@singh2299 - I saw the email you sent to Keith. Thank you for that. A key question is: how is the land model called in this system? Specifically, is the atmosphere the driver (that is, the main program), calling the land model from somewhere inside its physics? Or is there a separate driver that calls each of the components (a hub and spoke architecture)?
Sorry for the delayed reply.
For this application, the tool we have in development will likely be helpful to you. This tool, called LILAC (Lightweight Land-Atmosphere Coupler), is still in development, but we have a preliminary coupling between CTSM and WRF using this infrastructure. There are a couple of presentations here ESCOMP/CTSM that give you an overview (search for LILAC on that page). Here is some preliminary code illustrating what this coupling looks like for WRF: billsacks/WRF . A lot of that code is specific to WRF, but if you search for "call lilac" in that file, you'll get a sense of the interface. Note, though, that the interface is still changing somewhat, and that WRF code is already a bit out of date (I need to update it next week). Another example you can look at is the demo atmosphere driver we put together here, which feeds CTSM with fake atmospheric forcing data: ESCOMP/CTSM . Again, search for 'call lilac' in that file.
There are some big caveats here:
For this application, the tool we have in development will likely be helpful to you. This tool, called LILAC (Lightweight Land-Atmosphere Coupler), is still in development, but we have a preliminary coupling between CTSM and WRF using this infrastructure. There are a couple of presentations here ESCOMP/CTSM that give you an overview (search for LILAC on that page). Here is some preliminary code illustrating what this coupling looks like for WRF: billsacks/WRF . A lot of that code is specific to WRF, but if you search for "call lilac" in that file, you'll get a sense of the interface. Note, though, that the interface is still changing somewhat, and that WRF code is already a bit out of date (I need to update it next week). Another example you can look at is the demo atmosphere driver we put together here, which feeds CTSM with fake atmospheric forcing data: ESCOMP/CTSM . Again, search for 'call lilac' in that file.
There are some big caveats here:
- This coupling infrastructure is still undergoing some changes, and is not yet documented. We can't provide much support for it at this point, but will be able to provide support within a year or so.
- The build system is awkward right now, though we aim to improve it in the next month or so. (I can post an update to this thread once this is done.)
- Namelist settings are also awkward right now, though we also aim to improve that in the next month or so.
- Model initialization can be a big issue for weather forecasting applications. We are still working on a strategy and implementation for initializing CTSM in a weather forecasting context.
@sacks Thank you very much for the detailed reply. I will look into LILAC and the demo atmosphere driver for coupling CLM with the GFS atmosphere in IITM-ESM. I understand that since this is still in development the support might not be available, but request you to help me with the fake atmosphere driver. One more thing I wanted to mention was that :
At present we use the FMS coupler for coupling atmosphere-ocean models, as I understand I should be able to keep that configuration and couple CLM with these tools you have mentioned. But please correct me if I am wrong.
At present we use the FMS coupler for coupling atmosphere-ocean models, as I understand I should be able to keep that configuration and couple CLM with these tools you have mentioned. But please correct me if I am wrong.
@singh2299 Sorry that it has taken me longer than anticipated to get back to you on this. The LILAC infrastructure is now on CTSM's master branch. This includes an easier to use build system and scripts to generate runtime inputs, as I mentioned above. There is some documentation of this here 3. CTSM-LILAC User’s Guide — ctsm CTSM master documentation . We haven't had time yet to document the Fortran API - i.e., how to actually call LILAC-CTSM from your atmosphere model. For that, the demo atm driver (ESCOMP/CTSM) together with the relevant WRF code (billsacks/WRF) are still the best references.
I should note that, while we have done substantial in-house testing at this point, LILAC has not been used extensively yet by others. Please let us know if you run into trouble.
I should note that, while we have done substantial in-house testing at this point, LILAC has not been used extensively yet by others. Please let us know if you run into trouble.