Full text loading...
-
Slope Estimation by Convolutional Neural Networks
- Publisher: European Association of Geoscientists & Engineers
- Source: Conference Proceedings, 82nd EAGE Annual Conference & Exhibition, Oct 2021, Volume 2021, p.1 - 5
Abstract
Local dip field has have been widely used in geophysical applications, such as structure prediction, seislet transform, trace interpolation and denoise. The plane-wave destruction (PWD) is the common method to estimate the local slope. However, the PWD is sensitive to strong noise. It is not easy to estimate an accurately local slope from noisy data by PWD algorithm. To estimate an accurate slope from noisy seismic data, we have proposed an architecture based on deep learning (DL). The architecture contains two sections: the convolutional and deconvolutional sections. The conventional section can learn the local features and the deconvolutional section constructs the output using the learned feature to match the target. Numerical tests on two examples demonstrate that the proposed method can obtain a relatively accurate dip field from noisy data.