land-atmosphere coupling, climate-change and extreme events +

Land-atmosphere coupling, climate-change and extreme events

+Activities with regard to land flux

estimations at ETH Zurich

Sonia I. Seneviratne

Institute for Atmospheric and Climate Science, ETH Zurich, Switzerland

LandFlux Meeting, Toulouse, FranceMay 29, 2007

Outline

• Land-atmosphere coupling, climate change, and extreme events (Seneviratne et al. 2006) – Land-atmosphere coupling: hot spot in Europe?– Dynamics with climate change– Links with extreme events

• Activities with regard to land flux estimations at ETH Zurich– Atmospheric-terrestrial water balance estimates– Some results on models’ estimates (land surface models, GCMs)– SwissFluxnet activities and Rietholzbach catchment site

Seneviratne et al. 2006,

Nature, 443, 205-209

L-A coupling in Europe


Koster et al., 2004, Science



No strong coupling in Europe? How about Mediterranean region?

NB: Results based on only one year SST conditions (1994)



No strong coupling in Europe? How about Mediterranean region?

NB: Results based on only one year SST conditions (1994)

(Koster et al. 2006, JHM)

T

Projected changes in To variability

Schär et al. 2004, Nature[%]

/

[ºC]

T

T

IPCC AR4 GCMs, JJA (2080-2100)-(1970-1990)

Seneviratne et al. 2006, Nature, suppl. inf.

PRUDENCE, CHRM, JJA (2070-2100)-(1960-1990)

Projected changes in To variability

Schär et al. 2004, Nature[%]

/

[ºC]

T

T

IPCC AR4 GCMs, JJA (2080-2100)-(1970-1990)

Seneviratne et al. 2006, Nature, suppl. inf.

PRUDENCE, CHRM, JJA (2070-2100)-(1960-1990) Large changes in To variability

What are the responsible mechanisms:Large-scale circulation patterns? Land surface processes?

Land-atmosphere coupling experiment

Aim:Investigate the role of land-atmosphere coupling for the predicted enhancement of summer temperature variability in Europe

Approach: Perform regional climate simulations within the same set-up with and without land-atmosphere coupling for present and future climate conditions

Standard deviation of the summer (JJA) 2-m temperatureSCENCTL

CTLUNCOUPLED SCENUNCOUPLED

Summer temperature variability

(Seneviratne et al. 2006, Nature)

Most of the enhancement of summer temperature variability in SCEN disappears in the SCENUNCOUPLED simulation

Standard deviation of the summer (JJA) 2-m temperatureSCENCTL

CTLUNCOUPLED SCENUNCOUPLED

Summer temperature variability


Climate change signal vs. LA couplingCLIMATE-CHANGE SIGNAL: SCEN-CTL

LA COUPLING STRENGTH IN SCEN:SCEN-SCENUNCOUPLED

CONTR. OF EXT. FACTORS TO CC SIGNALSCENUNCOUPLED-CTLUNCOUPLED

CONTR. OF LA COUPLING TO CC SIGNAL(SCEN-SCENUNCOUPLED)-(CTL-CTLUNCOUPLED)Strength of land-atmosphere coupling in

future climate is as large as 2/3 of the climate-change signal !


CLIMATE-CHANGE SIGNAL: SCEN-CTL



CONTR. OF LA COUPLING TO CC SIGNAL(SCEN-SCENUNCOUPLED)-(CTL-CTLUNCOUPLED)


Climate change signal vs. LA coupling

Contribution of land-atmosphere coupling to climate change signal: dominant factor in Central and Eastern Europe!

GLACE results for present climate

GLACE experiment (Koster et al. 2004; 2006): no high land-atmosphere coupling in Europe neither for temperature nor for precipitation

How is the strength of land-atmosphere coupling for present vs. future climate in our simulations?

(Koster et al. 2006, JHM)

T

(Koster et al. 2004, Science)

P

Present vs. future climate

land-atmosphere coupling strength parameter analogous to GLACE

• Locally strong soil moisture-To coupling in present climate (Mediterranean; ≠GLACE)• Shift of region of strong soil moisture-To coupling from the Mediterranean to most of Central and Eastern Europe in future climate

€

T (COUPLED)2 − σ T

2(UNCOUPLED)

σ T (COUPLED)2

percentage of To variance explained by coupling [%]


Comparison with IPCC AR4 GCMs

Indirect measure of coupling between soil moisture & To: Correlation between summer evapotranspiration and temperature (ET,T2M)

Negative correlation: strong soil moisture-temperature coupling (high temperature as result of low/no evapotranspiration)

Positive correlation: low soil moisture-temperature coupling (high temperature leads to high evapotranspiration)

Comparison IPCC AR4 GCMs: (ET,T2M) CTL time period SCEN time period Climate-change signal


RCM

All GCMs

3 “best” GCMs

L-A coupling, Europe: present / future

• Strong soil moisture-temperature coupling for the Mediterranean region in the CTL time period (≠GLACE)

• Shift of region of strong soil moisture-temperature coupling to Central and Eastern Europe in future climate (transitional climate zone)

• Qualitative agreement between RCM experiments and analysis of IPCC AR4 GCMs

Seasonal Cycle of Soil Moisture

Month

Soil

moi

stur

e [m

m]

Mechanism for To variability increase

no limitation wet climate

below threshold (“plant wilting point”) dry climate

CTL (1961-1990)SCEN (2071-2100)

transitional climate

Summary

• The projected enhancement of To variability in Central and Eastern Europe is mostly due to changes in land-atmosphere coupling

• Climate change creates a new hot spot of soil moisture - To coupling in Central and Eastern Europe in the future climate (shift of climate regimes): Dynamic feature of the climate system!

• LandFlux: Consider transient modifications with climate forcing (greenhouse gases, aerosols)

Outline

• Land-atmosphere coupling, climate changes, and extreme events (Seneviratne et al. 2006a) – Land-atmosphere coupling: hot spot in Europe?– Dynamics with climate change– Links with extreme events

• Activities with regard to land flux estimations at ETH Zurich– Atmospheric-terrestrial water balance estimates– Some results on models’ estimates (GSWP/GLDAS-type; GCMs)– SwissFluxnet activities and Rietholzbach catchment site

Atmospheric-Terrestrial Water Balance

• Terrestrial water balance:



• Atmospheric water balance:




• Combined water balance:measuredstreamflow(Rs+Rg)

reanalysisdata(ERA-40)




• Combined water balance:measuredstreamflow(Rs+Rg)

reanalysisdata(ERA-40)

The water-balance estimates depend only on observed or assimilated variables (≠ P,E)

Main limitation: valid only for domains > 105-106 km2 (Rasmusson 1968, Yeh et al. 1998)


Case Study: Mississippi & Illinois

Water-balance Estimates

Observations(soil moisture+groundwater+snow)

Seneviratne et al. 2004, J. Climate, 17 (11), 2039-2057

corr=0.8, r2=0.71

Dataset for Mid-latitude River Basins

• divQ & dW/dt: whole ERA-40 period (1958-2002)

• runoff data: Global Runoff Data Center (GRDC)

Hirschi et al. 2006, J. Hydrometeorology, 7(1), 39-60

Comparisons with soil moisture observations from the Global Soil Moisture Data Bank

Volga River basin (1972-85)

corr=0.8r2=0.64

“BSWB”http://iacweb.ethz.ch/data/water_balance/



• Combined water balance:


Estimation of large-scale ET Atmospheric water balance:

Louie et al. 2002

Mackenzie GEWEX Study (MAGS)Peace

Estimation of large-scale ET

The water-balance estimates depend only on observed P and assimilated variables

Main limitations:

- valid only for domains > 105-106 km2

(Rasmusson 1968, Yeh et al. 1998;Seneviratne et al. 2004, J. Climate,Hirschi et al, 2006, JHM)

- Imbalances, drifts of reanalysis data

Retrospective dataset! (1958-2001, ERA-40; 2001-2007, ECMWF operational forecast analysis; e.g. Hirschi et al. 2006, GRL)

http://iacweb.ethz.ch/data/water_balance/

Outline



Precipitation Forcing for LSMs

Oki et al 1999: a minimum of about 30 precipitation gauges per 106 km2 or about 2 gauges per 2.5o x 2.5o GPCP grid cell are required for accurate streamflow simulations

( Koster et al, 2004: GPCP product, 1979-93)

(Fekete et al. 2004)

Fekete et al. 2004: Range between 4 state-of-the-art precipitation datasets (CRU, GPCC, GPCP, and Willmott-Matsuura)

r2 vs. ground data, yrs within 1979-93 (anomalies)

Illinois

Neva

DonDnepr

Volga

Amur

Lena

Yenisei

Ob

Effects on Catchment LSM Output

Soil moisture + snow

Precipitation

LSM results strongly dependent on quality of forcing...

Water-holding capacity

Modelling: GCMs

(Seneviratne et al. 2006, JHM)

LAND

Soil moisture memory


Modelling: GCMs

Water-holding capacity LAND


Modelling: GCMs

PSignificant range in model behaviour…

(Koster et al. 2004, Science)

Land-atmosphere coupling

Modelling: GCMs

Outline



Observations: FLUXNET

• Worldwide CO2, water and energy flux measurements (integrating several projects such as AMERIFLUX, CARBOEUROPE, …)

• At present, about 200 tower sites

• however, still some serious limitations in temporal availability (in Europe, most measurements available after 1995 only)

• only few sites with soil moisture measurements

http://www-eosdis.ornl.gov/FLUXNET/

Observations: SwissFluxnet

X Rietholzbach catchment site (Lysimeter, isotope measurements)

Will also focus on soil moisture measurements (ETH Zurich)

Outline

• Land-atmosphere coupling, climate changes, and extreme events (Seneviratne et al. 2006a) – Land-atmosphere coupling: hot spot in Europe?– Dynamics with climate change– Links with extreme events


• Conclusions and outlook

Conclusions and outlook

• Land processes important in transitional climate zones (e.g. Koster et al. 2004): seasonal forecasting, extreme eventsNB: possible changes in hot spots’ location with greenhouse warming

• Several methods to estimate water storage or ET, atmospheric-terrestrial water estimates are promising (retrospective datasets)

• No perfect dataset: but synergies are available

Comparison: Land datasets

Ground measurements

Atmospheric water-balance

Satellite data (SMOS, GRACE)

LSM with observed forcing

Resolution Point measurements

300-1000 km(105-106 km2)

SMOS: 40kmGRACE: ~1000km

1km (LIS) - 50km

Main advantage

Ground truth (...)

Retrospective dataset (1958-present); large coverage

Global coverage Good results in regions with good forcing; higher resolution

Main limitation Point-scale measurements; limited temporal and geographical coverage

Dependent on quality of convergence data (radiosonde vs. satellite data, drifts)

Only recent data; short timeseries; products’ limitations (top soil, low res.)

Results depen-dent on quality of forcing data; models optimized for regions with validation data

Outlook

A new GEWEX study area for Europe? (hot spot of coupling)

Temporal Integration (3)

Integration over longer time ranges is not straightforward due to the presence of small systematic imbalances in the monthly estimates

Comparison with imbalances from other water-balance studies

€

∂S∂t

+ ∂W∂t

⎧ ⎨ ⎩

⎫ ⎬ ⎭

Observations (Illinois)Integrated estimates

G97: Gutowski et al. 1997Y98: Yeh et al. 1998BR99: Berbery and Rasmuson 1999

Long-term Imbalances and Drifts (1)

EuropeWestern RussiaAsiaNorth America

Domain size (km2)

Imba

lanc

es (m

m/d

)

Rasmusson (1968)threshold for radiosonde data (2.106 km2)

Hirschi et al. 2004

Illinois (2 .105 km2)

?

Illinois (1987-96)

Soil moisture - precipitation couplingCLIMATE-CHANGE SIGNAL: SCEN-CTL



CONTR. OF LA COUPLING TO CC SIGNAL(SCEN-SCENUNCOUPLED)-(CTL-CTLUNCOUPLED)• appears relevant for

variability enhancement in the Alpine region

• this link needs to be better investigated in future studies!


Modelling

Vegetation - CO2 interactions

Only few models explicitly include vegetation-CO2 relationships…(enhanced water-use efficiency?, CO2 fertilization?)

(Sellers et al. 1997)

Modelling

Vegetation - CO2 interactions

(Sellers et al. 1997)

Only few models explicitly include vegetation-CO2 relationships…(enhanced water-use efficiency?, CO2 fertilization?)

(Ciais et al. 2005, Nature)

NPP, 2003

(Gedney et al. 2006, Nature)

Direct CO2 effect on runoff ?

(Fischer et al. 2006, in preparation)

Soil moisture-temperature coupling in the European summer 2003: Spring soil moisture impacted summer temperature by up to 2 oC!

Soil moisture-temperature feedbacks

Summer 2003 heatwave

(Fischer et al. 2007, J. Climate, submitted)

Summer 2003 heatwave

(Fischer et al. 2007, J. Climate, submitted)

Dry or wet conditions in spring make up to 2oC difference in summer!

Variability increases

1) DMI, HC1, HS12) DMI, HC2, HS23) HC, HadRM3H4) HC, HadAM3H, ens15) HC, HadAM3H, ens26) ETH/CHRM, HC_CTL, HC_A27) GKSS, HC_CTL, HC_A28) MPI, 3003, 30069) SMHI, HC_CTL, HC_A210) UCM, control, a211) ICTP, ref, A212) KNMI, HC1, HA213) CNRM, DA9, DE614) CNRM, DE3, DE715) CNRM, DE4, DE816) DMI, ecctrl, ecsca2‹

(Vidale al. 2006)

∆(P) vs. ∆(To), PRUDENCE models (Central Europe)

Observations: Soil moisture

Global Soil Moisture Data Bank(Robock et al. 2000, Bull. Am. Met. Soc.)

• Current ground observations networks of soil moisture are very limited in space and time (no data for Europe; only few observations in the former Soviet Union after 1990)

Indirect measurements/estimates

Satellite measurements• Microwave remote sensing (e.g. SMOS)• GRACE (Gravity Recovery and Climate

Experiment)• NDVI (Normalized Difference Vegetation Index)

Land surface models with observational input• Global Soil Wetness Project (GSWP) • Global Land Data Assimilation (GLDAS) • Land data assimilation with Ensemble Kalman

Filter(Reichle et al. 2002, JHM)

GRACE twin satellites

Some new approaches

Other applications Estimation of Large-scale Evapotranspiration:

Atmospheric water balance:

Louie et al. 2002

Mackenzie GEWEX Study (MAGS)

Observations: Soil moisture

Global Soil Moisture Data Bank(Robock et al. 2000, Bull. Am. Met. Soc.)

• Current ground observations networks of soil moisture are very limited in space and time (no data for Europe; only few observations in the former Soviet Union after 1990)

land-atmosphere coupling, climate-change and extreme events +

Documents

summer jja

role of land

scienceno strong coupling

land surface models

land surface processes

land flux estimations

climate science

variability schr