future simulations by using CLM-BGC and CLM-BGCDV

[dengmingshan]

Dear all,

I am trying to run CLM-BGC in the future, CMIP6 datasets in "QIAN" format are used to force the model. There is an error when running CLM-BGC, while CLM-SP is completely right by using the same forcing data. After looking up the log file, there are no errors in atm.log.240208-032400 and lnd.log.240208-032400 file. After setting DEBUG=TRUE in the env_build.xml file, there is another error occurs. How can I solve this error?

Many thanks

Mingshan

 

[oleson]

The CNDV model is not longer supported, so it's use at your own risk and we can't provide any assistance, sorry.

 

[dengmingshan]

Thank you so much for your reply, how about CLM-BGC? When running CLM-BGC, I met the same error.

 

[oleson]

Yes, CLM-BGC is supported in the recent release versions of the model. Nothing comes to mind when I look at your screen shot however (the first screenshot is the CLM-BGC run?). To provide assistance, please add the information requested in Information to include in help requests, such as the exact model version you are using and other details of your configuration, as well as attaching the complete log files. Thanks.

 

[dengmingshan]

Thank you Oleson!

My CLM version is CLM5.0 and the compset long name is SSP245_DATM%QIA_CLM50%BGC_SICE_SOCN_MOSART_SGLC_SWAV. I running CLM-BGC from 2017-09-01, when running to 2017-09-02, the error occurs. Some logs are the in annexes.

In addition, by using the same forcing data (CMIP6 dataset in QIAN format), the compset SSP245_DATM%QIA_CLM50%SP_SICE_SOCN_MOSART_SGLC_SWAV didn't met any errors ran from 2017-09-01 to 2050-12-31.

 

[oleson]

Thanks for providing that information. It looks like you are running with custom atmospheric forcing on a custom grid so I can't replicate the error on our machine. The error is associated with Sl_t which is the radiative temperature from CLM (t_rad_grc). The radiative temperature is derived from the upward longwave:

lnd2atm_inst%t_rad_grc(g) = sqrt(sqrt(lnd2atm_inst%eflx_lwrad_out_grc(g)/sb))

so something is probably wrong with the upward longwave. Normally, I might expect this to possibly be an atm forcing problem but since you are having no trouble with the forcing in a corresponding SP simulation, the forcing is probably ok. But I still would examine the forcing around the time you get the error to see if it reasonable.
I guess I would suggest tracking the upward longwave (FIRE in your history files), preferably at the model time step, to see if you can spot when the upward longwave starts going bad.
Are the logs you attached for the DEBUG=TRUE case or for the DEBUG=FALSE case?

 

[dengmingshan]

Thank for your reply, the attached log files are for the DEBUG=FALSE, when I setting DEBUG=TRUE in the env_build.xml file, there is another error occurs as following.

 

[oleson]

Oh, I get it, sorry. Even though you are running CLM-BGC, CNVegetationFacade.F90 is calling CNDVType.F90 because of this:

! Initialize the memory for the dgvs_inst data structure regardless of whether
! use_cndv is true so that it can be used in associate statements (nag compiler
! complains otherwise)
call this%dgvs_inst%Init(bounds)

where line 265 is the last line above.

However, line 87 in CNDVType.F90 refers to this section of code:

if (use_cndv) then
call this%InitCold (bounds)
call this%InitHistory (bounds)
end if

with line 87 being the call to InitCold. That should only be called if use_cndv is true though. It should be false if you are using CLM-BGC. I would look at your lnd_in to see if it is defined and what its value is.

 

[dengmingshan]

Yes, of course. I attach my lnd_in file here. Thank you so much.

 

[oleson]

Well, actually I wanted you to check it. However, I don't see it defined and the default should be false, so I don't see why that call to InitCold is being made. I would print the value of use_cndv out in CNDVType.F90 to see what its value is.

 

[dengmingshan]

I also look up the lnd_in file, in CLM-BGC, there is no use_cndv = .true., while in CLM-BGCDV, use_cndv = .true.. I will print the value of use_cndv out in CNDVType.F90. Thank you!

 

[dengmingshan]

Hello Oleson, I guess when running CLM-BGC use_cndv is false. I print a string at Line 87 in CNDVType.F90 in both CLM-BGC and CLM-BGCDV, only CLM-BGCDV can print the Mingshan Deng.

if (use_cndv) then
write(iulog,*)' Mingshan Deng '
call this%InitCold (bounds)
call this%InitHistory (bounds)
end if

 

[oleson]

Right, that should be what happens. But in the traceback line 87 of CNDVType.F90 is calling InitCold in the code I thought you were using. What is line 87 of CNDVType.F90 in your own code?

 

[dengmingshan]

The line 87 of CNDVType.F90 is:

 

[oleson]

Ok, thanks. I'm afraid I still don't understand what's going on then. Maybe you are running out of memory when you run the CLM-BGC simulation and are getting strange behavior.

 

[dengmingshan]

Thank you so much.