Changes needed for GFS Update 2017?

[j0nes2k]

Hello,

next week the GFS model will change: http://www.nws.noaa.gov/os/notification ... pgrade.htm

Do you know if there are any changes necessary for EMS or will this update work seamlessly? I am running EMS v15.58.5 and I am using the 0.25 degree gribfiles from Nomads servers.

Best regards,

Jonas

[meteoadriatic]

I don't know about EMS but I highly suspect that answer is clearly yes, Robert will need to issue update.

There are three things that has to be updated: 1) ungrib binary, 2) Vtable.GFS, 3) METGRID.TBL.ARW

I guess there will be update for that from Robert soon.
Ivan

[Howard]

Hi FYI

I have done a relatively low resolution run of new GFS ... temperature looks a mess. I am running a new run at 7km resolution and another at 3km or

Howard

[j0nes2k]

Hello,

regarding the original topic: there is a UEMS update available from Robert for working with the updated GFS files.

@Howard, I cannot confirm this. For my domains the temperature looks good, I cannot see any obvious differences to the "old" GFS.

However for me the runs are slower - can anyone confirm this? I am running autopost and the model itself on the same server (which was not an issue before), and at the moment I am not sure whether autopost or the model core itself is running slower.

Best regards,

Jonas

[Soseki]

Hello

I have not noticed any slow down since the GFS update, the ptiles on the EMS servers do seem to be arriving later than they use to (has been for a few weeks now). Maybe that might explain why your runs appear slower? Just an idea anyway.

[meteoadriatic]

I don't see how there can be a difference caused by new gfs, because, after metgrid phase everthing else should be exactly the same in the rest of EMS chain. With that being said; I took a look inside update file and there are some other updates aside from new GFS support, like fixing bug regarding aer_opt setting, new UPP libraries and other small things. At first glance nothing of that smells bad in terms of simulating speed performance, but I didn't checked in details. Maybe..... especially if you didn't had latest version before this update, then you might get other bigger changes. You should try to compare old and new namelist to see if there is something new or changed default inside.

[theocarter2911]

Hi All,

Anyone have a copy of the newest install.pl script? And I suppose second question, is that version working, or should I wait for Robert to issue a new one?

I have been hanging onto an older version of the model so as to keep running nmm (for comparison reasons against arw), but the new gfs update has broken my model, getting rubbish surface humidity and temperature fields. So I am going to have to try and rebuild tonight if the last available install.pl version does in fact work.

Best regards,
Theo

[meteoadriatic]

Hello,
http://gamma.meteoadriatic.net/test/uems_install.pl.gz

But you can't run NMM with this...

Let me prepare for you changes to run NMM; just a minute.

[theocarter2911]

You are a lifesaver meteoadriatic!

[meteoadriatic]

OK, here are the changes for NMM

http://gamma.meteoadriatic.net/tmp/NMM/

This ungrib is from WPS 3.9 with patch for new GFS, compiled statically so it should work with any system. Also, that one from Robert's update should work also in NMM but didn't tested it myself. Vtable.GFS should be placed in data/tables/vtable and METGRID.TBL.NMM in data/tables/wps

That should be all you need to change. Let me know if it works.
Ivan