Research on Active Obstacle Avoidance in Seismic Acquisition Using Compressed Sensing and Deep Learning Reconstruction Technology

The rapid development of seismic exploration technology has led to the increasing popularity of wide azimuth, full-azimuth, and high-density 3D seismic acquisition techniques. However, the presence of diverse and large obstacles in the seismic working area, such as buildings, roads, protected areas, and lakes that prohibit or pose challenges to conducting seismic surveys, significantly impacts the quality of imaging during exploration activities. While conventional obstacle avoidance methods can partially enhance seismic acquisition near obstacles, their effectiveness is limited when dealing with large-area obstacles due to the high density of shot points near them. In this study, we propose an active obstacle avoidance optimization method based on compressed sensing theory to improve shot point placement near obstacles while considering the boundary constraints. This approach enhances imaging in obstructed areas and introduces a data reconstruction technique using conditional generative adversarial network models to enhance data reconstruction in obstacle areas within prestack CMP trace sets. Through simulation data and field experiments, we successfully demonstrate the effectiveness of this method in significantly improving seismic data imaging in obstacle areas. It can be widely applied to optimize shot point and receiver point designs during seismic acquisition in obstacle areas as well as other sampling scenes limited by obstacles.