Simultaneous seismic random noise attenuation and signal preservation by optimal spatiotemporal TFPF

•The Shapiro–Francia (S–F) test statistic is used to identify seismic events and construct optimal filtering trajectories.•We propose an optimal spatiotemporal time–frequency peak filtering method to attenuate random noise and preserve seismic signals simultaneously.•The OST–TFPF reduces bias of seismic signal estimate by reinforcing morphology characteristics of seismic events using S-F.

The time–frequency peak filtering (TFPF) algorithm has been successfully applied to seismic random noise attenuation. However, the time–frequency peak filtering with fixed-type spatiotemporal filtering trajectories fails to preserve reflected signals in seismic events which have complex geometric structure. An optimal spatiotemporal TFPF (OST–TFPF) is proposed here combining the Shapiro–Francia (S–F) statistic to reduce random noise and preserve seismic signals simultaneously. In the novel algorithm, the S–F statistic is first calculated for seismic data to detect seismic events based on the fact that the non-Gaussian seismic signals lead to smaller values of the S–F statistic comparing to seismic random noise which is general Gaussian. Then, optimal spatiotemporal filtering trajectory can be constructed based on the S–F statistic to coincide with the shape of each event. Finally, the optimal spatiotemporal TFPF de-noises seismic data along the optimal trajectories. Since the resampled signals along the trajectories matching the geometric structures of seismic events become more linear compared to signals in time, the OST–TFPF gives better signal estimation while attenuating random noise. Synthetic and field data examples demonstrate that the optimal spatiotemporal TFPF is effective in the denoising and signal-preserving of the seismic data with low signal-to-noise ratio. Moreover, the OST–TFPF also obtains good performance in preservation of seismic event with complex geometric structure.