WAVE OVERTOPPING AND AIR ENTRAINMENT AT DIKES WITH CROWN WALLS WITH VIDEO CLUSTERING TECHNIQUES

The experimental investigation of the wave-structure interaction processes is a very complicated issue, due to the highly non-linear turbulent flow conditions, including wave breaking and air entrainment see Raby et al. (2020). The Particle Image Velocimetry (PIV) measurements require many contemporary measurement devices to represent the whole processes and may not be recommended in small scale because of the potential disturbance introduced by the measurement devices themselves. Indeed, the investigation with PIV may be also completed by repeating the same test many times and by changing the measurement devices; this practice is however highly time consuming while only non-breaking waves are repeatable (Lim et al. 2015). A relatively economic, non-intrusive alternative to PIV is the well-established use of video imagery. The research presented here combines this low-cost approach with machine-learning algorithms and analyses the effectiveness and the potentialities of the video-cluster analysis to study the wave overtopping process, providing quantitative estimations of the main overtopping parameters (discharge, volumes, velocities) and quasi-quantitative estimations of the air entrainment during the wave propagation. Details about the methodology and the results can be found in Formentin et al. (2021).