Set generate_initialization_file=True in run_lsm_rapid_process method but nothing generates

[jlgutenson]

I was running a set of FLDAS land surface model outputs for several HydroSHEDS drainage basins using the run_lsm_rapid_process method.

After completing several RAPID runs, one of the initializations file was not generated by the code. This leads to the namelist pointing to the previously processed watershed and an error interrupting processing.

[snowman2]

@jlgutenson, can you share the function you were running so I can check the inputs?

[jlgutenson]

@snowman2 Sure, here you go. See below for GLDAS data example.

run rapid for each HydroSHED

run_lsm_rapid_process(
    rapid_executable_location='/var/lib/rapid/src/rapid',
    rapid_io_files_location='/home/user/rapid-io',
    lsm_data_location='/home/user/GLDAS_NOAH10_3H.2.0',
    simulation_start_datetime=datetime(2001, 1, 2),
    simulation_end_datetime=datetime(2001, 1, 31),
    generate_initialization_file=True,
    file_datetime_re_pattern = r'\d{8}\.\d{2}',
    file_datetime_pattern = "%Y%m%d.%H",
    convert_one_hour_to_three=False
)  

[snowman2]

Did you look in the input or output folder?

It should be in the input folder. https://github.com/erdc-cm/RAPIDpy/blob/a4764eb81659074efe7399d136538f1df6d30098/RAPIDpy/inflow/lsm_rapid_process.py#L836-L840

[jlgutenson]

Yes, it seems to work for a segment of the watersheds I am providing RAPID but fails at other instances. See the output regarding the error.

nohup.txt

[snowman2]

I noticed that the model is attempting to run from an initial flow file. Very odd... Not sure how that is happening unless the rapid_namelist file is not being removed. Can you try running the model after deleting the rapid_namelist file and any qinit files?

[snowman2]

@jlgutenson, you are correct. I will try to work in a fix for this before the next release.

[snowman2]

@jlgutenson, try the latest. It should be fixed, but if not, let me know. Thanks!