Copernicus (CAMS) boundary conditions - CMAQv5.3

Hello,

Have CMAQ (v5.3.1 and onward) been adapted to use the CAMS global reanalysis (EAC4) for the IC and BCs?
If so, how can we use these files?

Regards,
Matthieu

Hello Matthieu,

thanks for your question. Before I get into specifics regarding CAMS, let me first try to answer it in more general terms.

There is no fundamental reason you couldn’t use fields from CAMS (or GEOS-Chem or CAM-Chem or any other large-scale model) to generate initial and boundary conditions for the CMAQ chemical mechanism and aerosol scheme you are using, regardless of the version of CMAQ you are using. However, you would want to have a detailed understanding of how the gas phase and aerosol species can be mapped from the non-CMAQ model to CMAQ and of how the underlying meteorological fields used in the large-scale model differ from those used in CMAQ since you might have potential mis-matches in terms of flow patterns.

Because the CMAQ development team lacks the resources to develop an in-depth understanding of the gas phase and aerosol schemes used in these different large-scale models (some models may offer several options, and mechanisms may change between versions, just as the mechanisms used in CMAQ continue to evolve), we do not include support for such mechanism mapping as part of our CMAQ code repository. That said, some tools available within the CMAQ code repository (specifically ICON, BCON, and combine) could conceivably be used for parts of a potential workflow like this – this is an adaptation of the workflow described in our tutorial on how to develop CMAQ initial and boundary conditions from hemispheric CMAQ output files distributed through the CMAS data warehouse:

  1. User develops custom code to convert fields from the large-scale non-CMAQ model from their native format to files compliant with I/O API netCDF conventions, e.g. using Fortran code linking in I/O API libraries. The resulting files need to use I/O API conventions for grid definitions, vertical coordinates, variable names (e.g. 16 character max), etc. and should also include pressure and/or height in addition to chemical fields.
  2. Such fields could then be processed through ICON and BCON to accomplish the horizontal, vertical, and time transformations from the original files to the files for the CMAQ target domain
  3. The resulting files now need to be mapped from the species present in the large-scale model (i.e. the species names of the files used in steps 1 and 2) to the species for the desired CMAQ mechanism. Tool combine can be used to perform this mapping of the files generated by ICON and BCON, but crucially it is up to the user to actually develop the SpecDef species mapping file to be used by combine
  4. The species-mapped ICON and BCON files generated in step 3 can then be used as input to CMAQ

Community members may have developed their own tools for accomplishing these required steps, but to the best of my knowledge, such tools are often specific to only a subset of possible combinations of large-scale models, model versions, mechanisms, and CMAQ versions and mechanisms, so at the end of the day, using non-CMAQ large-scale fields to generate initial and boundary conditions for CMAQ likely involves significant effort for the user.

As for CAMS, we have used a subset of EAC4 fields custom-extracted by ECWMF for specific years and domains for one of our projects, and have developed a workflow like the one described above to construct CMAQ initial and boundary conditions from these files. However, the custom tool developed for step 1 and the species mapping developed for step 3 haven’t undergone the level of testing and review that would be required for wider release, so we have no plans for including them in the CMAQ repository. One aspect we tried to account for while performing the species mapping is that CAMS sea salt aerosols are expressed at 80% relative humidity while all other CAMS aerosols are expressed as dry mixing ratios (see this note on the CAMS user support page). In addition, at least for the subset of EAC4 fields used in our project, sea salt and dust aerosol bins were specified in terms of radius, not diameter. We advised the following to our project participants using CAMS fields to derive boundary conditions for their regional models (not just CMAQ): “The sizes specified … are in radius, not diameter – this needs to be taken into account when transferring … particle speciation to the participant model speciation. Furthermore, … for sea salt these radii are specified for wet aerosols at an ambient relative humidity of 80%. To calculate the dry radii, the nominal radii of the sea salt bin limits (0.03, 0.5, 5 and 20 microns) should be divided by a reduction factor of 1.99. The mass reported … as sea salt mass is the wet aerosol mass. In order to determine the corresponding dry sea-salt mass, the … values need to be divided by a factor of 4.3.”

If community members have developed sharable and documented tools to map CAMS reanalysis fields to some or all of the CMAQ gas phase and aerosol mechanisms, it’d be great if they could share how to access such tools here.

4 Likes

I’ve made some generalized tools for reading in and processing global models, but they all start with detailed knowledge of the gas/aerosol speciation and vertical/horizontal structure of the model. Right now, that system is largely assuming the old non-hybrid vertical coordinate.

aqmbc

3 Likes

Thank you very Christian Hogrefe for all these details.

I expected challenges as those listed (from 1 to 4) and I am aware about the issue on mapping the different chemical species from a model to another.
I also thank you for your tips on sea salt and aerosols bins in EAC4 since I did not know about these issues.
I wonder if there is a working version of your tools where I can contribute (if I am able to do it). I am interested to explore the possibility to use CAMS as BCs and ICs for CMAQ, but if possible, I‘d rather avoid starting from scratch.

Saying this, I also thank Barron H. Henderson for his link. However, I am not fully sure to understand how I can adapt the codes for another model.

Matthieu

1 Like