I was approached by a colleague of mine who was asking about the feasibility of calculating chemical tracer transport with OpenIFS. I did some research and found three different project in this direction.

Firstly there is IFS-CTM, an old project by Flemming et. al to coupled IFS with three different chemical tracer models, MOZART3, TM5 and MOCAGE: https://www.geosci-model-dev.net/2/253/2009/gmd-2-253-2009.pdf

then I found C-IFS also by Flemming et. al which is said to superseed IFS-CTM https://www.geosci-model-dev.net/8/975/2015/

and lastly TM5 coupling in EC-Earth by Twan van Noije: https://www.geosci-model-dev.net/7/2435/2014/


My question thus is, coming from standalone OpenIFS, which is the most easily realized implementation of aerosol tracers? Does Cy40 contain C-IFS? If not, will Cy43 do so? I found traces of IFS-CTM, is it still functional? 

If all of the above questions would have no as an answer, then I take it I would recommend the EC-Earth TM5 route to him.


Best regards,

Jan

16 Comments

  1. Unknown User (nagc)

    Hi Jan,

    It depends exactly what you mean by chemical tracer transport, though I interpret your question to mean a full chemistry+aerosol scheme. OpenIFS has a tracer transport capability. It is straightforward to define extra fields which are transported by the semi-Lagrangian dynamical core. This is the basis on which chemistry is added to the model. Extra code is then needed to initialize these fields, add emissions/deposition/chemistry etc.  A simple scheme could be coded into the model but a full tropospheric or stratospheric composition code would be alot of work.

    The IFS-CTM was the first IFS composition system which used OASIS to couple IFS with a chemistry transport model (CTM) such as TM5. This was superceded by C-IFS which called the chemistry code directly from the model for improved efficiency, though it means that the tracers are not formally mass conserving because of the semi-Lagrangian transport. A coupled CTM would use it's own transport scheme which could be formally mass conserving. Mass fixers are needed for the composition tracers in IFS.

    OpenIFS CY40 does not contain any of the OASIS coupling code nor the TM5 composition code. For OpenIFS CY43R3, we will be providing the interface code to OASIS which has been ported from the EC-Earth version. We are considering including the CAMS TM5 code into a future version of OpenIFS 43R3 but there are no solid plans as yet, and if we do it will not be in the first version of OpenIFS 43R3.

    Hope that helps.

    Regards,   Glenn

  2. Etienne Tourigny

    Hi ! Is there any of the C-IFS code in openifs cy43r3? Or is this part of a project which  Philippe Le Sager at KNMI  and others are involved in?

  3. Marcus Koehler

    Hi Etienne, yes there is currently the OpenIFS/AC project underway wherein we implement atmospheric chemistry in OpenIFS.  This is in collaboration with KNMI (and other member state scientists) and the project lead is Vincent Huijnen and Philippe Le Sager is involved too. The implementation is largely aligned with the model used by CAMS however this will also include alternative aerosol modules. This is still very much work in progress and the chemistry version of OpenIFS is not yet available to OpenIFS users outside the project participants.

    Regards, Marcus

  4. Etienne Tourigny

    Thanks Marcus Koehler, at this point I am collecting all available information. In the frame of EC-Earth4 (which uses oifsr43r3) I am interested in the global carbon cycle (land/ocean/atmosphere), while other EC-Earth colleagues (such as Philippe) are working on atmospheric chemistry.

    For the atmosphere that would require a CO2 tracer (and other GHG eventually) and also a mass fixer (see Recommendations for mass fixers in OpenIFS?).  Would having "new" passive tracers in oifs require something like OpenIFS/AC? I know in the ifs passive tracers are part of GEMS and I guess also in CAMS, but I am a bit in the dark on this wrt openifs.

  5. Marcus Koehler

    Hi Etienne, at the beginning of the OpenIFS/AC we have as a test case first implemented "passive" tracers into OpenIFS (Pb and Rn). These tracers had a simple e-folding mechanism and didn't require a full chemistry scheme. This is in a feature branch that I could share with you if this is useful.

    I have provided some more details on the mass fixers in OpenIFS in your other post.

    Cheers,  Marcus

    1. Etienne Tourigny

      Thanks Marcus Koehlerit's good to know it should be easy to implement GHG tracers in openifs. I will get in touch with you sometime in the future about this.

      We would replace the e-folding mechanism with actual fluxes with the ocean and land (which we have working now in EC-Earth3 with NEMO/PISCES ocean model, LPJ-GUESS vegetation model and TM5 for the transport of CO2).

  6. Etienne Tourigny

    Hi again Marcus! I was digging into the openifsr43r3v1 code, and there are numerous references to GHG tracers such as CO2, CH4 and N2O in the GFL routines, which were added initially for the GEMS project (as far as I can understand).

    Also, the mass fixers discussed in Recommendations for mass fixers in OpenIFS? also support the GHG tracers .

    So does this mean we could use the existing interface and setup openifs to use a small  number of GHG tracers, and using the simple passive tracers you mention above?

    1. Marcus Koehler

      Hi Etienne, I cannot say how straightforward it would be to reactive the pre-existing interface to the GHG tracers. Typically, IFS code that is not used for OpenIFS is replaced by dummy routines. If these were reinstated again with the prior existing code I see no reason why this shouldn't work with OpenIFS. 

      If it helps, the KNMI team who work on the OpenIFS/AC project have begun their work by reinstating the existing chemistry-related code that was taken out from IFS 43r3 during the conversion to OpenIFS.  Vincent Huijnen has led this work and he would know exactly how time intensive this has been, but I understand from our conversations that at least initially it didn't pose any major unexpected challenges.

      I am not sure if this was helpful, but if you have any specific questions to this part of the code I could put you in touch with colleagues in the CAMS team?

      1. Etienne Tourigny

        This does help, but still it would be great to have a general idea of what was missing in openifs to enable the passive tracers, considering Unknown User (nagc)'s comment  below that none of the code concerning passive tracers was removed.  Thanks!

        1. Marcus Koehler

          Etienne, I will email you later to discuss how we implemented the passive Radon and Lead tracers in OpenIFS.

          Regards, Marcus

  7. Unknown User (nagc)

    I'm not exactly sure what code is being referred to here, but I can tell you that there is no code missing that would prevent passive tracers from being used in the model. And as Marcus says, I'm sure we could get any code you might need quickly.

    Glenn

  8. Etienne Tourigny

    Hi again, after a few offline discussions and some thinking, I realize that what I require should be already included in the GEMS interface, which is part of the GFL.

    There is support for enabling GHG tracers (NGHG in namgfl.nam.h) and activating mass fixers for each GHG species (see YGHG_NL and TYPE_GFL_NAML in yom_ygfl.F90), and even output.   

    However, do you know if any crucial code was removed in openifs43r3?  Thanks and have a good holiday season!

    1. Unknown User (nagc)

      Hi Etienne,

      As I mentioned in this thread already, tracers have already been run with OpenIFS 43r3. I am not sure what you mean by crucial code. The numerics code in OpenIFS should be the same as IFS, it's largely only the data assimilation code that's removed.  If you do find any missing routines, it's trivial to put them back in.

      There will be an update to OpenIFS 43r3v1 in early 2021 which will update the tracer mass fixers to the same as used in later IFS cycles.

      Glenn

      1. Etienne Tourigny

        Sorry for bothering you both with this topic again. I was under the impression that you were not talking about the CAMS/GEMS GHG tracers.  I will try testing this during the next few weeks, and report back if I have found anything.

  9. Marcus Koehler

    As I did not do the implementation of the chemical tracers myself I am not sure how to answer your question about "crucial code". During January I plan to create a Confluence documentation page on the implementation of the passive tracers, and I hope that this will be of help.

    Regards, Marcus

    1. Etienne Tourigny

      thanks Marcus! have a nice holiday!