TSUNAMI RISK MEASURE WITH TSUNAMI STOCHASTIC SOURCE MODEL

• Main Author: Fadhel Atras, Muh
• Format: Theses
• Language: Indonesia
• Subjects: Karya Umum
• Online Access: https://digilib.itb.ac.id/gdl/view/47782
• Institution: Institut Teknologi Bandung

Tsunami is a natural disaster that can result in heavy casualties and financial losses. Recovery of a post-tsunami disaster area can involve enormous costs. Therefore, we need a methodology that able to predict the risk of financial loss due to a tsunami disaster. The method for predicting the amount of this loss begins with first determining the probability distribution and a risk measure of the wave height at the coast caused by an earthquake occurring at sea. In this thesis, the risk measurement used is the Value-at-Risk (VaR) of the tsunami wave heights distribution. To obtain the risk measurement, first, we model the relationship between the tsunami source parameter and earthquake strength; followed by modelling the slip distribution in the fault plane using the Gaussian Randomfield; modelling seabed displacements; modelling the initial wave height, and then model the propagation of the tsunami waves from the earthquake epicentre through the coast using the Shallow Water Equation. The whole model is called the tsunami stochastic source model. What distinguishes this research from previous studies is that the height of the initial waves at the epicentre is analytically calculated and is not directly equated with the displacement of the seabed in a vertical direction. In addition, the wave propagation model used is a two-dimensional model. The complexity of the two dimensions is enough to reduce the computational complexity because it only sees the wave trajectory at a bathymetric line resulting from the intersection of the seabed and the plane that perpendicular to the surface of calm water level through the epicentre and the wave height observation point. In this thesis, the methodology discussed is applied to earthquake and bathymetry data of Sendai, Miyagi prefecture of Japan. It was found that at ten observation points around the coast of Sendai, Japan’s Miyagi prefecture, the most appropriate distribution of wave height is the Generalized Extreme Value distribution and the VaR value of the distribution is quite useful if evaluated using VaR Backtesting.