How to change the start year in simulations under HIST compset?

[Song]

Hi all,

I am trying to run both land-only and fully-coupled simulations with transient forcing. Therefore, I plan to choose the compset IHistClm50Sp for land-only simulations and the compset BHIST for fully-coupled simulations, respectively. I want to run simulations starting from 1870. However, the default start year is 1850 in HIST compset.

In this case, I am not sure how to change the start year. Based on the existed forums, my current understanding is as follows:

1. IHistClm50Sp
In land-only simulations, I need to change RUN_STARTDATE, DATM_CPLHIST_YR_START, DATM_CPLHIST_YR_END, DATM_CPLHIST_ALIGN.
RUN_STARTDATE = 1870
DATM_CPLHIST_ALIGN = 1
DATM_CPLHIST_YR_START = 1870
DATM_CPLHIST_YR_END = 2014 (if I hope the simulations end in 2014)

Is that all I need to set?

2. BHIST
I am more confused about the settings in fully-coupled simulations. Should I change RUN_STARTDATE, RUN_REFCASE, and RUN_REFDATE?

When the compset is set as BHIST and RUN_STARTDATE is set as 1870, I found the default RUN_REFCASE = b.e21.B1850.f09_g17.CMIP6-piControl.001_v3hist, and the default RUN_REFDATE = 0501-01-01. Apparently, the forcing data with the default RUN_REFCASE and RUN_REFDATE would not be starting from 1870.

Therefore, should I go to '/glade/campaign/collections/cmip/CMIP6/restart' to find another case which include the year 1870 as the RUN_REFCASE? Any other parameters I need to change?

Thanks for your attention to this matter.

Best,
Song

 

[slevis]

1a. The following may be explained somewhere in the documentation, but here's a simple explanation:
We typically run the period 1850-1900 with
RUN_STARTDATE = 1850
DATM_CPLHIST_ALIGN = 1901
DATM_CPLHIST_YR_START = 1901
DATM_CPLHIST_YR_END = 1920
because the DATM files start in 1901 and not earlier. In your case, I would just change the first line to 1870. The rest should be the same.

1b. When the simulation gets to any year from 1901 to 1920, I make the following changes to my env_run.xml and user_nl_clm files (I added the suffixes with the years to these file-names only to use with the diffs):
diff env_run.xml_1921-2014 env_run.xml_1850-1900

147c147
<     <entry id="RUN_TYPE" value="hybrid">
---
>     <entry id="RUN_TYPE" value="startup">
235c235
<     <entry id="RUN_STARTDATE" value="1920-01-01">
---
>     <entry id="RUN_STARTDATE" value="1850-01-01">
1108c1108
<     <entry id="DATM_YR_END" value="2014">
---
>     <entry id="DATM_YR_END" value="1920">

diff user_nl_clm.1921-2014 user_nl_clm.1850-1900

20c20,21
<  finidat = '<the_last_restart_file_generated_by_the_same_simulation>'
---
>  use_init_interp = .true.
>  finidat = '<the_initial_file_that_you_use_for_starting_the_simulation_in_1870>'

 

[slevis]

2. I'm less experienced with BHIST, but I would do one of two things:
- Change the startdate to 1870 and leave the rest unchanged. Assuming this works, I think you end up with initial conditions from 1850. If initial conditions for 1870 exist and you can access them, then use those instead.
- Change nothing and just run the extra 20 years from 1850 to 1869 before continuing with 1870.

 

[Song]

2. I'm less experienced with BHIST, but I would do one of two things:
- Change the startdate to 1870 and leave the rest unchanged. Assuming this works, I think you end up with initial conditions from 1850. If initial conditions for 1870 exist and you can access them, then use those instead.
- Change nothing and just run the extra 20 years from 1850 to 1869 before continuing with 1870.

Thank you, Dr. Levis! Your detailed explanation, especially for the land-only simulations, is very helpful.

For the fully-coupled simulations, I have tried to create the case with only the startdate changed to 1870 according to your suggestion. The ncdata = 'b.e21.B1850.f09_g17.CMIP6-piControl.001_v3hist.cam.i.0501-01-01-00000.nc', and there is no other files with different year exist in this path (/glade/campaign/cesm/cesmdata/cseg/inputdata/cesm2_init/b.e21.B1850.f09_g17.CMIP6-piControl.001_v3hist/), so the first way may not work. I am thinking about finding a restart file near 1870 year to run the fully-coupled simulations I want. I hope this would work, otherwise, I can only run the extra years as the second way you suggested.

Thank you again for your help!