Presentation by Hélène Boisgontier (Deltares) at the wflow User Day 2018, during Delft Software Days - Edition 2018. Friday 09 November 2018, Delft.

1. 9 November 2018
Catchment/continental scale model of
sediment dynamics using wflow
Delft Software Days – wflow
Hélène Boisgontier
The research leading to these results has received funding from the European
Union’s Horizon 2020 research and innovation programme under the Marie
Sklodowska -Curie grant agreement No 643052 (CCASCADES project)

2. 9 November 2018
C-CASCADES
• A Marie Sklodowska-Curie Innovative Training Network
• Consortium of 9 academic institutions, 3 industrial partners and 1
international organisation
• Goal: Better understand, quantify and model the role of the Land
Ocean Aquatic Continuum (LOAC) and its dynamics in the global
carbon budget.
• 15 sub-projects done by 15 Early Stage Researchers
2

3. 9 November 2018
Goals of the project
• Goal: Model exports of terrestrial sediment/carbon through the
LOAC at the catchment/continent scale
In-stream processes
River modelSoil loss model
3

4. 9 November 2018
Context/Goals of the project
• Goal: Model exports of terrestrial sediment/carbon through the
LOAC at the catchment/continent scale
• Model and estimate soil loss and sediment yield by land
erosion
• Determine the fate and transport of the incoming sediments
along the river system
• Application across Europe
• Application for water quality issues
Coupling with a distributed hydrology model
 fine space resolution
 large space scale
 fine temporal resolution
4

5. Hydrology model
wflow SBM
9 November 2018 5

6. Why wflow SBM model
• Distributed hydrology model developed
by Deltares
• Open source model
• Global version in development
• Based on available global datasets
• Uncalibrated model
• Three-clicks-to-a-model framework
9 November 2018
Overview of the different processes and fluxes in
wflow_sbm model
6

7. GENERALITIES
Sediment model
9 November 2018 7

8. Framework and inputs
• Wflow module with the same structure as wflow SBM
• Runned after wflow SBM
• Required (dynamic) inputs from wflow SBM:
• Surface Runoff
• Water Level
• Rainfall interception
• Additional input data
• Topsoil clay, silt, organic carbon from Soilgrids
• Optional: canopy height from a global map from Simard, 2011
9 November 2018 Kinnell, 20108

9. Test basin: Rhine basin
• Resolution: 0.6 km2
• Global datasets
• Catchment/river: HydroSheds
• Lakes: GLWD
• DEM: SRTM 30m
• Precipitation from
eartH2Observe
• Land Use: Globcover
• Soil: SoilGrids
9 November 2018 9

10. SOIL LOSS PART
Sediment model
9 November 2018 10

11. Processes to take into account
• Soil detachment
• Splash/Rainfall erosion
• Overland flow erosion
• Wind erosion
• Mass wasting…
• Transport / delivery to the river or catchment outlet
• Deposition
• Valley bottom
• Floodplain
• Depressions
• Field boundaries…
26 October 2018 Kinnell, 201011

12. Erosion models available
• USLE-based family (semi-empirical)
• Most well-known and used erosion models
• USLE-RUSLE: more suited for yearly/monthly timestep
• MUSLE: event-based (~daily) but implied sediment delivery
ratio
𝑆𝑜𝑖𝑙 𝑙𝑜𝑠𝑠 = 𝑅𝑖𝑛𝑡𝑒𝑛𝑠𝑖𝑡𝑦 ∗ 𝐿𝑆 ∗ 𝐾 ∗ 𝐶 ∗ 𝑃
• Physics-based models
• ANSWERS
• EUROSEM
• SWAT (hourly): combination of the two
𝑆𝑜𝑖𝑙 𝑙𝑜𝑠𝑠 = 𝑅𝑎𝑖𝑛𝑓𝑎𝑙𝑙 𝑒𝑟𝑜𝑠𝑖𝑜𝑛 + 𝑂𝑣𝑒𝑟𝑙𝑎𝑛𝑑 𝑓𝑙𝑜𝑤 𝑒𝑟𝑜𝑠𝑖𝑜𝑛
9 November 2018 12

13. Final chosen erosion model: EUROSEM+ANSWERS
9 November 2018 Adapted from De Vente et al., 2008
Soil from Upslope
Detachment
by Rain
Detachment
by Flow
Detachment on Increment
Total Detached Soil Transport Capacity of Flow
Soil Carried Downslope
Compare
If Det < Trans If Det > Trans
EUROSEM ANSWERS
Govers or Yalin
13

14. Results: Rhine basin
9 November 2018
Simulated surface runoff and soil loss for the year 2010:
14

15. Results: Rhine basin
9 November 2018
Comparison with modelled soil loss from Panagos & al. (2010) and
PESERA (2003):
Panagos map
RUSLE (rainfall based)
PESERA map
Physics (surface runoff based)
15

16. Results: Rhine basin
9 November 2018
Comparison per land use type with soil loss plots data and other modelled studies:
16
Forest
[t.ha-1.yr-1]
Cropland
[t.ha-1.yr-1]
Grassland
[t.ha-1.yr-1]
Cerdan (Europe) 0.2 3.6 0.4
Maetens (Europe) 0.7 6.5 0.7
Panagos (Rhine, year 2010) 2.61 2.16 2.53
PESERA (Rhine, year 2003) 0.33 1.62 0.73
Mean modelled (Rhine, years 2010 to 2014) 0.28 1.50 0.45
Mean Soil Loss
per LUSources
Literature
(field data)
Other
simulations
wflow sed

17. RIVER PART
Sediment model
9 November 2018 17

18. Processes to take into account
• Transport (equation or capacity)
• River erosion (total or bed/bank)
• Deposition/settling
• River
• Reservoir / lake
• Floodplain
• Particle differentiation (for water quality modelling)
26 October 2018 18

19. Models available (hydrology coupling)
9 November 2018
• Delft3D-WAQ type:
• Krone-Partheniades
• Transport equation (Engelund and Hansen & advection-
dispersion equation)
• SWAT (default) based on a sediment concentration threshold
• Erosion if C < Cthres
• Else deposition
• SWAT (physics-based)
• Erosion of bed and bank with physics based equation
• Deposition from Einstein
• Separation of suspended and bed loads
• Multiple equations for transport
19
Possibility of no calibration
calibration
calibration

20. Final chosen river model: SWAT
9 November 2018
Sediment from Upstream
Sediment from
Previous Timestep
Sediment from
Land Erosion
Total Sediment Input
River Erosion
Transport Capacity of Flow
Total Sediment Load
Compare
If Input < TransEinstein
Camp (reservoir)
Bagnold
(possible others)
Soil model
Deposition
Knight, 1984
Julian & Torres, 2006
20

21. Issues with the river model
9 November 2018
• wflow SBM and rivers:
• River map with cells containing part of the main streams
• Surface runoff and water level adapted
• River width (empirical equation)
• But not the slope!
21
Need calibration for river transport

22. Results: Rhine at Lobith 2012 (calibration)
9 November 2018
0
1000
2000
3000
4000
5000
6000
7000
8000
03-11-11 23-12-11 11-02-12 01-04-12 21-05-12 10-07-12 29-08-12 18-10-12 07-12-12 26-01-13
Q [m3/s]
Qsim
0
20
40
60
80
100
120
140
03-11-11 23-12-11 11-02-12 01-04-12 21-05-12 10-07-12 29-08-12 18-10-12 07-12-12 26-01-13
SPM [mg/L]
SPM sim
22

23. Results: Rhine at Lobith 2010
9 November 2018
0
1000
2000
3000
4000
5000
6000
7000
03-12-09 22-01-10 13-03-10 02-05-10 21-06-10 10-08-10 29-09-10 18-11-10 07-01-11 26-02-11
Q [m3/s]
Qsim
0
20
40
60
80
100
120
140
160
03-12-09 22-01-10 13-03-10 02-05-10 21-06-10 10-08-10 29-09-10 18-11-10 07-01-11 26-02-11
SPM[mg/L]
SPM sim
23

24. Conclusions
9 November 2018
• Development of a sediment model coupled to wflow:
• Based entirely on global datasets
• No calibration for the soil loss part
• Calibration (minimum) for the river part
24

25. Thank you for your attention
9 November 2018 25

26. 9 November 2018
C-CASCADES project
Goal: Better quantify the role of the Land Ocean Aquatic Continuum
(LOAC) and its dynamics in the global carbon budget.
• Improved understanding of the processes controlling carbon transport
and transformations (from mountain stream to open ocean).
• Quantification of LOAC carbon fluxes and CO2 emissions in hotspot
regions.
• Integration of the LOAC carbon cycle in European Earth System
Models to better estimate the global CO2 budget.
26

27. Model equations: Soil loss
9 November 2018
• Rainfall erosion
• ANSWERS: 𝐷 𝑅 = 0,108 ∗ 𝐾 𝑈𝑆𝐿𝐸 ∗ 𝐶 𝑈𝑆𝐿𝐸 ∗ 𝐴 ∗ 𝑅𝑖
2
• EUROSEM: 𝐷 𝑅 = 𝑘 ∗ 𝐾𝐸 ∗ 𝑒−𝜑ℎ
• Overland flow erosion
• ANSWERS: 𝐷 𝐹 = 0,90 ∗ 𝐾 𝑈𝑆𝐿𝐸 ∗ 𝐶 𝑈𝑆𝐿𝐸∗ 𝐴 ∗ 𝑆 ∗ 𝑄
• Transport capacity
• Govers: 𝑇𝐶 = 𝑐 𝜔 − 𝜔𝑐𝑟
η
• Yalin (differentiation): 𝑇𝐶 =
𝐿
𝑄
(𝜌𝑠 − 𝜌 𝑓)𝐷50 𝑈∗ 𝑃
𝐾𝐸𝑙𝑒𝑎𝑓 = 15,8𝐻 𝑝
0,5
− 5,87
𝐾𝐸 𝑑𝑖𝑟𝑒𝑐𝑡 = 11,87 + 8,73𝑙𝑜𝑔10 𝑅𝑖
27

28. Model equations: River
9 November 2018
• Deposition
• Einstein: 𝑃𝑑𝑒𝑝 = 1 −
1
𝑒 𝑥 ∗ 100 𝑥 =
1,055∗𝐿∗𝜔 𝑠
𝑢∗ℎ
• Erosion
• Erosion rates: 𝐸 𝑅 = 𝑘 𝑑 ∗ (𝜏 𝑒 − 𝜏 𝑐𝑟)
• Knight: 𝜏 𝑒,𝑏𝑒𝑑 = 𝜌𝑔𝑅ℎ 𝑆 ∗ 1 − 𝑆𝐹𝑏𝑎𝑛𝑘 ∗ 1 +
2ℎ
𝑊
log 𝑆𝐹𝑏𝑎𝑛𝑘 = −1,4026 ∗ 𝑙𝑜𝑔
𝑊
ℎ
+ 3 + 2,6692
• Julian & Torres:
𝜏 𝑐𝑟,𝑏𝑎𝑛𝑘 = 0,1 + 0,1779 ∗ 𝑆𝐶 + 0,0028 ∗ 𝑆𝐶2
− 2,34 ∗ 10−5
∗ 𝑆𝐶3
∗ 𝐶𝑐ℎ
𝑘 𝑑 = 0,2 ∗ 𝜏 𝑐𝑟
−0,5
h
WL
u𝜔𝑠
28

29. Model equations: River
9 November 2018
• Transport
• Simplified Bagnold: TC = 𝑐 𝑠𝑝
𝑝𝑟𝑓∗𝑄
ℎ∗𝑊
𝑠𝑝 𝑒𝑥𝑝
h
WL
u𝜔𝑠
29