Abstract:
Diffractions are seismic events which are a common feature in seismic profiles, generated by a radial scattering of the incident wave at discontinuities in the subsurface such as faults, fractures, channels and rough surfaces. These seismic events are generally undesirable artefacts in conventional seismic processing sequences, however, they contain important information about the velocity of the medium which can be extracted. Hence, diffractions have been exploited for velocity analysis and indication of hydrocarbon traps among other applications. The Hikurangi Margin on the east coast of New Zealand is thought to contain large amounts of gas hydrate deposits, with a particular interest in the Tuaheni Landslide Complex (TLC) and the low velocity creep deformation that the landslide is displaying. The presence of gas hydrates is what is proposed to be the driving force behind this type of deformation in a submarine environment. The landslide debris generates numerous diffractions in the seismic data, making it an ideal opportunity for carrying out velocity analysis. A small 3-D dataset, acquired from high resolution short streamer dataset from the TLC region is considered in the development of Diffraction Velocity Analysis (DVA), Migration Velocity Analysis (MVA) and an unconventional Pseudo CDP-NMO method to analyse diffractions. This dataset was acquired using a P-cable system that does not allow for standard velocity analysis typically derived from normal moveout, because the offsets generated from the survey geometry are too short. Shorter offsets are more common in Ground Penetrating Radar (GPR) surveys, hence, similar methodologies are adopted, to analyse diffractions. These approaches are rarely applied to 3D seismic data. Several properties of diffractions such as their trajectory and its response to velocity inputs and their collapse after migration, are used in order to extract velocity information. The aim is to produce methods that provide realistic velocity estimates, and results are promising across all three methods.
