GNGTS 2014 - Atti del 33° Convegno Nazionale

94 GNGTS 2014 S essione 2.1 In this paper results until now achieved will be discussed also in the new perspective opened by the 2 nd year of the INGV-DPC-S3 Project which “ instead of a “silver bullet relative to any forecasting anomaly” , is presently looking for “ multiparametric indexes” (including TIR anomalies) “… able to capture short term variations of the regional strain field that have the earthquake as a possible effect” . The rst methodology. The Earth’s thermally emitted radiation (mostly in the 8-14 micron atmospheric window) measured from IR satellite sensors is very highly variable due to different natural/observational causes (see for instance Tramutoli et al. , 2005). By applying a Robust Satellite data analysis Technique [RST: Tramutoli (1998, 2005, 2007)] to multi-year TIR satellite records it was possible to define the concept of TIR anomaly itself as a statistically significant deviations (in the space–time domain) from a normal state, preliminarily defined for each image pixel and period of the year, on the basis of satellite observations collected during several years in the past under similar observational conditions. For earthquake (EQ) prone area monitoring, anomalous TIR patterns were identified by using a specific the index named RETIRA [Robust Estimator of TIR Anomalies: Filizzola et al. (2004), Tramutoli et al. (2005)] ⊗ ( r , t’ ), which is computed as follows: where: • r ≡ (x,y) represents the ground geographic coordinates of the image pixel center; • t’ is the time of acquisition of the satellite image at hand with t ∈ τ, where τ defines the homogeneous domain of satellite images selected during the same time-slot (hour of the day) and period of the year (month) of the image at hand; these restrictions on the time series are useful in order to reduce the signal variability connected both to seasonal and daily cycle; • ΔT( r , t ’) is the difference between the current ( t = t’ ) TIR signal value T ( r , t’ ) at location r and its spatial average T ( t’ ) computed in place on the same image at hand within a predefined RETIRA computation area ( RCA ) considering only cloud-free pixels, and considering only sea pixels if r is located on the sea, only land pixels if r is located over the land; note that the choice of such a differential variable Δ T (r, t’ ), instead 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. For this reason RCA has to be chosen much wider than common meteorological fronts usually affecting the investigating area (usually RCA≥10 6 km 2 ) • μ Δ T ( r ) and σ Δ T ( r ) are the temporal average and standard deviation values of Δ T ( r , t ) computed on a 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 RETIRA index ⊗ ( r , t’ ) gives the local excess of the current ∆ T ( r , t’ ) signal compared with its historical mean value, μ Δ T ( r ), and weighted by its historical variability, σ Δ T ( r ), at the considered location. Both μ ∆ T ( r ) and σ ∆ T ( r ) are computed for each location r , processing several years of historical satellite records acquired in similar observational conditions. Main results. In the 1 st year INGV-DPC-S3 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 behaviour (reference fields μ Δ T ( r ) and σ Δ T ( 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).