Improved Near-surface Velocity Models from Waveform Tomography Applied to Vibroseis MCS Reflection Data

Multichannel vibroseis reflection surveys are prevalent in the land exploration seismic industry because of benefits in speed and cost, along with reduced environmental impact when compared to explosive sources. Since the downgoing energy must travel through the shallow subsurface, an improved model of near-surface velocity can in theory substantially improve the resolution of deeper reflections. We describe techniques aimed at allowing the use of vibroseis data for long-offset refraction processing of first-arrival traveltimes and waveforms. Refraction processing of surface vibroseis data is typically limited to near-offset refraction statics. Velocity models of the shallow subsurface can be built to facilitate CDP stacking and migration, but these models are typically coarse and of limited use for interpretation. Waveform tomography combines inversion of first-arrival traveltime data with full waveform inversion of densely-sampled refracted arrivals. Since inversion of the waveform amplitude and phase is not limited by the ray-theory approximation, identification of low-velocity zones and small scattering targets is possible. Incorporating a wide range of offsets is critical for a more complete characterization of the near-surface. Because of the use of a non-linear frequency-domain approach to the solution of this inverse problem, low data frequencies are important in comparison with conventional reflection processing. Through the use of waveform tomography, we plan to build useful, detailed near-surface velocity models for both the reflection work flow and direct interpretation. Several difficulties exist in first-arrival analysis and waveform inversion of vibroseis data. The mixed-phase vibroseis source signature exhibits variations in phase with offset that are difficult to quantify without detailed a priori knowledge of the near-surface. This causes difficulties with picking and initial model building, which is critical for non-linear waveform inversion. A sufficiently accurate starting model must be provided to allow convergence to an accurate final model. The Q-filter and deconvolution effects are theoretically accounted for in the waveform inversion process, once a starting model of sufficient quality is realized. To make this possible, preprocessing for waveform inversion is also necessary. It is designed to allow the use of the 2D, acoustic approximation to the wave equation in the waveform inversion implementation. The use of a 2D approximation to the true 3D geometry introduces AVO (Amplitude Variation with Offset) errors that must be accounted for in order for attenuation inversion to be possible. The acoustic approximation means that elastic propagation modes and mode-converted arrivals must be considered as systematic noise, with appropriate preprocessing steps to reduce their effects. Careful analysis of the early-arriving waveforms is necessary to deal with approximations due to the waveform inversion implementation, which are not easily separable from the approximations implicit in vibroseis acquisition. However, the potential benefits in near-surface velocity characterization and their wide applicability make the results of this research important for seismic processing and near-surface geological interpretation.