Modulation of Lateral Transport Barriers in the Ocean by Submesoscale Eddies and Inertia-Gravity Waves

Lagrangian coherent structures (LCS) can help identify material transport barriers in turbulent ocean flows. A commonly used practical and robust tool for identifying LCSs is the finite-time Lyapunov exponent (FTLE). Most past studies of FTLEs have focused on the mesoscale stirring field, often using geostrophic velocities derived from satellite altimetry. In this study, we compute FTLEs using velocity fields and particle trajectories derived from a very high resolution (submesoscale-resolving) ocean general circulation model, permitting the investigation of the role of smaller-scale / higher-frequency motions in the FTLE field. We have performed two sets of Lagrangian trajectory simulations: one where we have allowed the particles to be advected by the full three dimensional velocity fields and the other where vertical advection of floats is suppressed. At such high spatial and temporal resolutions, we observe very few/short-lived truly “coherent” structures; the flow is mostly made up of small-scale meandering flow features. The effect of unresolved scales (submesoscale eddies and inertia-gravity waves) is then studied by advecting particles with model velocity fields smoothed in space and time. The diagnostics obtained from these filtered simulations are compared with the “true” (fully resolved, 3-D) simulations. Mean FTLE decreases with both suppressed vertical motion as well as temporal smoothing. At lower temporal resolutions, we observe smooth, well-defined FTLE ridges which are independent of whether the particles are advected vertically. (More well defined FTLE ridges imply more well defined material transport barriers.) In addition, we also observe that when the particles are allowed to move with the full 3D velocities, they lead to a more “noisy” FTLE field (i.e more smaller-scale structures) as opposed to when the particles stay at the same vertical level. The FTLE fields for both 2D and full 3D advected particles converge to similar value for worst temporal resolution. We therefore conclude that turning off vertical advection has the same relative effect as worsening temporal resolution, i.e. to filter out small scales. Since existing satellite altimetry data has the limitations of both poor temporal resolution as well the assumption of horizontally non-divergent flow, LCS diagnostics based on satellite-derived geostrophic surface velocity fields may possibly exhibit more well defined coherent structures (FTLE ridges) than the real flow. Our findings thus call into question widely held approaches for the detection of transport barriers from coarse-resolution observations.