Tsunami Propagation Research Articles

Submarine landslide-induced tsunami generation in Makassar strait, Indonesia using Non-hydrostatic wave model

Tsunami modelling of potential landslide-induced tsunami in Makassar Strait is carried out to quantify possible damage to the nearby cities. The numerical model is used to represent the wave generation by using NHWAVE model. The simulations consist of a series of scenarios based on distinct size of the landslide volume. Four landslides with volume 5, 8, 70, and 200 km3 are used as tsunami sources in the initiation stage. The sources are evenly distributed in the strait addressing different landslide location. Maximum wave heights of 1.5 m are found in the area between Palu and Bangkir from case 1 and around Talok from case 2 simulations. The empirical run-up calculation of 7.5 m is estimated at the land for the presented wave height. The value significantly elevates the case 3 and 4 proportional to the volume values. The waves impact more than half of coastline with maximum value found in the Sulawesi side. Interestingly, wide and narrow shelf next to Kalimantan Island plays an important role in reducing the tsunami hazard level.

Numerical Modelling of tsunami wave to assess the possible impacts along western coasts of India

Numerical modelling of tsunami waves has been made for the western coasts of India using TUNAMI N2 code. In this study the fault parameters are considered from earlier published literatures. Bathymetry data and possible tsunami generation locations have been obtained from the ETOPO2 (Global Relief Model) and General Bathymetric Chart of the Oceans (GEBCO) satellite data. For tsunami run-up the land topography data Shuttle Radar Topographic Mission (SRTM) is used. The present simulation consists of a duration of 6 hours (360 min). Possible arrival time with amplitude at various locations have been estimated. The paper also analyses the changes in the directivity of the generated tsunami waves towards western coasts of India by changing the dip and strike angles as different scenarios. Time series and height along the different parts of Gujarat coast and hourly travel-time chart of the tsunami wave are also discussed. After the earthquake and initial tsunami wave generation, it reaches at all the locations along the Gulf of Kachchh (Gujarat) in nearly 2 hrs to 5.30 hrs with amplitudes from 1 to 2.5 m, Mumbai in around 4.45 hrs with amplitude 2 m, Goa in around 3.08 hrs with amplitude 1 m, Karwar (Karnataka) in around 3.12 hrs and Mangalore in around 3.36 hrs with amplitudes 1 m each. The authenticity of the estimated tsunami phases of the 1945 tsunamigenic earthquake along the MSZ are corroborated with the available reports and published literatures.

Scattering of moving atmospheric pressure induced tsunamis by bathymetry and coastline

Tsunamis can be generated by a moving atmospheric pressure disturbance. The 2022 Tonga volcanic eruption and tsunami demonstrated the global relevance of such a tsunami generation mechanism. The generated locked and free waves are often scattered by bathymetric variations and coastlines, generating more free waves and resulting in a complex wave field. The physical processes involved can be described by the forced linear shallow water equations. Analytical solutions are sought after for simplified bathymetric variations and coastline configurations. In this paper, a moving atmospheric pressure front, which is uniform in the direction normal to its propagation direction, is considered. The water wave motion is assumed to start from the quiescent condition, so that the incident locked and free waves are related. However, since the wave scattering processes are linear, the solutions obtained in this paper can be used for analyzing scattering of the incident locked and free waves with uncorrelated amplitudes. For one-dimensional horizontal (1DH) problems, the water depth is either a constant or has a sudden change (i.e., a step) with or without a vertically walled coastline and the atmospheric pressure front always moves in the direction normal to the depth contours. In the case where a coastline is considered, the atmospheric pressure front can either move from the land to the sea or from the sea to the land. In the 2DH (two-dimensional horizontal) example, the wave field generated by a moving atmospheric pressure front, sweeping over the a circular island surrounded by a circular shelf, is investigated and discussed. For all the cases considered, analytical solutions in the integral form are obtained using the Fourier transform method. For the 1DH cases, analytical solutions are also presented in the form of infinite series, summing over infinite number of scattered and reflected waves from the bathymetric variations and coastlines. The new solutions reveal both the complexity of the wave scattering process and the significant differences between locked and free waves.

Origins of the Tsunami Following the 2023 Turkey–Syria Earthquake

AbstractOn 6 February 2023, a local tsunami was recorded in the southeastern Mediterranean Sea following the Mw 7.7 Turkey–Syria inland strike‐slip earthquake. Due to the lack of underwater observation, the tsunami generation mechanism remains mysterious. To understand the source mechanisms, we analyzed the tsunami waveforms of four nearby tide gauges and located possible sources using a backward tsunami ray tracing approach. We then conducted forward numerical modelings for a range of possible source parameters. We show that there were probably two tsunami sources, inside and outside Iskenderun Bay, which may be related to thick coastal sediments. A source inside the Bay with a characteristic length of 7 km produced dominant periods of 10–30 min with negative initial motion, possibly generated by a landslide. Another source of 6 km length outside the Bay produced dominant periods of 2–10 min with positive initial motions, possibly related with liquefaction.

Development of early warning system for tsunamis accompanied by collapse of Anak Krakatau volcano, Indonesia

Present tsunami warning systems have been specialized for earthquake-generated tsunamis, but rapidly evaluating the tsunamis caused by volcanic eruptions and/or volcanic sector collapses remains a challenge. In this study, we applied a numerical model to the 2018 Anak Krakatau tsunami event, which was generated by the sector collapse, investigated a tsunami prediction skill by the model, and developed a real-time forecasting method based on a pre-computed database for future tsunamis accompanied by such eruption of Anak Krakatau. The database stores spatiotemporal changes in water surface level and flux, which are simulated under various collapse scenarios, for confined areas in the vicinity of potential source. The areas also cover the locations of observation stations that are virtually placed on uninhabited island surrounding the source area. During an actual volcanic tsunami event, a tsunami is expected to be observed at the observation stations. For real-time tsunami forecasting, the most suitable scenarios to reproduce the observed waveforms are searched quickly in the database. The precomputed results under the identified scenarios are further provided as input for rapid tsunami propagation simulation. Therefore, an effective real-time forecasting can be conducted to densely populated coastal areas located at a considerable distance from the source, such as the coasts of Java and Sumatra. The forecasting performance was examined by applying the method for three hypothetical collapse scenarios assuming different sliding directions. We demonstrated that the tsunamis along the coasts were successfully forecasted. Moreover, we showed that the combination of a pre-computed database and the existence of observation stations near the source area was able to produce appropriate tsunami forecasting for the coastal area even in a volcanic event.

Design of Vertical Evacuation Building at Painan City Using Results From Tsunami Propagation Modeling

Fault under the Mentawai Islands has high potential to cause a large earthquake and triggering tsunami. Painan City, is located near the fault. With geographical condition on a bay surrounded by hills in the upstream area, a tsunami may cause a flow channeling effect and increase the inundation distance. An alternative of mitigation effort is to build a Vertical Evacuation Building (VEB) to help the community to reach a safe elevation within the evacuation time. Therefore, VEB should be designed to withstand tsunami impacts and to resist large earthquakes. This study conducts the design and analysis of the performance of multi-functional VEB structures at Painan City. The tsunami propagation simulation was carried out, and parameters obtained were used for structural design. Three VEB models were developed, i.e., (i) model with walls under tsunami and earthquake loads, (ii) model without walls under tsunami and earthquake loads, and (iii) model without walls under earthquake load only as the lateral load. Non-linear static pushover analyses were later conducted to evaluate the structural performances. The study revealed that debris force affects the most compared to other tsunami force components. Structural openings also affect the tsunami load received by the structure. Column moment force for models designed with tsunami load is greater than that of earthquake, while the difference is less significant for beam. The internal force differences occur along the tsunami inundated depth, thus influence the component design. From the pushover analysis, all models were found to meet the performance targets for earthquake.

A simplified approach for efficiently simulating submarine slump generated tsunamis

A simplified approach was proposed to efficiently simulate the tsunamis generated by a submarine slump. The landslide tsunami generation process was simulated using a long-wave model, simplifying the wave generation problem as a slump traveling down a plane slope with a prescribed trajectory. As a result, the landslide tsunami generation process was parameterized by 11 input parameters. The wave profile at the end of the wave generation process can then be specified as the initial conditions in any numerical tsunami propagation model to study the subsequent tsunami propagation. To demonstrate the capability of this new landslide tsunami generation approach, we used it in combination with an existing Boussinesq wave solver to simulate the 1998 Papua New Guinea landslide tsunami. The results based on the newly calculated physics-based wave profile compare reasonably well with field measurements and the results based on an existing tuning-based wave profile. Sensitivity tests were performed to examine the sensitivity of the runup results to each of the 11 input parameters. Six sets of 100-case Monte Carlo experiments were conducted to investigate the propagation of uncertainty from the input parameters to the runup results. Runup uncertainty was found to be approximately 1.5 times the parameter uncertainty, highlighting the uncertain nature of landslide tsunamis.

LANDSLIDE TSUNAMI HAZARD ASSESSMENT: A NUMERICAL MODEL FOR THE SIMULATION OF MULTIPLE LANDSLIDE-INDUCED TSUNAMIS SCENARIOS IN A MONTE CARLO FRAMEWORK

Submarine landslides can pose serious tsunami hazard to coastal communities. However, performing a comprehensive landslide tsunami hazard assessment for a given area is in general difficult in view of the large uncertainties associated with tsunamigenic source parameters. These are often known only approximately, on the basis of estimates of the landslide geometry, slide material properties, and kinematics. Here, we present an efficient model to perform such hazard analyses, based on solving the linear mildslope equation with a time-dependent source term representing the seafloor motion as detailed in Iorio et al. (2021). This approach allows carrying out many computations, for a large number of landslide scenarios, in a Monte Carlo (MC) approach framework, at a reduced computational cost compared to other available methods, while still providing physically accurate simulations of most landslide tsunami generation and propagation.

3D NUMERICAL MODELLING OF FIVE SUBMARINE LANDSLIDE SCENARIOS IN PERTH CANYON, AUSTRALIA TO ASSESS TSUNAMIGENIC HAZARD

Submarine canyons have been identified on nearly all margins around the world (Urlaub et al., 2013). Their configuration and morphology has been attributed to several factors including geology, tectonism, sea-level variations, and sediment supply to the region (Laursen and Normark, 2002) with processes occurring over varying temporal and spatial scales driving complex morphologies (Drexler, et al., 2006). A common process in submarine canyons is the mass wasting of sediment in the form of submarine landslides (SMLS) (Brothers, et al., 2013). A SMLS is a displacement of sediment or debris driven by gravity where the downslope forces are greater than the forces that are acting to resist the mass-failures (Mountjoy and Micallef, 2018). The potential tsunami hazard posed by these SMLS was assessed by Buller et al. (2021) using empirical calculations which showed that these SMLSs posed a tsunami threat to the adjacent coastline with calculated wave amplitudes ranging from 2.13 – 15.90 m. However, the tsunami risk assessed in their study was a conservative initial assessment and did not consider how local bathymetry influenced tsunami propagation.

TSUNAMI MODELLING IN A BUILT-IN COASTAL ENVIRONMENT WITH ADAPTIVE MESH RIFINEMENT AND VARYING BOTTOM FRICTION

The great Japan earthquake on March/11/2011provoked an extreme tsunami wave toward the northeastern Japanese coast. The tsunami inundation covered a wide range of territories stretching over 500 km2 and caused devastating influences that claimed more than 15,000 lives (Mori et al. (2011)). Miyagi Prefecture, the closest mainland to the epicentre, received massive tsunami waves that reached over 40 m in run-up height. Prasetyo et al. (2019) aimed to reproduce the tsunami wave inundation in Onagawa Town of Miyagi Prefecture by building a physical model using Hybrid Tsunami Open Flume in Ujigawa Open Laboratory (HyTOFU) at Kyoto University to study the inundation process by measuring the wave height and arrival time. For that purpose, two wave types were used, the (Pump-type) hydraulic bore, and the (Piston-type) solitary wave. Also, 2D and quasi-3D (Q3D) models were tested for their robustness and applied to emulate tsunami propagation in the complex coastal city model. In this study, we adopted adaptive mesh refinement (AMR) for its efficiency in tsunami modelling with Basilisk open-source flow solver for non-linear shallow water (Saint-Venant) equations and fully non-linear Boussinesq (Green-Naghdi) equations (Popinet (2015)). The numerical experiment was conducted to validate the model by comparing the wave height with the physical model experiment, the 2D, and Q3D model results by Prasetyo et al. (2019).

NUMERICAL SIMULATION OF TSUNAMI IMPACT FROM THE 1/15/22 ERUPTION OF THE HUNGA TONGA - HUNGA HA’APAI VOLCANO

Tsunamis from volcanic ‘explosive’ eruptions are rare, with the last catastrophic event being Krakatau in 1883 (Verbeek, 1885), during which, tsunamis were generated in the far-field by pressure shock-waves and in the nearfield of the volcano, in the Sunda Straits, by several potential geological mechanisms including pyroclastic flows, ash column, and/or caldera collapse. On 1/22/55, at about 4:15 UTC, a one in 1,000 year eruption of the Hunga Tonga-Hunga Ha’a-pai Volcano (HTHHV), that had started on12/20/21, reached its paroxysm with a series of large underwater explosions, releasing enormous energy (4-18 Mt of TNT), and ejecting a large ash plume 58 km into the stratosphere. We simulate both the near- and far-field tsunami generation from the eruption, but in this paper we focus on analyzing and validating the near-field impact against field data.

Indonesian “tsunami-generation” in a nutshell: systematic literature review

Publications related to tsunami-generation phase hold 42.8% of all the publications over all tsunami hazard literatures in Indonesia. Corroborating the fact, tsunami-generation studies are also vital in determining the type of propagation and inundation that may appear in the surrounding coastal regions, which in practice, can also help determining disaster management specifics and risk reduction activities. A literature review regarding tsunami-generation has been done for those purpose by using our CARI! Knowledge Engine, which includes collections of both International and Indonesian journals within the scope of Indonesian locus context. This study unpacks publication trend, the most influential articles, top contributing journals, top main authors, top institutions, and dominant countries as well as the science mapping of tsunami-generation subjects. To gain deeper understanding, we also develop tsunami-generation research framework within the basis of tsunami generation force, time occurrence, monitoring, and measurement approach, as well as hazard analysis approach. From the review, an interesting peak of tsunami-generation publications has been shown in 2020 (39 publications), following the occurrence of several Indonesian major tsunamis in the previous decade and the globally impacting Aceh Tsunami (or the Indian Ocean Earthquake and Tsunami) in 2004. Among all the publications of tsunami-generation force, tsunamigenic earthquake has become the most studied topic (67.7% of all publications), while the most discussed atypical sources mainly come from volcanic eruption (46.67% of all atypical literatures) and submarine landslide (33.33% of all atypical literatures). In terms of tsunami-generation source, it is found that the Sunda Megathrust has been the most-discussed location, which is mentioned within 39 publications, followed by the Palu-Koro Fault with 18 publications and the Krakatau Volcano by 16 publications. Apart from the notice of significant difference between research of historical tsunami and paleotsunami (120:5 publications), some gaps also can be found when discussing the meteotsunami-generation mechanism in Indonesia and the Global Navigation Satellite System’s Total Electron Content (GNSS-TEC) method, which, although proven beneficial for the development of tsunami early warning system, accuracy is still the main issue in capturing tsunami signals and positioning actual source location of tsunami. These were some highlighted limitations that arise in our process to understand tsunami-generation mechanisms and the exploitations of the knowledge. Therefore, future studies specifically within these areas are further encouraged.

Variable passing method for combining 3D MPM–FEM hybrid and 2D shallow water simulations of landslide‐induced tsunamis

AbstractWith a view to simulating the entire process of a landslide‐triggered tsunami, ranging from tsunami generation to offshore wave propagation, with relatively low computational costs, we present a 2D–3D coupling strategy to bridge 3D MPM–FEM hybrid and 2D shallow water (SW) simulations. Specifically, considering the difference in basis functions between the 3D and 2D analysis methods, we devise a novel variable passing scheme in the domain overlapping method, in which a slightly overlapped domain enables the generated wave to pass through the connection boundaries with as little discrepancy as possible. For the tsunami generation stage in the 3D domain, the hybrid method combining the finite element method (FEM) and material point method (MPM) is adopted. In this method, the 3D governing equation of the solid phase is solved with the MPM, whereas the well‐established 3D stabilized FEM is applied to that of the fluid phase in an Eulerian frame. Additionally, the phase‐field method is employed to track the free surface of the 3D fluid domain. On the other hand, the SW equation that represents the offshore wave motion in the 2D domain is solved by the 2D stabilized FEM. Several numerical examples are presented to demonstrate the effectiveness of the developed scheme in properly passing the data from 3D/2D to 2D/3D domains.

Sensitivity of slip distribution on tsunami trace heights and geological evidences: a case study of the 2011 Tohoku-Oki earthquake

We examined whether it is possible to estimate the tsunami source model of the 2011 Tohoku-Oki earthquake from a comparison of numerical simulations of tsunami propagation and sediment transport, the measured trace heights, and the sediment thickness of tsunami deposits. Twelve models with different subfault numbers were prepared based on a reference model inferred from tsunami waveform inversion. The reference model with 55 subfaults considering rupture propagation and the model with instantaneous slip successfully reproduced both the tsunami trace heights and sediment thickness distribution of tsunami deposits in the Idagawa Lowland and Sendai Plain. Other models with the same moment magnitude but fewer subfaults could not reproduce the observed trace heights, and the reproducibility of sediment thickness distribution strongly depended on the slip distribution. Models with increased slip amounts and moment magnitude could reproduce the trace heights; however, the simulated sediment thickness was underestimated for the Idagawa Lowland while overestimated for the Sendai Plain. Our results indicate that the combination of trace heights, sediment thickness of tsunami deposits, and numerical simulations of tsunami propagation and sediment transport can be used to estimate historical earthquakes and tsunamis. Efforts should be made to increase the number of subfaults for the study of historical events, although the obtained solutions may not be unique because of fewer trace heights or tsunami deposit data.

Propagation characteristics of tsunami waves in shear flow based on the Boussinesq model with constant vorticity

Propagation of tsunami wave in linear shear flow with constant vorticity is studied in this paper. The nonlinear and dispersive Boussinesq equations for water waves in shear flow are derived by introducing uniform vorticity into the Euler equations. Generally, the transoceanic propagation of tsunami can be well simulated by the weakly nonlinear Boussinesq model. Thus, the nonlinear and dispersive terms are retained to the first-order and second-order (ϵ,μ2) respectively in the Boussinesq equations with uniform vorticity in the present work, which satisfies the simulation accuracy especially the dispersion effects of the long-distance propagation of tsunamis and does not require huge computational cost for the simulation. The set of equations is solved with the total variation diminishing (TVD) Lax–Wendroff scheme which is second-order both in space and time. This TVD scheme is applicable to handle the dispersion effects of the N-waves and undular bores. Different patterns of tsunami wave in the water with presence of vorticity are simulated by the numerical model. Wave height and wavelength are affected by the current with vorticity. The pattern of solitary wave is significantly changed in upstream and downstream conditions. The wave deformation and propagation speed of N-waves and undular bores in the shear flow are also studied. The vorticity has impacts on the leading crest amplitude and the maximum wave appearance time for N-wave and changes the wave amplitude and the following water level for undular bores.

Probabilistic landslide tsunami modeling of the 2018 Palu Bay event

On September 28, 2018, a Mw 7.5 earthquake triggered near Central Sulawesi generated a highly destructive tsunami within Paul Bay (Indonesia). Field surveys and various studies conducted after the event showed that, as a result of the earthquake, several large submarine landslides were triggered along the shores of the bay, that significantly contributed to tsunami generation. The estimated geometry and other parameters for these slides, however, were affected by a large uncertainty. Here, we present a probabilistic tsunami hazard analysis of this event, based on Monte Carlo simulations using a linear Mild Slope Equation (MSE) model combined with a Green’s function approach, that allow efficiently simulating a large number of stochastic landslide tsunami generation and propagation scenarios within Palu Bay, for each of the identified landslides. In the MSE model, a space and time-dependent source term is used to represent the seafloor motion associated with each landslide scenario. In the Green’s function approach, a large database of elementary solutions and their tsunami elevation at a large number of coastal save points is pre-computed for tsunamis generated by a unit seafloor acceleration specified over a small area. Then, given an actual submarine landslide scenario, with a specific acceleration function, the tsunami elevation at the save points is simply and efficiently computed as a weighed linear superposition of the elementary solutions. Tsunami runup is finally obtained using a semi-empirical method, based on results computed at save points for each landslide scenario. The model is applied to the 2018 Palu Bay tsunami event, allowing to investigate how the uncertainty in landslide parameters affects tsunami hazard. A comparison with observed runups is made, which shows that these fall within the range of uncertainty of the simulated probabilistic runups.

Assessment of the 1693 tsunami wave generation and propagation simulation based on multiple focal mechanism scenarios for recent disaster mitigation in eastern sicily, Italy

The 1693 tsunami was the most extensive earthquake-tsunami event in Sicily, submerging Catania, Augusta, and Syracuse. However, the earthquake rupture, water level, arrival time, and furthest inundation distance of the tsunami waves are not yet known. This study aims to investigate the tsunamigenic source, run-up height, furthest inundation distance, and arrival time of the 1693 tsunami waves on the east coast of Sicily. Moreover, the assessment of tsunami-prone zones was also conducted based on worst-case earthquake-tsunami scenarios. Numerical modeling was applied by proposing six offshore focal mechanism scenarios using the shallow water equation in Delft3D and Delft Dashboard. The input parameters include length, width, strike, dip, slip, rake, and depth of the earthquake rupture. Meanwhile, the tsunami wave propagation onshore utilized XBeach and ArcGIS, considering the maximum run-up height, surface roughness analyzed from land use maps, slope, river existence, and coastline from Digital Terrain Model (DTM) identification. The results indicate that the worst possible impact of the 1693 tsunami was generated by an earthquake with a magnitude of Mw 7.13. The maximum water level, furthest inundation distance, and arrival time achieved 7.7 m, 318 m, and 9 min after wave generation offshore, respectively. This simulation is consistent with the discovery of 1693 tsunami deposits at a distance of less than 400 m from the coastlines of Augusta and Syracuse, but it is above the estimated furthest inundation distance in previous studies, which only reached around 100 m–200 m from the eastern coastline of Sicily. The results of the study are reliable as they align with the 1697 historical document where seawater inundated San Filippo Square, Catania.

Tsunami Wave Characteristics from the 1674 Ambon Earthquake Event Based on Landslide Scenarios

This study focuses on understanding the historical tsunami events in Eastern Indonesia, specifically the Banda Sea region, by extracting information from the limited and challenging-to-interpret historical records. The oldest detailed account of a tsunami in Indonesia dates back to 1674, documented in the book Waerachtigh Verhael Van de Schlickelijcke Aerdbebinge by Rumphius. The study aims to comprehend the primary source of the tsunami and analyze its characteristics to facilitate future tsunami risk reduction. The methodology includes collecting topography and bathymetry data, conducting landslide scenario analysis, employing a two-layer wave propagation model, and performing spectral analysis. The study utilizes comprehensive datasets, investigates potential landslide scenarios, simulates tsunami propagation, and analyzes frequency characteristics using the fast Fourier transform. The 1674 event yielded a runup height of approximately 50–100 m, whereas this study underestimated the actual runup. To illustrate the tsunami wave along the bay’s coastline, a Hovmöller diagram was employed. By analyzing the Hovmöller diagram, the power spectral density was computed, revealing five prominent period bands: 6.96, 5.16, 4.1, 3.75, and 3.36 min. The integration of these components provides a rigorous approach to understanding tsunami dynamics and enhancing risk assessment and mitigation in the study area.

Experimental analysis of Tsunami–Current interaction in river channel

The 2011 Tohoku Tsunami caused severe human and property damage along the east coast of Japan. In particular, the tsunami introduced through the river mouth damaged inland areas as it traveled up the river. In this study, a laboratory experiment was performed by applying funnel-shaped river mouth characteristics to investigate the hydraulic characteristics of a tsunami traveling up a river. In the river channel, the tsunami and wave heights were increased by the non-linear interaction between the tsunami and river current, which intensified as the river flow velocity increased. The river current in the opposite direction delayed tsunami propagation, thereby reducing the average celerity. The dispersion relation of the tsunami increased the time difference between the large first peak and small second peak. Additionally, the river current increased the effective volume of the tsunami wave peak. These hydraulic phenomena occurred because the flow characteristics of the tsunami are stronger than those of oscillating waves in the non-linear interaction between the tsunami and river current.

Tsunami waveform forecasting at cooling water intakes of nuclear reactors with deep learning model

The Fukushima nuclear disaster highlights the importance of accurate and fast predictions of tsunami hazard to critical coastal infrastructure to devise mitigation strategies in both long-term and real-time events. Recently, deep learning models allowed us to make accurate and rapid forecasts on high dimensional, non-linear, and non-stationary time series data such as that associated with tsunami waveforms. Thus, this study uses a one-dimensional convolutional neural network (CNN) model to predict waveforms at cooling water intakes of nuclear power plant at Uljin in South Korea. The site is particularly vulnerable to tsunamis originating from the west coast of Japan. Data for the CNN model are generated by numerical simulation of 1107 cases of tsunami propagation initiating from fault locations. The time series data for waveforms were predicted at 13 virtual gauges located in the nearshore region of the study area, 10 of which were classified as observation points and 3 gauges situated at the cooling water intakes were categorized as target locations. The performance assessment of the model's forecasts showed excellent results with rapid predictions. The study highlights two main points: (i) deep learning models can be based on sparse waveform in situ data (such as that recorded by deep-ocean assessment and reporting of tsunamis or any locally operating monitoring stations for ocean waves) or numerically simulated data at only a few points along the dominant wave propagation direction, and (ii) deep learning models are fully capable of accurate and fast predictions of complex geo-hazards that prompt rapid emergency response to coordinate mitigation efforts.