Multicomponent Seismic Processing for Coherent Noise Suppression and Arrival Identification

Multicomponent seismic processing provides significant potential advantages compared to conventional scalar wavefield processing. Triaxial sensors which measure three orthogonal components of the ground particle motion are able to determine the wave polarization and can thus record the complete vector wavefield with all of its modes (P, S and surface waves). This fact can be exploited to design polarization filters for wavefield separation and high resolution P- and S-wave imaging techniques. A successful illustration of the benefits of multicomponent processing is only possible if the concepts are extensively tested both on synthetic data and real data. Such tests were performed during the course of this project. The results obtained confirm that vector wavefield processing is indeed a valuable tool in the analysis of seismic data and provides information that supplements the information obtained by conventionally processed vertical component P-wave data. In this thesis, I show how polarization properties can be exploited to determine the direction of arrival or to build filters that suppress coherent or random noise. If moveout characteristics are considered in addition to the polarization information, a separation of the wavefield can be achieved. Also, the polarization information can be incorporated into migration algorithms to obtain separated P- and S-wave images of the subsurface. Several computer scripts were developed to test these concepts. All of them performed well on synthetic data. On real data, a successful reduction of ground roll and a partial separation of the wavefield was achieved. Additionally, P- and S-wave images could be obtained using a multicomponent migration routine. The dataset that is treated in this thesis was provided by the National Cooperative for the Disposal of Radioactive Waste (Nagra). It is a VSP dataset that was recorded in the Benken borehole in Switzerland. Although the results of the analysis probably do not represent the precise geological structure in the region, it is shown that further multicomponent analysis of this dataset could be profitable.Joint Master's in Applied GeophysicsGeoscience & EngineeringCivil Engineering and Geoscience