Use of sf_lake_physics

[GabrieleBentiv]

I am trying to run a WRF simulazione over the Alpine region, which works correctly unless I try to activate sf_lake_physics. In that case, it stops after a few seconds, with the last line in the output file being:
the total number of lake grid is : 0
(the result is the same with Noah and NoahMP)
What I did was using cglc_modis_lcz with 61 land uses and in the resulting wrfinput files I edited the LU_INDEX value of the water cells corresponding to lakes, by replacing their value of 17 with 21 (both domains have been edited).
I am linking the wrfinput files, the namelist, the ncl code I used and the rsl.out.0000 files (too big to be attached here): Proton Drive

The following is a screenshot of how the LU_INDEX field appears (as viewed with ncview):

 

[kwerner]

Hi,
Apologies for the long delay in response while our team tended to time-sensitive obligations. Thank you for your patience.

Since it's been a while, can you let me know if you're still experiencing this issue? If so, were you able to run wrf.exe to completion if you used cglc_modis_lcz as the geog data resolution and if :

1) you didn't turn on sf_lake_physics
2) you don't modify the wrfinput* files

 

[GabrieleBentiv]

Yes, I am curring running it correctly without the edited wrfinput files and the lake physics.

 

[kwerner]

What are the results of these two tests:

1) Using modified wrfinput* files, but turning sf_lake_physics OFF
2) Using UNMODIFIED wrfinput* files, with sf_lake_physics ON

And again, just to verify - when it's running correctly, you are still using cglc_modis_lcz for the geog data in wps?

 

[GabrieleBentiv]

The results in the two cases are:
1) The model runs with no issues, as if I didn't edit the file
2) Real is completed successfully but then, when running WRF, it stops immediately and the last lines in the rsl.out.0000 are the same as in the attachment provided in the initial question.

 

[kwerner]

Again, apologies for the delay. We've had back to back tutorials/workshops this summer and haven't had much opportunity to tend to forum inquiries.

I notice in the namelist.wps file, you have the following setting

geog_data_res = 'cglc_modis_lcz','cglc_modis_lcz',

What happens if you set that to

geog_data_res = 'cglc_modis_lcz+default','cglc_modis_lcz+default',

Does it make any difference when using lake physics?

 

[GabrieleBentiv]

In this case the operations fail at the real.exe step, with the attached rsl.out.0000 file

----- ERROR: sf_lake_physics = 1 AND flag_lake_depth = 0
-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 347
Either modify the namelist settings, or rebuild the geogrid/metgrid data

 

[kwerner]

Thanks for trying that. Upon further research, it looks like you may need to use 'cglc_modis_lcz+modis_landuse_20class_15s' instead of '...+default." See Ming's post in this thread, which explains that. She also provides some references that may be useful.

 

[GabrieleBentiv]

Thank you for the suggestion, now I get the files with the correct land use codes without the need to run the python code above. However, I still get the same message when trying to run real.exe
Indeed, the met_em files do not have the FLAG_LAKE_DEPTH set to 1. I am using the provided METGRID.TBL.ARW when running metgrid

 

[kwerner]

That's odd that the met_em* files have FLAG_LAKE_DEPTH=0. I've tried several different tests and can't seem to find a way to get it set to anything other than 1. Can you share the following files so I can test with your files? Since real.exe stops immediately, I only need to test with a single domain and just a few time periods.

• your latest namelist.wps file
• geo_em.d01.nc file
• FILE* intermediate files for just the first 3 hours
• met_em.d01* files for the first 3 hours
• The GEOGRID.TBL and METGRID.TBL being used - make sure to send the actual file that "GEOGRID.TBL" and "METGRID.TBL" are linked from

Finally, did you happen to make any modifications to any of the code in WPS?

 

[GabrieleBentiv]

I've added the files to the drive folder (I had to put % in place of : in file names because the files where passing through windows and it would not let me otherwise but that was not the original file name)

Proton Drive

I did not make any modification in the WPS code (v 4.6.0), just compiled it according to the instructions for my hpc system.

 

[kwerner]

Thanks for sending that. Do you happen to have the lake_depth static data in your WPS_GEOG directory? If you don't, you can get it from this section of the static downloads page. Unpack the file in the WPS_GEOG directory and there should then be a new lake_depth directory there. Then try running geogrid again. After, you can confirm whether lake_depth exists in the file, using this command:

ncdump -h geo_em.d01.nc | grep LAKE_DEPTH

If it prints out information, it means it's in there. Then it should be retained during the other executables and you shouldn't get the "sf_lake_physics = 1 AND flag_lake_depth = 0" error during real. If real works, then try WRF again and let me know if it runs.