Closed billsacks closed 2 years ago
@billsacks, I like your plan of adding a [glad] section to the cism.config file. Some of the variables in this section would be set based on the compset, correct?
Some of the variables in this section would be set based on the compset, correct?
Yes, that's right.
I think this is a good first draft. We should try it, and if people complain a year or so down the road (not enough clarity or confused about what to set where) then we can revisit this once we have a better feel for the overall CISM workflow with multiple ice sheet.
I second Kate and Bill. This is new territory and we won't know until we try it out. Thanks for thinking about these details @billsacks.
On Thu, Feb 25, 2021 at 3:21 PM Kate Thayer-Calder notifications@github.com wrote:
I think this is a good first draft. We should try it, and if people complain a year or so down the road (not enough clarity or confused about what to set where) then we can revisit this once we have a better feel for the overall CISM workflow with multiple ice sheet.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/ESCOMP/CISM-wrapper/issues/53#issuecomment-786267104, or unsubscribe https://github.com/notifications/unsubscribe-auth/AB6QV6HUJISJGWBSBUTZVITTA3EP7ANCNFSM4YHIGF7A .
-- Gunter Leguy, Ph.D (he/him) Project Scientist National Center for Atmospheric Research cell: (575) 418 1021 desk: (303) 497 1790
I have done the last piece of this on a branch: moving zero_gcm_fluxes out of cism_in to the config file. So I will close this issue as soon as that piece comes to master.
We currently have two namelist / configuration files in CISM: cism_in (which is read by cism-wrapper) and cism.config (which is read by CISM itself). I think it's relatively straightforward to introduce a separate cism.config for each ice sheet instance. However, cism_in is posing some problems for multiple ice sheets. Currently, there are some ice sheet-specific variables in there along with some variables that apply to all ice sheets. I think we'll always need an overall cism_in file, in part because that will tell cism which ice sheets it's running. But then we need to figure out how to handle the ice sheet-specific variables that are currently there. Among other variables, this will include zero_gcm_fluxes and cesm_history_vars (those are just the first two I noticed; they popped out to me because their defaults depend on cism_evolve, which we want to let differ between ice sheets).
My current tentative plan is to move any such variables into a glad-specific configuration section in the cism.config file. Then these can be read in glad initialization, and passed back up to the cism wrapper. It feels a little weird for the wrapper to rely on CISM to read some of its own configuration settings, but this feels cleaner and more intuitive to me from a scripting and user perspective – as opposed to, say, having separate cism_ICESHEET_in files in addition to cism.ICESHEET.config and the overall cism_in.
@Katetc @whlipscomb I'm interested in whether you have any thoughts on this – and @gunterl if you'd like to weigh in.