Title

Digital Hum Filtering

Abstract

Data may be overprinted by a steady-state cyclical noise (hum). Steady-state indicates that the noise is invariant with time; its attributes, frequency, amplitude, and phase, do not change with time. Hum recorded on seismic data usually is powerline noise and associated higher harmonics; leakage from full-waveform rectified cathodic protection devices that contain the odd higher harmonics of powerline frequencies; or vibrational noise from mechanical devices. The fundamental frequency of powerline hum may be removed during data acquisition with the use of notch filters. Unfortunately, notch filters do not discriminate signal and noise, attenuating both. They also distort adjacent frequencies by phase shifting. Finally, they attenuate only the fundamental mode of the powerline noise; higher harmonics and frequencies other than that of powerlines are not removed. Digital notch filters, applied during processing, have many of the same problems as analog filters applied in the field. The method described here removes hum of a particular frequency. Hum attributes are measured by discrete Fourier analysis, and the hum is canceled from the data by subtraction. Errors are slight and the result of the presence of (random) noise in the window or asynchrony of the hum and data sampling. Error is minimized by increasing window size or by resampling to a finer interval. Errors affect the degree of hum attenuation, not the signal. The residual is steady-state hum of the same frequency.

Department(s)

Geosciences and Geological and Petroleum Engineering

Keywords and Phrases

Digital Signal Processing; Fourier Analysis; Noise Attenuation; Notch Filtering; Powerline Leakage; Reflection Seismic Data Processing; Reflection Seismology; Attenuation; Data Processing; Signal Filtering And Prediction; Spurious Signal Noise; Hum

International Standard Serial Number (ISSN)

983004

Document Type

Article - Journal

Document Version

Citation

File Type

text

Language(s)

English

Rights

© 1994 Elsevier Limited, All rights reserved.

Share

 
COinS