Suppressing migration noise in reverse time migration of vertical seismic profiles by multiple stacking estimation

Abstract

Reverse time migration (RTM) has advantages of imaging steep dip structures but may inevitably produce migration swings when applied to vertical seismic profiles (VSPs). These artifacts often are not suppressed by stacking over shots. The noise is mainly generated from the limited coverage of VSP acquisition geometry, which leads to uneven superposition. The noise is processed by the imaging condition and then overlaps with the real layers, generating artificial images with different dip angles that are difficult to distinguish. The difference is that the real layers can be enhanced by the redundancy of multifold data, whereas the noise is usually generated by a smaller subset of the data. A VSP RTM processing flow is developed to suppress this type of noise. This method constructs the dip angle gather and estimates the redundancy of the coverage in the dip domain. In the stratigraphic dip domain, the false structure and the real structure are located at different dip angles. The real dip is determined by identifying the center of the stacking number. Then, the noise is suppressed by attenuating the imaging information of the other dip angles. The efficient wavefield decomposition method based on Hilbert transform-based instantaneous wavenumber method is adopted to determine the wave propagation direction. Compared with the Poynting vector method and the instantaneous wavenumber method, this approach alleviates the influence of multiwave arrivals and is suitable for complex media, ensuring the accuracy of image decomposition. The whole process is data-driven and does not require prior information. Theoretical and field examples indicate that this method can significantly reduce migration noise and improve the image quality of VSP RTM.