Probabilistic Tsunami Hazard Analysis Alone Chinese Coast

Since ancient times, the coastal areas of China have always been characterized by high population, developed economy and convenient transportation. However, there is meanwhile higher-level tsunami hazard than inner lands. For recent years, the studies about the tsunami hazard analysis have been progressing gradually, but most of them gave the qualitative analysis with the deterministic method. Because of some key steps of PTHA(probabilistic tsunami hazard analysis) need to be improved, the key problems need to be effectively solved, there have been less quantitative evaluations with the probability analysis method. So this thesis focuses on determining the key parameters in PTHA, improving PTHA method and evaluating the sensitivities in PTHA. Finally the tsunami hazard maps are complied along the coastal areas of China. The major works are detailed below.(1) Determine the key parameters in PTHA. At first several empirical relationships between magnitude and fault length, width, slip are summarized. By comparing their suitable magnitude range, suitable regions and actual application situation, it is suggested the one given by Papazachos et al.(2004) is suitable for the parameter estimation in local and region tsunami sources in China, in addition to compare its empirical values with actual values of several destructive tsunami occurred recent years. As scenario tsunamis with the maximum magnitude of the potential local tsunami sources, it simulates the largest influence along Chinese coastal areas by tsunami waves and calculates the affected range of coastal areas with the most dangerous cases. The affected ranges for each local tsunami source are proposed.(2) Improve the key PTHA method. The exceedance probabilities of the different tsunami waves in three sites in Dayawan are calculated by using the traditional PTHA method. It can be found there still some shortages in traditional PTHA method: the uncontinuity of the discrete magnitude interval can induce the discrete distributions of tsunami waves; there is no point considering the space randomness where earthquakes occur in potential tsunami sources; the uncertainties are not considered reasonably. In order to solve these problems above, this thesis propose a improved PTHA method based on Monte Carlo technique. The magnitude and the location of epicenter in potential sources are took random samples for many times instead of magnitude discretization step in traditional PTHA method. Finally, use new method to calculate the exceedance probability in the same sites for comparing. It suggests that the new method can effectively improvethe shortages of traditional method.(3) Analysis the sensitivities of input parameters in PTHA. We study the sensitivity of bathymetry data to the numerical simulation of earthquake tsunami in the South China Sea. These data sourced from SRTM, ETOPO, GEBCO has been used commonly in public. We take the Manila subduction zone as the potential tsunami sources. The results show that the differences between different kinds of bathymetry data have a negligible influence on tsunami numerical simulation; the errors of bathymetry data themselves do not affect the simulated wave amplitudes, but obviously affect the wave phases. It can be concluded that the bathymetry data has slightly influence on the tsunami simulation while large-grid(low-resolution) bathymetry data is used in numerical simulation. The present opening bathymetry data may satisfy the demand on numerical simulation of tsunami propagation in open seas.(4) Compile the tsunami hazard maps along coastal areas in China. The PTHA analysis in 1480 sites along southeast Chinese coasts is took by using the improved method based on Monte Carlo technique. Totally, there are 640 scenario tsunamis simulated in 2 potential region tsunami sources and 8 potential local tsunami sources. According to the simulated results, a series of maps associated with the exceedance probability curves for 1m, 2m, 3m, 4m and 5m in 1 year, 10 years, 50 years and 100 years are compiled respectively. And a map associated with the return period for 1m, 2m, 3m, 4m and 5m waves is also compiled.