Datensatz

Zusammenfassung

The mixed layer (ML) hosts an intense submesoscale turbulence playing a pivotal role for energy transfers. Representation of ML turbulence from observations and models, partly, relies on the knowledge of its spatio-temporal scales. Here, we physically-inform the need of high spatio-temporal resolutions (L ~ 1km; T ~1 hour) to accurately infer the ML turbulence. Based on a numerical simulation of the Drake Passage in winter, we combine a Lagrangian filtering and a Helmholtz decomposition to decompose motions (LPF: low vs. HPF: high frequency) and their dynamical components (rotational vs. divergent). The ML hosts a 'zoo' of motions including: energetics, primarily rotational, submesoscale currents (LPF) and less energetics internal-gravity waves (HPF), such as rotational inertial waves, divergent lee waves and an internal-wave continuum. The contributions of motions to kinetic energy transfers are driven by their partitioning into dynamical components and spatio-temporal scales. Purely rotational motions realise an inverse cascade and coupled rotational-divergent motions realise a forward cascade. Submesoscale currents are largely rotational and primarily realise an inverse cascade. Internal-gravity waves, roughly equipartitioned between rotational and divergent components, realise an inverse and forward cascade of close magnitudes when coupled to submesoscale currents. All motions spread up to small spatio-temporal scales (L < 10 km; T< 6 hours) and these ranges significantly contribute to the inverse (≥ 30 %) and forward (80 — 95 %) cascades. Our results show that all classes of motions should be represented at high spatio-temporal resolutions to comprehensively infer winter ML turbulence, which has implications for study strategies.