some questions about the hybrid run results

[11438023@zju_edu_cn]

Hi everyone,I'm new to CESM, I have some questions while running it, please give me some suggestions:(1)I use cesm1.2.2, E1850C5CN with CN turned off, and the resolution I chose is f19_g16.Firstly, I run the model 100 years to spinup, and I saw the surface air temperaure is equilibrium (about 15 degree) and the TOA net flux is about zeros during the last 70 years. Secondly, I setup a hyrbid run, using restart file from spinup experiment as my restart file, my setting is as follow:RUN_TYPE = hybrid,RUN_STARTDATE = 0101-01-01RUN_REFCASE = spinup_caseRUN_REFDATE = 0101-01-01All other settings are the same with spinup experiment. Then I run 100 years to equilibrium.My question is: after the first experiment spunup, the SAT is about 15 degree. But the first several years of my hybrid experiment shows SAT about 13 degree, and then goes up to 15 degree. I don't know if this is correct?  why the SAT suddently declined at the beginnig of the hybrid run? (since the initial state is given by restart file if i'm not wrong)(2)I set an experiment using the same compset and resolution with above, where i replaced all the tree types with grass types.  But the simulation results have large differences with previous results, which used an early version of CESM and a thermo-only ice component.  For example, the global and annual mean SAT changes and precipitation changes are twice as lower as previous results. I checked my setting and found no error... I'm not sure if different ice components and CESM versions could lead to such large differences. Does this seem reasonable?Thank you very much!   

 

[kdagon@fas_harvard_edu]

I'm having a similar issue with a hybrid run in CESM1.2.2 using CAM4-CLM4.5-slab ocean and f09_g16 resolution. I suspect it is the slab ocean forcing surface temperature back down when the model initializes as a hybrid restart. (I'm checking this using the variable TS in CAM output.) When I try a hybrid run with prescribed SSTs instead of slab ocean, the global temperature looks more like the spin-up.I'm using the SOM forcing file "pop_frc.b.c40.B1850CN.f09_g16.100513.nc", but I've also tried this with the newer forcing "pop_frc.b.e11.B1850C5CN.f09_g16.005.082914.nc" and the problem persists.However this problem goes away when I try a branch run, so I suspect it is also tied to the nature of a hybrid restart. I'm wondering if a spin-up is useless for a hybrid run if the temperature needs to re-equilibrate anyway. If anyone has any intuition on how the slab ocean works when doing a hybrid restart, including how the docn.r and docn.rs1 files are utilized, that would be very helpful!Thanks,Katie

 

[kdagon@fas_harvard_edu]

If anyone is interested, I think I fixed my problem by setting read_restart = .true. in docn_comp_mod.F90. This forces the slab ocean to start like a restart run, and use information from the docn restart files copied over from the spin-up. The global average temperature now looks continuous from the spin-up simulation.

 

[11438023@zju_edu_cn]

Hi Kdagon,Can you be more detailedly about this changing ?Thanks!Best 

 

[kdagon@fas_harvard_edu]

You can modify the source code file /models/ocn/docn/docn_comp_mod.F90 by copying it over to your SourceMods/src.docn folder in the case directory. Comment out the line that sets read_restart based on the run type. Then add a line "read_restart = .true."That's my approach, and it seems to help with global temperature by keeping it continuous.

 

[11438023@zju_edu_cn]

Hi,Sorry for responding later. I just got some free cores for testing and your method is very useful.Further, I find that when I choose "RUN_TYPE = hybrid", the variable "start_type" is set to startup while running. So in script "cesm1_2_2/models/drv/shr/seq_infodata_mod.F90", this "start_type = startup" leads to "read_restart" to be "FALSE", and the ocean component will not use restart file. Luckily the branch type does not have such problem so we only need to set this once.Best,Duan