Near-surface wavefield perturbations can be very complex and completely mask target reflections. Despite this complexity, conventional methods to compensate for these perturbations rely on simplified parameterizations and mainly focus on time shift corrections. We present a fully data-driven procedure which uses finite-impulse response filters to describe near-surface wavefield perturbations, hence allowing time shifts to vary with frequency. These filters are obtained using blind channel identification. Subsequently, the recorded data are compensated for the near-surface wavefield distortions using multichannel deconvolution. We demonstrate the applicability of this procedure both on synthetic and field data. The experiments show that the procedure successfully removes near-surface wavefield distortions including scattered ground roll and delineates reflection events which are difficult to detect prior to our proposed correction.
展开▼
