Development of noise reduction method for MT data processing using Multi-channel singular spectrum analysis at Boso Peninsula in Japan

Abstract

. In 2014-2015, we conducted MT survey to understand the subsurface electrical structure at Boso peninsula. The obtained data is found to be contaminated by artificial noise (e.g. leak current from DC driven trains, factory noise, etc.). The artificial noise leads to the false structure. To reduce the noise effects, we applied Multi-channel Singular Spectrum Analysis (MSSA). If the window length set 800, we could reduce uncorrelated signal in a certain frequency band. Further investigation on MSSA parameters or additional information on source waves for MT exploration is necessary to eliminate the noise over all frequency bands.