Reducing atmospheric noise in RST analysis of TIR satellite radiances for earthquakes prone areas satellite monitoring

• Robust Satellite Technique has been used in order to study the Aquila earthquake.
• In this work we use LST data products instead than just TIR radiances at the sensor.
• The results obtained confirms and improve the ones obtained by using TIR radiances.

Space–time fluctuations of the Earth’s emitted Thermal Infrared (TIR) radiation observed from satellite from months to weeks before an earthquake are reported in several studies. Among the others, a Robust Satellite data analysis Technique (RST) was proposed (and applied to different satellite sensors in various geo-tectonic contexts) to discriminate anomalous signal transients possibly associated with earthquake occurrence from normal TIR signal fluctuations due to other possible causes (e.g. solar diurnal–annual cycle, meteorological conditions, changes in observational conditions, etc.). Variations in satellite view angle depending on satellite’s passages (for polar satellites) and atmospheric water vapour fluctuations were recognized in the past as the main factors affecting the residual signal variability reducing the overall Signal-to-Noise (S/N) ratio and the potential of the RST-based approach in identifying seismically related thermal anomalies. In this paper we focus on both factors for the first time, applying the RST approach to geostationary satellites (which guarantees stable view angles) and using Land Surface Temperature (LST) data products (which are less affected by atmospheric water vapour variability) instead of just TIR radiances at the sensor.The first results, obtained in the case of the Abruzzo earthquake (6 April 2009, MW ∼ 6.3) by analyzing 6 years of SEVIRI (Spinning Enhanced Visible and Infrared Imager on board the geostationary Meteosat Second Generation satellite) LST products provided by EUMETSAT, seem to confirm the major sensitivity of the proposed approach in detecting perturbations of the Earth’s thermal emission a few days before the main shock. The results achieved in terms of increased S/N ratio (in validation) and reduced “false alarms” rate (in confutation) are discussed comparing results obtained by applying RST to LST products with those achieved by applying an identical RST analysis (using the same MSG-SEVIRI 2005–2010 data-set) to the simple TIR radiances at the sensor.