| 講演要旨(和文) | | 近年，海底電場観測による熱水鉱床探査が行われている．このような手法は自然電位探査と呼ばれており，古くから陸上の鉱床探査に用いられてきた．その際，測定した電場データにはノイズが混入しており，海底下の構造解析が困難となる場合があり，ノイズを適切に除去し信号を抽出する必要がある．従来は複数の並行観測データを足し合わせる手法（スタッキング）によりノイズの低減がなされてきたが，データ間に相関のあるノイズを除去することは困難である．そこで，多成分の混合信号を独立な信号に分解する独立成分分析（ICA）という手法を適用し，ノイズの抽出および除去を試みた．その結果，ノイズと思われる成分をICAにより抽出および除去することができた．その後，実データに海底下からの電場信号を模した仮想的な信号を与え，ICAによる抽出を試みた所，スタッキングよりも的確に、仮想的な信号の抽出を行うことができた． |
|
|
| | 講演要旨(英文) | | Recently, we are trying to discover buried hydrothermal deposit by mapping the marine electric field. It is so-called the self-potential (SP) survey, which can detect the anomalies due to redox potential frequently observed at land mineral deposits. Noise is normally included in the observed electric field, and often interrupts quantitative estimation of current sources of SP below the seafloor. Removal of noise in the time series data of electric field is essential. However, we cannot eliminate the noise by using the conventional simple stacking method because the SP signal has some correlations with the noise. Here, we propose noise removal with the Independent Component Analysis (ICA), recently used for breaking down multicomponent-mixed signals into some independent signals. We tested effectiveness of ICA through the application to observed data. As a result, we succeeded to extract and remove the noise by using ICA, whose efficiency is better than those analyzed by using simple stacking.. I also added an artificial signal to observed electric field similar to the anomaly due to a sub-seafllor mineral deposit. , We confirmed that the ICA could pick out the signal and eliminate the noise more clearly than the simple stacking. We conclude that ICA can be used as a new tool for the noise reduction in marine observations. |
|
|
|
| |
|