NOAH-MP error with GFS

[gverri]

Hi,
I recently moved to WPS V3.9.0.1 because I need to deal with the current version (July 2019) of NCEP GFS forecast files as forcing data.
WRF is still v3.6.1 and it works well when I perform forecast experiments while working with NCEP GFS analyses on July 2018 I face the following error and wrf.exe blows up:

critical problem: ZLVL <= ZPD; model stops
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 4160
STOP in Noah-MP
-------------------------------------------

I checked metem files and real.exe and everything seems to be normal.
Thanks for help

Giorgia

 

[kwerner]

Giorgia,

I am a little confused by your last sentence. Do you mean that you have no problem when using older data, along with WPSV3.9.0.1 and WRFV3.6.1?
Can you attach a couple times periods of your met_em* files, along with your namelist.input file so that I can try a test here? And just to verify - is your V3.6.1 version of WRF non-modified (as in, you have not made any modifications to the code)? Thanks!

 

[gverri]

Hi,
Thanks for your promptly reply.
To be more clear: I'm working with WPS 3.9.0.1 + WRF 3.6 (I did no changes to the code)
It seems to me that the model set up works correctly after the last release of GFS data. Thus I attached as an example an experiment I performed on August 2019
We fail when we try to perform simulation in the past, I attached an experiment on July 2018
I hope I clarified
Thanks
Giorgia

 

[kwerner]

Hi,
I ran the following tests:
1) your input, your namelist, V3.6.1 - I get the same error that you're seeing.
2) your input, your namelist, V4.1.2 - it runs fine without errors
3) your input, my own basic namelist (only modifying necessary dimesions, etc.), V3.6.1 - I still see the same error.
4) I created my own WPS input (with V3.9.0.1), with GFS files (0.25 degree data) for the exact same times, using your dimensions, etc., ran wrf with your namelist, V3.6.1 - it runs without a problem.

So it seems to be a problem with your particular met_em* files. I'm not sure what the difference was in our WPS set-up. If you'd like me to create WPS met_em* files, using your exact WPS namelist, attach that, and I'll give that a shot. Whatever the problem was, it seems that it was resolved, likely with Noah-MP code updates between V3.6.1 and V4.1.2. There have been a LOT of code changes in that routine since then, though, so it would be quite difficult to narrow that down. I don't suppose using WRFV4.1.2 is an option for you?

 

[gverri]

Hi,
thanks for support.
I'd like to catch up with your test 4. Please find enclosed WPS namelist, and also Vtable and METGRID.TBL
Maybe, am I using the wrong tables?
Thanks
Giorgia

 

[kwerner]

Hi Giorgia,
Okay, I used your namelist.wps file, your Vtable, and your METGRID.TBL, with the GFS files I downloaded, and when I run with WRFV3.6.1, it still runs okay. The only difference I made in the namelist.wps file is to use the default static landuse setting (geog_data_res = 'default', 'default') as I don't have your srtm and eudem datasets. So the only 2 differences seem to be:
1) the actual GFS files we are using
2) your geog static data

So I would like for you to do 2 tests.

1) I am going to provide the GFS files I'm using. It was too large to attach, so you'll need to download it from this link (look for gverri_gfs_files.tar.gz):
https://nextcloud.mmm.ucar.edu/nextcloud/index.php/s/o8nYwN3nHMQr1Ov
Can you use these for a short (3 hour) run to see if they make any difference? You'll need to re-run ungrib, metgrid, real, and wrf.

2) If the above doesn't make a difference, can you set geog_data_res = 'default', 'default', rerun geogrid, metgrid, real, and wrf, and let me know if there is any change.