Transcript www.oeb.harvard.edu

Andes-Amazon Project:
Hydrology Model-Data Intercomparison
Brad Christoffersen
Nov. 08, 2010
Moore Foundation
Key Questions

Water budget partitioning




The case of CLM & modifying hydrology structure
Was soil physics standardization across models
effective?
Do models capture observed ET, runoff partitioning
(Manaus site)?
Soil moisture dynamics across models


Are remaining model differences due to physics, or
biology?
Are vertical gradients & seasonal variability
corroborated by observations?
Hydrology in CLM3.5
When water table is below the
soil column
One additional water storage variable: Wa
dW a
= qrech arg e− q drai
dt
q rechaarg e =- k a
y zÑ − y 10
z Ñ − z 10
qdrai= 1− f imp qdrai,max exp − 2.5zÑ
z Ñ = z h ,10 25−
Wa
0 .2
When Wa = Wa,max = 5000 mm,
water table depth = depth of the bottom
soil layer
water table depth
Oleson et al., 2008, JGR
Modified CLM3.5 per Site Simulation Protocol
–> Aquifer storage held const = 0
water table depth
Oleson et al., 2008, JGR
Modified CLM3.5 per Site Simulation Protocol
–> Aquifer storage held const = 0
--> Water table depth const > 10 m
table depth
Zwt =water
constant
Oleson et al., 2008, JGR
Modified CLM3.5 per Site Simulation Protocol
–> Aquifer storage held const = 0
--> Water table depth const > 10 m
--> Drainage out bottom layer =
hydraulic conductivity of bottom layer
Free drainage (= kbot)
table depth
Zwt =water
constant
Oleson et al., 2008, JGR
Modified CLM3.5 per Site Simulation Protocol
–> Aquifer storage held const = 0
--> Water table depth const > 10 m
--> Drainage out bottom layer =
hydraulic conductivity of bottom layer
--> Soil depth set from 3.5 m to site
soil depth
Free drainage (= kbot)
table depth
Zwt =water
constant
Oleson et al., 2008, JGR
Observed ET
JF M A M J JA S ON D
Site Precip
JF M A M J JA S ON D
All y-axis units are mm/month
Differences in ET seasonality
Model-model Intercomparison
Data-Model Intercomparison
Modifying hydrology: The case of CLM
CLM3
JF M A M J JA S ON D
NOAH
JF M A M J JA S ON D
CLM3GW
JF M A M J JA S ON D
IBIS
JF M A M J JA S ON D
CLM3.5
JF M A M J JA S ON D
Fluxes:
Surface runoff
Subsurface runoff
Soil evap
Interception evap
Transpiration
∑Fluxes + ∆Soil
Moisture
Site Precip
Observed ET
JF M A M J JA S ON D
Subsurface Runoff
Surface Runoff
JF M A M J JA S ON D
400
300
200
All y-axis units are mm/month
Differences in ET seasonality
Model-model Intercomparison
Data-Model Intercomparison
Modifying hydrology: The case of CLM
CLM3
CLM3GW
CLM3.5
CLM3.5
Aquifer
3.5 m soil depth
JF M A M J JA S ON D
NOAH
JF M A M J JA S ON D
IBIS
100
0
F M A M J JA S ON
CorrectJ ET
seasonality
(wrong reason?)
D
JF M A M J JA S ON D
JF M A M J JA S ON D
Fluxes:
Surface runoff
Subsurface runoff
Soil evap
Interception evap
Transpiration
∑Fluxes + ∆Soil
Moisture
Site Precip
Observed ET
JF M A M J JA S ON D
Subsurface Runoff
Surface Runoff
JF M A M J JA S ON D
400
300
200
All y-axis units are mm/month
Differences in ET seasonality
Model-model Intercomparison
Data-Model Intercomparison
Modifying hydrology: The case of CLM
CLM3
CLM3GW
CLM3.5
Aquifer
3.5 m soil depth
CLM3.5
Free drainage
3.5 m soil depth
JF M A M J JA S ON D
NOAH
JF M A M J JA S ON D
IBIS
100
0
F M A M J JA S ON
CorrectJ ET
seasonality
(wrong reason?)
CLM3.5
D
JF M A M J JA S ON D
Fluxes:
Surface runoff
Subsurface runoff
Soil evap
Interception evap
Transpiration
∑Fluxes + ∆Soil
Moisture
Incorrect ET seasonality
(what's missing?)
JF M A M J JA S ON D
Site Precip
Observed ET
JF M A M J JA S ON D
Subsurface Runoff
Surface Runoff
JF M A M J JA S ON D
400
300
200
All y-axis units are mm/month
Differences in ET seasonality
Model-model Intercomparison
Data-Model Intercomparison
Modifying hydrology: The case of CLM
CLM3
CLM3GW
CLM3.5
Aquifer
3.5 m soil depth
CLM3.5
Free drainage
3.5 m soil depth
JF M A M J JA S ON D
NOAH
JF M A M J JA S ON D
IBIS
100
0
F M A M J JA S ON
CorrectJ ET
seasonality
(wrong reason?)
CLM3.5
D
CLM3.5
Free drainage
8m soil depth
JF M A M J JA S ON D
Fluxes:
Surface runoff
Subsurface runoff
Soil evap
Interception evap
Transpiration
∑Fluxes + ∆Soil
Moisture
Incorrect ET seasonality
(what's missing?)
JF M A M J JA S ON D
Correct ET seasonality
(right reason?)
Was soil physics standardization
effective?
Subsurface Runoff
Surface Runoff
Manaus K34 - Yes!
JULES
ED2
IBIS
CLM

Similar pattern & magnitude of water budgets

JULES, IBIS, CLM potentially overestimate ET
JULES
ED2
IBIS
Evergreen Tapajos K67
CLM
JULES
ED2
IBIS
CLM
Evergreen Tapajos K67
Transitional/Semideciduous Reserva Jaru
JULES
ED2
IBIS
CLM
Evergreen Tapajos K67
Transitional/Semideciduous Reserva Jaru
Savanna Pe de Gigante
Are predicted budgets consistent
with observations? (Manaus K34)
JULES
ED2
OBS Data from Tomasella & Hodnett 2008
IBIS
CLM
OBS
Interlude: Model structure I

Infiltration/Surface Runoff:




JULES: Infiltration excess + PFT-dependent
infiltration “enhancement factor”
ED2: Infiltration excess + Surface water storage (8hour lifetime)
IBIS: Green-Ampt wetting front
CLM: Infiltration excess + Maximum ponding depth
(10 mm H2O)
Interlude: Model structure I

Infiltration/Surface Runoff:




JULES: Infiltration excess + PFT-dependent
infiltration “enhancement factor”
ED2: Infiltration excess + Surface water storage (8hour lifetime)
IBIS: Green-Ampt wetting front
CLM: Infiltration excess + Maximum ponding depth
(10 mm H2O)
Green-Ampt
psi = H
water
soil
wetting front
theta.i
Standard Darcy
theta.s
z=0
z=f
ponding depth
theta.1, theta.1.sat
1
2
3
Interlude: Model structure II

Rooting dynamics and water extraction:



JULES, CLM (& IBIS?): proportional to water
availability & root fraction in each soil layer
ED2: proportional to water availability and
maximum rooting depth (by veg. cohort)
Consider special case: Homogeneous soil water
profile, (& for ED); mid- to late-successional forest:
Interlude: Model structure II

Rooting dynamics and water extraction:



JULES, CLM (& IBIS?): proportional to water
availability & root fraction in each soil layer
ED2: proportional to water availability and
maximum rooting depth (by veg. cohort)
Consider special case: Homogeneous soil water
profile, (& for ED); mid- to late-successional forest:
Fraction of transpiration
0
1
Fraction of transpiration
0
1
Depth (m)
JULES,
CLM (& IBIS?)
ED2
Soil Moisture Dynamics: Are
remaining differences due to physics
or biology?
Manaus K34
INFILTRATION
Darcy's Law
Darcy's Law
JULES
0.33
0.39
Green-Ampt
ED2
0.45
Free drainage
BOTTOM BOUNDARY
0.33
0.39
Darcy's Law
IBIS
0.45
Free drainage
Other mechanisms?
0.33
0.39
CLM
0.45
Free drainage
0.33
0.39
0.45
Free drainage
JULES
ED2
IBIS
CLM
K67
0.35
0.45
0.35
0.45
0.35
0.45
0.35
0.45
0.30
0.10
0.30
0.10
0.30
0.10
0.30
0.30
0.05
0.30
0.05
0.30
0.05
0.30
RJA
0.10
PDG
0.05
What happens in CLM when we implement an
ED-like root water uptake scheme?
Free drainage
soil depth 8 m
Root distribution-dependent water stress
Free drainage
soil depth 8 m
Root distribution-independent water stress
Summary & Conclusions



Overall water balance: Encouraging.
(Standardization of surface runoff necessary?)
ET Seasonality: All models predict peak of
transpiration in dry season; some (ED) predict
constant ET year-round.
Soil moisture dynamics: Interesting, important
differences among models.



IBIS has much higher surface water content
ED develops wet season pool near bottom
boundary
Discriminating among model mechanisms: