Noise subtraction from KAGRA O3GK data using Independent Component Analysis

Abstract During April 7–21 2020, KAGRA conducted its first scientific observation in conjunction with the GEO600 detector. The dominant noise sources during this run were found to be suspension control noise in the low-frequency range and acoustic noise in the mid-frequency range. In this study, we show that their contributions in the observational data can be reduced by a signal processing method called independent component analysis (ICA). The model of ICA is extended from that studied in the initial KAGRA data analysis to account for frequency dependence, while the linearity and stationarity of the coupling between the interferometer and the noise sources are still assumed. We identify optimal witness sensors in the application of ICA, leading to successful mitigation of these two dominant contributions. We also analyze the stability of the transfer functions for the entire two weeks of data to investigate the applicability of the proposed subtraction method in gravitational wave searches.