Singular value decomposition (SVD) is primarily used for attenuating marine noise, such as

• crossflow noise recorded during vessel turns
• excessive swell noise
• noise from in-sea devices attached to streamers.

Enhanced flexibility and fit for purpose

Noisy traces are detected by calculating the energy of the trace in a user-specified window. The user defines specific frequencies and energy levels for the tool to target. The user can also specify the energy threshold above which traces are deemed noisy; traces below the threshold are untouched by the process.

SVD enables the application of noise attenuation within specific, tapered time windows. Multiple passes—targeting different frequencies and energy levels—can therefore be run in a time-variant manner. Noise removal is carried out iteratively; the output after each application of filtering forms the input for the next iteration.

Suitability for full-azimuth acquisition data

SVD has proven to be a robust and useful technique in the presence of crossflow noise—a low-frequency, high-amplitude noise mode that results from the movement of water across the streamer. It can be caused by strong rip currents or the intentional towing of the streamer in a circular or nonlinear manner.