Statistical correlation analysis between thermal infrared anomalies observed from MTSATs and large earthquakes occurred in Japan (2005 ‐ 2015)

The literature often reports space‐time relations between the abnormal variations of different kinds of non‐seismological (i.e. geophysical, geochemical, atmospheric, etc.) parameters and the occurrence of earthquakes. The integration of such observations with seismological ones could improve the quality of the seismic hazard assessment in the medium‐short term (months to days). Each considered parameter has, in principle, its capabilities to provide useful (and utilizable) information about seismic processes. Therefore, to define a system based on different observations, the first step is to estimate the informative contribution that each considered parameter could provide. In this paper, we will evaluate the potential of Significant Sequences of satellite Thermal Anomalies (SSTAs). In particular, we adopted the broadly used Robust Satellite Techniques (RST) data analysis methodology to identify SSTAs over 11 years (June 2005 ‐ December 2015) of night‐time satellite images acquired by MTSAT satellites over Japan. Aiming at reducing the false‐positive rate, we introduced and tested an innovative configuration of the RST technique, which is here presented. We executed a correlation analysis between SSTAs and Japanese earthquakes with MJMA≥6 by applying suitable constraints concerning space, time, and magnitude. The analysis highlights: a) the occurrence of just 29 SSTAs in the 11‐year period of observation, b) 18 SSTAs (i.e. 62%) occur in an apparent space‐time relation to earthquakes, c) 13 of them occur before the quake. Results of the random test analysis, based on error diagrams, confirm a non‐casual correlation between ‘RST‐based satellite thermal anomalies’ and earthquake occurrences. In particular, for MJMA≥6.5 earthquakes, probability gain is up to better than 4.3 as compared with the random guess.