머신러닝을 이용한 탄성파 해양학 자료 잡음 억제
-
Title
- 머신러닝을 이용한 탄성파 해양학 자료 잡음 억제
-
Alternative Title
- Noise attenuation of the Sparker Seismic Oceanography data using Machine learning
-
Author(s)
- 전형구; 주형태; 이상훈; 문혜진; 전청균; 안신혜
- KIOST Author(s)
- Lee, Sang Hoon(이상훈); Moon, Hye Jin(문혜진); Jeon, Chung Kyun(전청균); Ahn, Shin Hye(안신혜)
-
Alternative Author(s)
- 전형구; 주형태; 이상훈; 문혜진; 전청균; 안신혜
-
Publication Year
- 2019-12-12
-
Abstract
- Seismic oceanography (SO) is a method of obtaining the structure and physical properties of ocean by using the seismic exploration and processing. The advantage of the SO is that the data acquired by using SO has higher lateral resolution than the data acquired by using the conventional oceanographic devices. Therefore, the SO has been used to study the distribution of the water mass, the dissipation of the turbulence, and characteristic of the internal waves in many regions. In most SO studies, the seismic data was obtained by using the air-gun, but recently the sparker was also used to generate higher frequency source wavelet. The use of the higher frequency components increases the vertical resolution of the seismic data, which can provide much detail information of the ocean. However, the low signal to noise ratio of the sparker seismic data is one of the biggest obstacles of using sparker source in SO study. The energy of the sparker sourceis much smaller than the energy of the air-gun source, thus the influence of the random noise is severer in sparker seismic data than in the air-gun seismic data. Therefore, the attenuation of the random noise in the sparker seismic data is one of the important issues in SO data processing. In this study, we applied convolutional neural network (CNN) to attenuate the random noise in the sparker seismic data. The Denoising Convolutional Neural Network (DnCNN) which extracts th
-
URI
- https://sciwatch.kiost.ac.kr/handle/2020.kiost/21027
-
Bibliographic Citation
- AGU 2019, 2019
-
Publisher
- American Geophysical Union
-
Type
- Conference
-
Language
- English
- Files in This Item:
-
There are no files associated with this item.
Items in ScienceWatch@KIOST are protected by copyright, with all rights reserved, unless otherwise indicated.