Utrecht, 16/02/2012
Quantifying the AMOC feedbacks during a 2xCO2 stabilization
experiment with land-ice melting
Swingedouw Didier
Utrecht, 16/02/2012
AMOC spread in projections
• Large spread of AMOC response to GHG emissions
• Gregory et al. 2005: heat flux and freshwater flux both play a role
• Need for understanding processes
• Even for a given forcing: large spread
Stouffer et al. 2006
Schneider et al., 2007
Utrecht, 16/02/2012
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advection
Utrecht, 16/02/2012
Actuel
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advection
Utrecht, 16/02/2012
ActuelFutur
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advection
Utrecht, 16/02/2012
ActuelFuturFutur
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advection
Utrecht, 16/02/2012
ActuelFuturFuturFutur
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advection
Utrecht, 16/02/2012
ActuelFuturFuturFuturFutur
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advectin
Utrecht, 16/02/2012
ActuelFuturFuturFuturFuturFutur
Opposing effects for the water coming from the Arctic and the tropics
AMOC and salinity
forcing
Swingedouw et al. 2007a
Evaporation
Precipitations
Runoff
Humidity transport
Salinity advectin
Utrecht, 16/02/2012
Effect –Effect +
Oceanic meridional advection
Temperature density flux
Salinity density flux
AMOCi FCMs
AMOC internal feedbacks
Stocker et al., 2001
Utrecht, 16/02/2012
Effect –Effect +
Oceanic meridional heat transport
Oceanic meridional advection
Meridional atmospheric Temperature gradient
Sea ice amount
Brine rejection
Ekman divergence
Freshwater meridional transport
Sea ice transport and melting
Temperaturedensity flux
Salinity density flux
Salinity density flux
THCe TCMOs
Stocker et al., 2001
AMOCi
AMOC internal feedbacks
Utrecht, 16/02/2012
Questions
• How to quantify the mechanisms explaining the response of the AMOC to GHG increase?
• Can an additional freshwater input lead to substantial AMOC weakening in projections?
• What are the roles of feedbacks and forcing for the response of the AMOC?
Utrecht, 16/02/2012
Experimental design
Snow
Land Ocean
Ice
IPSL-CM4 model:• OPA-ORCA2 (0.5°-2°)• LMDz (2.5° x 3.75°)
2xCO2 scenario lasting for 500 years:
• With ice sheet melting• No ice sheet melting
Net heat flux
Temps
(années)
CO2
(ppm)
0 70 500
280
560
CTL
No
With
Time (years))
(Swingedouw et al. 2007b)
Utrecht, 16/02/2012
AMOC and convection sites
Climatology: Boyer Montegut et al., 2005
IPSL-CM4: CTL
• In CTL simulation: AMOC max. of only 11 Sv (obs. around 18 Sv )
• No convection in the Labrador Sea
• Overflow of 5.6 Sv (obs. around 6 Sv)
JFM mixed layer depth
Utrecht, 16/02/2012
• Greenland ice sheet (GrIS) melting amounts to 0.13 Sv after 200 years and is then constant
• Equivalent melting of GrIS by more than 50% after 500 years=very agressive melting scenario
With
CTL
NoAMOC index
Temps
(années)
CO2
(ppm)
0 70 500
280
560
CTL
No
With
Time (years))
Temps
(années)
CTL
No
With
Global temperatureModel responses
Utrecht, 16/02/2012
AMOC and density in the convection sites
Correlation of 0.98 between density in the black box and the AMOC:
t=0
No-CTL
With-CTL
Name of the meeting, 20/06/2011
Influence of haline and thermal response on the AMOC
With melting: Changes in temperature (T) and salinity (S) weakens the density (and AMOC)
TS
T
S
Name of the meeting, 20/06/2011
T
S
T
S
Influence of haline and thermal response on the AMOC
TS
With melting: Changes in temperature (T) and salinity (S) weakens the density (and AMOC)
No melting: Salinity changes explain the recovery
Utrecht, 16/02/2012
Density budget in the convection box
dtdwSSddVSSdvzll
expansionDilutionDiffusionTransportEvolution
)().(.
Transport
Surface
l
Utrecht, 16/02/2012
-2,5
-2
-1,5
-1
-0,5
0
0,5
1
1,5
2
2,5
1 2 3 4 5 6 7
NIS2-CTRL
WIS2-CTRL
Density buget in the convection box after 500 years
AMOC increases
AMOC decreases
Transport Surface Residual
STTSTSResiduResiduSurfaceSurfaceTransportTransport
Budget
No-CTLWith-CTL
Utrecht, 16/02/2012
-2,5
-2
-1,5
-1
-0,5
0
0,5
1
1,5
2
2,5
1 2 3 4 5 6 7
NIS2-CTRL
WIS2-CTRL
Transport Surface Résidu
STTSTSResiduResiduSurfaceSurfaceTransportTransport
Bilan
-4
-3
-2
-1
0
1
2
3
1 2 3 4
SoverTransport
SgyreTransport
ToverTransport
TgyreTransport
Sans-CTRLAvec-CTRL
AMOC increases
AMOC decreases
Density buget in the convection box after 500 years
Utrecht, 16/02/2012
Synthesis of important factor explaining density budget in No ice sheet melting
• Over 500 years AMOC reduction mainly caused by:
o Warming in the convection sites (26 %)
o Salinity transport by the overturning (65 %)
• AMOC recovery mainly caused by:
o Transport of salinity anomalies by the gyre (40 %)
o Sea ice melting reduction in the convection sites (28 %)
Utrecht, 16/02/2012
Sea ice melting anomaliesCTL:
• Sea ice transport through Fram Strait
• Brine rejection in winter when sea ice forms
-0,1
-0,05
0
0,05
0,1
0,15
0,2
Local Transport Total
CTRLNIS2
+ =
No melting projections (after 500 years):
• Sea ice transport decreases
• Brine rejection decreases
Sv
Sea ice transport in CTL
CTLNo
Utrecht, 16/02/2012
Salinity anomalies in the Atlantic
SSS anomalies: No melting - CTL
Utrecht, 16/02/2012
Feedback amplification
Linear feedback model (Hansen et al., 1984, for climate sensitivity):
• G0: Static gain
• λi: Feedback factors
0G
1
i
n
+-
Utrecht, 16/02/2012
Feedback quantification methology
now stands for the difference between the projections
We isolate GrIS melting effect:
STTSTSResiduResiduSurfaceSurfaceTransportTransport
with
Utrecht, 16/02/2012
-0,4
-0,2
0
0,2
0,4
0,6
0,8
1
1 2
Quantification of feedback factors
5.2)(1
1
TS
G
Dynamical gain:
S T
Climatic system transfer
TemperatureDensity flux
Salinity Density flux
AMOC_in AMOC_out
)( S
)( T
+-+
Utrecht, 16/02/2012
-5
-4
-3
-2
-1
0
1
2
3
4
1 2 Transport Surface Résidu
Salinity Temperature
Transport Surface Résidu
Heat flux feedback strongly damps heat transport feedback
Ocean transport
Temperature density flux
Salinity density flux
AMOCin AMOCout+
-
Local atmospheric damping
Quantification of feedback factors
Utrecht, 16/02/2012
Quantifying the AMOC feedbacks among different AOGCMs
• For a given freshwater input, large spread among AOGCMs (Stouffer et al. 2006)
• Methodology of feedbacks quantification could be useful (Swingedouw et al. 2007)
• Application to the models from this project framework?
Climatic system transfer
TemperatureDensity flux
Salinity Density flux
AMOC_in AMOC_out
)( S
)( T
+ -+
Stouffer et al. 2006
Utrecht, 16/02/2012
THOR water hosing
• 6 models including one OGCM
• 1965-2005 with 0.1 Sv
Utrecht, 16/02/2012
THOR project: Swingedouw et al., in prep.
An hypothesis to explain AMOC spread
Rypina et al. 2011
Utrecht, 16/02/2012
Outlooks
• Quantifying AMOC feedbacks in the different EMBRACE models
• Evaluating impact of gyre asymmetry on gyre feedbacks factor
Limits/challenges• Computation of density budget in a model• Assumption of AMOC related to density in convection zone holds in other models
Utrecht, 16/02/2012
Predefined sections across the North Atlantic
26N RAPID
OVIDE
A25AR7W
42N
interface between the subtropical and subpolar gyres
overflows : connection between the subpolar gyre and the Nordic Seas
connection with the Arctic ocean
reference salinity : 34.8
Utrecht, 16/02/2012
FCVAR from Julie Deshayes:Diagnostic package of 3D metrics of the circulation
model configurations and simulations
GFDL CM3CNRM CM5IPSL CM5CCSM4+ HADGEM2 ? + HADGEM3 ?pre-industrial control runsmonthly output
+ ocean-only simulationsORCA2-OON2ORCA1-OCEP09ORCA025.L75-G85GLORYS2V1
load grid pointsscale factors
identify sections and areas as sequences of grid points
extract data (t,z,l)along sections and areas
load datau, v, θ, S(t,z,y,x)
predefined sections and areas defined by (lat,lon) of end points+ additional sections
calculate indices (t) for sections, ie components of the northward transport of mass, heat and salt:•net (with and without net mass flux)•overturning in vertical coordinates•overturning in density coordinates•barotropic•baroclinic (net-overturning-barotropic)•0 to1000m deep•1000 to 2000m deep•2000m deep to bottom•related to thermal wind (only from ρ)
calculate indices (t) for areas, regarding heat and freshwater:•volume changes•advective fluxes at ocean boundaries•eddy fluxes at ocean boundaries•diffusion, ice + atmospheric fluxes
Matlab package available
to the community
Utrecht, 16/02/2012
Convection sites density as a driver of AMOC?
• Strong assumption from the proposed model
• Gregory and Tailleux (2010): buoyancy fluxes over the convection sites as a production of Available potential energy then used for Kinetic energy production
Utrecht, 16/02/2012
Linear feedback factor and hysteresis
• Stommel et al. (1961): non linearity for the response of the AMOC to freshwater input
• This would appears in the feedback factor of the overturning terms for salinity and heat transport (dependent in the mean state)
Utrecht, 16/02/2012
Transport Chaleur Méridien Océanique
Advection Méridienne Océanique
Gradient TempératureMéridien Atmosphérique
Formation Glace
Rejet de Sel Glace
Divergence Ekman, locale
Transport Eau Douce Méridien
Transport, Fonte Glace
Flux DensitéTempérature
Flux Densité Salinité
Flux Densité Salinité
THCe TCMOs
-
Réchauffement climatique
+
+
Action -Action +
Réchauffement climatique
Utrecht, 16/02/2012
Avec fonte : convection disparaît
Sans fonte : renforcement de la convection en mer de GIN, diminution en mer d’Irminger
Réponses des sites de convection
Avec : année 500
CTL
Sans : année 500
Utrecht, 16/02/2012
Représentation de la THC dans le modèle
Fonction de courant latitude-profondeur en Atlantique
Latitude
Profondeur
2 cellules avec NADW et AABW
Maximum pour la NADW d’environ 11 Sv (Indice THC)
Plus faible que les estimations issues d’observations (14-18 Sv)
Sv