Execution of UNGRIB.exe using ERA5 data

[dqsantos]

Hello! I’m currently trying to make ungrib files using ERA5 data and eventually want to make metgrid files. However, there are 2 types of ERA 5 data: pressure level data and surface level data. Can you enlighten me by answering the following questions?

1. We need to make GRIBFILE.* (e.g. (GRIBFILE.AAA ~ GRIBFILE.ZZZ) by linking the ERA5 data we want to create an ungrib file of. Since there are 2 types of ERA5 data, should I do this step twice before running ungrib.exe if I want both types of data to be in the generated ungrib files?

2. Or…should I link the surface level data and run immediately the ungrib.exe, then repeat this for pressure level data? Hence, creating separate ungrib files for each type of data.

3. Before running ungrib.exe, we should also link our desired Vtable. I saw a tutorial of using Vtable.ERA-interim.pl to ungrib ERA5 data and what he did was the same as in my question in 2 (running ungrib.exe for each type of data). What is its difference with Vtable.ECMWF?

Ultimately, my goal is to create metgrid files with combined ERA5 pressure level data and surface level data. Is it possible?

 

[kwerner]

Hi,

1. We need to make GRIBFILE.* (e.g. (GRIBFILE.AAA ~ GRIBFILE.ZZZ) by linking the ERA5 data we want to create an ungrib file of. Since there are 2 types of ERA5 data, should I do this step twice before running ungrib.exe if I want both types of data to be in the generated ungrib files?

2. Or…should I link the surface level data and run immediately the ungrib.exe, then repeat this for pressure level data? Hence, creating separate ungrib files for each type of data.

For ERA5 data, you should be able to run ungrib only one time. You can link in both the surface and the pressure-level data and the Vtable should have the codes for both.

3. Before running ungrib.exe, we should also link our desired Vtable. I saw a tutorial of using Vtable.ERA-interim.pl to ungrib ERA5 data and what he did was the same as in my question in 2 (running ungrib.exe for each type of data). What is its difference with Vtable.ECMWF?

Originally, Vtable.ERA-interim.pl was intended to be used with the pressure-level ERA-interim data, and Vtable.ECMWF was for ECMWF data; however, I believe it has been found that both of these Vtables work with ERA5 data. Take a look at How to Process ERA5 Data, which describes the process, including the inclusion of an INVARIANT file that you'll need.

 

[dqsantos]

Hello kwerner! Thanks for answering. Can make some follow up questions? Sorry, I’m really new to this.

For ERA5 data, you should be able to run ungrib only one time. You can link in both the surface and the pressure-level data and the Vtable should have the codes for both.

Can you make a step-by-step procedure how to link both the pressure and level data in GRIBFILE.*? What I did was execute ./link_grib.csh for the surface level data and it generated the GRIBFILE.AAA. Then, I did the same for the pressure level data and it generated the GRIBFILE.AAB. Is this correct?

Originally, Vtable.ERA-interim.pl was intended to be used with the pressure-level ERA-interim data, and Vtable.ECMWF was for ECMWF data; however, I believe it has been found that both of these Vtables work with ERA5 data. Take a look at How to Process ERA5 Data, which describes the process, including the inclusion of an INVARIANT file that you'll need.

What does the INVARIANT FILE do? What difference does it make to the met*.nc files when running metgrid.exe without adding this in the namelist.input?

 

[kwerner]

Can you make a step-by-step procedure how to link both the pressure and level data in GRIBFILE.*?

Let's say you have multiple surface files (with a prefix of "sfc") and pressure-level files (with a prefix of "pl") in a single directory called "data." Because the files have different prefixes, you won't be able to link them in using their prefix, so instead you'll need to link in everything in the "data" directory. Just make sure there is nothing else in that directory that you won't need - just the files you'll use. Then from the WPS/ directory, issue

./link_grib.csh path-to-data-directory/data/*

The star ( * ) indicates you want to link all files in that directory.

What does the INVARIANT FILE do? What difference does it make to the met*.nc files when running metgrid.exe without adding this in the namelist.input?

I'm not sure of the specifics in the INVARIANT file, but it contains some important fields that don't vary with time. This is just something that has to be included with the ERA5 files. If you're missing it, you will likely get an error later during the real.exe process.

 

[dqsantos]

Thanks for replying again, kwerner.

I do have one more question: The ERA5 surface level data (sfc) and pressure level data (pl) are placed in different folders/directory for the sake of organized documentation. Is it possible to link those 2 directories simultaneously?

With regard to the invariant file, I didn’t include it while executing real.exe and I didn’t encounter any error. I was able to generate the wrfbdy_d* and wrfinput_d* files.

 

[kwerner]

I don't believe there is a way to link the two directories simultaneously. You could simply copy the surface data into that same directory with the pressure-level data, issue the link_grib.csh script, and then move the sfc data back out of the directory to keep it organized how you want it.

I suppose if you go through the entire wrf process and your results are reasonable, and as expected, you can get away with not using the invariant file. You'll just have to test that out.

 

[dqsantos]

Hi kwerner,

Thanks for patiently answering my questions. You were really of a big help.

Kudos to you!