Wavelet Based Machine Learning Approach for Spectral Seismic Signal Analysis: A Case Study North Tapanuli Earthquake.

Machine Learning of seismic signals is considered to automatize the analysis of years of the recorded signal. In this research, we considered using the wavelet transform based machine learning approach to analysis the spectrum signal which recorded from Broadband Seismic Network in North Tapanuli area. We use the signal seismic which recorded from GSI, MNSI, PBSI, PSI, SBSI, and TDNI which has been deployed in North Tapanuli, North Sumatera, Indonesia. The main aim of this paper to extract the different the value and spectrum of the seismic signal and identify the energy of signal of seismic which recorded from the sensor. The result shows that for all signal, which recorded form broadband seismic network information for detection and characterization is found in the instantaneous spectrum, and the makes seismic signal most useful is that this spectrum changes over time. The wavelet-based machine learning approach utilizing a complex Morlet analyzing wavelet is not only particularly suitable to clearly and simultaneously of seismic signals but also powerful for processing frequency-energy with variant signal seismic with a short period. [ABSTRACT FROM AUTHOR]