MODELLING TSUNAMIS FROM RECONSTRUCTED LANDSLIDE SCENARIOS USING SPH

OVERALL OBJECTIVES This project is to use Smoothed Particle Hydrodynamics (SPH) to examine the generation of tsunami waves by realistic landslides. SPH is an innovative meshless computational technique ideal for potentially violent free-surface hydrodynamics. We will address the important issues of identifying the best techniques for modelling landslides in the context of predicting the resultant hydrodynamics, validate with experimental lab data (Fritz et al. 2004), apply to cases using real field data (e.g. Masson et al. 2002), simulate and identify effects of different landslide mechanisms on tsunami generation. OVERVIEW The impacts of tsunamis can be catastrophic as demonstrated by the Indian Ocean tsunami of 2004. While many are the result of earthquake excitation of the sea-bed, equally dangerous are tsunamis generated from landslides such as the Stromboli tsunami. The importance of understanding and reducing the destructive effects of tsunamis cannot be overstated. The identification of submarine landslide prone areas and characterisation of tsunami sources are quite advanced, but the ability to model accurately combined failure mechanisms and tsunami generation from landslides still needs development. Until now, modelling of landslide-generated tsunamis has represented a formidable challenge requiring fully 3-D multiphase numerical models. SPH offers a unique approach to simulate such difficult problems. Based on a representation of the fluid body by large volumes of water subject to Newton's 2nd Law, SPH does not require fixed computational grids so the domain can be multiply connected making modelling of violent free-surface flows straightforward. Application of SPH to landslide-generated tsunamis is a natural extension of the method. Monaghan et al. (2003) modelled the 2-D wave runup from blocks sliding down slopes on runners. Rogers & Dalrymple (2007) have modelled tsunami-initiation using deforming bed slopes and sliding blocks with prescribed motion. The application to deformable sliding masses has also been attempted by SPH with only partial success and needs development. Thus, many areas of research still exist where our knowledge and predictive capability requires substantial improvement. Namely (i) to try to understand how different landslide mechanisms affect generation (single instantaneous failure vs more gradual retrogressive failure), and (ii) the optimum method for simulating landslides and interaction with surrounding water. WORK PROGRAMME Months 1-6 Literature Survey & Preliminary Work Research Student (RS) carries out literature survey of numerical modelling of landslides and SPH. RS familiarises themselves with existing SPH code, SPHysics (our recently released open-source code). Months 7-16 Modelling combined landslide & hydrodynamics RS investigates optimum method for simulating landslides in the context of predicting the resultant hydrodynamics: modelling landslide as heavy water, individual rocks or deforming boundaries. Investigate possibility of coupling SPH with the Discrete Element Model (for the landslide). 1st-year report prepared. Months 17-32 Model Verification & Application Apply to experimental laboratory data on landslides to validate model paying close attention to runup, generated wave heights, resolution for convergence studies and landslide dynamics. Apply validated model to a series of increasingly complex landslide cases based on real data provided by National Oceanography Centre investigating effects of different landslide mechanisms on tsunami generation. Months 24-28 (concurrent) Model Coupling & Parallelisation of new Code If necessary, RS will include coupling of SPH to a longwave propagation model (development already underway at Johns Hopkins University, an international collaborator on SPHysics). RS will add new code developments to previously developed SPHysics parallel code. Months 33-36 Dissemination RS prepares final thesis & papers