Seismic Characteristics Of Gas Hydrates In East China Sea And Study Of Full Waveform Inversion

The most gas hydrates and free gases in the marine strata were identified and predicted by using the seismic data for quantitative analysis over the past decades. Most of these quantitative studies relied on the P wave seismic velocities and the seismic reflection amplitude variation associated with the presence of the gas hydrates and free gases. Additional information was contained in the detailed waveforms. As a result, multi-attribute analysis of prestack seismic data for identification of gas hydrates was developed.Many mud diapirs had been recognized in southern Okinawa Trough by a multi-channel seismic 973 surveying on R/V KEXUE I in 2001. Gas hydrates had been found in the layers immediately above mud diapirs and in the host sediment. Formation of mud diapirs on the seismic line DMS01-5 and their seismic characteristics associated with gas hydrates were studied via stacking velocity analysis, heat flow calculation from the bottom simulating reflector (BSR), fluid potential analysis, density and interval velocity measurements, and impedance inversion. It has been found that the BSRs are not always be indicative of the base of the gas hydrate stability zone (BGHSZ), which could indicate the base of the saturated gas hydrate formation zone (BSGHFZ) and the top of the free-gas zone (TFGZ) in different mud-diapir-associated strata.In this thesis a prestack full waveform inversion using a genetic algorithm was presented to obtain the P and S wave seismic velocities of a horizontally, finely layered elastic medium. The generalized reflection transmission matrix method intruded by Kennett was chosen to calculate a complete seismic response for a stratified elastic half space, which including free-surface waves, mode conversions and multiple reflections. Genetic algorithm was a global optimization algorithm, which worked with a group of models simultaneously and used stochastic processes to guide the search for an optimal solution with little priori constrains, other than the conventional local linearization algorithms which required a good starting model and the derivative information.