Initialize MPAS with ERA5 data
- Thread starter Debjit_Paul
- Start date
Hi everyone,
Was anyone able or successful in using ERA5 to force or initialize MPAS?
I wish to use ERA5 datasets to force MPAS simulation, first using a ~60km uniform grid. I have read and followed through the discussions on this thread and was able to generate 'accepted' (no complaints or errors of missing variable(s)) intermediate files for MPAS initialization. However, the MPAS init_atmosphere_model stops with the error below:
At this point, I do not know how to proceed or overcome this error. I would be grateful if someone could help and come to my rescue.
I have attached a copy of the namelist, Vtable used, and the dump of a sample PRES and SST intermediate files generated by the ungrib from the ERA5 datasets.
Thanks
Makinde
Was anyone able or successful in using ERA5 to force or initialize MPAS?
I wish to use ERA5 datasets to force MPAS simulation, first using a ~60km uniform grid. I have read and followed through the discussions on this thread and was able to generate 'accepted' (no complaints or errors of missing variable(s)) intermediate files for MPAS initialization. However, the MPAS init_atmosphere_model stops with the error below:
At this point, I do not know how to proceed or overcome this error. I would be grateful if someone could help and come to my rescue.
I have attached a copy of the namelist, Vtable used, and the dump of a sample PRES and SST intermediate files generated by the ungrib from the ERA5 datasets.
Thanks
Makinde
nickheavens-cgg
New member
Dear all,
I think I have successfully managed to put together a reasonable MPAS-A initial condition file from ERA5 (at least one that does not have the problems identified by @wangevine). I'm just aggregating this information, because I had to figure out a working procedure from bits and pieces of information scattered around the Web, including these forums. I will assume familiarity with doing the analogous procedure with GFS. My version of MPAS-A is a slightly modified version of 8.0.1 that barely modifies the source code of init_atmosphere_model.
My procedure was this:
1. I downloaded ERA5 pressure level and surface field information separately from the new Copernicus Climate Beta repository using Amirhossein Nikfal's Python-based download codes under era5 in this package here: GitHub - anikfal/atmospheric_science: Codes and solutions for atmospheric data and modeling
[For simplicity, I edited them to download a single time slice. Note that you need to be set up properly with CDS API and accept the license for the data on the web site to access the data with the API. You also may need to do a few things to get around any firewalls you might have.] A. Nikfal has a helpful WRF-specific video here, but not all of the information is current or relevant to MPAS-A:
2. You need to concatenate the pressure level and surface field GRIB files, e.g.,
3. You then need to use ungrib.exe within WPS to ungrib the merged file:
a.
b. Change namelist.wps to the correct start and end dates; helpful but not necessary to change prefix under &ungrib to 'ERA5'
c. Link to the correct Vtable:
d. Run ungrib.exe
e. You get a file called ERA5:<timestamp>
4. You then can use init_atmosphere_model as you would for GFS, just changing the namelist.atmosphere setting config_met_prefix to a new file prefix if you used one.
I downloaded 10-25-2023, 0Z
Here is the surface_pressure field I get out of running convert_mpas.
Here is the qv field for level 10:
I think I have successfully managed to put together a reasonable MPAS-A initial condition file from ERA5 (at least one that does not have the problems identified by @wangevine). I'm just aggregating this information, because I had to figure out a working procedure from bits and pieces of information scattered around the Web, including these forums. I will assume familiarity with doing the analogous procedure with GFS. My version of MPAS-A is a slightly modified version of 8.0.1 that barely modifies the source code of init_atmosphere_model.
My procedure was this:
1. I downloaded ERA5 pressure level and surface field information separately from the new Copernicus Climate Beta repository using Amirhossein Nikfal's Python-based download codes under era5 in this package here: GitHub - anikfal/atmospheric_science: Codes and solutions for atmospheric data and modeling
[For simplicity, I edited them to download a single time slice. Note that you need to be set up properly with CDS API and accept the license for the data on the web site to access the data with the API. You also may need to do a few things to get around any firewalls you might have.] A. Nikfal has a helpful WRF-specific video here, but not all of the information is current or relevant to MPAS-A:
2. You need to concatenate the pressure level and surface field GRIB files, e.g.,
Bash:
cat level.grib single.grib > merge.grib3. You then need to use ungrib.exe within WPS to ungrib the merged file:
a.
Code:
./link_grib.csh <location of merge.grib>c. Link to the correct Vtable:
Bash:
ln -sf ungrib/Variable_Tables/Vtable.ERA-interim.pl Vtablee. You get a file called ERA5:<timestamp>
4. You then can use init_atmosphere_model as you would for GFS, just changing the namelist.atmosphere setting config_met_prefix to a new file prefix if you used one.
I downloaded 10-25-2023, 0Z
Here is the surface_pressure field I get out of running convert_mpas.
Here is the qv field for level 10:
Could you please tell me how you solved this problem? I also encountered the same problem.
@xiaoyu
Would you please clarify your issue? If it is solely related to HGT, then you only need to get the correct ERA5 file that contains SOILGHT.
Note that MPAS and WRF employ different vertical coordinate, and thereby this variable won't affect WRF initialization but it is important for MPAS initialization.
Would you please clarify your issue? If it is solely related to HGT, then you only need to get the correct ERA5 file that contains SOILGHT.
Note that MPAS and WRF employ different vertical coordinate, and thereby this variable won't affect WRF initialization but it is important for MPAS initialization.
I have not pursued this. The post above by @nickheavens-cgg describes a method that works for him. Good luck!