Hi

I'd like to do a sensitivity experiment with OpenIFS 43r3 using smoother orography.
The plan is to use Tco199 but use orography at either Tco95 or coarser resolution.

Since the input file "sporog" is spectral, I'm thinking it should suffice to set high wavenumbers to 0 to smooth the orography. But I don't understand how the file is structured.
Is there a way to do this using either ecCodes, CDO or python?

Cheers
Joakim

5 Comments

  1. Glenn Carver

    Hi Joakim,

    I don't know off-hand what the layout of spectral coefficients is for stored GRIB data - have you looked in the IFS manuals for IFS CY43R3?  https://www.ecmwf.int/en/publications/ifs-documentation  If not there, I can find someone who would know.

    If you alter the orography the surface pressure field will then be out of balance and the model will generate some strong gravity waves at the start of the forecast until balance is restored. Might be worth reducing the timestep for the first couple of hours.

    Have a look in the IFS manuals at the description of the processing of the climatological fields. If memory serves me well, I believe there's a description in there of how the spectral filtering for orography is done and how smoothing is applied. In the past, I've applied a spectral filter over the higher wave numbers for spectral data (unrelated to IFS).  Depends exactly what you are trying to achieve with a smoother orography. Is this over the whole globe or just a specific region?


    Glenn

  2. Joakim Kjellsson

    Hi Glenn

    Thanks for your reply! I found the "sp2sp" function in CDO which upscales spectral data by padding with zeros. Exactly what I need.

    I see that IFS has both "lsmoro" and "sporo" files for input. One has orography on the reduced grid and the other is spectral. Are they both used, or is only the "sporo" file used?

    Cheers
    Joakim

  3. Glenn Carver

    Hi Joakim,

    Ah, good I didn't know about sp2sp.

    I would alter the orography field in the ICM* files for the experiment rather than lsmoro & orog which are in the climate.v015 directory.  Those files are only there for a few people who have their own way of creating the initial files, but we're now working on a project to provide a web portal at ECMWF for users to create their own initial files (bit like ERA-5 download).

    Glenn


  4. Joakim Kjellsson

    Hi Glenn

    Yes you were right. Turns out sporo is ignored, and only ICMSH seems to matter.

    I managed to use ecCodes to replace my orography with the "z" in the ICMSH file. And it works!

    There are some insane gravity waves in the first few hours, but I did not need to reduce the time step.

    A simple web portal would be awesome!

    Cheers
    Joakim

  5. Glenn Carver

    Glad the model worked ok. Another thing you could do reduce the orography successively, though this would involve creating multiple copies of the ICMCL file and doing restarts.

    You'll be on our list of test users to try the web portal, I promise!

    Cheers,  Glenn