response of the amoc to global warming: role of ice sheets melting and amoc feedbacks

Didier Swingedouw

CERFACS, Toulouse, France

Response of the AMOC to Response of the AMOC to globalglobal

warming: role of ice sheets warming: role of ice sheets

melting and AMOC melting and AMOC feedbacksfeedbacks

Thermohaline circulation (THC)

Ocean circulation related to salinity and temperature gradients

NADW

AABW Rahsmtorf 2002

Rahsmtorf 2002

Quadfasel 2005

Effect –Effect +

Meridional advection

Heat transport

Salt transport

THCi THCo

THC internal feedbacks

Stocker et al., 2001

Stocker et al., 2001

Heat transportMeridional advection

Merdional temperature gradient in the atmosphere

Sea ice budget

Sea ice production

Ekman divergence

Atmopheric freshwater transport

Sea ice transport

Heat transport

Salt transport

Surface salinity flux

THCi THCo

THC internal feedbacksEffect –Effect +

Heat transportMeridional advection

Merdional temperature gradient in the atmosphere

Sea ice budget

Sea ice production

Ekman divergence

Atmopheric freshwater transport

Sea ice transport

Heat transport

Salt transport

Surface salinity flux

THCi THCo

Global warming

Global warming

Future of the THC (or AMOC): what role for ice sheet melting?

Schneider et al., 2007

Ice sheets melting neglected in most of the « IPCC » coupled models

Polar ice sheets

Greenland Ice volume equivalent to 7

meters of sea-level rise Surface area of 2 millions km²

(81% ice covered)

Antarctica Ice volume equivalent to 61

meters of sea-level rise Surface area of 14 millions km²

(98% ice covered) Massive ice shelves

Problematic

Can the Greenland ice sheet (GIS) melting accelerate the weakening of the AMOC?

How to quantify the key feedbacks of the AMOC?

Can the Antarctic ice sheet (AIS) melting stabilize this weakening?

How works the bipolar ocean seesaw?

Outlines

Impact of future GIS melting on the AMOC

Quantifying the AMOC feedbacks

Impact of future AIS melting on the AMOC

Revisiting the concept of the bipolar ocean seesaw

Outlines

Impact of future GIS melting on the AMOC

Quantifying the AMOC feedbacks

Impact of future AIS melting on the AMOC

Revisiting the concept of the bipolar ocean seesaw

Tool No1: IPSL-CM4 coupled model

LMDz

GCM, 3.75°x2.5°

19 levels

ORCA-LIM

GCM, 2° horizontal

31 niveaux verticaux

ORCHIDEEIce sheet

thermodyn.

module

OASIS

AMOC in IPSL-CM4

Atlantic meridional overturning streamfunction

Latitude

Depth

∗ Two oceanic cells in the Atlantic: NADW and AABW

∗ Maximum for NADW cell: 11 Sv

∗ Smaller than observations-based estimates (13-23 Sv)

∗ No convection in the Labrador Sea in the model

Sv

Mixed layer depth in JFM

Experimental design 1

Two versions of the IPSL-CM4:

1) With ice sheet melting Snow

Land Ocean

Ice

Net surface heat flux

2) Without ice sheet

melting

Simple thermodynamical parametrisation of ice sheet melting

Response of the AMOC after 500 years at

2xCO2

CO2

(ppm)

0 70 500

280

560

CTL

NoGIS

GIS

Time (Year)

Swingedouw D. and Braconnot P., Effect of Greenland ice-sheet melting on the response and stability of the AMOC in the next centuries, AGU monograph "Ocean Circulation: Mechanisms and Impacts" by Schmittner A., 383-392, (2007).

GIS

CTL

NoGIS

AMOC max.

GIS melting = 0.13 Sv after 200 ans. More than half of GIS melted in 500 years

AIS melting < 0.02 Sv = negligible

CTL

NoGIS

GIS

Global temperature

AMOC and convection

€

AMOC = γΔρ

Correlation of 0.98 between density

anomalies and THC variations

€

AMOC

t=0

NoGIS-CTL

GIS-CTLSwingedouw et al. Quantifying the AMOC feedbacks during a 2xCO2 stabilization experiment with land-ice melting. Climate Dynamics, 2007

Density budget in the convection sites

€

TransportS + Δρ Transport

T + ΔρSurfaceS + ΔρSurface

T + Δρ ResiduT + Δρ Residu

S ≈ Δρ

Transport

Surface

∗ In NoGIS: the main decreasing term for the AMOC is

the change in heat flux in the convection sites

∗ Main processes that help the AMOC to recover:

∗ Transport of salinity anomalies from the tropics

∗ Decrease of sea-ice melting in the convection sites

In GIS => Role of feedbacks?

Outlines

Impact of future GIS melting on the AMOC

Quantifying the AMOC feedbacks

Impact of future AIS melting on the AMOC

Revisiting the concept of the bipolar ocean seesaw

AMOC related feedbacks

TOT

Tsurface

TTransport

SOT

Ssurface

STransport

We consider differences between the scenarios to isolate feedbacks effects

01

1

ii

iii

with

i

i 0 0with Land-ice melting anomaly

AMOC feedbacks quantification

€

G =1

1− (λ S + λT )

⎛

⎝ ⎜

⎞

⎠ ⎟= 2.5

Dynamical gain

S T

Feedback gain

Temperaturechanges

Salinity changes

THCe THCs

)( S

€

(λT )

+-

€

T

€

S

A model dependent result

Using LOVECLIM model, Driesschaert et al. (2007) found a very moderate effect of ice sheet melting (at 4XCO2)

Causes ?

1. Different GIS melting?

2. Different THC dynamics?

3. Different GIS melting localisation ?

4. Different AIS melting?

After 500 years the GIS melting represents:

• 8% in LOVECLIM

• 50% in IPSL-CM4

1

AMOC max. (Driesschaert et al. 2007)

Outlines

Impact of future GIS melting on the AMOC

Quantifying the AMOC feedback

Impact of future AIS melting on the AMOC

Revisiting the concept of the bipolar ocean seesaw

Tool No2: LOVECLIM

ECBILTQG, T21 with 3 levels

CLIOGCM, 3°x3° 20 levels

VECODE

ISM

10 km

31 levels

LOVECLIM climatology

Two cells with larger magnitudes than in IPSL-CM4

Convection in the labrador Sea

Better agreement with observations

Mixed layer depth JFM

Atlantic meridional streamfunction

Experimental design 2

∗ Anaysis of 4XCO2 projections:

∗ Without polar ice sheets melting (fixed)

∗ With Greenland and Antarctic ice sheet melting (AGIS)

∗ With Greenland ice sheet melting only (GIS)

∗ With Antarctic ice sheet melting only (AIS)

Sans

CO2 (ppm)

0140 3000

280

1120

CTRL

Year

4xCO2

AGISGISAIS

AABW response in the projections

∗ AABW cell is weakened the first 300 years

∗ Then it increases

∗ The cell stabilizes around its inital value with Antarctic ice sheet melting (AGIS, AIS)

∗ And 25% above without (GIS, fixed)

∗ AIS looses mass and put around 0.1 Sv in the SO

Export AABW at 30°S

CTRL

AGIS

AIS

GIS

fixed

NADW cell response in the projections

Without AIS melting the

weakening of the NADW cell is

larger

CTRLNoISAGISGISAIS

Swingedouw et al., AIS melting provides negative feedbacks for global warming. GRL, 2008

AMOC max.

NADW export at 30°S

Outlines

Impact of future GIS melting on the AMOC

Quantifying the AMOC feedback

Impact of future AIS melting on the AMOC

Revisiting the concept of the bipolar ocean seesaw

The ocean bipolar seesaw

In ocean model: Stocker et al. (1992)

In paleoclimatic reconstructions: Broecker (1998)

Confirmed in OGCMs: Seidov et al. (2001)

Not true in AOGCM in transient phase: Stouffer et al. (2007)Why?

NADWAABW

NADWAABW

Stouffer et al.’s experimental design

∗ Freswater input of 1 Sv south of 60°S during 100 years (Southern Ocean Hosing: Hos1)

∗ Equivalent freshwater amount larger than GIS volume

∗ Stouffer et al. noticed a slight weakening of the NADW cell

∗ They attributed this weakening to the salinity anomalies transport from the Southern Ocean to the North Atlantic

AMOC max. (Stouffer et al. 2007)SSS anomalies after 100 yrs

Response of NADW cell to a freshwater input in the Southern

Ocean

∗ We reproduce the same experiement using LOVECLIM (Hos1) en utilisant le modèle LOVECLIM (sans calotte polaire)

∗ A dipole of streamfunction anomalies:

∗ Weakening of the northern cell

∗ Enhancement of the cell south of 30°N

Swingedouw et al., Impact of transient freshwater releases in the Southern Ocean on the AMOC and climate. Climate Dynamics, 2009

Atlantic meridional overturning streamfunction

Hos1 – CTRL (100 yr mean)

Climatic response to a freshwater input in the

Southern Ocean Cooling of the Southern Hemisphere

Increase of the meridional temperature gradient

Increase in westerly winds

Potential impact on NADW cell (Toggweiler and Samuels 1995)

We test this effect with a similar experiment to Hos1 but with fixed wind (CTRL) called HosWind

Surface temperature and wind

Hos1 – CTRL (100 yr mean)

Three main processes

Three main processes influence the NADW cell:

The wind increase explains part of the NADW cell increase

Hos1 - HosWindHosWind - CTRLHos1 - CTRL

1. Bipolar ocean seesaw: a weakening of AABW cell enhances the NADW cell

2. Salinity anomalies advection, from the south up to the North Atlantic convection sites

3. Increase in the Southern winds, which pushes (ekman) surface water towards the Atlantic basin

Atlantic meridional streamfunction

F S F N

F N

AABW

NADW

30°S Latitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

Quantification of the impact of each process

DF=t

Vb

We use the density binning analysis, which quantifes the formation-consumption of water masses

Which reconcile dynamical and thermodynamical approach

F S F N

F N

NADW

30°S Latitude

NF

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°S Latitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

NNNtAtl DFV= 1

Budget North of 32°S in the Atlantic for water denser than 27.6 kg/m3

F S F N

F N

AABW

NADW

30°SLatitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

Process 1: bipolar ocean seesaw

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

+4.5 Sv

F S F N

F N

AABW

NADW

30°SLatitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

-3 Sv

Process 2: salinity anomalies advection

F S F N

F N

AABW

NADW

30°SLatitude

SF NF

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

F S F N

F N

AABW

NADW

30°SLatitude

Ψ

D1

2D

Pycnocline

Depth

+1.5 Sv

Process 3: Southern wind increase

Phase diagramm

Impact of the rate for the freswater release for a given amount (100 Sv.yr)

Rate < 0.2 Sv = process 2 is very small

In the projections, we are in this side of the phase diagramm

Hosing Perturbation (Sv)0 0.4 0.8 1.2 1.6 2

Conclusions

∗ GIS melting induces a collapse of the AMOC in the IPSL-CM4 after 500 years

∗ This is due to a large positive salinity feedback and a weak temperature negative one

∗ In LOVECLIM the AIS melts after a few centuries at 4XCO2, which stabilises the AMOC weakening

∗ The mechanisms for the response of the AMOC to a Southern Ocean hosing are not trivial and their effects depend on the rate of the hosing

Quantifying the AMOC feedbacks among different CGCMs

For a given freshwater input, large spread among AOGCMs (Stouffer et al. 2006)

Methodology of feedbacks quantification useful (Swingedouw et al. 2007)

To be done using the Thor project framework (FP7)

Feedback gain

Temperaturechanges

Salinity changes

THCi THCo

)( S

€

(λT )

Stouffer et al. 2006

Outlooks

Compare the AMOC feedbacks in different CGCMs: THOR project.

Coupling of ice sheets in a higher resolution cimate model (IPSL-CM4…) can improve our: Projections of sea-level rise Projections of THC changes

Compare the trend in salinity in both observations and models

Thank you

Mailto: swingedo@cerfacs.fr Web: http://dods.ipsl.jussieu.fr/dssce/public_html/index.html