The role of land surface processes
for the water cycle
How sensitive are weather and climate to land conditions?
Ryan Teuling, Thierry Corti, Martin Hirschi, Eric Jäger, Andreas
Roesch & Sonia Seneviratne
Institute for Atmospheric and Climate Science
ETH Zurich, Switzerland
HERC-iLEAPS seminar
24 April 2008, Helsinki
Water cycle
1 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
Land-climate interactions group
2 / 37
Research themes
I
Role of land feedbacks and
processes in the global and
regional climate systems
I
Impacts of land-climate
interactions for extreme events
(e.g., 2003 heat wave), climate
variability and climate change
I
Evaluation of land-atmosphere
coupling characteristics in
models and observations
I
Water cycle
Group picture, Nov 2007
Development of new
observations-based datasets
Land-climate interactions group
3 / 37
Research tools
I
I
I
I
Suite of numerical models
Diagnostic dataset of basin-scale terrestrial water storage changes
Rietholzbach catchment (30+ years of observations), now being
upgraded to FLUXNET site
Soil moisture monitoring network across Switzerland (starting June
2008)
Rietholzbach lysimeter, October 2005
Water cycle
Land-climate interactions group
4 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
The land surface energy budget
5 / 37
Coupling between water- and energy balance
(1 − α)Rs + Rl,net − λE − H = c∆Tsoil
P − E − Q = ∆S
1g water
vapour
2450 J
1g liquid
water
(20oC)
Water cycle
1g liquid
water
(21oC)
4.2 J
1g liquid
water
(20oC)
The land surface energy budget
1g air
(21oC)
1J
1g air
(20oC)
6 / 37
Impacts of land cover change vs. land variability
Climate conditions
Are these two different research fields? . . .
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Impacts of land cover change vs. land variability
Climate conditions
Are these two different research fields? . . .
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Impacts of land cover change vs. land variability
Climate conditions
Are these two different research fields? . . .
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Impacts of land cover change vs. land variability
Climate conditions
Are these two different research fields? . . .
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Impacts of land cover change vs. land variability
Are these two different research fields? . . .
Climate conditions
?
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Impacts of land cover change vs. land variability
Are these two different research fields? . . .
Climate conditions
?
Land conditions
. . . maybe not!
Water cycle
The land surface energy budget
7 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
Land impacts on climate
8 / 37
Land use changes (anthropogenic)
I
Past land cover changes largest in North-America and Eurasia
I
Large future land use changes projected in the tropics
I
Associated changes in albedo and evapotranspiration depend on
region
Davin et al., GRL, 2007
Water cycle
Land impacts on climate
9 / 37
Radiative forcing and temperature changes
I
Strong regional dependence of temperature response
Davin et al., GRL, 2007
Water cycle
Land impacts on climate
10 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Model experiments: current approach
I
Isolation of impact of land-atmosphere coupling from other effects
(circulation, SST)
I
Twin experiments with GCMs or RCMs
I
Soil moisture is assumed to be the main driver of land-atmosphere
coupling through its impact on latent heat flux
I
One simulation where soil moisture freely interacts with the
atmosphere, while in the other soil moisture is replaced with the grid
point climatology for every timestep
I
Difference in atmospheric variability between experiments indicates
strength of land-atmosphere coupling
⇒ Model coupling strength cannot be validated directly
Water cycle
Land impacts on climate
11 / 37
Soil moisture
Model experiments: principle
Time
Water cycle
Land impacts on climate
12 / 37
Soil moisture
Model experiments: principle
Time
Water cycle
Land impacts on climate
12 / 37
Autocorrelation of precipitation
I
I
Variance and autocorrelation of precipitation likely influenced by soil
moisture
Simulations with soil moisture feedback closer to observations
Koster et al., GRL, 2003
Water cycle
Land impacts on climate
13 / 37
Land-atmosphere coupling: GLACE
Koster et al., Science, 2004
Water cycle
Land impacts on climate
14 / 37
Land-atmosphere coupling: GLACE
Koster et al., Science, 2004
Seneviratne et al., Nature, 2006
Water cycle
Land impacts on climate
14 / 37
Land-atmosphere coupling: European heat waves
Number of hot days can
only be reproduced when
soil moisture impact is
accounted for
Fischer et al., GRL, 2007
Water cycle
Land impacts on climate
15 / 37
Land-atmosphere coupling: temperature variability
I
Changes in mean and
variability
I
Over Europe, impact of
coupling increases
Schär et al., Nature, 2004
Seneviratne et al., Nature, 2006
Water cycle
Land impacts on climate
16 / 37
Changes of land-atmosphere coupling
Correlation between temperature and evapotranspiration in RCM and 3
IPCC GCMs
Seneviratne et al., Nature, 2006
Water cycle
Land impacts on climate
17 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
Water & energy balance variability
20 / 37
Soil moisture
Importance of anomalies in coupling
Time
Water cycle
Water & energy balance variability
21 / 37
Soil moisture
Importance of anomalies in coupling
Time
Water cycle
Water & energy balance variability
21 / 37
Soil moisture
Importance of anomalies in coupling
Time
Water cycle
Water & energy balance variability
21 / 37
Albedo: 2003 MODIS anomalies
I
Albedo depends on moisture availability
+100
0
+0.02
A
−0.02
+0.02
0
−0.02
B
0
−100
−0.02
C
0
+0.02
+7
0
D
−7
Teuling and Seneviratne, GRL, 2008
Water cycle
Water & energy balance variability
22 / 37
Albedo: anomalies vs. land cover (% cropland)
Strong relation between albedo anomalies and cropland occurence
Vegetation dominates soil (moisture) effects
⇒ Croplands are albedo anomaly “hot-spots”
I
I
A
−7
0
+7
B
0
50
100
Teuling and Seneviratne, GRL, 2008
Water cycle
Water & energy balance variability
23 / 37
Evaporation: decay timescales
ET related to available storage:
(A) FIFE (Konza Prairie)
ET [mm d−1]
ET (t) ∝ S(t)
Decay in absence of rainfall:
−
t−t0
λ
4
3
2
1.5
2
1.5
1
1
0.5
0.5
8
5
1991
3
2
1
1990
1995
190
ET [mm d−1]
2
1.5
275
285
(D) Rheindahlen
0.5
e-folding time λ characterizes
system response
(C) Little Washita
4
3
Mixed
Grass
1
265
ET [mm d−1]
ET (t) = ET0 e
(B) Ghanzi
4
3
75
85
(E) Ione
0.5
160
95
1
0.5
Tonzi
Vaira
0.1
130
150
170
(H) Roccarespampani
6
5
4
3
2
1.5
1
0.5
1998
1999
240 250 260 270
(I) GLOWA Volta
3
2.5
2
1.5
2
1.5
0.2
220
6
4
3
2
5
3
2
1
0.5
210
230
250
(G) Audubon
1997
1998
180
200
(F) Bondville
Ejura
Tamale
1
1
0.1
ET [mm d−1]
KP
BV
IO
AU LW
RD
RC
SB
TA
EJ
110
130 150
(K) Twizel
170
3
2
1.5
1
170
190
210
(L) SEBEX Savanna
3
2
2
1.5
1.5
1
1
0.2
355
GZ
320 340 360 380
(M) SEBEX Tiger−bush
4
5
4
3
0.5
0.5
365
375
280
300
320
DOY
340
280
300
320
340
TW
Teuling et al., GRL, 2006
Water cycle
Water & energy balance variability
24 / 37
I
e-folding time of ET between 12 (grass/crops) and ∼35 days (forest)
I
ET can significantly reduce during long heat waves
I
Rapid switch between “wet” and “dry” states unlikely
I
Timescale land surface property; forcing of less importance
I
Upper limit to climate predictability due to soil moisture?
(D) Rheindahlen
8
1991
ET [mm d−1]
5
3
2
1
0.5
190
Water cycle
1990
1995
210
230
250
Teuling et al., GRL, 2006
Water & energy balance variability
25 / 37
Evaporation: atmospheric water balance
I
Evaporation among the most uncertain hydrological fluxes
I
An alternative is provided by considering the large-scale atmospheric
water balance:
(
I
I
Water cycle
n o
P
n
∂W
∂t
)
n
o
n
~ − P −E
= − ∇Q
o
from observations
o
n
o
~ from
∂W /∂t and ∇Q
atmospheric re-analysis (e.g.,
ERA40)
Water & energy balance variability
26 / 37
Evaporation: atmospheric water balance
I
Weekly average ET
I
Good correspondence with
independent FLUXNET
observations (blue lines)
I
Weekly anomalies as high
as 40 W m−2 (or ∼50%)
2003 ET (W m−2)
150
100
50
0
0
50
Water cycle
150
200
250
300
350
2003 ET anomaly (W m−2)
50
See also
Seneviratne et al., JHM, 2004
Hirschi et al., GRL, 2006
Hirschi et al., JHM, 2006
www.iac.ethz.ch/data/water_balance
100
0
−50
160
Water & energy balance variability
180
200
220
240
27 / 37
Evaporation: local land cover impacts
I
Inverse relation
between surface
temperature and
evaporation
I
Increase in
temperature
difference between
forest and cropland
I
Consistent with
slower response of
forests to moisture
deficits
Zaitchik et al., IJC, 2006
Water cycle
Water & energy balance variability
28 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
Model validation
29 / 37
Evaporation: decay times in GSWP2 models
I
I
Models disagree at decay times, especially in Sahel (B)
Variability between models at single site larger than variability
between all observations (!)
(A)
(B)
SB−s
5
10
20
λ [d]
50
SB−t
100
200
Teuling et al., GRL, 2006
Water cycle
Model validation
30 / 37
Evaporation: storage capacity of climate models
I
I
Soil moisture storage capacity in current GCMs is very different
Potential impact on simulated climate variability
Seneviratne et al., JHM, 2006
Water cycle
Model validation
31 / 37
Soil moisture dynamics
I
Two distinct modes in summer soil moisture (Illinois, USA)
Modes not reproduced by ERA-40 reanalysis product
θhθw
θc
θf
θs
(A)
5
15
θw
15
12
9
2
6
1
3
0
0
(B)
10
May
August
θs
15
12
9
9
6
6
3
3
0
0
0.1
0.2
0.3
0.4
0
0.5
θ [−]
8
p(θ) [−]
θc=θf
ERA40, L=0.28 m
ERA40, L=1 m
Sim., L=0.5 m
12
p(θ) [−]
3
χ(θ) [mm d−1]
p(θ) [−]
4
χ(θ) [mm d−1]
I
6
Teuling et al., GRL, 2005
4
2
0
0
0.1
0.2
0.3
0.4
0.5
θ [−]
Water cycle
Model validation
32 / 37
Water storage changes in 2003
I
I
Validation of storage variations in RCM during 2003 heat wave
Early spring drying captured by both BSWB and RCM
BSWB 2003
BSWB 1973−2000
BSWB +/− 1 stdev
−2
−1
0
1
2
CTL 2003
CLIM 1970−2000
CLIM +/− 1 stdev
−3
Terrestrial water storage variations [mm/d]
3
French basin
Jan
Feb
Mar
Apr
May
Jun
Jul
Aug
Sep
Oct
Nov
Dec
month
Jäger et al., 2008, in prep.
www.iac.ethz.ch/data/water_balance
Water cycle
Model validation
33 / 37
Soil moisture and evaporation in 2003
I
2003 annual cycle
3
2003 annual cycle
2003 annual cycle
Vielsalm
CLM044
6
Amplero
CLM044
SanRossore
CLM044
6
6
6
1.0
8
2.0
I
Validation of evaporation and soil moisture in RCM simulations
during 2003 heat wave
Evaporation more consistent than soil moisture
Representative soil moisture measurements still lacking
1.5
I
6
5
5
2
6
5
0.5
1.0
7
8
4
4
7 5
7
4
4
11
1
9
−0.5
0
12
−1
11
0
2
12
1
2
0.5
9
3
1
−0.5
11
10
1
11
−2
−1
0
SMI
2
11
1
12
2
SMI
10
9
1
12
1
2
−2
−1
0
1
12
−1.0
2
12
4
2
10
11
8
7
7
8
3
−1.0
−1
3
9
10
10
−2
8
8
3
−3
10
7
4
3
latent heat flux
5
0.0
5
9
0.0
latent heat flux
1
latent heat flux
3
9
1
SMI
Jäger et al., 2008, in prep.
Water cycle
Model validation
34 / 37
Outline
Land-climate interactions group
The land surface energy budget
Land impacts on climate
Water & energy balance variability
Model validation
Summary
Water cycle
Summary
35 / 37
Summary
I
Extreme events (e.g. 2003 heat wave) offer unique opportunity to
study strength of land-atmosphere coupling and its impacts
I
The projected enhancement of temperature variability in Central and
Eastern Europe is mostly due to changes in land-atmosphere coupling
I
Evaporation rather than albedo is the main process in
land-atmosphere coupling
I
Some significant uncertainties remain in the assessment of
land-atmosphere interactions and their representation in climate
models
Water cycle
Summary
36 / 37
Summary
I
Extreme events (e.g. 2003 heat wave) offer unique opportunity to
study strength of land-atmosphere coupling and its impacts
I
The projected enhancement of temperature variability in Central and
Eastern Europe is mostly due to changes in land-atmosphere coupling
I
Evaporation rather than albedo is the main process in
land-atmosphere coupling
I
Some significant uncertainties remain in the assessment of
land-atmosphere interactions and their representation in climate
models
Water cycle
Summary
36 / 37
Summary
I
Extreme events (e.g. 2003 heat wave) offer unique opportunity to
study strength of land-atmosphere coupling and its impacts
I
The projected enhancement of temperature variability in Central and
Eastern Europe is mostly due to changes in land-atmosphere coupling
I
Evaporation rather than albedo is the main process in
land-atmosphere coupling
I
Some significant uncertainties remain in the assessment of
land-atmosphere interactions and their representation in climate
models
Water cycle
Summary
36 / 37
Summary
I
Extreme events (e.g. 2003 heat wave) offer unique opportunity to
study strength of land-atmosphere coupling and its impacts
I
The projected enhancement of temperature variability in Central and
Eastern Europe is mostly due to changes in land-atmosphere coupling
I
Evaporation rather than albedo is the main process in
land-atmosphere coupling
I
Some significant uncertainties remain in the assessment of
land-atmosphere interactions and their representation in climate
models
Water cycle
Summary
36 / 37
Thank you for your attention
View from the office, Zürich, 21 January 2008
[email protected]
Water cycle
Summary
37 / 37