) global ensemble system (also I'd rework the SREFs -- no ensemble system should have as extreme of a spread as we often see there on short lead times).

 

Out of curiosity (since this has come up from time to time), what are your views of running the GFS to 384-h for four cycles?  I've often wondered what if it's run to 384 at 00Z and 12Z, but only out to something like 240 at 06/18.  I know some of the reasons for the GFS ops and ensemble to run out the entire 384 is for the climate people, from what I understand.

Link to comment
Share on other sites

Out of curiosity (since this has come up from time to time), what are your views of running the GFS to 384-h for four cycles? I've often wondered what if it's run to 384 at 00Z and 12Z, but only out to something like 240 at 06/18. I know some of the reasons for the GFS ops and ensemble to run out the entire 384 is for the climate people, from what I understand.

There are others who may have more direct experience with how these decisions are made but based on my experiences with other parts of our scientific community is that managers are reluctant to change something that's already produced operationally, even if proven that those resources could be used elsewhere.

Link to comment
Share on other sites

Can we please move technical model discussion out of here, it's sickening! Meanwhile, GFS 0z will make northern folks happy.

 

Awww, come on!  Nothing like a little discussion of partial differentials, integration, boundary layer conditions, Laplacians, etc., especially when I've had a glass or so of wine!!! :P

 

(ETA:  Wait...Laplacians?  Who let the Laplacians in here??? :lol: )

Link to comment
Share on other sites

There are others who may have more direct experience with how these decisions are made but based on my experiences with other parts of our scientific community is that managers are reluctant to change something that's already produced operationally, even if proven that those resources could be used elsewhere.

 

Thanks...I've truly appreciated your input on all this stuff.  And I'm all too aware of changing something that's been kind of entrenched for some time.

Link to comment
Share on other sites

Absolutely ridiculous how different the GFS is compared to almost all other models. If it's right, well it's been superior with this storm, if it's wrong then it has been clueless and sucks!

It's been very steady since 6z.. and arguably prior if you discount the wonky qpf. That said, the Euro has crushed it on southern stream waves so I'm assuming it's wrong at least in part.

Link to comment
Share on other sites

doesn't seem very hard to make it perfect...run the GFS first

 

    I agree, but the NWS regions were very clear years ago that they wanted the regional model first.   The GFS has to wait for all of the global data to become available, and we can't run the NAM and GFS simultaneously, so you wouldn't get the NAM starting until very late.   

Link to comment
Share on other sites

For what its worth and probably not a whole lot...comparing current radar to the initialization of both the GFS and the NAM, it looks a little closer to the GFS to me....Anyway, Im expecting a similar tally to 3/3 but I am not sure what to think about snow ratios and if they are better or worse.

Link to comment
Share on other sites

Well, I like the RGEM

 

I know this form will B & M if you don't get snow up there, but we could use a nice event down here. 

 

Like Hi Res NAM and RGEM bullzeye this far out.

 

Interesting is that if you look at soundings, at KLYH wet bulbs are already below freezing at 7am. (hope that is correct)

 

I'm planning ratios at 7-1 once snow is falling. I figured between .5 and .8 falls as snow, so that gives me 3.5 to 5.6

NAM soundings have us below freezing (or very close) everywhere but the surface 3pm tomorrow. 

Link to comment
Share on other sites

    I agree, but the NWS regions were very clear years ago that they wanted the regional model first.   The GFS has to wait for all of the global data to become available, and we can't run the NAM and GFS simultaneously, so you wouldn't get the NAM starting until very late.   

imho, accuracy is worth the 2-3 hour wait

sure would remove from the NWS menu humble pie after so many busts from a flawed (or at least stale) model run

Link to comment
Share on other sites

Archived

This topic is now archived and is closed to further replies.

  • Recently Browsing   0 members

    • No registered users viewing this page.

×
×
  • Create New...