GNGTS 2017 - 36° Convegno Nazionale

280 GNGTS 2017 S essione 2.1 Stucchi M., Camassi R., Rovida A., Locati M., Ercolani E., Meletti C., Migliavacca P., Bernardini F., Azzaro R. (eds.) (2007). DBMI04, il database delle osservazioni macrosismiche dei terremoti italiani utilizzate per la compilazione del catalogo parametrico CPTI04. Quaderni di Geofisica, 49, INGV, Roma, 38 pp., doi:10.6092/ INGV.IT -DBMI04. Synergy Software (2014). Kaleidagraph, Tools for discovery, version 4.5. Intermediate distance correlation between events P. Harabaglia Scuola di Ingegneria, Università degli Studi Basilicata, Potenza, Italy Up to date a considerable effort has been carried out to evaluate inter-event time intervals, in order to better estimate time dependent, non poissonian, hazard. Probably the best model available is that of Corral (2005), even though it is not completely satisfactory. Conversely spatial inter-event investigations have not created great interest in the scientific community. The availability of the last version (4.0, released on January 26, 2017) of the ISC-GEM catalogue (Storchak et al. , 2013) allows finally to have a complete catalog for Mw ≥ 6.0 since 1927. This means that the data set is made by 9014 earthquakes with depth equal or less than 60 km : it is large enough to investigate also the spatial terms of seismic hazard. The basic approach is similar to the temporal one: inter-event distances are computed, ordered in terms of increasing value and than the survival probability function is determined. The survival function at a given distance d , is given by the ratio between the number of time intervals n d ≥ d with respect to the total number of time intervals. For sake of simplicity, the choice was to compute the inter-event distance along the straight line passing between the two hypocenters, since this allows to consider events with similar epicentral locations but different depths. The regular angular distance approach would be in fact more imprecise at short distances. The first step consists in determining the shape of the survival functions of randomly located events. It turns out that it has the shape of a quadratic. It retains the same shape, obviously with different parameters, even if we keep the actual hypocentral locations, just randomly varying the sequence of occurrence. As far as the survival function P S of actual data is concerned (Fig. 1), at distances greater then about 2000 km it can still be represented in terms of a quadratic but it is markedly different at shortest distances. Fig. 1 - Survival functions for Mw ≥ 6.0 and depth less then 60 km. Black: 1927-2013; yellow: 1927-1949; green: 1950-1969; light blue: 1970-1992; violet 1993-2013; red: best model; thicker dark blue: quadratic fit in spatial interval 1850-12750 km.