Past ENSO : observations and simulations with the IPSL model for

the Holocene and the last Millennium

P. Braconnot, Laboratoire des Sciences du climat et de l’envrionnement,

Gif-sur-Yvette, France

With input from different collegues

High resolution proxy records to infer ENSO variability

From T. Correge, G. Cabioche, C. E. Lazareth, M. CarréS. Tudhope, M. Helliot

Massive corals

ENVIRONMENTALRECORDS FROMMASSIVE CORALS:

Annual banding in coral cores provides ~monthly resolution

over lifespan of the coral

ENVIRONMENTALRECORDS FROMMASSIVE CORALS:

Annual banding in coral cores provides ~monthly resolution

over lifespan of the coral

Chemical analyses of massive corals: A window on tropical climate

S. Thudope

T. Correge

~8000 yr old porites coral

Tracers in corals: Sr/Ca

8,8

8,85

8,9

8,95

9

9,05

9,1

9,15

9,2 21

22

23

24

25

26

27

28

1993 1994 1995 1996 1997 1998 1999 2000

Sr/

Ca

(m

mo

le/m

ole

)

SS

T (°C

)

Year

21

22

23

24

25

26

27

288,8 8,85 8,9 8,95 9 9,05 9,1 9,15 9,2

y = 169,26 - 16,092x R= 0,95994

SS

T (

°C)

Sr/Ca (mmole/mole)

Calibration of Sr/Ca vs SST in a live coral from New Caledonia

Sr/Ca appears to be a robust SST tracer ("proxy")

from Correge

(Tudhope et al., unpublished)

-4 -3 -2 -1 0 1 2 3 4sea surface tem perature anomaly (degrees C)

NINO3.4 regionJarvis

-4 -3 -2 -1 0 1 2 3 4sea surface tem perature anomaly (degrees C)

NINO3.4 regionJarvis

r2 = 0.62 for monthly data

Jarvis coral 18O and NINO3.4 SST

Marine bivalves

ENSO strength from corals and bivalves

0.00

0.05

0.10

EN SO'strength '

(s.d.,‰ 1 8 O )

130 ± 2 ka85kaM O D ER N 2-3 ka

6.5ka 38 - 42 ka

112ka

118-128ka

H95

-64*

DT

91-7

Dur

-2

M93

TB

FC

Lain

g F

C2

H95

-16

H96

-36*

H96

-18

H96

-6

H96

-27

H97

-10*

H95

-14*

H94

-2*

H95

-58*

H95

-18

H95

-25*

H95

-6

AG E (Tudhope et al, 2001)

From Living and fossil corals

Comparison with bivalvesCabioche and Correge

0 20 40 60 80 100 120 140 160

20

0

-20

-40

-60

-80

-100

-120

-140

sea-level curve from deep sea core data (Shackleton, 1987)

thousand yea rs be fo re presen t

sea-

leve

l, m

‘GLACIAL’‘INTER-

GLACIAL’‘INTER-

GLACIAL’‘GLACIAL’

tim e intervals from which multi-decadal coral cores have been collected and analysed

Paleo-ENSO from mollusk shells

• Ex East Pacific

R=0.74

TC

alla

o

Annual Niño1+2 index

50-70% reduction of ENSO activity

M. Carré

Multi proxy reconstruction for the early to mid Holocene

• Need to combine different proxy to reconstruct mean-state, seasonality and variability

• Variability reduced in Early to mid-Holocene• Link with the insolation forcing through Bjerknes feedback (Clement

et al. 1999, 2000, etc…) or monsoon activity (Liu et al. 2003) • La Niña type mean state ?

From literature : Braconnot et al. subm

ENSO in PMIP 6ka simulations

Almost all simulations show reduced ENSO The increased monsoon counteracts the development of ENSO Case where the change in large scale dynamics dominate the signal different from future or LGM where thermodynamics seems more important.

Zheng et al 2008

Niño 3

ENSO reduction underestimated? Pb model? Pb proxy interpretation?

Mid and Early HoloceneResponse to insolation forcing and

fresh water fluxes

Results from Yihu Luan’s PhDCollaboration S. Brewer, W. ZhengBraconnot et al. Subm.

Early vs mid-Holocene?

• Differences in the response depending on precession?

Insolation 6ka-0ka Insolation 9.5ka-0ka

Latit

ude

Latit

ude

> 20 W/m2 > 25 W/m2

- 6 kyr

Présent

- 9.5 kyr

- 6 kyr

Présent

- 9.5 kyr

From Berger 1978

Sum

mer

sol

stic

e

Sum

mer

sol

stic

e

Simulations with IPSLCM4 OAGCM

Oceans26%

2.3 Pg y-1

Oceans26%

Oceans26%

2.3 Pg y-1

LAND ATMOSPHERE OCEAN & SEA I CE

Oceans26%

2.3 Pg y-1

Oceans26%

Oceans26%

2.3 Pg y-1

LAND ATMOSPHERE OCEAN & SEA I CE

water & energy carbon cycle

Land29%

2.6 Pg y-1

Land29%Land29%

2.6 Pg y-1

CO2

carbon cycle

Land29%

2.6 Pg y-1

Land29%Land29%

2.6 Pg y-1

CO2

Aerosols & chemistry

chemistry

Aerosols & chemistry

chemistry

•Ocean-atmosphere-sea-ice-land surface•Same version as in AR4 and PMIP2•Resolution atm : 96x72xL19•Ocean : 2° (refined at eq) xL31

Simulation CTRL 6ka 9.5ka

Length (yrs) 1000 650 300

Test to insolation forcing only

Marzin and Braconnot 2009

Role of fresh water in the North Atlantic?

• Fresh water flux on present day mean state enhances ENSO in a multimodel ensemble simulations (Timmermann et al. 2007)

• Test for 9.5ka and 6ka using IPSL-CM4 and a crude representation of interactive ice-sheet melting with climate

Following Swingedouw et al. 2006

Simulation 6kawF 9kawF

Length (yrs) 800 280

Test to insolation forcing and fresh water forcing

Ice-sheet Fresh water interactive with climate, uniformly distributed over the ocean > 40°N in Atlantic and Arctic

Change in DJF interannual variability

SST : standard deviation

0k

6k

9.5k

Insolation Insolation +wF•Variability larger in model than in observations

•Variability increases from 9.5k to 0k

•Fresh water flux in NA enhances variability

Change in seasonality

Niño 3 box : annual mean cycle

Wind and precipitation changes in July-August

mm/d

Insolation

Fresh water

Braconnot et al. Subm. Luan et al. In prep

ENSO composites Typical evolution in Niño3 box

Rms computed over this periodSST isolinesPr : color

Months Yr previous Yr after

Ex: role of EH insolation

Niño

Niña

Normal= seasonal cycle

mm/d

Relative changes induced by changes in seasonality, El-Niño and La Niña on SST

Braconnot et al. Subm.

The last millenium: ENSO response to volcanic forcing

From M. Khodri

IPSL-CM4 ModelIPSL-CM4 Model

Calendar time

Gao et al. 2008Khodri et al. In prep

Influence of external (volcanic and solar) forcing on SST? Influence of external (volcanic and solar) forcing on SST?

Warm Medieval Period Warm Medieval Period (≈ 950-1250 Year A.D.)(≈ 950-1250 Year A.D.)

simulated cooling in Niño 3 in respone to volcanic forcing

Persistant cooling in data (multi-year La Niña or La Niña like state?

Transition to LIA Transition to LIA (≈1400-1700?)(≈1400-1700?)

Higher Low Frequency (decadal) Variability

Modulation of El Niño events?

Volcanic ForcingVolcanic Forcing

Mann et al 2009

SST (ENSO) response to Volcanic ForcingSST (ENSO) response to Volcanic Forcing

SS

T A

no

ma

ly (

°C)

SS

T A

no

ma

ly (

°C)

Niño 4 IndexNiño 4 IndexControl Run

Millennium Run

Volcanic Forcing (W/m2)

SS

T A

no

ma

ly (

°C)

SS

T A

no

ma

ly (

°C)

Niño 3 IndexNiño 3 IndexControl Run

Millennium Run

Volcanic Forcing (W/m2)

Selected Eruption (max >2.5 W/m2)Selected Eruption (max >2.5 W/m2)

Random SST with same variance () as in CTRL

Simulated SSTSST > for at least 9 months

Conclusion

• Insolation and fresh water flux damp the SST seasonal cycle in the east Pacific (link with large scale dynamics)

• Increase ENSO magnitude from 9.5 ka to 0k in response to insolation

• Early and mid-Holocene ENSO development is damped in Autumn

• Fresh water flux enhances ENSO magnitude • Volcanic eruptions favors increased decadal variability at

the MWP/LIA transition in last millennium simulation with insolation and volcanic forcing

• Colder SST (more La Niña? ) after volcanic eruption? What about the hypothesis of El-Niño 1 year after?

• Need to better understand relative impact of seasonality and variability on relationship between proxy records and ENSO variability to produce reliable model-data comparisons.