1. Indian Monsoon, Indian Ocean dipoles and ENSO Pascal Terray LOCEAN/IPSL, France Fabrice Chauvin CNRM/Météo-France, France Sébastien Dominiak LOCEAN/IPSL, France Herve Douville CNRM/Météo-France, France

2. GOALS AND QUESTIONS The focus of this study is to revisit the role of the south subtropical Indian Ocean in the transitions of the (Indian) monsoon-ENSO-TIOD climate system during recent decades. Our hypothesis is that South East Indian Ocean (SEIO) SST anomalies during late boreal winter may trigger coupled air-sea processes (e.g. the wind-evaporation-sst and wind-thermocline- sst feedbacks) in the tropical eastern Indian Ocean during the following boreal spring, summer and fall which are fundamental for the turnabouts of the whole monsoon-ENSO- TIOD system. Tools : data analysis and coupled simulation experiments.

5. Role of the fluctuations in the SST background state Differences between the April-May mean SST before and after the 1976-1977 climate regime shift

6. Background the 1976-1977 climate regime shift was accompanied by a remarkable change in the lead-lag relationships between Indian Ocean SSTs and El Niño evolution.

7. * The statistical significance of the difference of the correlations observed before and after the 1976-1977 climate shift was estimated through a permutation test with 9999 shuffles. In the figure, the 1977-2001 correlations are plotted only if they are significantly higher (90% level) during the 1977-2001 period according to this permutation test.

8. The SEIO index is the time series constructed from the SST anomalies area averaged over the domain 90°-122°E and 5°-45°S for the February- March season

11. After the 1976-1977 regime shift, southern Indian Ocean SSTs observed during late boreal winter are a more suitable precursor in predicting El Niño evolution than the traditional oceanic heat content anomalies in the equatorial Pacific or zonal wind anomalies over the equatorial western Pacific ! Cross-validated correlation (COR) and root-mean-square-error (RMSE) skills for predicting the El Nino index NINO3.4 from simple regression models for the period 1977-2001.

12. The CNRM climate model (CM2) ARPEGE global atmospheric model version 3 (spectral model, T63 truncation, 45 vertical levels), see Geleyn et al. (1995); OPA global ocean model version 8 (finite differences model, 31 vertical levels with a higher resolution in the boundary layer, horizontal grid with a variable resolution, 2° interval in longitude and a latitude interval of 0.5° at the equator which increases poleward), see Madec et al. (1997); GELATO dynamic and thermodynamic sea ice model, see Salas Mélia (2002); In this study, we use a « control » simulation spanning 50 years. The simulation is initialized with oceanic temperature and salinity profiles from Levitus (1982) and with atmospheric trace gas concentrations observed in the 1950’s. The greenhouse gases (GHG) and sulfate aerosols concentrations are updated each year according to observations from 1950 to 1998.

13. Climatological annual cycle of the dynamical Indian Monsoon Index (IMI, Wang et al., 2001) and of the All India Rainfall (AIR) Index in the CNRM coupled model compared to wind reanalyses and precipitation observations from NCEP, ERA40 and CRU (top panel), as well as climatological spatial distribution of the JJAS mean precipitation over South Asia (right panel) in the CNRM coupled model compared to the GPCP climatology.

18. Coupled sensitivity experiment (ARPEGE/OPA-CM2) Prescription of a temperature anomaly of +/- 1°K on the first two levels and of +/- 0.5°K on the third level of the OPA model in the SEIO (71-121°E, 10°-30°S), with a weight of 0.5 on the limits of the domain. This has been done for every years (1950-1998) in the february’s restarts of the coupled model.

19. Positive-negative SEIO SST perturbations SST differences; all years included.

20. Positive-negative SEIO SST perturbations (rainfall and 850 hPa wind differences; all years)

21. Positive-negative SEIO SST perturbations (evaporation differences; all years)

22. Positive-negative SEIO SST perturbations SST differences; « Warm » SEIO years included

23. Positive-negative SEIO SST perturbations (rainfall and 850 hPa wind differences; warm SEIO years)

25. Conclusions Southern Indian Ocean SSTs during late boreal winter is a highly significant precursor of the Indian monsoon-ENSO- TIOD system evolution after the 1976-1977 regime shift; These SSTs anomalies are mainly generated by Mascarene High pulses occurring in late boreal winter (in the observations) or through the « atmospheric bridge » (in the CNRM-CM2 coupled model); These SST anomalies trigger coupled air-sea processes in the tropical eastern Indian Ocean (in the observations), modulate the western North Pacific anomalous cyclone/anticyclone (in both the observations and the CNRM-CM2 coupled model) and produce a persistent remote forcing on the whole monsoon-ENSO-TIOD system during the following seasons.