Analyses and Applications of Seismic Surface Waves Using Machine Learning Algorithm

Date

2018-12

Journal Title

Journal ISSN

Volume Title

Publisher

Abstract

Surface waves are energetic signal components in the seismic inversion for near-surface elastic properties. However, they are noise components in reflection seismology in hydrocarbon exploration. Detecting and outlining the region of surface waves in seismic gathers facilitates either their inversion for sub-surface structures or their noise removal. I propose an unsupervised machine-learning algorithm using K-means clustering to automatically identify surface waves in the raw seismic data according to their common features, including low frequency, low velocity, and high amplitude when compared to body waves. The local attributes of frequency, amplitude, and velocity, which are obtained by the Gabor frequency and structure tensor, are used in K-means analyses to detect and outline surface waves on seismic shot gathers. The performance of this new algorithm is evaluated on three datasets, including two synthetic datasets with random noise and missing traces, and a field dataset. In these examples, the proposed algorithm is stable and can successfully outline the surface waves. Then I propose to integrate this approach into a procedure to automatically suppress and/or invert surface waves from raw seismic shot gathers. In the first stage, K-means analysis detects and outlines surface waves on raw shot gathers by using the seismic properties of local frequency, amplitude, and velocity. Then I calculate a high-resolution dispersion image from the outlined surface waves via nonlinear signal comparison. From the high-resolution image, the dispersion curves are obtained by an auto-picking algorithm. The auto-picked dispersion curves are used for both surface-wave suppression and inversion. v Surface waves are predicted and then subtracted from the raw shot gather to enhance the signal-to-noise ratio of body waves. Moreover, the picked dispersion curves can be used to invert for the shallow S-wave velocity model. We applied our approach on a ten-layered synthetic shot gather and field data. The results indicate that our approach is capable to automatically suppress and invert surface waves on raw seismic shot gathers.

Description

Keywords

Machine learning, Surface Wave Detection, Surface Wave Suppression, Surface Wave Inversion

Citation