GNGTS 2013 - Atti del 32° Convegno Nazionale

of T ( r , t ’) is expected to reduce possible contributions (e.g. occasional warming) due to day-to-day and/or year-to-year climatological changes and/or season time-drifts; • μ ΔT ( r ) and σ ΔT ( r ) are the temporal average and standard deviation values of Δ T ( r , t ’) computed on the homogeneous data-set, of cloud-free satellite records collected at location r in the same time-slot (hour of the day) and period of the year (month) of the image at hand. The ( r , t’ ) index gives the local excess of the current Δ T ( r , t ’) signal compared with its historical mean value, μ ΔT ( r ), and weighted by its historical variability, σ ΔT ( r ), at the considered location. Both μ ∆T ( r , t’ ) and σ ∆T ( r , t’ ) are computed for each location r , processing several years of historical satellite records acquired in similar observational conditions. In this project, RST approach has been implemented on MSG-SEVIRI data. Nine years of TIR satellite images acquired from 2004 to 2012 in the same time of the day (00:00 GMT) for each months of the year were used to characterize the expected signal behavior (reference fields μ ΔT ( r ) and σ ΔT ( r ) at each location r ) in unperturbed conditions. On this basis RETIRA index has been computed for all the TIR SEVIRI/MSG images covering the whole project duration (July 2012 – June 2013) and testing areas (Po Plain and Southern Apennines; Fig. 1). Main results. The test was performed only considering significant the spatial – temporal persistence of TIR anomalies, as explained in Tramutoli et al. (2005). Following the code reported in Fig. 2 persistence has been evaluated considering spatially extended (D>150km) TIR anomalies (RETIRA≥3) reappearing one (AS) or more (AVS) times, approximately in the same area, within the following 7 usable days (i.e. excluding images affected by a “ cold spatial average effect” or having the RETIRA computation area affected by more than 80% of cloudy pixels 1 ). Data analysis has been performed in order to evaluate to which extent TIR anomalies belonging to classes AS or AVS can be related or not to EQs of M>4 occurred within a distance D (spatial window) 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). Following our previous experience (see for instance Corrado et al. , 2005; Genzano et al. , 2007; Tramutoli et al. , 2012 and reference herein) 150 km<D<R D has been considered being R D = 10 0.43M the Dobrovolsky radius (Dobrovolsky et al. , 1979). The analysis enhanced only 4 cases of significant TIR anomalies (3 persistent and only one very persistent). In 3 cases (2 over the Southern Apennine, 1 over the Po Plain) the anomalies appeared (within the considered space-time windows) few days before (2 cases) or after (1 case) the main shocks (Tab. 1). In particular: • 29/30 Sept 2012 and 1 Oct 2012. AS/AVS TIR anomalies were observed in the 3 days following the seismic event of Benevento (27 Sept 2012, Ml 4.1); • 12/13 Oct 2012. AS TIR anomalies (Fig. 3) were observed 13 and 12 days before the seismic event of Pollino (25 Oct 2012, Ml 5); • 12/14 Nov 2012. The AS TIR sequence recorded in Southern Apennines as to be considered a false positive with reference to the pre-established validation rules. • 3/7 Jan 2013 AS TIR anomalies were observed 22 and 18 days before the seismic event of Garfagnana (25 Jan 2013, Ml 4.8). In terms of reliability (assumed here as the fraction of significant TIR anomalies occurred in the investigated area/period appearing effectively related to the occurrence of an earthquake) a score of 75% (67% on the Southern Apennine, 100% over the Po Plain) was reached during the considered period, which means 25% of false positive (33% on the Southern Apennine, 0% over the Po Plain: Tab. 2). It should be noted that meteorological clouds prevent Earth’s emitted TIR signal to reach the satellite sensor, so that corresponding portions of the scene represent an information gap (data missing) both in the space and time dimension. Under this circumstance it is impossible to give 1 note that such analyses are performed separately for pixels belonging to the land or sea class. 149 GNGTS 2013 S essione 2.1