http://chineseinput.net/에서 pinyin(병음)방식으로 중국어를 변환할 수 있습니다.
변환된 중국어를 복사하여 사용하시면 됩니다.
Estimation of Change in Global Mean Temperature Based on Machine/Deep Learning of Global Data
Desy Caesary,Soo Jin Jang(장수진),Seo Young Song(송서영),Myung Jin Nam(남명진) 한국신재생에너지학회 2021 한국신재생에너지학회 학술대회논문집 Vol.2021 No.7
Continuous increase in anthropogenic CO<SUB>2</SUB> emission can cause rise in global mean temperature (GMT). Many studies have been made to build climate models for the estimation of change in GMT based on several climate factors such as greenhouse gas (GHG) emissions and net ecosystem exchange (NEE). However, the complexity of the climate factors possibly causes uncertainties in climate models leading to an inaccurate estimation of GMT. To overcome the uncertainties, this study applies machine and deep learning (M/DL) methods to the analysis on the relationship links between climate factors and changes in GMT. The pattern learning ability supports more objective analysis on climate models. The input data of climate factors include CO<SUB>2</SUB> and non-CO<SUB>2</SUB> concentration, and NEE global data during 1999~2012, which are currently available. The available data were mostly annual, thus were interpolated to generate more data for the training process of DL. The open-source library of support vector machine (SVM) and artificial neural network (ANN) methods were used. A better-trained model was observed from the ANN with more interpolated data, whereas SVM had difficulty in finding optimal prediction parameters due to the high non-linearity input data.
심층 학습을 이용한 물리탐사 자료 잡음 제거 기술 소개
Caesary Desy,조아현,유희은,정인석,송서영,조성오,김빛나래,남명진 한국지구물리.물리탐사학회 2020 지구물리와 물리탐사 Vol.23 No.3
Noises can distort acquired geophysical data, leading to their misinterpretation. Potential noises sources include anthropogenic activity, natural phenomena, and instrument noises. Conventional denoising methods such as wavelet transform and filtering techniques, are based on subjective human investigation, which is computationally inefficient and time-consuming. Recently, many researchers attempted to implement neural networks to efficiently remove noise from geophysical data. This study aims to review and analyze different types of neural networks, such as artificial neural networks, convolutional neural networks, autoencoders, residual networks, and wavelet neural networks, which are implemented to remove different types of noises including seismic, transient electromagnetic, groundpenetrating radar, and magnetotelluric surveys. The review analyzes and summarizes the key challenges in the removal of noise from geophysical data using neural network, while proposes and explains solutions to the challenges. The analysis support that the advancement in neural networks can be powerful denoising tools for geophysical data.
Geophysical Monitoring on Controlled CO<sub>2</sub> Released Field Experiments: a Review Study
( Desy Caesary ),( Seo Young Song ),( Huieun Yu ),( Myung Jin Nam ) 대한지질공학회 2019 대한지질공학회 학술발표회논문집 Vol.2019 No.2
Geologically sequestrated carbon dioxide (CO<sub>2</sub>) in deep subsurface can leak through leaky well or fault, thus the storage fails. Leaked CO<sub>2</sub> contaminates groundwater when reaching to the atmosphere. Many field experiments of releasing CO<sub>2</sub> into shallow depth in test bed sites, have been made not only to understand impacts of CO<sub>2</sub> in shallow groundwater or atmosphere but also to test monitoring methods for detecting the leaked CO<sub>2</sub>. Among various monitoring methods, geophysical methods have been widely utilized to detect CO<sub>2</sub> plume migration in shallow depth before CO<sub>2</sub>’s reaching to the atmosphere. This study reviews and analyzes geophysical monitoring experiments and results in seven existing field experiment sites. In several fields, geophysical measurements were carried out in time-lapse manners to monitor changes in physical properties of subsurface due to the presence of CO<sub>2</sub> before and during injection, while a couple of fields conducted CO<sub>2</sub> monitoring even after injection has been finished. After the analysis, this study not only summarizes changes in subsurface physical properties such as bulk electrical resistivity, complex resistivity and permittivity, but also analyzes effects of geological conditions on the changes in physical properties; whether the CO<sub>2</sub> injection zone is saturated or vadose zone, contains calcite minerals or clay, precipitation rate, etc. Further, this paper will also introduce tests of geophysical monitoring for shallow CO<sub>2</sub> injection experiment in Korea. This research was supported by KEITI (Project Number: 2018001810002), and partly by KETEP (No. 20194010201920).