Ground roll detection and attenuation by 3Cpolarization analysis

摘要

Ground roll is one of the most troublesome noises of land data and its elimination is often a challenge.rnThree-component (3C) seismic recording offers techniques to suppress it by exploiting its ellipticalrnpolarization nature. We present a flexible method that detects and subsequently removes the groundrnroll using polarization analysis. The polarization analysis is carried out by doing singular valuerndecomposition (SVD) of the matrix formed by a sliding time window (data-adaptive) of every single-stationrn3C recording. We derive a new attribute from the singular values of the SVD to quantify thernellipticity and energy of the recorded events and use it to detect the ground roll. The detected groundrnroll is attenuated from single-station or multi-station data by truncated eigenimages. The single-stationrnfiltering has the advantage of no dependency on the spatial sampling of the data. The economicalrnimpact of this feature is that the survey design may be less constrained by the sampling requirement ofrnground roll removal and thus makes the acquisition more cost-effective. We show that the method canrnreliably recognize and efficiently attenuate the ground roll with a land data example.