Problems with simulation start times

[liu-13]

I used ERA5-land data for driving the Noah-MP land surface model. The start time is set to 12:00 on July 28, 2016 and no errors are reported when running Prepare_ERA5-Land_Forcing.sh and extract_era5_init.py. But after I run create_forcing.exe, the first momentary LDASIN file generated, the 2016072812.LDASIN_DOMAIN1 file, does not have the VEGFRA and LAI variables in it. And the Q2 obtained from later simulation moments has a large number of default values. Whereas, if the simulation starts at 00:00, everything works fine and does not have these problems with the simulation starting at 12:00. I don't know what's the problem, please help, thanks!The following two graphs show the Q2MV at the same moment from the simulation starting at 12:00 and 00:00, respectively.

[cenlinhe]

which dynamic vegetation option did you use? If you use the option that requires time series of vegetation fraction and LAI from input files, you will need to add the vegetation fraction and LAI variables to every forcing files based on the time. The current ERA5 preprocessing scripts and create_forcing.exe will not do this. They only generate the required meteorological forcing variables.
Also, please double check your generated forcing files and HRLDAS_setup files to make sure the variables inside the files have reasonable values.

[liu-13]

Do you use same HRLDAS_setup file for two runs? Or different ones?

different ones. However, I found that as long as the initial moment is simulated, regardless of whether the initial moment is 12:00 or 00:00, the Q2MV and Q2MB in the LDASOUT file for the first simulated moment I get are -9999e+03. but they are all normal in the LDASIN file.

[cenlinhe]

Ok, to answer your previous questions:

1. because you used differnet HRLDAS_setup file for your two simulations starting at different times, your results will be different.
2. The Q2MV and Q2MB are diagnostic variables from model which are set to the default missing value -9999 at the very initial/starting time point in the output file, because the model has not executed any physics to compute Q2MV and Q2MB yet.
3. The Q2 from model output will be different from the specific humidity in the forcing (even if you use Q2 for the forcing variable), because the specific humidity in the forcing is the value at the model forcing level (ZLVL set in the namelist.hrldas). The model will further use M-O theory to compute a new Q2 at the 2-m level (as I explained earlier).

[liu-13]

Ok, to answer your previous questions:

1. because you used differnet HRLDAS_setup file for your two simulations starting at different times, your results will be different.
2. The Q2MV and Q2MB are diagnostic variables from model which are set to the default missing value -9999 at the very initial/starting time point in the output file, because the model has not executed any physics to compute Q2MV and Q2MB yet.
3. The Q2 from model output will be different from the specific humidity in the forcing (even if you use Q2 for the forcing variable), because the specific humidity in the forcing is the value at the model forcing level (ZLVL set in the namelist.hrldas). The model will further use M-O theory to compute a new Q2 at the 2-m level (as I explained earlier).

Thank you. But I also found a problem, with normal forcing data, no matter if the simulation starts from 00:00 or 12:00, the Q2MB and Q2MV in the LDASOUT file for the first moment of the simulation is -9999, and in the whole simulation area it's -9999. but not for the subsequent moments, it's just normal. I don't know what is the reason for this.

[cenlinhe]

the reason is my 2nd point mentioned above.

[liu-13]

2. The Q2MV and Q2MB are diagnostic variables from model which are set to the default missing value -9999 at the very initial/starting time point in the output file, because the model has not executed any physics to compute Q2MV and Q2MB yet.

I'm sorry, I read it wrong. Thank you for your answer, I understand.