Detection of Nonvolcanic Tremors using Spectral Cross-Correlation

Wednesday, 24 February 2016
Jorge Arturo Real Sr., Vladimir Kostoglodov and Allen L Husker, UNAM National Autonomous University of Mexico, Mexico City, Mexico
Abstract:
The hypothesis that the nonvolcanic tremor (NVT) is a swarm of the low frequency earthquakes is not still widely confirmed. This endorses further attempts to develop new more rigorous methods of the NVT detection and analysis. There are several approaches and algorithms already implemented to detect, locate, and measure the main characteristics of the tremor (e.g., its duration, energy etc.). Nevertheless a complex nature of the NVT, its long duration and its low amplitude over the noise level makes it difficult to obtain unbiased estimates of tremor parameters.

This paper presents a new method for detection of the NVT bursts that is able not only to determine the NVT duration, but also to differentiate the type of tremor comparing the amplitude spectra of the continuous seismic record with different NVT spectral templates extracted from the real tremor spectra.

The method uses cross-correlation functions sequentially computed between the spectra of characteristic NVT signal (spectral template) and the spectra of seismic record within a moving window (15-25 sec). The maximum of the cross-correlation functions, the frequency lag, and the energy (V2 in 2-10 Hz band) then plotted together as functions of time.

The results for the records from the low nose seismic stations show that the method can distinctly detect episodes of different families of NVT within the same tremor burst. Moreover, applying the constraint of “zero lag” the duration of the NVT corresponding to the particular spectral template is essentially unbiased (does not need to establish any threshold for the amplitude of the cross-correlation functions).

The proposed method is very useful for the routine, near real time compilation of catalogs of different NVT families. The method can also detect the presence of micro seismicity within the NVT episodes, which is crucial for locating NVT bursts without the "contamination" that these events involve, thus facilitating the study of the NVT.

<< Previous Abstract | Next Abstract >>