Scheduled Downtime
On Friday 21 April 2023 @ 5pm MT, this website will be down for maintenance and expected to return online the morning of 24 April 2023 at the latest

Change in soil variables from ECMWF Cycle 49r1

Hi Ana,
I checked your METGRID.TBL file and it is ok. I inserted it into my system and it worked, everything was calculated without errors. I am sending my Vtable file and rd_grib2.F file for WRF 4.6.1 model just in case. Please check the calculations with these files. It must work. It works in my case.
Mariusz
 

Attachments

  • rd_grib2.F
    51.6 KB · Views: 10
  • Vtable.ECMWF_GRIB2.txt
    3.9 KB · Views: 12
Hi @figurski

The problem was in the decompressed grib2 files. The method not decompresse correct.

The use of the Vtable and the rd_grib2.F are completes for the use of ECMWF IFS files.

Thank you for the assistence.

Ana Teena
 
I was trying to use the eccodes library in Python, but the problem was decompressing the GRIB2 files. The script did not create the process correctly, and the file size remained the same as the original file.
The search in another thread on the WRF forum mentions the ECMWF dataset GRIB2 in C3 compression. However, I modified the Python script and was able to decompress the GRIB2 file correctly.

Ana Teena
figurski said:
How do you decompress files?
M.
Click to expand...
 
weathermen said:
Can someone explain exactly, step by step, how to load the files from the European model? That is, as far as I understand, one can use the files from:
by making changes to the WPS using the rd_grib2 and Vtable files.
Or these files need to be converted first. I would be happy if you could send the scripts
Click to expand...

Use Mariusz's rd_grib2.F and Vtable and compile WPS.
In relation of the grib2 data this link Error when ungrib.exe the ECMWF IFS 0p25 dataset #257 explain the decompress grib2 files.

Ana Teena
 
Hi, everyone

I'm also encountering a similar problem here where I'm using ECMWF IFS dataset. I've followed the instructions as @figurski wrote earlier on. I've successfully decompressed the IFS grib2 files and managed to run WPS fine, but when I try to run ./real.exe I get this error.

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2815
Not enough soil temperature data for SSIB LSM scheme.
-------------------------------------------

I've checked the metgrid outputs and it says NUM_METGRID_SOIL_LEVELS = 0. I'm not sure what went wrong, do I have to add a supplementary SST or external GFS output data?

Appreciate the help, thanks
 
Last edited:
aimuni said:
Hi, everyone

I'm also encountering a similar problem here where I'm using ECMWF IFS dataset. I've followed the instructions as @figurski wrote earlier on. I've successfully decompressed the IFS grib2 files and managed to run WPS fine, but when I try to run ./real.exe I get this error.

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2815
Not enough soil temperature data for SSIB LSM scheme.
-------------------------------------------

I've checked the metgrid outputs and it says NUM_METGRID_SOIL_LEVELS = 0. I'm not sure what went wrong, do I have to add a supplementary SST or external GFS output data?

Appreciate the help, thanks
Click to expand...
Which compilers are you using. I had a similar issue with the new Intel ones.
 
William.Hatheway said:
Sadly I don't.

But I would use the tools in the wps/Util folder to check your files to see if they have all the data.
Click to expand...
That's unfortunate. Is there a particular reason why metgrid interpolates to zero levels for soil whereas the original IFS data contains 4 levels
 
Hi everyone, I just want to share my experience of using the current IFS dataset. What I did to correct the soil level = 0 was utilize GFS dataset incorporated with the IFS dataset. I followed this tutorial https://www2.mmm.ucar.edu/wrf/users/docs/user_guide_v4/v4.3/users_guide_chap3.html#_Using_Multiple_Meteorological to combine both GFS and IFS dataset. Everything ran well and my soil level was corrected again and could run real accordingly.

But when I try to run wrf.exe, the process stops halfway and I get a segmentation error. I've attached my rsl and namelist files just in case someone might know a few tips to handle this. Thanks!
 

Attachments

  • rsl.error.0000
    4 KB · Views: 2
  • rsl.out.0000
    3.1 KB · Views: 1
  • namelist.input
    2.4 KB · Views: 2
Last edited:
aimuni said:
Hi everyone, I just want to share my experience of using the current IFS dataset. What I did to correct the soil level = 0 was utilize GFS dataset incorporated with the IFS dataset. I followed this tutorial https://www2.mmm.ucar.edu/wrf/users/docs/user_guide_v4/v4.3/users_guide_chap3.html#_Using_Multiple_Meteorological to combine both GFS and IFS dataset. Everything ran well and my soil level was corrected again and could run real accordingly.

But when I try to run wrf.exe, the process stops halfway and I get a segmentation error. I've attached my rsl and namelist files just in case someone might know a few tips to handle this. Thanks!
Click to expand...
That's a good work around, i appreciate it, but somehow we still need to figure out how to use the IFS soil data, because it might be different then GFS
 
aimuni said:
Hi, everyone

I'm also encountering a similar problem here where I'm using ECMWF IFS dataset. I've followed the instructions as @figurski wrote earlier on. I've successfully decompressed the IFS grib2 files and managed to run WPS fine, but when I try to run ./real.exe I get this error.

-------------- FATAL CALLED ---------------
FATAL CALLED FROM FILE: <stdin> LINE: 2815
Not enough soil temperature data for SSIB LSM scheme.
-------------------------------------------

I've checked the metgrid outputs and it says NUM_METGRID_SOIL_LEVELS = 0. I'm not sure what went wrong, do I have to add a supplementary SST or external GFS output data?

Appreciate the help, thanks
Click to expand...
Hi @aimuni,

This problem is related to the ECMWF IFS dataset. Did you decompress the GRIB2 file?
Setting NUM_METGRID_SOIL_LEVELS=0 means that the soil variables are not being read by the model.
Try decompressing the GRIB2 files using ecCodes and then run WPS and WRF.

Ana Teena
 
Ana Teena said:
Hi @aimuni,

This problem is related to the ECMWF IFS dataset. Did you decompress the GRIB2 file?
Setting NUM_METGRID_SOIL_LEVELS=0 means that the soil variables are not being read by the model.
Try decompressing the GRIB2 files using ecCodes and then run WPS and WRF.

Ana Teena
Click to expand...
Hi, Ana

Thanks for your help. I've actually managed to figure it out as stated in post #73 but I'm having trouble with running wrf.exe. I get a segmentation error. Have you encountered similar problem?
 
aimuni said:
Hi, Ana

Thanks for your help. I've actually managed to figure it out as stated in post #73 but I'm having trouble with running wrf.exe. I get a segmentation error. Have you encountered similar problem?
Click to expand...
Hi, @aimuni
I had a similar problem, which was caused by the OpenMPI library used to compile the WRF model with gcc/gfortran on a machine with AMD EPYC processors. In my case, the error occurred when running real.exe and wrf.exe. The solution was to use MPICH instead of OpenMPI and recompile the model.
In your case, the problem might be related to the combined use of ECMWF and GFS datasets, or perhaps you're using other initial conditions?
 
hi, this is how we got it working- using the wonderful contribution from @figurski above as the basis.

Worth saying this is for the data on the ECMWF website- our interest in in the forecast products only. e.g. https://data.ecmwf.int/forecasts/20250528/00z/ifs/0p25/oper/

Source code change
The source code changes are in the rd_grib.F file in WPS (in WPS/ungrib/src), so swap that out for the one provided by @figurski and recompile WPS.

Prepare new Vtable
The new Vtable to read the data is also provided by @figurski above- rename that to whatever you want before linking to it and running ungrib.

Install eccodes
You need this to access the grib_set tool which decompresses the grib data first. I found this easiest to do in a conda environment like so:

# Make the conda environment
conda create -n eccodes_env python=3.9
conda activate eccodes_env
conda install -c conda-forge eccodes

Step 5: decompress the grib files
# Activate the conda environment
conda activate eccodes_env

then run the grib_set commands, passing in the grib2 files and setting the desired output names e.g.
1748528734964.png
This example shows the command for 4 IFS grib files from the link above, the output file has ecmwf_ prepended to the beginning of the filename.

The output files above are what you need to link to with the link_grib.csh step in WPS.

Step 6: namelist.wps settings
I set my namelist.wps like this to account for the 14 vertical levels in the IFS and the 4 soil levels.
1748528763543.png

I think that’s all I did- massive thanks to Mariusz/@figurski for providing the solution. Hope my small contribution helps someone!
 
You must log in or register to reply here.
Top