Abstract
New Caledonia, in the southwestern tropical Pacific, has recently been identified as a hot spot for energetic semidiurnal internal tides. In a companion paper, the life cycle of coherent internal tides, characterized by fixed amplitude and phase, was investigated in the region through harmonic analysis of a year-long, hourly time series from numerical simulation output. In this study, we investigate the temporal variability of the internal tide by decomposing the semidiurnal signals into coherent and incoherent components. Semidiurnal barotropic-to-baroclinic energy conversion is largely governed by the coherent component (> 90 %), amplified by a factor of 3 to 7 from neap to spring tides through the interaction of M2 and S2 barotropic tidal currents. Incoherent conversion – negligible in the annual mean – can explain on monthly to intraseasonal scales a notable fraction of variability, modifying semidiurnal conversion by up to ±20 %. The latter is largely explained by local effects, particularly the work of the coherent barotropic tide on incoherent baroclinic bottom pressure amplitude variations, linked to mesoscale-eddy-induced stratification changes. Away from the generation sites, tidal incoherence increases, evident through altered orientation of tidal beams and increasing phase variability, caused mainly by interactions with mesoscale currents. Variations in conversion are not consistently proportional to those in energy flux divergence, suggesting that variations in energy dissipation are linked to additional mechanisms that deserve further investigation. The incoherent sea surface height signature, with a root mean square amplitude of 1–2 cm, is widespread across the domain and introduces limitations in disentangling balanced (near-geostrophic) and unbalanced (wave-like) motions in spectral space. Transition scales – the length scale at which unbalanced motions become dominant over balanced motions – are proven meaningful when inferred from altimetry tracks that align with the main propagation direction of internal tides. However, when not aligned the method is flawed as it does not take into account the anisotropy of internal-tide dynamics.