CAM5 within CESM1.2.2 yields RELHUM as high as 140%

[wadewei]

Hi everyone,

I am trying to run CAM5-CESM1.2.2 to compare with my previous CAM4 results regarding LGM climate. However, using the same topography file, surfdata, SST and sea ice condition while starting up a fresh run, the model blew up after around 3 yearswith the following error message. The problem persists when restarting from the last restart file.

I looked briefly into the history output. Everything seems fine including TS, precipitation and U. However, relative humidity exceeds 100% at a bunch of places every year, reaching as high as 140%.

This is very confusing. Any help is appreciated.

Best regards,
Wade



pLCL does not converge and is set to psmin in uwshcu.F90
3.640891690722548E-006 128.376220645550 102752.983156971
pLCL does not converge and is set to psmin in uwshcu.F90
4.420935291246052E-006 -20.4007801912152 102849.741094839
Impossible case2 in instratus_condensate
1.00000000000000 0.000000000000000E+000 0.000000000000000E+000
NaN NaN

 

[cacraig]

It is unclear from your posting exactly how you setup your run. If you need further assistance, please include the information requested at: Information to include in help requests

I will attempt to answer your question with the information you've currently supplied.

If you are setting -phys cam4 in the CAM5-CESM1.2.2 setup, you should be aware that we do not support CAM4 in this configuration. The simple reason for this is that while the switch still works, we have not had the resources to tune all the changes which went into CAM5 so that they work robustly with CAM4.

If on the other hand you are running CAM5, but attempting to use data files created for CAM4, this too would be problematic as CAM5 was developed and tuned with specific input data requirements.

If your setup is either of the above scenarios, you are using an unsupported configuration which could require extensive tuning by you to get it to run properly.

I will ask a couple of scientists if they have any insights and if so, they will post a reply here. Again, it would be helpful to know at the very least the create_newcase, xmlchange and user_nl_cam settings you are using, along with knowing if you've made any modifications to the code.

 

[andrew]

Greetings,

I would not recommend doing what you are doing. There is no reason why it should work since so many things have changed. I guess you want to compare to an LGM simulation you already have? You might consider trying to manually interpolate the old fields into the new field grids. RH in CAM5 might in isolated regions get up to 140% in instantaneous data.

What is the science motivation? To see how different CAM4 and CAM5 are for the LGM? Is the topography or SSTs in a standard configuration from CCSM4 and CESM1 different enough you need to do this? The Paleo group might know or have dealt with this, perhaps ask the question there.