Water demand and allocation in `sbm` model concept #402

verseve · 2024-04-24T14:09:21Z

Issue addressed

Fixes #223 and fixes #102

Explanation

Support water demand and allocation in the sbm model concept. Water demand (gross and net) for the sectors industry, domestic and livestock can be provided (optional) to the wflow_sbm model as input. Irrigation demand for paddy areas (Paddy) or other crops (NonPaddy), both optional, is computed during the simulation. Water is first abstracted from surface water, first locally (grid cell) and then for allocation areas. For the remaining water demand groundwater (first locally and then for allocation areas) is abstracted. Return flows are computed for water demand sectors industry, domestic and livestock.

Checklist

Updated tests or added new tests
Branch is up to date with master
Tests & pre-commit hooks pass
Updated documentation if needed
Updated changelog.md if needed

Rendered docs in e.g. https://deltares.github.io/Wflow.jl/previews/PR402/model_docs/vertical/sbm/#sbm_demand_allocation

as part of vertical SBM concept

for each sector a struct is defined

- all structs part of SBM (domestic, industry, livestock, nonpaddy and paddy) - for now assumption is that water demand arrays have the same lenght as SBM (can change later)

- soil water pressure heads of the root water uptake (rwu) reduction function (Feddes) can be provided (netCDF) - potential transpiration is partioned over depth using the root fraction - h1 of the rwu reduction function can be provided, either as 1.0 (rwu not constrained during saturated conditions) or 0.0 (rwu constrained during saturated conditions)

was missing in previous commit

- update unit tests - limit nonpaddy irrigation gross demand by soil infiltration capacity - include irrigation efficieny in gross demand calculation

now after snow and before recharge update

total_gross_demand, fraction groundwater used (irrigation and non irrigation) and allocation areas.

and added field `volume` to lateral kinematic wave subsurface flow

to kinematic wave for river flow, this is computed actual surface water abstraction from `WaterAllocation`. A check is maded for external negative inflow (local abstraction), and local available surface water volume for surface water abstraction is adjusted.

- set surface water `abstraction` from update of `WaterAllocation` - fix update of `act_surfacewater_abst` at allocation area level (should be added to local abstraction).

Errors related to variable names and indexing.

actual groundwater abstraction from `WaterAllocation` is added to `recharge` of the model type `sbm`.

- use one value for fraction surface water (no subdivision between irrigated and non-irrigated demands) - livestock can abstract from both surface and groundwater (was limited to surface water)

- groundwater demand calculation not correct (units) - reset `surfacewater_alloc` (0.0) for all cells

- add allocated irrigation water to SBM - compute return flow for non irrigation water withdrawal - add units/descriptions to fields of structs water demand

- split waterallocation between river and land domain, this is also easier for writing output - add non irrigation returnflow to river and overland flow kinematic wave routing

- fix check to apply `irri_alloc` in `SBM` - update waterdemand settings (read from TOML)

Use root fraction (length) for each unsaturated zone layer to scale required irrigation demand (only root zone). Fix computation infiltration_capacity and irri_dem_gross for nonpaddy irrigation (removed from unsaturated zone layers loop).

exclude lake and reservoir areas

Model parameter `rootfraction` and partitioning of potential transpiration over soil layers.

visr · 2024-06-25T13:21:05Z

src/flow.jl

@@ -29,6 +30,7 @@ abstract type SurfaceFlow end
    lake_index::Vector{Int} | "-"                # map cell to 0 (no lake) or i (pick lake i in lake field)
    reservoir::R | "-" | 0                       # Reservoir model struct of arrays
    lake::L | "-" | 0                            # Lake model struct of arrays
+    waterallocation::W | "-" | 0                 # Water allocation


Maybe just allocation? It's also abstraction.

I agree with replacing waterallocation by allocation.

visr · 2024-06-27T08:10:42Z

src/utils.jl

+function divide(x, y; max = 1.0, default = 0.0)
+    z = y > 0.0 ? min(x / y, max) : default
+    return z
+end


Perhaps it is more explicit to name this something like bounded_divide. Would be good to add a docstring and unit tests as well.

Good point, changed the function name and added a docstring and unit tests as well.

visr · 2024-06-27T08:15:38Z

test/groundwater.jl

        fill(0.1, ncell), # specific storage
        fill(1.0, ncell),  # storativity
        fill(0.0, connectivity.nconnection),  # conductance
+        fill(0.0, ncell), # total volume that can be released
    )


It looks like something like Deltares/Ribasim#1566 may be useful for you as well.

Yes, for most structs we already make use of this, probably good to extend this to smaller structs as well as part of the v1 release.

visr · 2024-06-27T08:29:27Z

test/vertical_process.jl

-    )
+    head = Wflow.head_brooks_corey(0.25, 0.6, 0.15, 10.5, -10.0)
+    @test head ≈ -90.6299820833844
+    h3 = Wflow.feddes_h3(-300.0, -600.0, 3.5, Second(86400))


I feel like these functions are a bit undertested. The functions have several if branches, but we test only one.

I agree, tests are extended.

visr · 2024-06-27T09:03:45Z

docs/src/model_docs/params_vertical.md

@@ -61,6 +61,13 @@ profile `kv` is used and `z_layered` is required as input.
 | **`waterfrac`** | fraction of open water (excluding rivers) | - | 0.0  |
 | **`pathfrac`** | fraction of compacted area | - | 0.01  |
 | **`rootingdepth`** | rooting depth  | mm | 750.0  |
+| **`rootfraction`** | root fraction per soil layer (relative to the total root length) | - | - |


Is the total root length here the length from the surface to the bottom of the longest root? Or is it the total length of all roots, such that it can take into account that some depths have a higher root density?

It can indeed take into account that some depths have a higher root density, changed the description to make this more clear.

visr · 2024-06-27T10:52:03Z

src/sbm.jl

+    # correct rooting depth for soilthickness
+    rootingdepth = @. min(0.99 * soilthickness, rootingdepth)


You could also consider throwing an error. Perhaps more annoying at first, but also less magical.

Not sure if I agree with an error (due to different underlying datasets of soilgrids and the LULC mapping tables of hydromt_wflow). But perhaps showing a warning message indicating (in how many cells) this correction occurs would be nice

I think the preferred approach is to catch and correct this during model building as a pre-processing step. By the way, this correction or magical step was already implemented in the wflow Python version.

visr · 2024-06-27T10:53:32Z

src/sbm.jl

+                availcap =
+                    min(1.0, max(0.0, (sbm.rootingdepth[i] - sbm.sumlayers[i][k]) / usl[k]))
+            end
+			maxextr = usld[k] * availcap


Removed tabs.

visr · 2024-06-27T10:53:45Z

src/sbm.jl

+            end
+			maxextr = usld[k] * availcap
+            # the rootfraction is valid for the root length in a soil layer, if zi decreases the root length
+            # the rootfraction needs to be adapted           


Trailing whitespace

visr · 2024-06-27T11:02:00Z

src/sbm.jl

+
+Update water demand for vertical `SBM` concept for a single timestep. Water demand is
+computed for sectors `industry`, `domestic` and `livestock`, and `paddy` rice fields and
+`nonpaddy` (other crop) fields. 


Suggested change

`nonpaddy` (other crop) fields.

`nonpaddy` (other crop) fields.

Might be good to look into pre-commit to remove trailing whitespace.

visr · 2024-06-27T12:57:16Z

src/flow.jl

+    @unpack subdomain_order, topo_subdomain, indices_subdomain, upstream_nodes, area =
+        network


Unrelated, but I'd replace all unpack macros by the now built in destructuring, (; a, b) = c

Yes, good suggestion, definitely something for v1 work!

Before updating surface water and groundwater allocations and abstractions reset to zero, so it is independent of the update at local and allocation area levels.

Water bodies were excluded in the computation of `groundwater_demand`.

JoostBuitink

Very nice work! A couple of comments on top of @visr's comments.

docs/src/model_docs/params_vertical.md

docs/src/model_docs/vertical/sbm.md

JoostBuitink · 2024-07-10T08:28:05Z

src/sbm.jl

+    # correct rooting depth for soilthickness
+    rootingdepth = @. min(0.99 * soilthickness, rootingdepth)


Not sure if I agree with an error (due to different underlying datasets of soilgrids and the LULC mapping tables of hydromt_wflow). But perhaps showing a warning message indicating (in how many cells) this correction occurs would be nice

src/sbm.jl

src/water_demand.jl

test/groundwater.jl

Unit of this variable is mm/t so `maximum_irrigation_rate` is a better variable name. Also fixed a couple of typos in docs.

visr

Don't really have time for another review round, but saw that my comments were being addressed, so for me this is good to go.

JoostBuitink

Thanks for addressing all comments (also quickly looked at the responses to Martijn's comments)! Looks good to me!

vers-w added 30 commits January 11, 2024 10:34

initialization water demand (non-irrigation)

c082441

as part of vertical SBM concept

file water_demand

23bd36e

change netto to net

2cb6449

non-irrigation water demand sectors

415f798

for each sector a struct is defined

irrigation structs (not complete)

dce4d74

water demand structs

2b6017a

- all structs part of SBM (domestic, industry, livestock, nonpaddy and paddy) - for now assumption is that water demand arrays have the same lenght as SBM (can change later)

fix: add ust option

bb4ea77

was missing in previous commit

use crop coefficient kc (and not et_reftopot)

589b31b

run JuliaFormatter

709ece5

water demand calculation

eaeb7bb

irrigation demand and tests

650c493

- update unit tests - limit nonpaddy irrigation gross demand by soil infiltration capacity - include irrigation efficieny in gross demand calculation

move water demand (and allocation) calculation

059cd30

now after snow and before recharge update

add fields WaterAllocation

6dc92b8

total_gross_demand, fraction groundwater used (irrigation and non irrigation) and allocation areas.

surface water abstraction and allocation

d0f3ed2

groundwater abstraction and allocation

9f5939a

and added field `volume` to lateral kinematic wave subsurface flow

surface water abstraction (set and fix)

0c99ff4

- set surface water `abstraction` from update of `WaterAllocation` - fix update of `act_surfacewater_abst` at allocation area level (should be added to local abstraction).

fix running update water demand and allocation

900864e

Errors related to variable names and indexing.

add groundwater abstraction to recharge of SBM

3f12e69

actual groundwater abstraction from `WaterAllocation` is added to `recharge` of the model type `sbm`.

fraction surface water and livestock

aac650f

- use one value for fraction surface water (no subdivision between irrigated and non-irrigated demands) - livestock can abstract from both surface and groundwater (was limited to surface water)

fix groundwater demand, reset surfacewater_alloc

37f964b

- groundwater demand calculation not correct (units) - reset `surfacewater_alloc` (0.0) for all cells

precalculate area as part of model.network.land

d296eab

irrigation allocation, returnflow, units

3613d2e

- add allocated irrigation water to SBM - compute return flow for non irrigation water withdrawal - add units/descriptions to fields of structs water demand

non irrigation returnflow and split waterallocation

b17a67a

- split waterallocation between river and land domain, this is also easier for writing output - add non irrigation returnflow to river and overland flow kinematic wave routing

fix check (add irrigation) and update waterdemand settings

22176e2

- fix check to apply `irri_alloc` in `SBM` - update waterdemand settings (read from TOML)

update water demand settings

846a565

local surface water abstraction

2e5b559

exclude lake and reservoir areas

update metadata structs water demand

10b5465

vers-w added 2 commits June 13, 2024 16:38

Update changelog

21d845c

Model parameter `rootfraction` and partitioning of potential transpiration over soil layers.

Update docs and changelog

f393cea

verseve marked this pull request as ready for review June 14, 2024 11:38

verseve requested review from JoostBuitink, CFBaptista and visr June 14, 2024 11:38

verseve linked an issue Jun 17, 2024 that may be closed by this pull request

improvement reduction function Feddes #102

Closed

Fix adding nonirri_returnflow to overland flow

9a0927c

shartgring mentioned this pull request Jun 24, 2024

Setting downstream boundary condition for local-inertial Deltares/hydromt_wflow#282

Open

visr reviewed Jun 27, 2024

View reviewed changes

vers-w added 4 commits July 2, 2024 08:44

Reset allocations and abstractions before update

61640d7

Before updating surface water and groundwater allocations and abstractions reset to zero, so it is independent of the update at local and allocation area levels.

Small fix unit act_surfacewater_abst

495c6d9

Fix indexing nonirri_returnflow for river cells

c590219

Compute groundwater_demand for all land cells

c2ca984

Water bodies were excluded in the computation of `groundwater_demand`.

verseve removed the request for review from CFBaptista July 3, 2024 11:33

Simplify computation groundwater_demand local

a24c531

JoostBuitink reviewed Jul 10, 2024

View reviewed changes

vers-w added 7 commits July 15, 2024 10:06

Adress review comment: replace waterallocation by allocation

ce9ecd4

Adress review comment: replace divide by bounded_divide

b66a3df

Adress review comment: extend tests Feddes reduction function

6b5f746

Addres review comments: docs

486d4aa

Adress review comment: rename variable maximum_irrigation_depth

67ca647

Unit of this variable is mm/t so `maximum_irrigation_rate` is a better variable name. Also fixed a couple of typos in docs.

Address remaining review comments

6d15db1

Small docs improvements

d08def6

verseve requested review from visr and JoostBuitink July 16, 2024 08:00

visr approved these changes Jul 16, 2024

View reviewed changes

JoostBuitink approved these changes Jul 17, 2024

View reviewed changes

verseve merged commit 4f8b156 into master Jul 17, 2024
10 checks passed

verseve deleted the water-demand branch July 22, 2024 11:49

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Water demand and allocation in `sbm` model concept #402

Water demand and allocation in `sbm` model concept #402

verseve commented Apr 24, 2024 •

edited by visr

Loading

visr Jun 25, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

JoostBuitink Jul 10, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

visr Jun 27, 2024

verseve Jul 15, 2024

JoostBuitink left a comment

JoostBuitink Jul 10, 2024

visr left a comment

JoostBuitink left a comment

		# correct rooting depth for soilthickness
		rootingdepth = @. min(0.99 * soilthickness, rootingdepth)

	`nonpaddy` (other crop) fields.
	`nonpaddy` (other crop) fields.

		@unpack subdomain_order, topo_subdomain, indices_subdomain, upstream_nodes, area =
		network

Water demand and allocation in sbm model concept #402

Water demand and allocation in sbm model concept #402

Conversation

verseve commented Apr 24, 2024 • edited by visr Loading

Issue addressed

Explanation

Checklist

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

JoostBuitink left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

visr left a comment

Choose a reason for hiding this comment

JoostBuitink left a comment

Choose a reason for hiding this comment

Water demand and allocation in `sbm` model concept #402

Water demand and allocation in `sbm` model concept #402

verseve commented Apr 24, 2024 •

edited by visr

Loading