Not finding the GDAS elevation file during LDT “Ensemble restart processing”

[Gayathridevi123]

Hi,

I'm trying to generate an ensemble restart file for data assimilation. Forcing is GDAS and LSM is Noah3.9. But while running LDT “Ensemble restart processing” it is not finding the GDAS elevation file:-

Reading the GDAS forcing data
GDAS forcing directory :./input/MET_FORCING/GDAS
[INFO] Initializing GDAS grid-1 for: 1991-2000 grid (T126)
[INFO] Initializing GDAS grid-2 for: 2000-2002 grid (T170)
[INFO] Initializing GDAS grid-3 for: 2002-2005 grid (T254)
[INFO] Initializing GDAS grid-4 for: 2005-2010 grid (T382)
[INFO] Initializing GDAS grid-5 for: 2010-2015 grid (T574)
[INFO] Initializing GDAS grid-6 for: 2015-present grid (T1534)
Projection and Transform: gaussian, bilinear
Reading forcing elev for GDAS_T126
GDAS elevation file missing:
Opening/reading GDAS elev file::
[ERR] error grib_open_file in read_gdas_elev Stopping. -11
endrun is being called

In the terminal it is showing ECCODES ERROR: IO ERROR: Bad address: (null) (Bad address)
Wondering why this is happening. For LDT "LSM parameter processing" the same files have been read by LDT and LIS open-loop outputs are generated.

Could anyone help me with this? Please find the ldt.config, ldtlog and lis_input_noah39.d01.nc files.
ldt.config.txt
ldtlog.txt
lis_input_noah39.d01.txt

Thank you,
Gayathri

[bmcandr]

Hi @Gayathridevi123,

Based on your ldtlog file it looks like the elevation filepath itself is not being read into LDT properly which naturally fails a "does this file exist?" check. Here is what the log should contain when initializing GDAS, note the presence of the filename:

...
 Reading forcing elev for GDAS_T126
 Opening/reading GDAS elev file::
 ./input/LS_PARAMETERS/metforcing_parms/GDAS/global_orography.t126.grb
 #Regridding: Applying bilinear interp to input parameter
 Done reading forcing elevation field.
...

What is odd is that we check that some path is present in the config file when reading config entries (see here). This should fail if no path is present, causing LDT to exit. Your GDAS entries are identical to the ldt.config in our GDAS testcase so I don't see a problem there.

Let's start troubleshooting this by modifying lines 53-57 in ldt/metforcing/gdas/readcrd_gdas.F90 to print another log message containing the filepath just after LDT reads it:

    call ESMF_ConfigFindLabel(LDT_config,"GDAS T126 elevation map:",rc=rc)
    do n=1,LDT_rc%nnest
       call ESMF_ConfigGetAttribute(LDT_config,LDT_rc%gdasT126elevfile(n),rc=rc)
       call LDT_verify(rc,'GDAS T126 elevation map: not specified')
       write(LDT_logunit,*) "T126 filepath: ", LDT_rc%gdasT126elevfile(n)
    end do

Re-compile, try to run again, and report back the contents of your log.

---

I do see there are some tab characters in your ldt.config file. See the ^I characters in the screenshot below:

I don't think these would cause a problem given other paths are read, but it couldn't hurt to replace these with spaces. To see these characters open the file in vim and type :set list.

[bmcandr]

Hi @Gayathridevi123,

Thanks for the clarification regarding the runmode you are using.

Can you try setting Number of met forcing sources: 0 to disable metforcing source? As far as I can metforcing data do not need to be processed for upscaling/downscaling LSM restart files. Here is an example ldt.config file for this runmode: ldt.config_ensrst.txt

Note that metforcing is disabled in this config file.

@karsenau, should/can metforcing sources be disabled when using LDT's Ensemble restart processing runmode?

[Gayathridevi123]

Hi bmcandr,

Thank you for the config file. By setting Number of met forcing sources: 0 LDT ran successfully.

I just want to clarify about the metforcing data. Is it not needed for ensemble restart file generation?

Thanks
Gayathri

[bmcandr]

I am not very well versed in this area of LISF, but my understanding is that LDT is using an existing LIS restart file to generate a new restart file. It is not creating a new LIS input parameter file and therefore the metforcing inputs are not needed for this runmode.

[Gayathridevi123]

Thank you Brendan for your help.

[karsenau]

Yes -- to confirm that you want to set the number of met forcing sources to "0" when working on generating a new restart file using LDT. The restart file only cares about the LSM or hydrological model states that are needed to initialize or restart the same model runs.