1. Groundwater – SM – climate interactions:
EGU 2016 – CL4.11 – April 21st, 2016
16/09/2018
Groundwater – SM – climate interactions:
Lessons from idealized model experiments with forced water table depth
Agnès Ducharne1, Min-Hui Lo2, Bertrand Decharme3, Rong-You Chien2, Fuxing Wang4, Josefine Ghattas5, Frédérique Cheruy4, Jeanne Colin3 , Sophie Tyteca3, Chia-Wei Lan2
1 METIS/IPSL, Paris, France
2 Department of Atmospheric Science, NTU, Taiwan
3 GAME, CNRM, Toulouse, France
4 LMD/IPSL, Paris, France
5 IPSL, Paris, France

2. GW with long residence times  buffering effects
1. Introduction
2/12
Kollet & Maxwell, 2008
GW with long residence times  buffering effects
on stream flow
on SM and LA coupling where the water table is close enough to the surface
As an introduction, I’d like to remind you that GW tends to flow horizontally in river catchments, with rather long residence times, and this comes with important buffering effects
Objective
Identify where the WT can influence SM, ET, and LA coupling
through idealized model experiments

3. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

4. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

5. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

6. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

7. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

8. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

9. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

10. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Point at the underground !!

11. 3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
2. The numerical experiment
3/12
3 state-of-the-art LSMs: CLM4, ISBA/SURFEX, ORCHIDEE
Off-line or coupled to their parent climate model
following LMIP/AMIP-like protocols for intercomparability
Off-line forcing = PGF (1°, 3-hourly, , Sheffield et al. 2006) + GPCC bias corrrection
Reference simulations with standard configuration
+ 7 simulations with forced water table depth (WTD) between 0.5 and 10 m
CLM
SUR
ORC
Modifier SURFEX : rajouter le drainage libre !!!

12. Land averages – Sensitivity to WTD
3. Off-line results
4/12
Land averages – Sensitivity to WTD
REF
(Rodell et al., 2015)
REF
REF
REF
Total runoff  -Qforce
Dynamic LAI in ORCHIDEE, which increases with SM, and contributes to the slower increase in ET via a complex interplay with soil evaporation and transpiration

13. The critical WTD 3. Off-line results Qle = f(WTD) Variation rate
5/12
The critical WTD
Qle = f(WTD)
Variation rate
Variation rate in % of Qle(REF)
To map the spatial variations of this sensitivity, we introduced a new variable : the critical WTD, which I will illustrate here based on land averages.
All values are land averages

14. The critical WTD 3. Off-line results Qle = f(WTD)
5/12
The critical WTD
Qle = f(WTD)
WTDc = depth at which Qle response becomes small
Variation rate in % of Qle(REF) 5%
WTDc similar to extinction depth of Shah et al 2007
WTDc
= 2.5m
WTDc
= 4m
WTDc
= 6.5m
Deeper WTDc  higher sensitivity to WTD
All values are land averages

15. The critical WTD : 5% threshold Strong inter-model difference
3. Off-line results
6/12
The critical WTD : 5% threshold
Patterns // aridity
Strong inter-model difference
DIRECTLY MOVE TO 1% !!!

16. The critical WTD : 1% threshold Strong inter-model difference
3. Off-line results
7/12
The critical WTD : 1% threshold
Patterns // aridity
Strong inter-model difference
VG vs Brooks Corey ?

17. 3. Off-line results REF: Aridity index P/Rn 1% WTDc CLM ORC SUR 8/12
Arid zones
ORC
Transition zones
SUR
Arid zones

18. 3. Off-line results 1% WTDc
9/12
1% WTDc
The reasons for inter-model differences are not clear yet but may involve:
Dynamic LAI in ORC, combined with different sensitivities of soil evaporation and transpiration
Different models of unsaturated hydraulic parameters (BC for CLM & SUR, VG for ORC) cf. Decharme et al. 2011
Different ways to link the soil and deep WTs
CLM
Arid zones
ORC
Transition zones
We’ve also started to compare with many other possible explanatory factors, and the reasons for the inter-model differences are not clear yet. But we think they may involve…
SUR
Arid zones

19. 3. Off-line results 1% WTDc Simulated WTD (Fan et al., 2013) CLM
10/12
1% WTDc
Simulated WTD (Fan et al., 2013)
CLM
Major aquifer systems (Whymap, 2008)
ORC
We also compared our WTDc maps to the simulated WTD map published at the global scale by Fan et al in 2013, which shows that the WTD is deep in arid zones, which is a well known fact.
The map of the major aquifer systems, just below, tells us that in many part of the world, these deep simulated WTD do not correspond to an aquifer formation so there is no WTD…
Yet, there are also many places where deep WTD exist in unconfined aquifers, and the critical WTD diagnostic suggests they can influence the surface fluxes by capillary rise.
SUR

20. 3. Off-line results 1% WTDc Simulated WTD (Fan et al., 2013) CLM
10/12
1% WTDc
Simulated WTD (Fan et al., 2013)
CLM
Major aquifer systems (Whymap, 2008)
ORC
We also compared our WTDc maps to the simulated WTD map published at the global scale by Fan et al in 2013, which shows that the WTD is deep in arid zones, which is a well known fact.
The map of the major aquifer systems, just below, tells us that in many part of the world, these deep simulated WTD do not correspond to an aquifer formation so there is no WTD… GREY, like in central Asia and parts of Sahara
Yet, there are also many places where deep WTD exist in unconfined aquifers, and the critical WTD diagnostic suggest they can influence the surface fluxes by capillary rise.
SUR

21. We miss simulations with deeper WTD to define the WTDc
4. Coupled simulations
11/12
Short overview
Only 3 1, 2, 3m
Variation rates around 1.5 and 2.5m
CLM and ORC have a larger reference Qle in coupled mode than offline
CLM shows smaller variations rates to shallow WTD in coupled mode
We miss simulations with deeper WTD to define the WTDc
All values are land averages

22. 5. Conclusion and perspectives
12/12
Off-line results
The critical WTD helps comparing the sensitivity of surface fluxes to GW between different regions and models
Models need WTDs down to m to represent the effect of GW on SM and ET in arid and semi-arid zones
Coupled results
« Deeper » analysis is needed
Limits and perspectives
Fixed WTD over the entire grid-cells  highly unrealistic
Same experiments with forced WTD over fractions of grid-cells (coupled mode)
Comparison of the three LSMs with dynamic WTD parametrization

23. Thank you for your attention Thank you for your attention
EGU 2016 – CL4.11 – April 21st, 2016
16/09/2018
Thank you
for your attention
Thank you
for your attention
Whymap 2008
Thank you !
Myself
My talk today is about…

24. Thank you for your attention Complementary slides
EGU 2016 – CL4.11 – April 21st, 2016
16/09/2018
Thank you
for your attention
Complementary slides
Whymap 2008
Thank you !
Myself
My talk today is about…

25. Land averages – Reference simulations
3. Off-line results 4
Land averages – Reference simulations
Evapotranspiration
Total runoff
Point at the underground !!
OBS from Rodell et al. (2015)

26. Land averages – Reference simulations
3. Off-line results
Land averages – Reference simulations
Point at the underground !!
Total ET has two major components: transpiration & soil evaporation:
In CLM and SURFEX, the changes of total ET are driven by soil evaporation.
In ORCHIDEE, the changes of total ET are driven by the transpiration (until WTD1)

27. ESoil
TVeg
ORCHIDEE
WTD1-WTD2
Esoil and TVeg in mm/d
LAI

28. ESoil
TVeg
ORCHIDEE
WTD05-WTD1
Esoil and TVeg in mm/d
LAI

29. SM-ET coupling strength (from Dirmeyer, 2013)
3. Off-line results
SM-ET coupling strength (from Dirmeyer, 2013)
C(REF) Annual, W/m²
C(WTD1) Annual, W/m²
Coupling strength decreases with 1m because it reduces the SM and ET variability
ORC keeps more coupling strength because ET increases less, meaning that it keeps more water stress than the other models

30. C(WTD1) – C(REF) Annual, W/m²
3. Off-line results
C(WTD1) – C(REF) Annual, W/m²
1% WTDc
CLM
ORC
Striking similarity between patterns of WTDc and change in C
SUR

31. 3. Off-line results 9
1% WTDc
CLM
ORC
VG vs Brooks Corey ?
SUR

32. 3. Off-line results 9
1% WTDc
CLM
ORC
VG vs Brooks Corey ?
SUR

33. 3. Off-line results 9
1% WTDc
CLM
ORC
VG vs Brooks Corey ?
SUR

34. In arid zones, sandy soils exhibit a shallower WTDc
3. Off-line results 9
1% WTDc
ORCHIDEE
CLM
ORC
The sensitivity to sandy soil is opposite to what’s reported in Shah et al 2007 for the extinction depth, but with climate from Florida.
In arid zones, sandy soils exhibit a shallower WTDc
SUR