Is GEOSldas_LDAS.rc ignored?

[weiyuan-jiang]

The default values in GEOSldas_LDAS.rc are not captured in this function _produceExeInp
Is that intentional or a bug ?

@saraqzhang @gmao-rreichle

[gmao-rreichle]

Not sure what exactly isn't captured. There are no defaults for [EXP_ID, EXP_DOMAIN, ..., BCS_RESOLUTION].
Then there's the block:

GEOSldas_GridComp/GEOSldas_App/ldas_setup

Line 1778 in dd54caa

# add defaults from GEOSldas_LDAS.rc

where the behavior depends on ladas_cpl. Which case are you asking about?

[weiyuan-jiang]

Maybe this is a bug introduced by me. The default values should be added to the output out_dict

But somehow the lines saved here are not used. lines

[gmao-rreichle]

Maybe this is a bug introduced by me. The default values should be added to the output out_dict

This particular line is related to GEOS_SurfaceGridComp.rc

But somehow the lines saved here are not used. lines

This particular line is related to GEOSldas_LDAS.rc

How are these two code sections connected? There may be nothing wrong here

[weiyuan-jiang]

Maybe this is a bug introduced by me. The default values should be added to the output out_dict

This particular line is related to GEOS_SurfaceGridComp.rc

But somehow the lines saved here are not used. lines

This particular line is related to GEOSldas_LDAS.rc

How are these two code sections connected? There may be nothing wrong here

The values in GEOSldas_LDAS.rc are not in out_dic. So my original question still holds: Is this intentional?

[gmao-rreichle]

The values in GEOSldas_LDAS.rc are not in out_dic. So my original question still holds: Is this intentional?

Thanks, @weiyuan-jiang! I see the issue now. The "out_dict" argument for _ProduceExeInp() was added to create a copy of the sample exeinp file inside memory for the coupled land-atm DAS setup (i.e., avoid writing and reading back the sample exeinp file). It is meant to capture all default values that come with the tag (including those in GEOSldas_LDAS.rc and GEOS_SurfaceGridComp.rc).

The setup for the coupled land-atm DAS is done by fvsetup, which calls ldas_setup. If I understand this correctly, the LDAS.rc file that is created during the coupled land-atm DAS setup is now missing the defaults from GEOSldas_LDAS.rc for the rc variables that are not otherwise overwritten during ldas_setup. This is not the desirable outcome, even though it may be working, probably because the defaults are also in the Fortran source code.

The fix for this needs to go into a separate PR, assuming the ldas_setup refactoring will take longer. I'm desparately trying to wrap up #111 and #118 so I can create a new GEOSldas_GridComp release for use in the coupled land-atm DAS.

cc: @saraqzhang

[saraqzhang]

that is not intentional. but it worked, because the parameters defaults currently in GEOSldas_LDAS.rc are reset by ldas_setup in case of ladas_cpl .

[gmao-rreichle]

that is not intentional. but it worked, because the parameters defaults currently in GEOSldas_LDAS.rc are reset by ldas_setup in case of ladas_cpl .

But for ladas_cpl>0, ldas_setup does not reset all of the defaults in GEOSldas_LDAS.rc. So there are some rc variables that remain undefined in the LDAS.rc of the coupled land-atm DAS. We still need to capture them

[saraqzhang]

I wondered where ldas_setup get the value of parameters such as CATCHMENT_OFFLINE: 1 ( written in LDAS.rc when ladas_cpl > 0 ), it is one of the defaults in GEOSldas_LDAS.rc that ldas_setup does not overwrite. maybe the function call in ldas_setup : status = ld.createRCFiles() get the default values after all ?

[gmao-rreichle]

I wondered where ldas_setup get the value of parameters such as CATCHMENT_OFFLINE: 1 ( written in LDAS.rc when ladas_cpl > 0 ), it is one of the defaults in GEOSldas_LDAS.rc that ldas_setup does not overwrite. maybe the function call in ldas_setup : status = ld.createRCFiles() get the default values after all ?

I'm afraid I can't say what createRCFiles() does exactly, and how it relates to the dictionary returned by _ProduceExeInp(). It's possible the defaults are captured in createRCFiles(). @weiyuan-jiang ?

[weiyuan-jiang]

The crateRCFiles() function indeed creates LDAS.rc that includes all the defaults. But I have two questions:

1. Is ladas trying to avoid LDAS.rc ?
2. more seriously, LDAS.rc probably uses GEOSldas's defaults instead of GEOSgcm's defaults.

[weiyuan-jiang]

The crateRCFiles() function indeed creates LDAS.rc that includes all the defaults. But I have two questions:

1. Is ladas trying to avoid LDAS.rc ?
2. more seriously, LDAS.rc probably uses GEOSldas's defaults instead of GEOSgcm's defaults.

I take my second point back. LDAS.rc uses GEOSgcm's defaults if ladas_cpl > 0

[gmao-rreichle]

Is ladas trying to avoid LDAS.rc ?

The LADAS does use LDAS.rc because it also runs GEOSldas.x (i.e., an offline LDAS), except the latter is governed by and coupled with what is going on in the ADAS, hence the modifications in ldas_setup.

[weiyuan-jiang]

So now the question is why in the first place we need the default values ( out_dict) if they are only used during run from LDAS.rc.