Comprehensive prediction of subaerial landslide-tsunamis via slide model- and water body geometry-invariant machine learning techniques

Jenkins, DG, Heller, V and Giannakidis, A ORCID: https://orcid.org/0000-0001-7403-923X, 2025. Comprehensive prediction of subaerial landslide-tsunamis via slide model- and water body geometry-invariant machine learning techniques. Ocean Engineering, 320: 120197. ISSN 0029-8018

Text 2340734_Giannakidis.pdf - Post-print Full-text access embargoed until 8 January 2026. Download (9MB)

Abstract

Subaerial landslide-tsunamis (SLTs) are generated by mass movements impacting water bodies. These phenomena are responsible for substantial loss of life and damage to properties making their prediction vital. There is limited literature on deriving cross-slide model and water geometry prediction equations. Herein, state-of-the-art machine learning methods were utilised for building slide model- and water body geometry-invariant models, using data sets of four laboratory studies involving a wide range of experimental conditions. Models on individual data sets are also built. Other novel contributions include the development of: (i) ensemble models for predicting maximum SLT characteristics, (ii) sequence models for predicting wave decay, and (iii) a neural network for categorising wave types. All proposed models trained on individual data sets outperformed or matched performance of existing empirical equations. The slide model- and water body geometry-invariant models held performance across different slide models and water body geometries. The developed models are successfully validated by predicting the maximum height of the 1958 Lituya Bay tsunami with 2.47% error and the maximum wave run-up of the 2007 Chehalis Lake case with 9.52% error. This study, based on a range of idealised experimental conditions and several studies, is expected to shape preliminary hazard assessment of SLTs.

Item Type:	Journal article
Publication Title:	Ocean Engineering
Creators:	Jenkins, D.G., Heller, V. and Giannakidis, A.
Publisher:	Elsevier BV
Date:	March 2025
Volume:	320
ISSN:	0029-8018
Identifiers:	Number Type 10.1016/j.oceaneng.2024.120197 DOI 2340734 Other
Divisions:	Schools > School of Science and Technology
Record created by:	Laura Borcherds
Date Added:	24 Jan 2025 09:32
Last Modified:	24 Jan 2025 09:32
URI:	https://irep.ntu.ac.uk/id/eprint/52914

Actions (login required)

Edit View

Statistics

Views

Downloads