203 Synthesis image quality that should be given directly by the residuals between field data and seismograms computed with the elastic images. The last example presented in this chapter is based on data from a one-shot seismic survey in the Paris basin. This study includes processed acoustic and multi-component data collected at two different wells with inter-well distances of about 100 m. Only upgoing S-S and S-P reflected events were used for the tomography. The target zone included three sand reservoir levels between depths of 575 and 600 m. Due to insufficient source and receiver coverage of the target zone (because it was only a one-shot survey), the diffraction tomography produced unreliable images with strong artifacts on the upper section, i.e. above 560 m. However, tomography proved capable of producing high-resolution (≈ 3-5 m) images for the reservoir region. The comparison of both density images with a pseudo-density log, produced by a density log convolved with a characteristic signal with the same bandwidth as the density image, is quite satisfactory within the reservoir region. In conclusion, based on the good results obtained from the field studies, it can be said that seismic tomography has the potential to provide superior images that are capable of addressing the problem of near-surface structure characterization. Chapter 4 is dedicated to near-surface reflection surveying. After a short review of the design of conventional 2D and 3D surveys, we briefly summarize the main steps of a processing sequence. With two field datasets, we showed that it is possible to obtain very high-resolution 3D blocks for near-surface applications with very light seismic spreads (48 channel recorders, a single geophone per trace, and a light seismic source). Near-surface studies require specific test phases to define the optimum parameters (minimum offset, geophone interval) for the acquisition. The processing sequence must be carefully adjusted to the field data, especially for the wave separation. In the example of the imaging of the near-surface karstic reservoir (Hydrogeological Experimental Site of Poitiers), we showed that the velocity distribution obtained by refraction tomography in the first 30 m can be merged with the velocities extracted from the amplitude of the reflected events to obtain a continuous velocity model from the surface to a depth of 120 m. Chapter 5 discusses the huge potential of Full Waveform Inversion (FWI) in terms of quantitative seismic image interpretation. In practice, the applicability of the method depends on the quality of the data, as well as on the most appropriate pre-processing (for example to preserve the low frequency data) and on the correct physical understanding of the wave propagation phenomena. In future elastic FWI will replace the acoustic approach, and the technique will be able to extract more than a single parameter (e.g. velocity and attenuation). Also, it may be possible to incorporate higher frequencies. On the exploration scale, FWI is still in its infancy. We hope that this chapter will positively encourage the reader to evaluate for themselves the use of FWI on near-subsurface data sets. Chapter 6 discusses handling different types of waves present in the same set of experimental data. We have underlined some of the advantages of hybrid seismic imaging strategies to provide efficient, accurate and reliable subsurface models, in
RkJQdWJsaXNoZXIy NjA3NzQ=