Improvements of the Robust Satellite Techniques for monitoring the Earth thermal emission in seismic areas: an analysis on 11 years (2005-2015) of MTSAT TIR observations over Japan

Looking toward the development of multi parametric approaches, which could be able to improve our capability to assess the seismic hazard in the short-term (from weeks to days before the earthquake), a preliminary step is to identify those parameters (chemical, physical, biological, etc.) whose anomalous variations can be, to some extent, associated with the complex process of preparation of earthquakes. The fluctuations of Earth's thermally emitted radiation, measured by satellite sensors operating in the thermal infrared (TIR) spectral range, have been proposed since eighties as a potential earthquake precursor. Since 2001, the general change detention approach Robust Satellite Techniques (RST), used in combination with RETIRA (Robust Estimator of TIR Anomalies) index, showed good ability to discriminate anomalous TIR signals possibly associated to seismic activity, from the normal variability of TIR signal due to other causes (e.g. meteorological). In this paper, the RST (Robust Satellite Technique, Tramutoli 1998) data analysis approach has been implemented on TIR satellite records collected over Japan by the geostationary satellite sensor MTSAT (Multifunctional Transport SATellites) in the period June 2005 - December 2015 in order to evaluate its possible contribute to an improved multi parametric system for a time-Dependent Assessment of Seismic Hazard (t-DASH). For the first time, thermal anomalies has been identified comparing the daily TIR radiation of each location of the considered satellite portions, with its historical expected value and variation range (i.e. RST reference fields) computed using a moving window (i.e. 30 days) instead than fixed monthly window. Preliminary results of correlation analysis among the appearance of Significant Sequences of TIR Anomalies (SSTAs) and time, location and magnitude of earthquakes (M≥5), performed by applying predefined space-temporal and magnitude constraints, show that 74% of SSTAs were in an apparent space-time relations with earthquake with a false alarm rate of 26%.