2017 Volume 70 Pages 42-55
Measurements of spontaneous electrical potential (or self-potential, SP) for mineral exploration have been conducted on land. More recently, for exploration of hydrothermal ore deposits (by observing marine electric fields), underwater SP surveys are also conducted. Observed electric field data have smaller amplitudes than those obtained on land. Therefore, the noise must be removed to emphasize self-potential signals in the observed data. However, conventional "stacking" techniques cannot eliminate coherent noise which has a coherent component among the parallel observed data. This paper describes the use of a proposed noise removal technique using independent component analysis (ICA), which can decompose multi-component mixed signals into independent signals.
We tested the application of ICA to electric field data measured using deep-towed self-potential exploration. The data were obtained far from hydrothermal active areas and ore deposits. Results show that we extracted and removed noise that is difficult to extract by conventional stacking. Then we found no large marine spontaneous electric field, and inferred the causes of noise. Subsequently, to the observed electric field data we added a hypothetical marine electric signal from ore deposits to simulate the SP field. After applying ICA to the artificial electric field data we identified and eliminated the noise The hypothetical signal was extracted more clearly than when using conventional stacking.
Based on successful application to the test data, we applied ICA to electric field data obtained by an autonomous underwater vehicle cruising in a hydrothermally active area. Results show that we extracted the marine electric field deriving from sub-seafloor sources. Such extraction was not possible using conventional stacking This report also presents discussion of the causes of noise.
