This study presents a method for the detection of the most likely tsunami scenario among a set of possible scenarios using an observational wave sequence based on a sequential Bayesian update scheme. The proposed method consists of two phases: an offline preliminary learning phase and an online real‐time detection update phase. The innovation of this study is that proper orthogonal decomposition (POD) and Bayesian update are used together with an established tsunami simulation technique. In the offline reinforcement learning process, a series of tsunami simulations are carried out based on geophysically feasible scenarios, and the spatial modes of wave data calculated at predefined synthetic gauge locations are extracted through the application of POD. When a real tsunami event occurs and observational ocean data are obtained, the online process can then be performed as follows: using the stored spatial modes along with their component coefficients, pseudocoefficients are repeatedly estimated from the obtained wave data and used to sequentially update the most likely tsunami scenario according to the posterior probability through Bayesian update. A verification analysis is carried out to illustrate the procedure of the proposed method, and a validation analysis is conducted to demonstrate both the capabilities and applicability of the process with reasonable accuracy. A comprehensive discussion details the characteristic features of the proposed method in terms of the real‐time prediction of tsunami hazards and risks.

Abstract. Studies have indicated that submarine landslides played an important role in the 2018 Sulawesi tsunami event, damaging the coast of Palu Bay in addition to the earthquake source. Most of these studies relied on observed coastal subaerial landslides to reproduce tsunamis but could still not fully explain the observational data. Recently, several numerical models included hypothesized submarine landslides that were taken into account to obtain a better explanation of the event. In this study, for the first time, submarine landslides were simulated by applying a numerical model based on Hovland's 3D slope stability analysis for cohesive–frictional soils. To specify landslide volume and location, the model assumed an elliptical slip surface on a vertical slope of 27 m of mesh-divided terrain and evaluated the minimum safety factor in each mesh area based on the surveyed soil property data extracted from the literature. The soil data were assumed as seabed conditions. The landslide output was then substituted into a two-layer numerical model based on a shallow-water equation to simulate tsunami propagation. The tsunamis induced by the submarine landslide that were modeled in this study were combined with the other tsunami components, i.e., coseismal deformation and tsunamis induced by previous literature's observed subaerial coastal collapse, and validated with various post-event field observational data, including tsunami run-up heights and flow depths around the bay, the inundation area around Palu city, waveforms recorded by the Pantoloan tide gauge, and video-inferred waveforms. The model generated several submarine landslides, with lengths of 0.2–2.0 km throughout Palu Bay. The results confirmed the existence of submarine landslide sources in the southern part of the bay and showed agreement with the observed tsunami data, including run-ups and flow depths. Furthermore, the simulated landslides also reproduced the video-inferred waveforms in three out of six locations. Although these calculated submarine landslides still cannot fully explain some of the observed tsunami data, they emphasize the possible submarine landslide locations in southern Palu Bay that should be studied and surveyed in the future.

Abstract. Studies have indicated that submarine landslides played an important role in the 2018 Sulawesi tsunami event, damaging the coast of Palu Bay in addition to the earthquake source. Most of these studies relied on visible landslides to reproduce tsunamis but could still not fully explain the observational data. Recently, several numerical models included hypothesized submarine landslides that were taken into account to obtain a better explanation of the event. In this study, for the first time, submarine landslides were simulated by applying a numerical model based on Hovland’s 3D slope stability analysis for cohesion-frictional soils. To specify landslide volume and location, the model assumed an elliptical slip surface on a vertical slope of 27 m of mesh-divided terrain and evaluated the minimum safety factor in each mesh area based on the surveyed soil property data extracted from the literature. The landslide output was then substituted into a two-layer numerical model based on a shallow-water equation to simulate tsunami propagation. The results were combined with the other tsunami sources, i.e., earthquakes and observed coastal collapses, and validated with various postevent field observational data, including tsunami runup heights and flow depths around the bay, the inundation area around Palu city, waveforms recorded by the Pantoloan tide gauge, and video-inferred waveforms. The model generated several submarine landslides, with lengths of 0.2–2.0 km throughout Palu Bay. The results confirmed the existence of submarine landslide sources in the southern part of the bay and showed agreement with the observed tsunami data, including runups and flow depths. Furthermore, the simulated landslides also reproduced the video-inferred waveforms in 3 out of 6 locations. Although these calculated submarine landslides still cannot fully explain some of the observed tsunami data, they emphasize the possible submarine landslide locations in southern Palu Bay that should be studied and surveyed in the future.

No abstract