Author(s): David Jenkins; Valentin Heller; Archontis Giannakidis
Linked Author(s): Valentin Heller
Keywords: No Keywords
Abstract: Landslides impacting a body of water can generate large landslide-tsunamis. Therefore, producing reliable and fast methods of predicting such waves is vital. No single empirical equation exists for universally predicting the landslide-tsunami characteristics involving both granular and block slide models. To fill this gap, we created a machine learning model using the Gradient Boosting method to predict the relative maximum tsunami amplitude aM/h and height HM/h for both slide types, where h is the still water depth. Our model produced an R2 score of 0.919 for aM/h and 0.937 for HM/h. Our method has shown promise and opens us possibilities to employing machine learning in real-world landslide-tsunami predictions.
Year: 2022