Publication Date:
2017-04-04
Description:
Active volcanoes generate sonic and infrasonic signals, whose investigation provides useful information for both monitoring purposes and the study of the dynamics of explosive phenomena.
At Mt. Etna volcano (Italy), a pattern recognition system based on infrasonic waveform
features has been developed. First, by a parametric power spectrum method, the features
describing and characterizing the infrasound events were extracted: peak frequency and quality factor. Then, together with the peak-to-peak amplitude, these features constituted a 3-D ‘feature space’; by Density-Based Spatial Clustering of Applications with Noise algorithm
(DBSCAN) three clusters were recognized inside it. After the clustering process, by using a common location method (semblance method) and additional volcanological information
concerning the intensity of the explosive activity, we were able to associate each cluster to a
particular source vent and/or a kind of volcanic activity. Finally, for automatic event location,
clusters were used to train a model based on Support Vector Machine, calculating optimal
hyperplanes able to maximize the margins of separation among the clusters. After the training phase this system automatically allows recognizing the active vent with no location algorithm and by using only a single station.
Description:
Published
Description:
253-264
Description:
1.4. TTC - Sorveglianza sismologica delle aree vulcaniche attive
Description:
JCR Journal
Description:
reserved
Keywords:
Time series analysis
;
Volcano monitoring
;
Volcano seismology
;
04. Solid Earth::04.06. Seismology::04.06.08. Volcano seismology
Repository Name:
Istituto Nazionale di Geofisica e Vulcanologia (INGV)
Type:
article
Permalink
