CAM and CLM output frequency

[San Diego WX]

Hi there. I am new to CESM and need help figuring out how to control output frequency. I am doing a 1-yr FHIST run and would like output for get daily average precipitation and daily average near-surface temperature. Below are my edits so far

1) added these lines to user_nl_cam
fincl2='PRECC','PRECL'
nhtfrq=0,-24
mfilt=1,1

2) added this to user_nl_clm
hist_fincl2 = 'TSA'
hist_nhtfrq = 0,-24
hist_mfilt = 1,1

I can build and run the model but I am not sure I am getting what I wanted. Are my changes correct? Am I getting daily averages or instantaneous values?
Thank you!

 

[hannay]

Yes, this will give you daily output of the variable of the variables
'PRECC' and 'PRECL' for CAM
'TSA' for CLM

If you want to make sure to have daily average you can add the flag ':A'
Also, I recommend that you increase mfilt for the daily output. The current setting will give one file per day. It will be many files and it is not the best way to handle the space.

for instance, you could try:

fincl2='PRECC:A','PRECL:A'
nhtfrq=0,-24
mfilt=1,30

2) added this to user_nl_clm

hist_fincl2 = 'TSA:A'
hist_nhtfrq = 0,-24
hist_mfilt = 1,30

 

[San Diego WX]

Thank you!

 

[CGL]

Hi, where can i find the guidence for output in clm. I didn't see the detail in clm teachnote. Maybe i miss it?

 

[CGL]

I modified the use_nl_clm as :
hist_mfilt=12 ，hist_nhtfrq=0
But i got the monthly netcdf file still. @hannay

 

[zarzycki]

hist_nhtfrq=0 means monthly output, so that is what you would expect. See @hannay's example again, the first integer in hist_nhtfrq, hist_mfilt, corresponds to the fincl1 (h0) files, the second integer corresponds to the fincl2 (h1) files, etc.

hist_fincl2 = 'TSA:A'
hist_nhtfrq = 0,-24
hist_mfilt = 1,30

 

[CGL]

Yes, I want to output a netCDF file that contains 12 month for 1 year. But the code seems didn't work.
hist_mfilt=12 ，hist_nhtfrq=0

 

[zarzycki]

The monthly averaged files (nhtfrq=0) are hardcoded to have one timestamp by default. I do not know the formal origin of this decision, but I assume it has to do with the diagnostics packages that are used to evaluate the model.

E.g., in cam_history.F90.

        ! Only one time sample allowed per monthly average file
        !
        if (nhtfrq(t) == 0) then
          mfilt(t) = 1
        end if

I see two viable options.
1.) Make sourcemods to remove this restriction on monthly average files.
2.) Use a tool such as ncrcat to "glue" monthly averages files together as a postprocessing step.

module load nco # or conda install nco -c conda-forge
ncrcat mycamfiles.h0.*nc myconcatenatedfile.nc

The latter is probably easiest (and preferred).

 

[zarzycki]

FYI, the relevant code in CTSM (histFileMod.F90)

  integer, public :: &
       hist_mfilt(max_tapes) = (/ 1, (30, ni=2, max_tapes)/)        ! namelist: number of time samples per tape

 

[CGL]

The monthly averaged files (nhtfrq=0) are hardcoded to have one timestamp by default. I do not know the formal origin of this decision, but I assume it has to do with the diagnostics packages that are used to evaluate the model.

E.g., in cam_history.F90.

        ! Only one time sample allowed per monthly average file
        !
        if (nhtfrq(t) == 0) then
          mfilt(t) = 1
        end if

I see two viable options.
1.) Make sourcemods to remove this restriction on monthly average files.
2.) Use a tool such as ncrcat to "glue" monthly averages files together as a postprocessing step.

module load nco # or conda install nco -c conda-forge
ncrcat mycamfiles.h0.*nc myconcatenatedfile.nc

The latter is probably easiest (and preferred).

You are right. I see the hist_mfilt must equal 1 when we set the hist_nhtfrq=0 in the Tutorials. The ncrcat may be the best way.