Internal tide interactions in the Bay of Biscay: Observations and modelling

Abstract

The paper addresses three-dimensional dynamics and interactions between internal waves in the Bay of Biscay, observed during an oceanographic survey conducted in summer 2006. These interactions are modelled and compared with observations. The effect of the internal tide (IT) beam structure in the deep ocean, on the interfacial IT along the seasonal thermocline, is analysed.To determine the regions of strong three-dimensional interactions and explain the observed IT structures, numerical simulations are used. Model results and observations show evidence of IT following analytical ray paths, but also interactions between beams. IT currents higher than 0.20m.s−1, are measured at 3000m near the bottom of the French continental slope, and are clearly modelled. In addition, in the mid-bay of Biscay, both the model and the data highlight a strong internal tidal current from 0 to a depth of 1500m, generated by three-dimensional interactions.Large interfacial ITs, associated with short wavelengths, are shown to be generated by deep ocean ITs when the latter reach the surface and impinge on the seasonal thermocline. It is indeed demonstrated that:•In the model and in the observations, interfacial ITs are intensified where the beam of the deep ocean IT reaches the seasonal thermocline.•The vertical variations of currents, measured in the surface layers, are related to short wavelengths observed in the seasonal thermocline, and also to the velocity modelled along the beam of the deep ocean IT.•These short wavelengths are generated in the middle of the Bay of Biscay where the deep IT impinges on the seasonal thermocline.Drifting thermistor chains were launched to follow the evolution of the seasonal thermocline. Along their trajectories, internal tide events occur at locations where the modelled surface tidal current has maximum values, and during spring baroclinic tides. The model clearly reproduces these events, but with a delay of one or two days compared to the observations. This delay is most likely due to a mixed layer too deep in the model.