Tidal turbines are sensitive to the high turbulence of tidal flows. Large and intense vortices are generated at the seabed and cause extreme loads, fatigue damage and degraded power production. Thus, these vortices must be characterised prior to the turbine installation. The flow characteristics can be assessed through ADCP measurements, but these measurements are sparse. The vortex characteristics are strongly affected by the seabed macro-roughness (rocks, faults) that induces spatial variations of the flow characteristics at a local scale. These local variations are difficult to catch through measurements [1]. Numerical simulations, that cover large areas, can fill in the gaps of measurements. Reynolds Averaged Navier Stokes simulations cover wide domains, but do not simulate the turbulent motions. Large Eddy Simulations (LES) do simulate the turbulent motions, but are computationally expensive, which reduces their spatial and temporal coverages. LES has been validated for the simulation of turbulence at a tidal power site, with a spatial coverage of about 0.5 km² and a temporal coverage of about 30 minutes [2,3]. However, the high turbulence intensity complicates the analysis of turbulent motions. The tracking and characterisation of vortices is complex and time-consuming. New methods to automate this work would be very welcome. In this work, we use Large Eddy Simulations to simulate the vortices generated at the rocky seabed of the Paimpol-Bréhat tidal turbine test site (France). A tracking method is used to follow the movement of turbulent motions (see Figure 1). This tracking highlights the long durability of turbulent motions. The impact of the most intense motions on fictive turbines is assessed and extreme flow variations are observed. This confirms the interest of the method for an easy detection of the most troublesome vortices and the locations where they are generated. It paves the way for the identification of the locations where turbine installation should be avoided due to potential damaging turbulent motions. [1] Togneri, M. & Masters, I. Micrositing variability and mean flow scaling for marine turbulence in Ramsey Sound. Journal of Ocean Engineering and Marine Energy, 2016, 2, 35-46 [2] Mercier, P.; Grondeau, M.; Guillou, S.S.; Thiébot, J. & Poizot, E. Numerical study of the turbulent eddies generated by the seabed roughness. Case study at a tidal power site. Applied Ocean Research, 2020, 97 [3] Mercier, P. & Guillou, S.S. The impact of the seabed morphology on turbulence generation in a strong tidal stream. Physics of Fluids, 2021, 33