Noise reduction of seismic signal for Llaima volcano

Abstract

The aim of this paper is to remove the Gaussian Noise present in seismic signals obtained in the Llaima volcano, one of the most active volcanoes in Chile. In order to classify subsequently, the methods used to clean the seismic signals are usually band pass Butterworth filter for its low ripple to reduce interference from other sources of mechanical energy to provide low frequency components. However, these methods are unable to eliminate white Gaussian noise present in seismic signal samples, which affects the observation signals that are too weak to show slightly above the noise, thereby preventing classification. The methods used in this study to eliminate the white noise spectral subtraction are consisting of narrow samples and then subtract noise in the frequency domain and the modified Wiener filter that using the statistical method of least squares generates clarification signal. When comparing both methods, it was observed that the Wiener filter has greater effectiveness in cleaning the noise when owning an iterative noise estimation. The development of these methods applied to the investigation of seismic signals provides information of seismic events that are of low intensity for further observation.