search

ESMCI / cime

Common Infrastructure for Modeling the Earth
http://esmci.github.io/cime
Other
160 stars 207 forks source link

Remove CMIP5 RCP scenarios #3026

Closed ekluzek closed 5 years ago

ekluzek commented 5 years ago

I think we should remove the CMIP5 RCP scenarios from cime as we start adding in the CMIP6 SSP-RCP scenarios (see #3025). This means the following changes

In cime/src/components/data_comps/datm/cime_config/config_component.xml

Setting for CO2 removed for each RCP prescribed aerosols for each RCP CPLHIST forcing for each RCP

In addition

cime/src/drivers/mct/cime_config/config_component_cesm.xml needs the following:

Description for RCP removed as well as CCSM_BGC set for RCP's.

jedwards4b commented 5 years ago

Are they in the way? I think that there still may be some desire to run cmip5 scenarios in the future.

ekluzek commented 5 years ago

I started removing them in CTSM, so that ssp_rcp and rcp wouldn't get confused. Here I think it might be OK to have both, but it's not immediately obvious to the uninitiated that one is CMIP5 and one is CMIP6. But, if components like CTSM start removing the capability to utilize them, it won't work for the full system anyway. I'm pretty sure not all the CMIP5 RCP's even still exist, nor do components necessarily respond to them. So I don't think you get something useful now, which is why I think they should go away (as well as making it more clear for what you should use for CMIP6 future scenarios).

If there is a desire to run both CMIP5 and CMIP6 as a requirement, we need to hear that that is the case, so we can plan on that. That ability is definitely not there now.

jedwards4b commented 5 years ago

Thanks for checking before removing them.

jgfouca commented 5 years ago

@ekluzek , is this done?

ekluzek commented 5 years ago

@jgfouca no it's not. But, I did get approval for it to happen.

ekluzek commented 5 years ago

Here's the message I got from Gokhan on March 6th for the CESM side...

Erik,

As far as I know, we are not planning to carry along CMIP5 RCPs in CESM2. If we were to do so, it goes beyond just carrying RCP data sets as the PI and 20C simulations also need "consistent" data sets, I believe. The only set of simulations that we said we would do is a few ensemble members of RCP8.5 simulation with CESM2 to compare the solutions with RCP8.5 and SSP5-85. These are just one-off simulations and do not necessarily require official support.

Best, Gokhan

rljacob commented 5 years ago

telecon: this is done in maint-5.6