Source separation of seismological signals using blind deconvolution and independent component analysis techniques

Abstract

Blind source separation and independent component analysis techniques have been implemented in many fields to analyze signals, in order to reveal the hidden factors and components that underlie. Analyzing some basic problems in seismology, it is seen that these techniques can play an important role in analyzing field data. This work aims to apply current natural gradient signal processing techniques on seismological signals to separate independent components and separate the source and medium effects present in a seismological observation. Two algorithms are implemented for this purpose with a small modification and corrections proposed to one. The results obtained indicate that instantaneous independent component analysis methods can be employed to preprocess seismological observations to separate the effects of remote and local seismic events. In addition, single channel blind deconvolution techniques may be used to separate the effects of propagation medium from the seismic source.