Imprint of chaotic ocean variability on transports in the southwestern Pacific at interannual timescales

Abstract

Abstract. The southwestern Pacific Ocean sits at a bifurcation where southern subtropical\nwaters are redistributed equatorward and poleward by different ocean\ncurrents. The processes governing the interannual variability of these\ncurrents are not completely understood. This issue is investigated using a\nprobabilistic modeling strategy that allows disentangling the\natmospherically forced deterministic ocean variability and the chaotic\nintrinsic ocean variability. A large ensemble of 50 simulations performed with\nthe same ocean general circulation model (OGCM) driven by the same realistic\natmospheric forcing and only differing by a small initial perturbation is\nanalyzed over 1980–2015. Our results show that, in the southwestern Pacific, the\ninterannual variability of the transports is strongly dominated by chaotic\nocean variability south of 20 ∘ \u2009S. In the tropics, while the\ninterannual variability of transports and eddy kinetic energy modulation are\nlargely deterministic and explained by the El Nino–Southern Oscillation (ENSO),\nocean nonlinear processes still explain 10\u2009 % to 20\u2009 % of\ntheir interannual variance at large scale. Regions of strong chaotic variance\ngenerally coincide with regions of high mesoscale activity, suggesting that a\nspontaneous inverse cascade is at work from the mesoscale toward lower frequencies\nand larger scales. The spatiotemporal features of the low-frequency oceanic\nchaotic variability are complex but spatially coherent within certain\nregions. In the Subtropical Countercurrent area, they appear as\ninterannually varying, zonally elongated alternating current structures, while\nin the EAC (East Australian Current) region, they are eddy-shaped. Given this\nstrong imprint of large-scale chaotic oceanic fluctuations, our results\nquestion the attribution of interannual variability to the atmospheric forcing\nin the region from pointwise observations and one-member simulations.