The differences between using hybrid and branch to start the CESM2.1.5 BHSIT restart case

[xli585]

Hello! I am currently running a BHIST case using the cesm2.1.5 version, with the 2000-year restart file. The expectation is to run the simulation with fixed 2000-year emissions and external conditions until a stable equilibrium state is reached. My first startup settings are as follows:

./xmlchange RUN_REFCASE=b.e21.BHIST.f09_g17.CMIP6-historical.010
./xmlchange RUN_REFDATE=2000-01-01
./xmlchange RUN_STARTDATE=2000-01-01
RUN_TYPE=hybrid
Then I set: ./xmlchange STOP_OPTION=nyears
./xmlchange STOP_N=10.

After running for every ten years, I would stop to check, and then continue the run until reaching the equilibrium position. For the second run,
I set: ./xmlchange CONTINUE_RUN=TRUE.

I would like to know if, when running in such a intermittent manner, the output would be continuous or if it would always re-hybridize? Would there be any climate drift? And what would be the difference if using branch to spin up to the climate equilibrium state?

 

[manishvenu]

Hello @xli585,

Let me take a stab at these questions!

when running in such a intermittent manner, the output would be continuous or if it would always re-hybridize?

CONTINUE_RUN takes precendence over any other argument! So, you will get bit-for-bit output when you run this simulation further. After the first run, the RUN_TYPE=hybrid is not relevant and won't re-hybridize.

Would there be any climate drift?

Could you elaborate on what you mean by this?

And what would be the difference if using branch to spin up to the climate equilibrium state?

A branch takes the exact restart files from the previous case (so it is very strict), while hybrid allows you more flexibility, including changing the compset and ref date vs. start date. This link is a good documentation page in the CESM tutorial to learn more about the differences. It seems like you need some flexibility with your setup (and are not trying to reproduce a run), so hybrid likely makes more sense.

One other thing to note is that the "BHIST" might not be the best compset for your usecase of running to climate equilibrium. "HIST" has data from 1850 to 2014 (I'm not sure, but you may error out on the second 10-year run). Spin-up is often done with B1850, which repeats the year 1850 for the forcings. I would recommend searching B2000 in the CESM forums for some insight.

Please let me know if you have any additional questions about this.

 

[xli585]

你好 @xli585，

让我来试着回答这些问题！



CONTINUE_RUN比其他任何论点都更重要！所以，当你继续运行这个模拟时，你会得到逐位的输出。第一次运行后，RUN_TYPE=杂交就不重要了，也不会重新杂交。


你能详细说明一下你的意思吗？



分支会使用上一个案例的完全相同的重启文件（所以非常严格），而混合分支则提供更多灵活性，包括更改复合集和参考日期与起始日期。这个链接是CESM教程里的一个很好的文档页面，可以了解更多差异。看起来你需要在配置上有点灵活性（而且不是想重现一次），所以混合型可能更合理。

还有一点需要注意，“BHIST”可能不是你达到气候平衡用例的最佳组合。“HIST”有1850年至2014年的数据（我不确定，但你可能会在第二个十年时间里出错）。旋转启动通常用B1850完成，B1850对强迫数据重复1850年。我建议你在CESM论坛搜索B2000，获取一些见解。

如果你对此有其他问题，请随时告诉我。

 

[xli585]

Thank you for this！ I have understood.