Год издания: 2020

Land seismic data are often quite challenging for reliable seismic imaging. The mechanism responsible for difficulties in data processing may be identified as severe frequency-dependent phase and amplitude perturbations making events incomprehensible in multi-channel records. We present a new approach combining nonlinear beamforming and time-frequency masking to compensate for such effects in two steps. First, we construct a crude signal "guide" using data-driven nonlinear beamforming. Second, we employ trace-by-trace time-frequency masking to repair damaged phase and amplitude. Phase masks are of paramount importance to heal the wavefield and make desired events coherent and trackable. Amplitude masks aim to suppress noise from the amplitude spectra. Introduced methods avoid smearing of amplitude information across channels and preserve frequency bandwidth of desired signals.