Irrigation Module

[stefanocasirati]

Overview:

• Adds new land BCs version (v14)
• Allows Irrigation on Catchment ("land") tiles

Repositories affected by updates:
_GEOSgcm_GridComp, GEOSldas_GridComp_

New Files:

• GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSirrigation_GridComp/GEOS_IrrigationGridComp.F90
• GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/GEOSland_GridComp/GEOSirrigation_GridComp/irrigation_model.F90
• GEOSgcm_GridComp/GEOSagcm_GridComp/GEOSphysics_GridComp/GEOSsurface_GridComp/Utils/Raster/makebcs/module_irrig_params.F90

Brief description:
New land BCs version:
v14 : v13 (i.e., NL3 + JPL veg height + PEATMAP + MODIS snow alb v2 + mean land elevation fix) + Irrigation

The bcs generation package reads different irrigated area datasets and generates a new file "irrigation_IMxJM_DL.dat" containing, for each tile, a static: irrigation fraction, Paddy fraction, Rainfed fraction, crop irrigation fraction, crop irrigation type, crop planting and harvest dates, sprinkler fractions, drip fractions, flood fractions, and minimum and maximum LAI.

Irrigation module:
Adapted from: feature/SM_irrigation_model
The irrigation module computes irrigation rates by applying three different irrigation schemes: sprinkler, drip, and flood (furrow). Irrigation is triggered based on location, growing season (Implemented alternative options: LAI-based and Crop Calendar-based), and soil moisture.

New Output Variables: each variable is weighted based on the specific irrigated fractions:
SPRINKLERRATE: sprinkler_irrigation_rate (kg m-2 s-1)
DRIPRATE: drip_irrigation_rate (kg m-2 s-1)
FLOODRATE: flood_irrigation_rate (kg m-2 s-1)
IRRLAND: total irrigation rate (kg m-2 s-1) = SPRINKLERRATE + DRIPRATE + FLOODRATE

To do:

• Verify stop/start regression.
• Fix GNU build (see Irrigation Module #959 (comment))

LDAS nightly test

Runtype	Clone	Build	Build Time	Model Run/Compare	Assim Run/Compare
-------	-----	-----	----------	-----------------	-----------------
conus	pass	pass	43 min	pass/pass	-- / --
global	--	--	--	pass/pass	pass/pass
globalcs	--	--	--	pass/pass	pass/pass
globalcnclm4	--	--	--	pass/pass	-- / --

[stefanocasirati]

The latest commits updated flood irrigation by differentiating it between furrow irrigation (now applied only to irrigated fractions) and flood irrigation (now applied only to paddy fractions).
The main difference consists of where the water is applied: furrow irrigation rate is still applied to the root zone, while flood irrigation rate is now applied to the surface excess.

Output Variables: each variable is weighted based on the specific irrigated fractions:
SPRINKLERRATE: sprinkler_irrigation_rate (kg m-2 s-1)
DRIPRATE: drip_irrigation_rate (kg m-2 s-1)
FURROWRATE: furrow_irrigation_rate (kg m-2 s-1)
FLOODRATE: flood_irrigation_rate (kg m-2 s-1)
IRRLAND: total irrigation rate (kg m-2 s-1) = SPRINKLERRATE + DRIPRATE +FURROWRATE+ FLOODRATE

[gmao-rreichle]

@stefanocasirati : I just pushed several commits:

Primarily, it's an attempt to make the names of the internals (irrigation model prognostics and parameters) clearer and more consistent. Prefacing each name with IRRG_ helps keep these parameters separate from those of other model components.

I think there was a bug in the handling of MIDS_LENGTH, see 43167ab

I recognize that the renaming of the internals will create havoc with the bcs. To test these changes, we'd either need to regenerate the bcs or create a copy of the current ones and then surgically rename the impacted, irrigation-related nc4 variables. For now, though, this is probably premature. I haven't got to al the variables yet, and I still have questions re. the meaning of some variables.

As of right now, I just pushed my commits, and I don't even know if everything builds ok. I'll keep working as time permits

[gmao-rreichle]

A few more comments, see below

[gmao-rreichle]

Do we really want to hardwire defaults here and in GEOS_SurfaceGridComp.rc ?

[gmao-rreichle]

get help from SI team to avoid duplication of "to_radian()"

[gmao-rreichle]

get help from SI team to avoid duplication of "haversine()"

[gmao-rreichle]

If I understand correctly, EFCOR is introduced because irrigation water is "lost" during irrigation (presumably via direct evaporation to the atmosphere or by running off the field into natural land). Do we keep track of this water somehow? Does the model as is close the water balance? Not sure if there's a difference re. EFCOR and the water balance between sprinkler and flood irrigation. Since SRATE is added to precipitation, I suppose the sprinkler water lost will be accounted for (via interception loss and runoff). But does the same apply to flood irrigation water? This may need more thought.

[gmao-rreichle]

Need to document somewhere that SPIRRG must be included in the irrigation water balance. Also need to verify that irrigation water balance (and total water balance) closes.

[gmao-rreichle]

If I understand the code correctly, for LAI-triggered irrigation, SRATE and DRATE are effectively 1-dimensional, but FRATE has 2 elements in the 2nd dimension, one for FURROW and one for PADDY irrigation. I think it would be much cleaner to split FRATE into two one-dim arrays: FRWRATE and PDYRATE for LAI-triggered irrigation. The problem with this approach is that existing restart files would need to be translated accordingly. But because of my efforts to rename just about every variable, we probably have to deal with fixing up restarts anyway. Something to consider.