A critical analysis of existing semi-empirical predictive equations for the initial amplitude of landslide-generated waves

The recent potentially landslide tsunami in Palu, Sulawesi (Indonesia, ~500 deaths) in September 2018 highlighted the significant risk associated with landslide tsunamis. In terms of landslide-generated wave's hazards, the wave height and in particular the initial maximum crest amplitude is of essential importance for hazard prevention. Several researches over the past years tried to shed light on the characteristics of underwater landslide waves through numerical, analytical and empirical studies. Here, we analysed the existing predictive equations for the initial amplitude of the landslide-generated waves and studied their performances in reproducing real-world landslide incidents. Moreover, the effects of different parameters on the wave amplitude are studied. The semi-empirical equations, used in this study, have been presented by different researchers and are published in the literature. The 1994 Skagway, Alaska (USA) landslide tsunami event, which destroyed the railway dock and claimed the life of one construction worker, was used as the benchmark event. The initial data for the equations were the landslide thickness, width, length, specific gravity, velocity, the still water depth, and slope angle. Our results showed that various predictive equations result in wave amplitudes of 1.73 - 27.7 m, as compared to actual wave measurement of 1.0 m made by a tide gauge record. It was found that there was a poor agreement between measurement and the predictions of those semi-empirical equations that do not explicitly use the still water depth as one of the predictive parameters. The initial wave amplitude is strongly affected by the slide volume and slope angle.