Scott McDougall

Tsunamis generated by long and thin granular landslides in a large flume

In this experimental study, granular material is released down slope to investigate landslide-generated waves. Starting with a known volume and initial position of the landslide source, detailed data are obtained on the velocity and thickness of the granular flow, the shape and location of the submarine landslide deposit, the amplitude and shape of the near-field wave, the far-field wave evolution, and the wave runup elevation on a smooth impermeable slope. The experiments are performed on a 6.7 m long 30° slope on which gravity accelerates the landslides into a 2.1 m wide and 33.0 m long wave flume that terminates with a 27° runup ramp. For a fixed landslide volume of 0.34 m3, tests are conducted in a range of still water depths from 0.05 to 0.50 m. Observations from high-speed cameras and measurements from wave probes indicate that the granular landslide moves as a long and thin train of material, and that only a portion of the landslide (termed the “effective mass”) is engaged in activating the leading wave. The wave behavior is highly dependent on the water depth relative to the size of the landslide. In deeper water, the near-field wave behaves as a stable solitary-like wave, while in shallower water, the wave behaves as a breaking dissipative bore. Overall, the physical model observations are in good agreement with the results of existing empirical equations when the effective mass is used to predict the maximum near-field wave amplitude, the far-field amplitude, and the runup of tsunamis generated by granular landslides.

A non-hydrostatic model for the numerical study of landslide-generated waves

This paper proposes and demonstrates a two-layer depth-averaged model with non-hydrostatic pressure correction to simulate landslide-generated waves. Landslide (lower layer) and water (upper layer) motions are governed by the general shallow water equations derived from mass and momentum conservation laws. The landslide motion and wave generation/propagation are separately formulated, but they form a coupled system. Our model combines some features of the landslide analysis model DAN3D and the tsunami analysis model COMCOT and adds a non-hydrostatic pressure correction. We use the new model to simulate a 2007 rock avalanche-generated wave event at Chehalis Lake, British Columbia, Canada. The model results match both the observed distribution of the rock avalanche deposit in the lake and the wave run-up trimline along the shoreline. Sensitivity analyses demonstrate the importance of accounting for the non-hydrostatic dynamic pressure at the landslide-water interface, as well as the influence of the internal strength of the landslide on the size of the generated waves. Finally, we compare the numerical results of landslide-generated waves simulated with frictional and Voellmy rheologies. Similar maximum wave run-ups can be obtained using the two different rheologies, but the frictional model better reproduces the known limit of the rock avalanche deposit and is thus considered to yield the best overall results in this particular case.