Canceling stationary sinusoidal noise

Abstract

Sinusoidal noise often contaminates seismic data. When this noise is large compared to seismic signals, it adversely affects prestack seismic processing and subsequent interpretation. We develop a digital least‐squares filtering algorithm for canceling stationary sinusoidal noise in seismic data. The method effectively cancels sinusoidal noise when the noise is stationary, which is typical for recordings of a few seconds in length. This procedure differs from the usual notch‐filtering techniques because the sinusoidal noise is canceled without notching the signal spectrum. Since the method requires that the line frequency be accurately known, the algorithm can automatically search the trace spectrum to find the exact sinusoidal frequency value needed for filter design. The algorithm is highly automated and requires no input parameters when the interference comes from power lines or generators. We use model and field data to quantify the algorithm’s performance.