Volcanic signals segmentation

Yoiner Jeús Erazo Bravo, Universidad de Nariño, San Juan De Pasto - Colombia 2019


When dealing with signal processing problems, segmentation is an indispensable preliminary stage in case of characterization, classification and/or clustering are subsequently required -which are conventional phases for several applications. Segmentation is defined as the process of dividing a signal into its main characteristic parts, and it is traditionally performed in conjunction with a process called signal filtering. Different seismological studies have been carried out on different volcanoes, which have allowed registering and observing a variety of volcanic signals. The appropriate processing and analysis of these signals allows the extraction of knowledge about an earthquake-volcanic event. It often happens that an observation (a signal) contains several impulses (discharges) related to a seismic event. For a more detailed analysis of several pulses in a signal, the joint analysis is not feasible and, therefore, each shock must be isolated and seperately analyzed. To achieve this separation, the process called segmentation is performed.

In this work, we propose to study segmentation algorithms (one in time-domain and another of spectral type) in order to determine the techniques with greater efficiency, precision and lower computational cost compared to this type of signals, so that a step can be taken forward in the challenge of achieving a generalization of an automatic system aimed at the segmentation of earthquake-volcanic signals in real time.

                                                                                                   

         
                        Figure 1. S
egmentation performed by STA / LTA algorithm.
          
Figure 2. 
S
egmentation performed by TF&SM algorithm.
  
                                                         
                         
 

Figures 1 and 2, show the result of the segmentation performed by the two algorithms implemented in this work to 9 concatenated tectonic volcano (VT) signals taken at random from the database.


   METHODOLOGY

Publications

Thesis

"ESTUDIO COMPARATIVO DE ALGORITMOS PARA LA SEGMENTACIÓN DE SEÑALES VOLCÁNICAS ORIENTADO A LA CLASIFICACIÓN DE SISMOS"


Graphic interface

Figure 3.  Overview interface for volcanic signals segmentation.

Video Tutorial


About author

 Yoiner Jesús Erazo Bravo

 Electronic Engineer
 yoinererazo@gmail.com
ynr.e.b@hotmail.com




Comments