FSDS output data is abnormal in Fluxnet sites model

[wjp1997]

Dear scientist：

Hi! I am trying to establish a model in some Fluxnet sites to calculate the changes in stomatal conductance of vegetation over the years. I used the fluxnet2015 data to make atmosphere forcing data. I choose compset with 2000_DATM%1PT_CLM50%BGC- CROP_SICE_SOCN_MOSART_SGLC_SWAV. I try to use DATA_MODE =CLM1PT and data_mode=GSWPS3V1.
But, When I used DATA_MODE =CLM1PT, the output FSDS had an error different from the input FSDS. Every day FSDS is abnormally high at 17:00-18:00, this leads FSH to an abnormal high in output result.
Then， I tried to change data_mode=GSWPS3V1, FSDS was correct, and the output FSDS was the same as the input FSDS. but it has error Bottom layer specific humidity sent from the atmosphere model is less than zero. I have checked all RH in the input data, It didn't have a value less than zero, I think this may be induced by PSFR and TBOT.
I have mainly three questions:
1, Why does DATA_MODE =CLM1PT make output FSDS become abnormally ? input FSDS is different from putout FSDS. Could you give me some suggestions?
2, when I used data_mode=GSWPS3V1， PSFR（pressure） and TBOT (temperature) lead Bottom layer specific humidity sent from the atmosphere model is less than zero， how to deal with input data to solve the question?
3, compset 2000_DATM%1PT_CLM50%BGC- CROP_SICE_SOCN_MOSART_SGLC_SWAV，Is the compset I have chosen reasonable? Is this compset using the Medlyn method to calculate leaf stomatal conductance? Can I use satellite phenology (I1PtClm50SpGs)?
4，If I choose compset I1PtClm50SpGs，can I use other kind DATA_MODE, such as GSWPS3v1?


The following is my code to run the model:
./create_newcase --case ${SITENAME} --compset 2000_DATM%1PT_CLM50%BGC-CROP_SICE_SOCN_MOSART_SGLC_SWAV --res CLM_USRDAT --mach myintel --compiler intel --run-unsupported
cd ${SITENAME}

./xmlchange DATM_MODE=CLM1PT
./xmlchange CALENDAR=GREGORIAN
./xmlchange DIN_LOC_ROOT_CLMFORC=/home/wjp/cesm/inputdata/fluxnet/
./xmlchange CLM_USRDAT_NAME=${SITENAME}
./xmlchange ATM_DOMAIN_FILE=domain.lnd.1x1pt_${SITENAME}_1x1pt_${SITENAME}.240109.nc
./xmlchange LND_DOMAIN_FILE=domain.lnd.1x1pt_${SITENAME}_1x1pt_${SITENAME}.240109.nc
./xmlchange LND_DOMAIN_PATH=/home/wjp/cesm/inputdata/fluxnet/${SITENAME}/
./xmlchange ATM_DOMAIN_PATH=/home/wjp/cesm/inputdata/fluxnet/${SITENAME}/
./xmlchange RUN_STARTDATE=${yearstart}-01-01
./xmlchange DATM_CLMNCEP_YR_START=${yearstart}
./xmlchange DATM_CLMNCEP_YR_ALIGN=${yearstart}
./xmlchange DATM_CLMNCEP_YR_END=${yearend}
./xmlchange STOP_OPTION=nyears
./xmlchange STOP_N=${yearnum}
./case.setup

echo "fsurdat='/home/wjp/cesm/inputdata/fluxnet/"${SITENAME}"/surfdata_1x1pt_"${SITENAME}"_78pfts_CMIP6_simyr2000_c240109.nc'" >> user_nl_clm
echo "hist_nhtfrq = -1" >> user_nl_clm
echo "hist_mfilt = 8760" >> user_nl_clm
echo "hist_fincl1= 'GPP'" >> user_nl_clm
echo "hist_fincl1='FV'" >> user_nl_clm
./preview_namelists
cp CaseDocs/datm.streams.txt.CLM1PT.CLM_USRDAT user_datm.streams.txt.CLM1PT.CLM_USRDAT

cat CaseDocs/datm_in >> user_nl_datm

./preview_namelists
./check_input_data
./case.build --skip-provenance-check
cd ~/cesm/scratch/${SITENAME}/run
mkdir -p timing/checkpoints
mpirun -np 1 ../bld/cesm.exe
tar -cf ~/${SITENAME}.tar -P $(find . -name ""${SITENAME}".clm2.h0.*.nc")
cd ~
let ist=ist+1
done

 

[wjp1997]

there are some other files, Very Thanks for any help! I have confused for a long time~

 

[wjp1997]

there are some other files about DATA_MODE= GSWPS3V1, Very Thanks for any help! I have been confused for a long time~

 

[wjp1997]

@olesen

 

[wjp1997]

@oleson

 

[wjp1997]

I have solved the error that the Bottom layer specific humidity sent from the atmosphere model is less than zero, I changed the variables of input data, and I used QBOT to replace RH in CLM1PT_data. Most Fluxnet site results simulations have good accuracy performance. But, all sites in the southern hemisphere perform very badly （The correlation coefficient is almost 0， including EFLX_LH_TOT、FSH、GPP）. Taking the AU-How site as an example, I have checked all hourly output data, and I found that GPP gradually becomes 0, and the H changes very little in one day. May I ask if someone can give me any help or advice?

 

[oleson]

Glad to see you fixed your humidity problem. Not sure why you would get bad results just in the southern hemisphere. I suggest to check that your latitude/longitudes in your domain, surface dataset, and forcing data files are all correct and consistent. Also, make sure that the forcing data time is aligned with the model's GMT. For example, I see you've posted a plot of CLM FSDS versus FLUXNET FSDS but I can't tell if they are aligned.
To answer one of your other questions, you can look at your lnd_in and your lnd log to see which stomatal conductance model is being used. I think the namelist item is stomatalcond_method. You could also use the SP (satellite phenology) compset instead of BGC-CROP, both should be using Medlyn.

 

[wjp1997]

HI， Oleson， Very thanks for your help！ I have solved my problem！ the forcing data time is not aligned with the model's GMT， FLUXNET data use UTC， and the AU-How site is later than 9 hours， I checked FSR data, and it found that from 9:30 in the morning, the solar radiation reflected on the ground is equal to the direct solar radiation. After I changed the input data and compared the observation data to the CLM data, I got the model simulation results that seemed to be right! The following is the result of my re-change（H、LE and GPP）!
Thank you again for your help!

 

[oleson]

Great, thanks for letting us know.