Automatic Noise Removal from Seismic Data using Time-Frequency Analyses

S. Mostafa Mousavi, Stanford University

Wednesday, June 21, 2017 at 10:30 AM

Location:
Building 3, Rambo Auditorium

Seismic data recorded by surface arrays are often contaminated by unwanted noise. In many conventional seismic methods, the reliability of the seismic data and accuracy of parameter extraction, such as onset time, polarity, and amplitude, are directly affected by the background noise level. As a result, the accuracy of event location and other attributes derived from seismic traces are also influenced by the noise content. Therefore, there is a great need for developing suitable procedures that improve signal-to-noise ratios allowing for robust seismic processing. In this presentation, I introduce four different methods for automatic denoising of seismic data. These methods are based on the time-frequency thresholding approach. The efficiency and performance of the thresholding-based method for seismic data have been improved significantly. Proposed methods are automatic and data driven in the sense that all the filter parameters for denoising are dynamically adjusted to the characteristics of the signal and noise. These algorithms are applied to single channel data analysis and do not require large arrays of seismometers or coherency of arrivals across an array. Hence, they can be applied to every type of seismic data and can be combined with other array based methods. Results show these methods can improve detection of small magnitude events and accuracy of arrival time picking.

Closed captions are typically available a few days after the seminar. To turn them on, press the ‘CC’ button on the video player. For older seminars that don’t have closed captions, please email us, and we will do our best to accommodate your request.

Video Podcast