confused about RUN_STARTDATE & RUN_REFERDATE

[yangx2]

Hi everyone,
After looking through the thread (https://bb.cgd.ucar.edu/cesm/threads/suggestions-for-modifying-the-start-date-cesm2.4921/#post-35387), there are other questions pop up regarding their discussion.
1. BHIST
Case descripition:
./create_newcase --case ~/cases/bhist --compset BHIST --res f19_g17
I found:
RUN_REFEDATE=0301-01-01
RUN_REFCASE=b.e20.B1850.f19_g17.release_cesm2_1_0.020
RUN_STARTDATE=1850-01-01
RUN_TYPE =hybrid
Question:
1. RUN_REFCASE is used to provide the initial condition for the model. And RUN_REFDATE does not need to be matched (same as) with RUN_STARTDATE.
2. If I need to run the model with different initial conditions, all I need to do is chang RUN_REFDATE (e.g. RUN_REFDATE=0209-01-01) if I chose to use the same simulation (e.g. b.e20.B1850.f19_g17.release_cesm2_1_0.020). Also, it is ok if I am using a different simulation.
3. If I want to get a historical run starting from 1979, I only need to change RUN_STARTDATE=1979-01-01. No need to make changes of RUN_REFCASE and RUN_REFDATE.

Is my understanding of the above correct?
Thanks for the help!
Best,
Skylar

 

[katec]

Hi Skylar. Some of these questions are a little confusing for me, but I'll try to answer them the best I can. There's some documentation on CESM ref cases and start dates here: 5. Running a Case — CIME master documentation

1. The ref case provides initial condition in a hybrid run. The ref date points to the date or folder containing the restart files that you wish to use as those initial conditions. The Start date for your model run does not have to match the refdate.

2. If you want to run the model with a different year from the same case as your initial conditions, then, yes, all you need to do is change the refdate. If you want to use a different year from a different simulation, you will need to update your refcase and refdate.

3. Historical runs require different forcing than the B1850 Pre-Industrial (control) run provides. For these cases, you should look for a BHIST run to use as your refcase. The B1850 refcase that you are pointing to here just runs year 1850 forcing over and over again. So you would be using year 1850 forced simulation to start your run at year 1979. It would probably be better to use year 1979 files from the BHIST control run.

 

[yangx2]

Hi Skylar. Some of these questions are a little confusing for me, but I'll try to answer them the best I can. There's some documentation on CESM ref cases and start dates here: 5. Running a Case — CIME master documentation

1. The ref case provides initial condition in a hybrid run. The ref date points to the date or folder containing the restart files that you wish to use as those initial conditions. The Start date for your model run does not have to match the refdate.

2. If you want to run the model with a different year from the same case as your initial conditions, then, yes, all you need to do is change the refdate. If you want to use a different year from a different simulation, you will need to update your refcase and refdate.

3. Historical runs require different forcing than the B1850 Pre-Industrial (control) run provides. For these cases, you should look for a BHIST run to use as your refcase. The B1850 refcase that you are pointing to here just runs year 1850 forcing over and over again. So you would be using year 1850 forced simulation to start your run at year 1979. It would probably be better to use year 1979 files from the BHIST control run.

Hi Katec,
Thank you so much for the detailed explanation. It solves all my confusion. Just one small question, if I want to run a historical run starting from 1979, is there any RUN_REFCASE that I can "hybrid" on Cheyenne? Running BHIST from 1850 is too long for me.
There is another question related to making ensemble runs. From my perspective, there are two methods. One is to slightly vary the CAM initial condition, which can be done with the namelist variables: pertlim. Another one is "hybrid" on different RUN_REFCASE. So my question here is which one is the right one or the popular one to make ensembles?
Best,
Skylar

 

[katec]

There are very many BHIST restart initial conditions on Cheyenne here: /glade/p/cesmdata/cseg/inputdata/cesm2_init

As for your ensemble perturbation, your choice depends on your scientific goals. I suggest a review of the literature to help you make your choice.

 

[yangx2]

There are very many BHIST restart initial conditions on Cheyenne here: /glade/p/cesmdata/cseg/inputdata/cesm2_init

As for your ensemble perturbation, your choice depends on your scientific goals. I suggest a review of the literature to help you make your choice.

Thank you so much!