Seismic exploration is a continuous innovation domain since more than one century. A significant part of the studies consists in separating the various waves propagating in the medium, especially surface waves. In the near-surface, surface waves are useful for tomography. Near-surface imaging becomes possible if they are well modelised. When exploration is dedicated to depth – meaning more than 95% of the seismic exploration business – the surface waves mainly hide body waves, which contains the informations related to the depth. Body and surface wave separation then becomes a fundamental task. In these situations, the surface waves can nevertheless be used to better know the near surface. It allows computing parameters usable to better the depth imaging. Research knew recent developments in this domain due to the recent impulsion given by the passive seismic imaging from ambient noise and the study of new acquisition designs with high spatial density. In parallel, the oil fields study for better exploitation is growing as a new industrial development axis. 4D (i.e. 3 spatial dilensions + time) imaging mastering becomes a key research activity, in which sub-surface parameters are estimated and monitored. This PhD thesis comes from the following remarks: - Despite rich works, surface waves are still an important research issue in seismic exploration. - Laboratory scale experiments know relatively few investigations, especially for high density acquisition design. The first step has been dedicated to the set up and the validation of a complete acquisition environment in the laboratory, adapted to surface wave study and high spatial density. Using Agar-agar phantoms, a mix of S body waves and Rayleigh surface waves comparable to the on-field P body waves and Rayleigh wave mix has been highlighted. Then, using array processing, wave separation has benne successfully demonstrated. After waves separation, it becomes possible to follow their arrival time variation in presence of surface and/or depth variation in the medium, as in reservoir monitoring conditions. A complete 4D study has been performed, allowing not only the arrival time monitoring but also amplitude and arrival and launch directions. A method has been proposed to compensate the near-surface spurious variations. An adaptation of the method on a field data set is then performed. Generally, velocity profiles on the field show weak velocities in the sub-surface. As a consequence, the various waves coming from the depth have weak and comparable incidences angles. Classical separation method using array processing are usually insufficient to work with such incidence angles set. For this reason, a complete part of this work has been dedicated to the study of high resolution algorithms in the frame of seismic exploration and their adaptation. At the end, taking advantage of the high spatial density allowed by the laboratory environment, a comparative study of two designs – the first one theoretically ideal but somewhat unrealistic and the second one more viable economically but less efficient – has been performed to address the scattered waves filtering issue. For the second design, a new filtering method has been proposed to enhance the scattered waves filtering.
