How do I choose the TropChem simulation type and the 47 levels for a C48 simulation with GCHP v14.2.2?
Closed this issue · 6 comments
Name: Herizo Narivelo
Institution: Observatoire Midi-Pyrénées - Laboratoire d'aérologie (LA UMR 5560) - Université Toulouse III Paul Sabatier
Dear all,
I am a newcomer to using the GCHP model. In my project, I am going to run a long simulation (multi-year) but before doing these extended simulations, I first tested a one-month simulation with c48 horizontal resolution on our local cluster. For information, I installed/used the GCHP v14.2.2 model and ran it on our local cluster, which does not have enough computing resources. In the previous version, for example, the GCHP v12.3.2, we could use the TropChem simulation type with 47 vertical levels and the UCX flux from the stratosphere only. However, in the case of GCHP v14.2.2, I cannot specify a TropChem but rather fullchem and by default 72 vertical levels when creating my run directory. I would like to ask you if it will be possible to use a TropChem and 47 vertical levels with GCHP v14.2.2 and only the stratosphere UCX flux to optimise my simulations as we do not have many ressources in terms of processors and nodes. If this is the case, how can I achieve this?
Thanks in advance for your advice.
Herizo Narivelo
Thanks for writing @herizonarivelo. GCHP cannot use the reduced 47L grid, as the MAPL I/O software that GCHP uses cannot do vertical regridding. All GCHP simulations have to be done at 72L.
You can try to save memory by reducing the number of diagnostic outputs that are saved to disk.
You also might explore using GCHP on the AWS cloud computing environment. Tagging @yidant.
Agreeing with @yantosca - if you do want to run with only 47 layers you would need to switch the internal parameters in GCHP to 47 layers and manually regrid the input meteorology. This would be a significant undertaking but is in theory doable - we have run GCHP with 26 layers using CAM input, for example. There are plans for vertical regridding to be performed online in GCHP but there is no timeline for that change as yet.
Hi @herizonarivelo, since you will be running at C48 you could also consider using GC-Classic at 2x2.5. This is the roughly equivalent resolution on lat-lon grid and would give you the option of running on a 47L vertical grid.
Hi everyone, thank you very much for your advice @yantosca, I will try to reduce the number of diagnostic outputs. Thanks also to you @lizziel , I did a one month fullchem type simulation on Geos-Chem Classic with 47L and 2x2.5 horizontal resolution but with only one node and 16 CPUs, my one month simulation took a lot of time, that's why I am going to use GCHP to be able to use several nodes for my multi-year simulations. Thanks to others for their advice. I will come back to you if I have any questions. Thanks again.
Closing out this issue. Feel free to open a new issue if you are stuck.
Hi @herizonarivelo, please create a new issue for this question. Thanks!