How to set env_run.xml

[wbr2023]

Hi! I am currently using the FHIST compset of CESM2.1.3 for simulation, with a planned simulation period from 2000 to 2014, but I am still confused about the setting of the env_run.xml.

Because I have modified the parameterization scheme for dust emission in the model, it may not be good for me to use the default run type “hybrid” of env_run.xml in FHIST and initial run from the restart files on 1979-01-01 (or 2000-01-01). The default settings for env_run.xml are as follows:

<entry id="RUN_TYPE" value="hybrid">
<entry id="RUN_REFDIR" value="cesm2_init">
<entry id="RUN_REFCASE" value="f.e20.FHIST.f09_f09.cesm2_1.001_v2">
<entry id="RUN_REFDATE" value="1979-01-01"> （or 2000-01-01）
<entry id="GET_REFCASE" value="TRUE">
<entry id="RUN_STARTDATE" value="1979-01-01"> （or 2000-01-01）
<entry id="STOP_OPTION" value="nyears">
<entry id="STOP_N" value="36">

Should I change RUN_TYPE to startup to initial run with the new dust emission scheme. The corresponding settings in env_run.xml are as follows:

<entry id="RUN_TYPE" value="startup">
<entry id="RUN_REFDIR" value="cesm2_init">
<entry id="RUN_REFCASE" value="f.e20.FHIST.f09_f09.cesm2_1.001_v2">
<entry id="RUN_REFDATE" value="1979-01-01">
<entry id="GET_REFCASE" value="FALSE">
<entry id="RUN_STARTDATE" value="1999-01-01">
<entry id="STOP_OPTION" value="nyears">
<entry id="STOP_N" value="16">

Here, I set 1999 as the spin-up year, but after startup running from 1999-01-01, I found that the simulation effect from 2000 to 2014 was not very good.

Questions:
(1) May I ask which method I should use to set env_run.xml?
(2) If using hybrid run, is it best to set RUN_REFDATE and RUN_STARTDATE to be the same in order to simulate the correct historical year?
(3) Can hybrid run be used without considering setting a spin-up year?

Sincerely hope to get your advice.
Best wishes!

 

[peverley]

Hi,

1. We recommend using ./xmlchange to set your xml variables. You can edit the file directly, but if you use xmlchange, it'll prevent you from making a typo that will break things and will also validate the options you are trying to set. For example, if you want to change your run type to hybrid, you'd go to your case directory and then run:

./xmlchange RUN_TYPE=hybrid

2. In a hybrid run, you can use the same or a different RUN_STARTDATE (compared to the reference case). More on hybrid runs here: Variables Related to Run Type — CESM Tutorial

3. Can you elaborate on what you mean by "without considering a spin-up year"?

Courtney

 

[wbr2023]

Hi,

1. We recommend using ./xmlchange to set your xml variables. You can edit the file directly, but if you use xmlchange, it'll prevent you from making a typo that will break things and will also validate the options you are trying to set. For example, if you want to change your run type to hybrid, you'd go to your case directory and then run:

./xmlchange RUN_TYPE=hybrid

2. In a hybrid run, you can use the same or a different RUN_STARTDATE (compared to the reference case). More on hybrid runs here: Variables Related to Run Type — CESM Tutorial

3. Can you elaborate on what you mean by "without considering a spin-up year"?

Courtney

Hi Courtney,

I am very sorry that I may not have made my questions clear. Because I have modified the parameterization scheme for dust emission in my model, if I use the default restart files to hybrid run my model, will it cause some problems with the simulation results. Should I startup run the model based on my modified dust emission scheme.

 

[wbr2023]

Hi,

1. We recommend using ./xmlchange to set your xml variables. You can edit the file directly, but if you use xmlchange, it'll prevent you from making a typo that will break things and will also validate the options you are trying to set. For example, if you want to change your run type to hybrid, you'd go to your case directory and then run:

./xmlchange RUN_TYPE=hybrid

2. In a hybrid run, you can use the same or a different RUN_STARTDATE (compared to the reference case). More on hybrid runs here: Variables Related to Run Type — CESM Tutorial

3. Can you elaborate on what you mean by "without considering a spin-up year"?

Courtney

In addition, many papers have pointed out that it takes a certain amount of time for the model to reach equilibrium after a compele restart, which is called spin-up time. I want to know how long spin-up time it takes to use hybrid and spinup run respectively.

 

[slevis]

@wbr2023 it is difficult to predict how the model will behave with another developer's code modifications, so your best bet is to test things:
You may wish to test whether the default restart works at all, and whether you see problems in your simulation relative to a startup case without the restart.

Your question makes me wonder whether you need to add new fields to restart to preserve exact restarts. That's something that you may need to consider in your code modifications.

I will let others offer guidance on the length of spin-up needed when you modify something like dust emission.