Image clustering for overtopping volume measurements

The representation of the wave overtopping is of unquestionable relevance for the assessment of the coastal and harbor safety and for the design of coastal defenses. The direct measurement of the wave overtopping volumes and discharges is a common practice in the laboratories, but it requires the installation of specific devices and instrumentation and may cause disturbance to the experiments. This contribution presents a procedure based on the image clustering to estimate the individual and the total overtopping volumes at coastal structures, while capturing the time-spatial evolution of the overtopping flow and avoiding any disturbance to the process itself. The procedure is validated against laboratory tests of focused waves on a sloping dike with a gentle and shallow foreshore and provided accurate and reliable estimations of the wave overtopping. The procedure automatically detects the individual overtopping volumes and its adaptability to different test conditions and structural configurations is proved. Requiring essentially a video camera to film the wave overtopping, the image clustering procedure can be applied also to prototype and field experiments, where the measurement of the wave overtopping may result complicated or unaffordable.