Using jittered sampling in designing geometry and imaging in shallow 3D seismic surveys
When the depth of the shallow three-dimensional seismic exploration is less than 100 m, one often encounters very low velocities for the target and high frequencies in the data. Following Nyquist–Shannon sampling theorem, the permissible maximum receiver interval can be smaller in this case compared to relatively deeper seismic exploration. This suggests that there are still issues to be addressed in the design of geometry and in data processing in shallow three-dimensional seismic exploration. This paper addresses these problems by applying the theory of compressed sensing for signal processing to shallow-seismic geometry designing and data processing. Theoretical research shows that random sampling of data can better reconstruct the wavefield than undersampled data. Random sampled data can transform the coherent aliasing to non-coherent noise, which turns the seismic data interpretation problem into a data denoising problem. The jittered random sampling method avoids the situation when the spatial data points of a randomly sampled dataset are too concentrated or too sparse. Our proposed approach was tested on simulated and real seismic data. The results show that if the jittered random undersampling method is used in shallow three-dimensional seismic data acquisition, then a wider range of observation with fewer receivers in the layout is possible. This greatly improves the data collection efficiency in the field. In addition, the random sampling method has more flexibility in the field environment.When using the regular samplingmethod, an open survey area without large obstacles is needed. However, the random sampling method can be adapted to rugged terrains. When obstacles are encountered, the receiver spacing can be increased appropriately. In open areas, the receiver spacing can be decreased to compensate for the reduced data.