no ensemble average for catchCN

[weiyuan-jiang]

GEOSldas/src/Components/GEOSldas_GridComp/GEOS_LdasGridComp.F90

Lines 921 to 933 in 8ad174f

	if (LSM_CHOICE == 1) then
	! collect cat_param
	call ESMF_GridCompRun(gcs(ENSAVG), importState=gex(igc), exportState=gex(ENSAVG), clock=clock,phase=3, userRC=status)
	VERIFY_(status)
	call ESMF_GridCompRun(gcs(ENSAVG), importState=gex(igc), exportState=gex(ENSAVG), clock=clock,phase=2, userRC=status)
	VERIFY_(status)
	if( mwRTM ) then
	! Calculate ensemble-average L-band Tb using LAND's output (add up and normalize after last member has been added)
	call ESMF_GridCompRun(gcs(LANDASSIM), importState=gex(igc), exportState=gex(LANDASSIM), clock=clock,phase=3, userRC=status)
	VERIFY_(status)
	endif
	endif

I think at this point there is no ensemble average of LSM_CHOICE =2 and 3. Do you want to add those average ? @gmao-jkolassa

[gmao-jkolassa]

@weiyuan-jiang Can you please clarify for me what this means in terms of running the model? When we run Catchment-CN in ensemble mode, it does currently write out ensemble average output files. So I am not quite sure what it is that is missing?

[gmao-rreichle]

I'm pretty sure we need to add the ensavg functionality for the CatchCN exports. Otherwise, the CatchCN ensemble runs would have to write out data for all ens members separately and the ens avg would have to be done in post-proc. For Catch, all of our ens runs so far have only ever output the ens avg. I think one of Manuela's students was interested in outputting individual members and Weiyuan created a script to generate the appropriate HISTORY file. But we didn't hear back after the initial effort.
@weiyuan-jiang : Do we still have the script to create a HISTORY file that outputs individual ens members? If so, we should add it to the GEOSldas utilities in the Applications dir.

[gmao-jkolassa]

Do we still have the script to create a HISTORY file that outputs individual ens members? If so, we should add it to the GEOSldas utilities in the Applications dir.

I know this question was for Weiyuan, put Trent actually needs to output individual ensemble members for Catchment-CN for the MAP project he is working on. He has manually created the appropriate HISTORY.rc file, but in the future it would be convenient to have a script that could do that. The LDAS runs within the LADAS are another example where this currently has to be done manually.

[gmao-rreichle]

Thanks, @gmao-jkolassa, for the updates.
@weiyuan-jiang, I still think we need the ensavg functionality for CatchCN exports. And we also have to revisit/update/test the script that generates HISTORY.rc for outputting individual members and add it into the utilities. Thanks!

[weiyuan-jiang]

I am confused. How can you get the average output ?

[gmao-jkolassa]

Not sure, but Trent was running CatchmentCNCLM45 in ensemble mode and it wrote out ensemble average files (but not individual ensemble members). I haven't double-checked whether those files have any reasonable data in it though.

[gmao-rreichle]

I am confused. How can you get the average output ?

@weiyuan-jiang:
I don't quite understand your question.
Ens avg output is working fine for Catch, and the "LADAS" mentioned by @gmao-jkolassa is for Catch only (so far).
I'm not aware of anyone having got ens avg output for CatchCN. Jana's contact (Trent) uses CatchCN but has only been writing output from individual members.

So, unless I'm misunderstanding the issue here, we do need to:

1. add the ens avg functionality for CatchCN-specific variables
2. add the utility script that creates a HISTORY.rc file for output from individual members into the GEOSldas repo

When you get a chance, please create a branch and start a PR. Let me know if you have any questions, thanks!

[gmao-jkolassa]

Just to clarify, before Trent started writing out individual ensemble members he did run a CatchCN simulation in ensemble mode and it did generate output files in the /ens_avg directory. As I mentioned, I did not check whether the values in those files are in any way reasonable. I have sent @weiyuan-jiang the directory where this experiment is located, so he can take a look himself.

[gmao-rreichle]

I highly doubt that Trent's output in ./ens_avg is indeed ens average data.