3D distributed and dispersed source array acquisition and data processing

Abstract

Numerous field acquisition examples and case studies have demonstrated the importance of recording, processing, and interpreting broadband land data. In most seismic acquisition surveys, three main objectives should be considered: (1) dense spatial source and receiver locations to achieve optimum subsurface illumination and wavefield sampling; (2) coverage of the full frequency spectrum, i.e., broadband acquisition; and (3) cost efficiency. Consequently, an effort has been made to improve the manufacturing of seismic vibratory sources by providing the ability to emit both lower (approximately 1.5 Hz) and higher frequencies (approximately 120 Hz) and of receivers by utilizing single, denser, and lighter digital sensors. All these developments achieve both operational (i.e., weight, optimized power consumption) and geophysical benefits (i.e., amplitude and phase response, vector fidelity, tilt detection). As part of the effort to reduce the acquisition cycle time, increase productivity, and improve seismic imaging and resolution while optimizing costs, a novel seismic acquisition survey was conducted employing 24 vibrators generating two different types of sweeps in a 3D unconstrained decentralized and dispersed source array field configuration. During this novel blended acquisition design, the crew reached a maximum of 65,000 vibrator points during 24 hours of continuous recording, which represents significantly higher productivity than a conventional seismic crew operating in the same area using a nonblended centralized source mode. Applying novel and newly developed deblending algorithms, high-resolution images were obtained. In addition, two data sets (i.e., low-frequency and medium-high-frequency sources) were merged to obtain full-bandwidth broadband seismic images. Data comparisons between the distributed blended and nonblended conventional surveys, acquired by the same crew during the same time over the same area, showed that the two data sets are very similar in the poststack and prestack domains.