Impacts of Indian and Atlantic Oceanson ENSO

in a comprehensive modelingframework

Pascal Terray, Sébastien Masson,

Mathew Koll Roxy, Chloé ProdhommeLOCEAN/IPSL, Paris, France and IITM, Pune, India

Past sudies on impacts

of Indian and Atlantic Oceans on ENSO

Indian Ocean:

1) Tropical Indian Ocean: Yu et al (2002,2005,2009), Wu and Kirtman (2004), Yeh et al (2007), Dommenget et al (2006), Jansen et al (2009), Frauen and Dommenget (2012), …

2) Indian Ocean Basin mode (IOD): Kug and Kang (2005), Kug et al (2006a,2006b), Kug and Kang (2006), Ohba and Ueda (2007), Dayan et al (2014), …

3) Indian Ocean Dipole (IOD): Luo et al (2010), Izumo et al (2010,2014), Dayan et al (2014), …

4) South Indian Ocean Dipole (SIOD): Terray and Dominiak (2005), Dominak and Terray (2005), Yoo et al (2010), Terray (2011), Boschat et al (2013), …

Atlantic Ocean:1) Tropical Atlantic Ocean: Dommenget et al (2006), Jansen et al (2009) , Frauen

and Dommenget (2012), …2) Atlantic zonal mode: Rodriguez-Fonseca et al. (2009); Losada et al (2010), Ding

et al (2012), Keenlyside et al (2013), Polo et al (2014), …3) North tropical Atlantic: Ham et al (2013a, 2013b), Dayan et al (2013), …4) South Atlantic Ocean Dipole (SAOD): Terray (2011), Boschat et al (2013)5) Extra-tropical North Atlantic: Wang et al (2010)

Motivations and questions

Due to the complexity of ENSO teleconnections, results based on observations arealways ambiguous, especially those from short time series.

Many of the previous modeling investigations are based on short simulations, or partlysimplified (in oceanic or atmospheric components) coupled models or use differentdecoupling strategies.

Very few studies have focused on the relative impacts of the Atlantic and Indian Oceanson ENSO and most of them focus on a specific region inside the Atlantic and/or Indianbasins. Moreover, the relative roles of equatorial vs extra-tropical Indian and Atlantic SSTvariability on ENSO are not known.

It is not yet understood if the Atlantic and/or Indian Oceans have significant impacts on :1) onset, peaking or decaying phases of ENSO, or only in one of these phases;2) only in the transition from El Nino to La Nina and not in El Nino onsets;3) and what are the mechanisms at work for each phase?

Significant improvements of ENSO in coupled models in the lastdecade justify to revisit these questions with a state of the artcoupled model

Use of a high

resolution

ocean-

atmosphere

coupled model

without flux

corrections and

a realistic

ENSO

variability

ocean/sea ice:

NEMO v3.2

ORCA05-

LIM2

722×511×31

atmosphere:

ECHAM5.3

T106 L31

320×160×31

ORCA05 mesh (1 point over 10)

OASIS 3

The SINTEX-F2 coupled model

Extra-Tropics too cold, Tropics too warm,especially in the Atlantic Nino region;but only small cold tongue bias in theequatorial Pacific.

Realistic ENSO variability pattern, butslightly weaker than observed.

Key role of the SeasonalFootprinting Mechanism (SFM)and the North Pacific MeridionalMode (PMM) in ENSO onset forREF simulation

Realistic SST evolution andteleconnections

Sensitivity coupled experiments

✔ Suppression of the SST variability in either the Indian or Atlantic

Oceans by applying a strong nudging of the SST toward a SST

climatology computed from a control experiment (or observations).

Modification of the non-solar heat flux provided by the

atmosphere to the ocean by adding a correction term that scales

with the sst error of the model

Qns = Qns + dqdt*( sst-sst_clim)Qns: the non solar heat flux received from atmosphere

dqdt: -2400 W m-2 °K-1 (corresponds to the heat flux needed to warm a

50m thick layer of 1° during 1 day)

Experience Name REF FTIC (-obs) FTAC (-obs)Oceanic/

Atmospheric vertical resolution

31 levels(10 m)

31 levels(10 m)

31 levels(10m)

Coupling frequency 2 h 2 h 2 h

Nudging domain None Tropical Indian Ocean(30°S – North Boundary)

Tropical Atlantic Ocean(30°S – 30°N)

Length (years) 210 110 (50) 110 (50)

Mean state,ENSO seasonal phase-locking,

variability and statistics in REF and sensitivity experiments

No change of the mean state if the nudging is toward the SSTclimatology of REF, but improvement of the mean state andseasonal cycle (especially during boreal spring) in the Tropics ifthe nudging is toward the observed SST climatology!

Large increase of ENSO variability when Indian Ocean SSTvariability is excluded, but only a small increase when Atlantic SSTvariability is excluded!

Modification of the seasonal phase-locking of ENSO variabilityand large improvement of the seasonal phase-locking of ENSO inFTIC-obs and FTAC-obs experiments!

Shift of the ENSOperiod towards longerperiods in both theFTIC and FTACexperiments, butespecially in FTACexperiments!

Changes in ENSO onset phasewithout

Indian or Atlantic SST variability

ENSO onset twoyears before theENSO peak in FTICand FTACexperiments!

Key-role of the SPCZduring boreal springand of the low-levelwind circulation overthe North-WestPacific (NWP) duringboreal fall in thedevelopment ofENSO in FTAC!

Key-role of anextended trough overthe South Pacificduring austral winterin ENSO onset, 18months before theENSO peak in FTIC!

Changes in ENSO decaying phase without Indian or Atlantic

SST variability

Very slow decay ofENSO in FTIC andFTAC experimentsdue to very weakanomalous low-level easterlies overthe westernequatorial Pacific

Fast ENSO transitionin REF with cold SSTanomalies over theequatorial Pacific assoon as June-July ofyear+1

• No change of the Pacific mean state or seasonal cycle when excluding the Indian orAtlantic SST variability (not true if nudging toward an observed SST climatology).

• Changes in the the phase-locking to the seasonal cycle (less robust), but Indian andAtlantic SSTs do matter for ENSO seasonal phase locking!

• Indian or Atlantic SST variability tends to damp the ENSO amplitude, especially theIndian SST variability (robust).

• Shift of the ENSO period towards longer periods, especially when the Atlantic variabilityis decoupled (robust).

• ENSO onset occurs one year before when the Indian or Atlantic variability is decoupledand the nature and relative importance of the other atmospheric ENSO precursors areentirely different (robust).

• During decaying ENSO phase, Indian and Atlantic warm SST anomalies induce easterlysurface wind anomalies over the western equatorial Pacific, which fasten the transitionfrom El Nino to La Nina, promote ENSO biennial variability and a shorter ENSO cycle(robust).

Evidence for the importance of both Indian and Atlantic SST variability forimproving the coupled simulation, the seasonal phase-locking of ENSO,especially during the onset and decaying phases!

Testing the role of one ENSO precursor in isolation has no meaning, we needto adopt a global perspective!

CONCLUSIONSrelated to Indian and Atlantic SST impacts on ENSO