Abstract
The principles of tomography are well-established, and geotomography has been in use for several decades. Seismic tomography has previously been used principally in the oil and nuclear industries, but it has occasionally been applied in civil engineering site investigation (SI). The technique can provide a qualitative or a quantitative representation of the distribution of the seismic velocity properties within a surveyed region. Seismic tomography has been successfully used to image, for example, zones of fracturing. The technique has been less successful when used to detect cavities. This thesis presents a theoretical and experimental examination of the usefulness - or otherwise - of seismic tomography in SI. It is found that geotomography has a role to play when one is interested in, for example, general subsurface variability. An explanation is offered for the low success rate of cavity detection surveys utilizing seismic velocity tomography. It is shown that, of the numerous and diverse fields of application of the tomographic method, geotomography offers what is, perhaps, the most inherently ill-posed form of the reconstruction problem. Significant difficulties are encountered because, in general, measurements can be acquired from only a restricted set of positions. Additional problems are introduced through the use of seismic wave energy, which follows hard-to-predict routes across the surveyed zone. Observational errors lead to inconsistencies within a tomographic system. In consequence, the geotomographic reconstruction problem is typically under-determined, inconsistent and non-linear. An heuristic method of reconstructing seismic velocity tomograms from typical field data is developed and presented.