GNGTS 2017 - 36° Convegno Nazionale

GNGTS 2017 S essione 2.2 353 Tab. 1 - Distances of the stations from the epicenter in km, according to their membership to the two groups. Group 1 75.1 98.6 64.7 79.8 16.1 42.4 106.5 Group 2 198.9 279.8 173.8 143.8 117.1 135.8 212.8 As the signals start at different times, in a preprocessing step the time is standardized and the waves are synchronized; then the alignment is applied in order to retain the underlying pattern.The algorithm provides the classification of the stations into two groups, characterized by different amplitude variability: starting with randomly selected clusters, the algorithm provides further allocations in order to optimize the classification; in Fig. 1 e Fig. 2, the aligned waves are represented in a small interval of time, for a better understanding of the dynamics of the waves and of their differences: Fig. 1 shows the initial allocation: the grouped signals are represented for the three dimensions respectively in the column on the left, cluster 1 and in the column on the right for the cluster 2. Fig.2 show the results for the optimal allocation. Fig. 3 shows the clusters for the entire sequences of the three dimensions. The distances of the stations from epicenter are reported in Tab. 1, where their spatial separation is confirmed. Fig. 3 - Clusters for the entire sequences of the three dimensions.