Near-Offset Noise Characterization and Attenuation Using Orthogonal Matching Pursuit

Near-offset noise in land seismic data (i.e., less than 300 m) can be very challenging to deal with as it is dominated by high-amplitude surface waves and acoustic noise from vibrators. In a seismic data set acquired to test a novel 5C sensor, we observed a high-frequency coherent noise mode (up to 80 Hz) dominating the near offsets. This mode was very energetic and aliased at frequencies above 20 Hz. We formulated a cost function that enabled picking the noise mode dispersion curve by combining the vertical component and the gradient measurement provided by the 5C sensor. An iterative scheme known as the Orthogonal Matching Pursuit was implemented to invert for a sparse set of coefficients describing the noise mode.

In addition to the phase velocity retrieved from the dispersion curve, we also estimated the horizontal-over-vertical (H/V) amplitude ratio and the horizontal amplitude decay with depth of the noise mode. These attributes can potentially be combined to invert for near-surface properties.