GNGTS 2014 - Atti del 33° Convegno Nazionale

GNGTS 2014 S essione 2.1 93 (i.e. TIR anomalies apparently not related to earthquakes) together with an effectiveness of more than 66% (i.e. less than 33% of earthquakes occurred being not closely preceded nor followed by the appearance of TIR anomalies) were obtained by using predefined temporal- spatial correlation windows. In that case only space-time persistent and relatively intense (S/N>3σ) TIR anomalies were considered. Following previous experience (see for instance Corrado et al. , 2005; Genzano et al. , 2007; Tramutoli et al. , 2012 and references herein), the correlation with earthquakes having M>4 was admitted if they occurred within a distance 150 km <D<R D [being R D = 10 0.43 M the Dobrovolsky radius, (Dobrovolsky et al. , 1979)] and a time period within 30 days after (pre-seismic TIR anomalies) and 2 weeks before (post seismic TIR anomalies) the occurrence of such anomalies (temporal window). It should be noted that the used rules for success/failure evaluation were substantially the same accepted for the planned experimentation at CSEP (USC, http://www.cseptesting.org/ ). Even though based on a very short period of time and only on two testing areas, results achieved during the 1 st year of the project were not surprising confirming results already achieved by independent, 3 years long, studies funded by the German Space Agency [DLR: Halle et al. (2008)] and by NASA [together with USGS and University of South California, Eneva et al. (2008)] which successfully applied the same method (RST approach and RETIRA index) on longer data sets, to different satellite sensors, different geographic areas and tectonic setting in Europe and in Asia (details on these studies are given in the already quoted INGV-S3 UR8 first report available at https://sites. google.com/site/ingvdpc2012progettos3/) . Results appeared moreover to confirm the ones achieved during the 2 years long EU-FP7 project PRE-EARTHQUAKES (www.pre-earthquakes.org ) where the considered techniques played a major role (trough the integration with several independent observations) for demonstrating, in different geographic area of the world, the potential of an improved system of time-dependent seismic hazard assessment. A similar analysis (Genzano et al., 2013) recently performed on a longer time series of all TIR observations collected around 00 GMT by MSG-SEVIRI over the same area (Italy) 2004 and 2012 (about 9 years of continuous observations) gave an increased percentage (45% ) of significant TIR anomalies apparently not related (following the quasi-empirical rules previously described) to earthquakes (M>4) occurrence, however confirming the possible role of such kind of observations in the framework of a multi-parametric system devoted to a time-Dependent Assessment of Seismic Hazard (t-DASH). Tab. 1 - Results of the reliability (top) and sensitivity (bottom) analyses performed on the testing areas of the first DPC-INGV-S3 project during the considered July-2012/June 2013 testing period. Reliability Analysis Southern Apennines Po Plain Total Number of TIR anomaly sequences detected 3 1 4 Corresponding to EQ 2 1 3 (67%) (100%) (75%) NON Corresponding to EQ (false) 1 0 1 (33%) (0%) (25%) Sensitivity Analysis Number of seismic sequences (M ≥ 4) 3 3 6 Associated to significant TIR anomalies 3 1 4 (100%) (33%) (67%) NON Associated to significant TIR anomalies 0 2 2 (missed) (0%) (66%) (33%)