Traveltime tomography for land seismic data

Abstract

Full wave form inversion (FWI) uses the full recorded wavefield to determine the physical properties of the subsurface. Due to the lack of wide aperture and low-frequency data, a reliable low-frequency starting velocity model is often required. A velocity model derived from pre-stack depth migration (PSDM) is typically used as the starting model for FWI. Iterative PSDM is an expensive process, and it also requires a velocity macro model. Traveltime tomography methods are efficient algorithms to construct a smooth velocity model using refraction and reflection traveltimes. However, with limitations caused by acquisition constraints, data quality and assumptions used in traveltime tomography methods, the velocity model determined by these methods can be sub-optimal and results in degradation in the depth image. In this thesis, I address several strategies to improve tomography and to incorporate error measurements from refraction and reflection waveforms into the tomographic inversion kernels. My goal is to help to alleviate the inherent limitations in traveltime tomography methods.