Spectral nudging with ERA5

[Abhi_nav]

Hi,

I am preparing to run a nested domain with chemistry (WRF-Chem V4.0) with outermost domain nudged by ERA5 reanalysis product. I have gone through namelist options for spectral nudging however, I am not sure how to provide ERA5 files as input during the WRF run. How will I make sure WRF read those files and use the namelist options? Thanks.

 

[Ming Chen]

In order to turn on spectral nudging, you will need to set grid_fdda = 2. The wavenumber nudged is selected in namelist (e.g., xwavenum, ywavenum, e.g. =3). Please choose the number so that (domain size)/(wavenumber)=~1000 km in each direction.

When you run real.exe with grid_fdda=2, in addition to wrfinput_d0* and wrfbdy_d01 files, a file named “wrffdda_d0* will be created. This file will bee read when wrf.exe is running.

 

[Abhi_nav]

Dear Ming,

I followed your suggestion and was able to create wrffdda file for the domain. I wish to ask you whether the new variables in wrffdda files are based on information available in the met_em files? Since I have prepared my met_em files from GDAS FNL (0.25 degree), does the simulation gets nudged with this information? Is there any way where I can create boundary conditions from GDAS FNL (met_em) and nudge the simulation with other dataset like ERA5 (ERA5 being used to create wrffdda files)? Thanks.

 

[Ming Chen]

Hi,
You are right that the information in wrffdda is derived from met_em files. If your met_em files are produced based on GDAS FNL, then yes WRF will nudge information from this GDAS FNL.
Technically you are able to produce wrfinput and wrfbdy based on FNL, and produce wrffdda based on ERA5. However, note that there might exist inconsistency between FNL and ERA5, and such inconsistency will affect your simulation.
Personally I prefer to run the same case using the same forcing data.

 

[Abhi_nav]

Hi,
You are right that the information in wrffdda is derived from met_em files. If your met_em files are produced based on GDAS FNL, then yes WRF will nudge information from this GDAS FNL.
Technically you are able to produce wrfinput and wrfbdy based on FNL, and produce wrffdda based on ERA5. However, note that there might exist inconsistency between FNL and ERA5, and such inconsistency will affect your simulation.
Personally I prefer to run the same case using the same forcing data.

Dear Ming,

Thanks a lot for clearing my doubts. You are correct about the inconsistencies that may get introduced from using two different datasets. I will follow your suggestion of using GDAS FNL for nudging as well as BC.

Since you said it is possible to use ERA5 for nudging and FNL for BC, I would like to know how to produce wrffdda files this way as an experiment. Thank you.

 

[Ming Chen]

To produce wrffdda form ERA5, you need to run ungrib.exe to extract all WPS-required data from ERA5, run metgrid.exe to put data on WRF grids. Then you need to run real.exe, which produces wrfinput, wrfbdy and wrffdda. You can discard wrfinput and wrfbdy since you have produced these two files from GDAS FNL, but you need to keep wrffdda and use it for grid nudging in your sensitivity experiment.
Hope this is helpful for you.

 

[Abhi_nav]

Dear Ming,

Yes I understood the steps. Thank you for your help.

 

[janac4]

Dear Ming,

Yes I understood the steps. Thank you for your help.

Hello Abhi_nav,
Have you tried to merge GDAS FNL files and ERA5 files in a run with nudging?
I work with GFS input files for my real-time forecasts, and was looking for a way to incorporate GDAS FNL files (with nudging) to improve my initial conditions. Do you think this could work? Would you recommend me a better workflow to achieve a warm state of the atmosphere to avoid wasting spin-up time at the beggining of every simulation?
Thanks a lot.

 

[Ming Chen]

if the GDAS FNL is observational data, I would say that many analysis datasets including ERA5 and GFS have already include GDAS data. I don't see any necessity to repeat the assimilation procedure.