Publisher's Synopsis
In the last two decades a lot of effort has been directed towards methods that have the potential to succeed in imaging complex 3D subsurface structures from multi-coverage seismic data. The necessity of an estimate for the wave propagation velocities, required for transforming the data from the time domain to the depth domain, poses one of the fundamental problems in seismic imaging. Inadequate velocity models distort the final depth image. So-called data-oriented approaches are a class of imaging methods that avoid the explicit parameterisation of a velocity model in the first imaging steps. Instead, the data-oriented approaches parameterise the reflection events in the time domain and try to obtain as much information as possible from the measured data. The extracted information is then used to transform the seismic data into depth. The common-reflection-surface (CRS) stack is one of the data-oriented imaging approaches. This method makes use of second-order traveltime approximations in order to describe seismic reflection events in the time domain. For the processing of data from a 3D acquisition, the traveltime equations can be used as stacking operators to simulate a zero-offset (ZO) volume of high accuracy and high signal-to-noise ratio from multi-coverage prestack data. During the stack, reflection energy from the entire five-dimensional data hyper-volume enters into the construction of one ZO sample. The eight parameters, which express the traveltime approximation for the ZO case, relate to kinematic wavefield attributes. These locally describe the propagation directions and curvatures of specific wavefronts at the Earth's surface which have travelled through the subsurface. Thus, the kinematic wavefield attributes constitute integral quantities of the medium's parameters and are suitable to estimate the properties of the Earth's interior. The accurate determination of the wavefield attributes is, therefore, a crucial step in the CRS processing. In this thesis the derivation of the traveltime approximations is presented. The kinematic wavefield attributes are introduced by means of concepts known from geometrical optics. The determination of the eight kinematic wavefield attributes for the ZO case from 3D multi-coverage seismic data is elaborated. The applications of the attributes to support and facilitate 3D seismic imaging are discussed. In this context emphasis is put on the utilisation of the kinematic wavefield attributes for the 3D CRS stack. The proposed search algorithms are validated on a synthetic data example and have shown to be successful. Finally, the 3D CRS stack is applied to a real marine dataset. In this way the functionality of the search algorithms on complex data is verified. Moreover, the imaging quality of the 3D CRS stack is checked by migrating the simulated ZO volume to depth and comparing the obtained result with the result from a prestack depth migration. The comparison shows that the CRS based result is competitive to the result of the prestack depth migration. Thus, CRS based imaging is an alternative to prestack depth migration due to the good imaging quality and also due to the provided information in form of the kinematic wavefield attributes.
