1. RSMAS, University of Miami, The South Pacific Meridional Mode: A Mechanism Miami, Florida, USA 2. NCAR, Boulder, Colorado, USA for ENSO-like Variability and A Precursor for ENSO 3. IPRC, University of Hawaii, Honolulu, Hawaii, USA 1* 1 2 3 2 This work is supported by the Office of Science, U.S. DOE Honghai Zhang , Amy Clement , Clara Deser , Pedro Di Nezio , and Robert Tomas and by NSF Climate and Large-scale Dynamics Program. * [email protected]

Motivation Figure 2. Multi-model SPMM-NPMM interhemispheric asymmetry (slab) mean composite of anomalous SST and SLP 1. Hypothesis The North Pacific Meridional Mode (NPMM) has been identified as (left), cloud radiation and total cloud cover (CLT)  The asymmetry (see Section 1 Definition of SPMM) is due to the an effective mechanism through which North Pacific atmospheric (middle), and latent heat asymmetric distribution of trade winds in the tropical Pacific. variability impacts tropical Pacific climate (Chiang and Vimont, flux and surface winds (right) at times of -12, -6, 0 2. Climate model experiments 2004). Further, it has been argued that the NPMM can trigger and +6 months relative to the peak of the SEP index CAM4‐SOM Control run (cntl) Sensitivity run (Qa) ENSO events (Chang et al., 2007). Here we explore the linkages (SST averaged over 25°S– 15°S, 110°W–90°W). SST Simulation length 100 years 100 years between the equatorial and South Pacific. and heat fluxes are plotted in shading while SLP and Qflux (prescribed) Qflux(cntl) Qflux(cntl) + Qanomaly CLT in contours (positive Data and methods solid and negative dashed) Qanomaly with intervals of 40Pa/K and 2%/K, respectively.  CMIP3/5 Archive: Models Slab/Picntrl (yr) Observations Gray hatched area indicates significance (see Fig. 1 caption). The heat  AGCM-slab ocean CCCma 30 / 1001 flux is positive downward. All fields are normalized by the standard deviation of the SEP index before (50m): no ENSO performing composite. CCCma T63 30 / 350  Fully coupled: surface winds GFDL CM2.0 50 / 500 2. Mechanisms of SPMM preindustrial Figure 5. Qanomaly added to Qa. Positive (NCEP),  The SPMM has nearly identical physics to the NPMM. values denote cooling of the slab ocean scenario (dT/dt=Q -Q ). Figure 6. Spatial patterns (shading, SST regression as in Fig. GFDL CM2.1 100 / 500 precipitation net flux  The composite warm event in the southeast (SE) Pacific (left in 1) of the NPMM and SPMM in cntl and Qa, respectively.  Observations (GPCP v2.2), 3. Experiment results Vectors denote mean surface winds. Dashed-purple lines are INM‐CM3 60 / 330 Fig. 2) is initiated by the weakening of trade winds (Fig. 2c, t=-12). the zero-contours of mean meridional winds.  Climate model SLP (Hadley MIROC(hires) 20 / 100  The surface warming propagates northwestward (Fig. 2a & Fig. 4a)  The imposed Qanomaly changes mean trade winds, shifting ITCZ to experiments center), SST via the wind-evaporation-SST (WES) feedback (Fig. 2c) onto the the south of the equator (Fig. 6). MIROC(medres) 60 / 500 (HadISST and  Analytical solution equator, leading to the ENSO-like pattern (Fig. 2a, t=6) in the  In both experiments, The SPMM-NPMM asymmetry is in ERSST v3); MPI ECHAM5 100 / 506 AGCM-slab models. agreement with the asymmetry of the mean trade winds. (Fig. 6)  Remove seasonal MRI 150 / 150  A positive cloud feedback in the east Pacific also contributes to the cycle and apply a 1979 to date  The propagation of the two PMMs cannot extend beyond the ITCZ growing of the event, but it is model-dependent, and explains the low pass filter (1.5 NCAR CCSM4 540 / 500 where the mean meridional winds switch direction (V=0) (Fig. 7) inter-model difference of the SPMM amplitude (Fig. 3). years) HadGEM1 70 / 172 Figure 7. a1-d1: Mean surface  The composite event decays primarily from the SST-induced latent winds along the dashed-blue lines in Fig. 6, with wind speed heat flux anomaly (Fig. 2c, t=0 and 6). in blue, U in green and V in red; South Pacific Meridional Mode (SPMM)  The SPMM operates on seasonal-to-decadal time scales. a2-d2: anomalous composite Hovmoller diagrams along the same position for NPMM and 1. Definition of SPMM SPMM, respectively. Shading is 3. Role of active ocean dynamics SST. Colored contours are latent  Fig. 1a illustrates the well-established NPMM, while Fig. 1b heat flux (downward in green) basically shows a mirror image of the NPMM. We define the pattern  The SPMM is also active in and black contours are wind speed (weaker in dash). Note in Fig. 1b as the SPMM. fully coupled models (Fig. 1d that time axis goes from right to left, opposite to Fig. 4.  Interhemispheric asymmetry between the NPMM and the SPMM: and Fig. 4b) and observations (Fig. 1f and Fig. 4c-d). 4. Analytical solution  T  C T  C T  bT  The NPMM appears to be confined to the northern hemisphere,  fv  Tx  Au t x x y y   The Bjerknes feedback  fu  Ty  Av while the SPMM extends more onto the equator (Fig. 1a VS 1b).  Conceptual model: a  C  ( fu  Av) amplifies the equatorial signal T  a(uu  vv) /U bT y 2 2  Lindzen-Nigam  t U ( f  A ) that resembles ENSO (Fig. 1d Figure 3. Cloud feedback (estimated in the thick- atmosphere (Liu and Xie, 1994) C y (u) ~  f & f, and Fig. 4b-d). blue box in Fig. 2) VS SPMM amplitude: the  Asymmetric meridional stronger the feedback, the larger the amplitude. C (v) ~ v  Slab ocean (Wang 2010) propagation is caused by V: y  The SE Pacific warming leads the equatorial warming, suggesting that the SPMM can trigger ENSO events. Conclusions and implications  The SPMM has a larger equatorial signature than the NPMM.  The meridional mode exists in the South Pacific and explains the  The two PMMs may be related with different ENSO flavors: the ENSO-like variability in the absence of active ocean dynamics. equatorial signal of the SPMM resembles Eastern Pacific El Nino;  The SPMM is a potential precursor for ENSO events. while that of the NPMM resembles Central Pacific El Nino (Fig. 1c-f)  The SPMM has a stronger impact on tropical Pacific than the NPMM, owing to asymmetric mean trade winds in tropical Pacific. Figure 4. Composite Hovmoller diagrams (SST  Including observations from the data-poor South Pacific could in shading and SLP in contours) along a tilted improve the ENSO predictability. path in SE Pacific (dash- blue line in Fig. 2a,t=6) in References (a) AGCM-slab, (b) fully coupled models, and Chiang, J. C. H., and D. J. Vimont, 2004: Analogous Pacific and Atlantic meridional modes of tropical observations (c) NOAA atmosphere–ocean variability. J. Climate, 17, 4143–4158. Figure 1. Regression of anomalous SST (shading), SLP (contours, negative dashed, positive solid, zero ERSST v3, and (d) Chang, P., and co-authors, 2007: Pacific meridional mode and El Nino–Southern Oscillation. heavy solid) and surface winds (vectors) onto normalized SST time series averaged in Northeast (21°N– HadISST, respectively. Geophys. Res. Lett., 34, L16608, doi:10.1029/2007GL030302. 25°N, 138°W–142°W) (left) and Southeast (19°S–15°S, 103°W–107°W) (right) Pacific, respectively. Top Hatched area in the top Liu, Z., and S.-P. Xie, 1994, Equatorward Propagation of Coupled Air–Sea Disturbances with (a, b) are the multi-model mean of AGCM-slab models, middle (c, d) are fully coupled version and row denotes significance Application to the Annual Cycle of the Eastern Tropical Pacific. J. Atmos. Sci., 51, 3807–3822. bottom (e, f) are observations. SLP contour interval is 10Pa/K. Gray hatched area in (a)-(d) means the (see Fig. 1 caption for multi-model mean is not significant, where the significance is defined when at least 8 out of 11 models definition). Time axis goes Wang, F., 2010, Thermodynamical coupled modes in the tropical atmosphere-ocean: An analytical agree in sign. from left to right. solution, J. Atmos. Sci., 67, 1667-1677.