Influence of submarine landslide failure and flow on tsunami genesis

Thomas Zengaffinen | NGI


A bunch of mass failure mechanisms can cause a tsunami. One quite rare mechanism is submarine mass movements. However, these movements can cause disasters with widespread outcomes. At present, a broad understanding of how different submarine mass movements affect tsunami genesis is lacking. In our project a crucial task is to find out how different failure mechanisms, disintegration, and flow dynamics govern tsunami genesis. Our focus will be on submarine slumps and translational landslides. Slump and translational landslide induced tsunamis result from different formulations for the material behaviour. We shall, therefore, analyse viscoplastic flows using numerical slide models, which have to be tested against benchmark tests from the literature first.

One mechanism to trigger tsunamis is the slump. Slumps are impulsive gravitational mass failures that move with large accelerations and relatively short run-out distances. Present tsunami modelers prescribe simply the block motion to reproduce observed parameters: tsunami run-up height, slump volume and slump run-out distance. To this day, nobody has made an attempt to describe the slump with a numerical slide model, which incorporates viscoplastic flows. As a first direct representation, we shall link the slump motion and tsunami generation to mechanical parameters, such as the soil yield strength. The new slide models in BingClaw, that researchers have developed as part of the GeoClaw software, will run the simulations. We shall revisit the following historical cases with the aim of reproducing field observations. The Papa New Guinea tsunami from 1998 is the largest tsunami disaster in the 1990’s. That slump induced tsunami received most attention from the scientific community. Another case is the Grand Banks tsunami from 1929, which this research group has recently found due to an even larger slump source. We shall publish the results of the slump study with its link to these historical cases in a peer-reviewed journal article, which may have large impact in the tsunami community. The main motivation of the slump study is to constrain the significance of the mechanical slump parameters with respect to the tsunami observations. This first study of slump tsunamis serves as an introductory exercise, as slumps are just one end member of landslides.

Submarine landslides need a much broader study than slumps. At present, researchers have used kinematic block models to simulate the landslide. As a consequence, we cannot link tsunami genesis to landslide parameters yet. The task in our study is to use numerical slide models in BingClaw to be able to link landslide parameters to landslide dynamics and tsunami metrics. We shall carry out a large number of simulations in a systematic manner of the following key soil and geometrical parameters: initial yield strength, remoulding rate, remoulded yield strength, slope gradient, landslide geometry and volume. Two types of outputs are envisioned. The first type is a systematic representation of landslide characteristics as a function of the landslide parameters listed above. Landslide characteristics are usually slide acceleration, turbidity forming, and run-out distance. The second type is a systematic representation of tsunami characteristics as a function of the landslide parameters. Tsunami characteristics necessitate a functional description of the tsunami shape and strength, which are linked to the wave length and amplitude. We shall publish the results of the landslide study in a peer-reviewed journal article, which is a novel contribution to the landslide tsunami modelling community. Our study is a new way of linking tsunami genesis to mechanical parameters. In a next part of our project, we shall give particular attention to the remoulding rate of the landslide parameters, which controls the rate of failure.

Submarine landslides are described by a few soil parameters, mentioned in the previous paragraph. One particular parameter, the remoulding rate, is the key parameter to retrogressive and top-down landslide failures. In the retrogressive case the pre-failure remoulding stimulates the lower part of the landslide to fail, and then the failure moves up the slope. On contrary, in the top-down case, the pre-failure remoulding stimulates the upper part of the landslide to fail, which causes down-slope masses being pushed downwards; low values may result in cascading failures and envelopes of inefficient tsunami generation. No researchers have studied the complex phenomenon of top-down failure and its adverse effect on tsunami generation yet. We shall use the viscoplastic BingClaw code to study top-down failures of slide masses with the attention being on two of the largest ever recorded submarine landslides: the 8150 BP Storegga Slide, and the 4000 BP Trænadjupet Slide. The former is believed to have failed retrogressively, while the latter involved both retrogressive and top-down failure mechanisms. Together with the University of Durham, we shall retrofit the slide mechanism of the Trænadjupet Slide towards the University’s field data. The results of this study about cascading landslides will be published in a peer-reviewed journal article. Taken together, the focus of this project is on the link between tsunami metrics and parameters of submarine mass movements, including slumps and landslides. In addition, the dynamics of a landslide are crucial for the tsunami generation, which is best described by the remoulding rate. This analysis is important, as it will shed light on possible landslides that produce hardly any tsunamis, despite having enormous volumes. The last part will adapt this study about the submarine landslide tsunamis to probabilistic tsunami hazard analysis (PTHA).

The link between tsunami characteristics and landslide parameters will be linked to PTHA. Tsunami hazard depends, among others, on geological settings. We shall develop a set of idealised tsunami sources as a function of the landslide inputs. These sources will be placed as initial conditions in a numerical tsunami model. The results will be convolved with tsunami probability functions, and cast into an offshore or a coastal hazard model. This study will be linked to a regional hazard assessment in the Gulf of Cadiz, whereat University of Barcelona is responsible. My workplace will be in Barcelona and I shall be a companion author of a peer-reviewed paper led from researchers from Barcelona, including another early stage researcher (ESR). However, I shall prepare a technical report based on my work on the hazard assessment.

After the entire project, we shall be able to understand how the different mass movement mechanisms affect tsunami genesis. Key soil and geometrical parameters will be able to be linked to landslide characteristics and tsunami metrics by using the numerical slide model BingClaw. In addition, the retrofit of slide mechanisms to historical giant tsunami events will emphasis the theory from the models and reproduce field observations. A probabilistic tsunami hazard assessment will result from the analysis of the landslide tsunamis. In accordance with this project, the tsunami community will move a step further with the new way of linking tsunami genesis to mechanical soil parameters. Tsunami triggering will be understood better.


Hosting institute
Working Group
  • Computational Geomechanics


Thomas Zengaffinen
NGI Oslo
PhD student – ESR 13
Carl Bonnevie Harbitz
NGI Oslo
Principal Investigator – ESR 13