OPTIMAL SETTING OF EARTHQUAKE-RELATED IONOSPHERIC TEC (TOTAL ELECTRON CONTENT) ANOMALIES DETECTION METHODS: LONG-TERM VALIDATION OVER ITALY AND MEDITERRANEAN AREA

In this thesis an optimal setting and long-term validation of the methodological inputs used for the detection of earthquake-related anomalies of the ionospheric-Total Electron Content (TEC) was made. The setting was optimized using own made R-coded machine learning techniques and using for the first time multi-year time series of TEC satellite data and multi-year time series of seismic data relating to the Italian and Mediterranean area. Analyses on seismic-connected parameters have been developed for some decades now, during which national research groups have been established in various countries of the World (e.g. Japan, former USSR, China, Taiwan and USA [1]), space missions specifically dedicated to research on these phenomena have been carried out and several thousands of scientific publications on various ionospheric, atmospheric and lithospheric parameters, proposed as potential seismic precursors, have been produced (here are some exhaustive reviews that collect the main scientific works of the last decades: [2], [3], [4], [5], [6], [7], [8], [9]). However, during the first periods of research very mixed results were found, the proposed analyses were often not confirmed by subsequent observations and this increased scepticism around the research sector and consequently often led to funding cuts by the governments. During the last decade, thanks to the availability of historical satellite observations, which have begun to be significantly large, and thanks to the exponential growth of machine learning techniques, which allow processing on large amounts of data, a considerable amount of statistically robust study (multi-year analysis) on seismic-related parameters have been developed, relaunching the research sector. In 2018, in the wake of this newfound enthusiasm on the subject, as a result of an international collaboration between China National Space Administration (CNSA) and the Italian Space Agency (ASI), CSES (China Seismo-Electromagnetic Satellite) was launched into orbit, the second satellite ever (after the French micro-satellite DEMETER launched in 2004) specifically dedicated to the study of atmospheric and ionospheric seismic-related phenomena. Today the observations of earthquake-related parameters using satellite techniques are the most widespread and performing (also for the measurement on the Earth's surface), thanks to the fact of ensuring adequate spatial resolution of historical data. The literature references that include the most substantial and statistically significant observations of seismic-related parameters mainly concern: ionospheric parameters (total electron content [10], [11], [12], electron density [13], etc.), electromagnetic perturbations [14] and ground temperature parameters, i.e. increases in the ground thermal emission observed in the infrared band (Thermal InfraRed emission, TIR), see [15], [16] and references within [17]. The TEC (Total Electron Content) is probably the atmo-ionospheric parameter that has most contributed to the growth of studies on seismic-related anomalies in recent years, as it is the only one measurable through the GNSS (Global Navigation Satellite System) constellations, which today provide an over twenty-year data series on a global scale from a number of satellites that currently (February 2022) reaches 134 units. The difficulties historically encountered in the correlation studies between TEC anomalies and seismic events are mainly two: 1. Establishing when the behaviour of the TEC parameter (as well as the other ionospheric parameters) can be defined anomalous is extremely complicated, because the ionosphere noise sources are many, of various nature (known and unknown) and produce disturbances (mainly governed by the influence of the Sun) in time and space, which can be even stronger than the anomalies themselves. 2. It is equally complicated to establish when the identified anomalies are actually correlated to the seismic activity in progress, since the correlation should be established in the spatial, temporal and magnitude domains. To address these problems, many methods have been proposed over time for the study of seismic-related ionospheric anomalies, which in turn contain further diversified methodological inputs (filters, thresholds, domain limits). Some methods have recently been validated on a large number of seismic events, analysing the weeks/months preceding and following earthquakes. What was still missing were analyses that also involve the remaining periods, that is those of seismic inactivity, this represents the most important innovation that we have introduced in our research project. The validation even on periods of seismic inactivity (refutation) makes the findings more consistent, as it drastically reduces the chances that the results obtained are random. Moreover, despite a large amount of studies, the theoretical basis on the subject are very few, because the theoretical hypotheses formulated are difficult to prove. Consequently, the methods for the detection of seismic-related anomalies are based on a mostly empirical approach. This is the reason why the proposed methodological inputs are so many and so different. Sorting out the multitude of input variables proposed would allow research efforts to be focused in the right direction. Today, thanks to the enormous availability of historical data and to the great added value that machine learning has brought to the world of research, it is possible to overcome these problems. In this thesis work, after an in-depth review of the existing literature on the subject, having recognized the previously described methodological non-uniformities and employing as a working basis the most widely used method in the literature for the identification of TEC anomalies, named InterQuartile Range (IQR), we proceeded to test various input elements in the method, either chosen from the literature researches that returned the best performances, or proposed by us on the basis of analyses made on the TEC variation sources. To this aim, we used our R-coded machine learning techniques combining 11 multi-year time series of TEC satellite data and the related multi-year time series of seismic catalogues, simulating their behaviours in tens of thousands of possible combinations according to a predetermined set of input elements and classifying these methodological inputs according to criteria established a priori. The main elements of innovation made with respect to the state of the art are the following: - For the first time, multi-year time series (mostly ~20-year, overall in the time interval from 2001 to 2021) are analysed without time interruptions (i.e. also involving periods of seismic inactivity); - For the first time an optimal setting of the methodological inputs for the detection of seismic-related anomalies is realized; - For the first time a long-term TEC earthquake-related anomalies detection method is applied over Italy and Mediterranean area; - It is proposed and tested a new filter to eliminate/minimize the effects of solar activity on the TEC. Three categories of inputs were tested: - Inputs used to define the statistically anomalous behaviour of the TEC parameter; - Inputs used to set the earthquakes magnitude-space-time (MST) domain; - Thresholds to mark the difference between standard and anomalous behaviour. The IQR is a method of detecting anomalies based on the comparison between the observed TEC value and comparison samples containing data collected in the same time slot (and geographical location) of the data under investigation and in a certain number of previous days (typically 15) which define the sample size. The comparison samples express the parameter standard behaviour. Given the empirical nature of the study, several conclusions coming from the results of our analysis were drawn. Our conclusions should be considered valid for applications at mid-latitudes, since the behaviour of the parameter is extremely variable as a function of latitude. The main of these conclusions are set out below. The following are the conclusions that our results firmly confirm. - Within the IQR method, due to the strong solar activity influence on the TEC, the 27-day comparison sample proves to be significantly more efficient than the (most widely used) 15-day sample for the detection of the TEC anomalies; - The detection of seismic-TEC anomalies can be more efficient looking for punctual rather than persistent phenomena (from the temporal point of view); - Earthquakes with hypocentral depth greater than 50 km are less likely to affect the ionosphere. - Due to the fact that the IQR comparison samples are non-Gaussian (mainly right-tailed) the thresholds used (positive and negative) for the anomalies detection should be set independent of each other; The following conclusions require further confirmation. - Data obtained from the individual GNSS receivers are useful for capturing local earthquake-ionospheric effects (Magnitude ≥ 4; Distance ≤ Dobrovolsky radius); - Data obtained from multiple GNSS receivers (portions of Global Ionospheric Maps, GIMs) prove to be particularly effective in detecting large-scale earthquake-ionospheric effects. Following the multitude of analyses carried out for the setting of the MST (magnitude-space-time) domain, we propose the use of the following inputs:  Magnitude ≥ 5.5;  Radius of influence equal to 1250 km;  Depth of the hypocenter less than 50 km;  Alert time window ranging from 90 days before (after) to 30 days after (before) the seismic event (the ionospheric anomaly).   REFERENCES 1. S. Uyeda and T. Nagao (2018) “Historical Development of Pre-Earthquake Phenomena Studies” in Pre‐earthquake processes. A multidisciplinary approach to earthquake prediction studies. Editors D. Ouzounov, S. Pulinets, K. Hattori, and P. Taylor (John Wiley & Sons), 19–39. https://doi.org/10.1002/9781119156949.ch1 2. Park, S. K., Johnston, M. J. S., Madden, T. R., Morgan, F. D., and Morrison, H. F. (1993). Electromagnetic Precursors to Earthquakes in the Ulf Band: A Review of Observations and Mechanisms. Reviews of Geophysics 31 (2), 117–132. https://doi.org/10.1029/93RG00820 3. Geller, R. J. (1997). Earthquake Prediction: a Critical Review. Geophys. J. Int. 131 (3), 425–450. https://doi.org/10.1111/j.1365-246X.1997.tb06588.x 4. Johnston, M. J. S. (1997). Review of Electric and Magnetic fields Accompanying Seismic and Volcanic Activity. Surveys in Geophysics, Netherlands: Springer. 18. Issue 5, 441–476. https://doi.org/10.1023/a:1006500408086 5. Tronin, A. A. (2006). Remote Sensing and Earthquakes: A Review. Phys. Chem. Earth, Parts A/B/C 31 (4–9), 138–142. https://doi.org/10.1016/j.pce.2006.02.024 6. Helman, D. S. (2020). Seismic Electric Signals (SES) and Earthquakes: A Review of an Updated VAN Method and Competing Hypotheses for SES Generation and Earthquake Triggering. Phys. Earth Planet. Interiors 302, 106484. https://doi.org/10.1016/j.pepi.2020.106484 7. Sorokin, V. M., Chmyrev, V. M., and Hayakawa, M. (2020). A Review on Electrodynamic Influence of Atmospheric Processes to the Ionosphere. Open J. Earthquake Res. 09 (02), 113–141. https://doi.org/10.4236/ojer.2020.92008 8. Picozza P, Conti L and Sotgiu A (2021) Looking for Earthquake Precursors From Space: A Critical Review. Front. Earth Sci. 9:676775. https://doi.org/10.3389/feart.2021.676775 9. Conti, L., Picozza, P., and Sotgiu, A. (2021). A Critical Review of Ground Based Observations of Earthquake Precursors. Front. Earth Sci. Sec. Geohazards Georisks. https://doi.org/10.3389/feart.2021.676766 10. Kon, S., Nishihashi, M., and Hattori, K. (2011). Ionospheric Anomalies Possibly Associated with M⩾6.0 Earthquakes in the Japan Area during 1998-2010: Case Studies and Statistical Study. J. Asian Earth Sci. 41 (4), 410–420. https://doi.org/10.1016/j.jseaes.2010.10.005 11. Liu, J. & Chen, C.H. & Tsai, Ho-Fang. (2013). A Statistical Study on Seismo-Ionospheric Anomalies of the Total Electron Content for the Period of 56 M≥6.0 Earthquakes Occurring in China during 1998-2012. Chin J Space Sci. 33. 258-269. 12. Liu J.Y., Chen C.H., Tsai H.F. A statistical study on seismoionospheric precursors of the total electron content associated with 146 M 6.0 earthquakes in Japan during 1998-2011, Earthquake Prediction Studies: Seismo Electromagnetics, ed. Hayagawa, M (TERRAPUB, Tokyo, 2013), 17-30. 13. De Santis, A., Marchetti, D., Pavón-Carrasco, F.J. et al. Precursory worldwide signatures of earthquake occurrences on Swarm satellite data. Sci Rep 9, 20287 (2019). https://doi.org/10.1038/s41598-019-56599-1 14. Zeren Zhi-Ma, Shen Xu-Hui, CAO Jin-Bin, ZHANG Xue-Min, Huang Jian-Ping, Liu Jing, Ouyang Xin-Yan, Zhao Shu-Fan. Statistical analysis of ELF/VLF magnetic field disturbances before major earthquakes[J]. Chinese Journal of Geophysics, 2012, 55(11): 3699-3708. https://doi.org/10.6038/j.issn.0001-5733.2012.11.017 15. Filizzola, C.; Corrado, A.; Genzano, N.; Lisi, M.; Pergola, N.; Colonna, R.; Tramutoli, V. (2022) RST Analysis of Anomalous TIR Sequences in Relation with Earthquakes Occurred in Turkey in the Period 2004–2015. Remote Sens., 14, 381. https://doi.org/10.3390/rs14020381 16. Genzano, N.; Filizzaola, C.; Hattori, K.; Pergola, N.; Tramutoli, V. Statistical correlation analysis between thermal infrared anomalies observed from MTSATs and large earthquakes occurred in Japan (2005–2015). J. Geophys. Res. Solid Earth 2021, 126. https://doi.org/10.1029/2020JB020108 17. Valerio, Tramutoli & Corrado, R. & Carolina, Filizzola & Genzano, N. & Lisi, Mariano & Pergola, Nicola. (2015). From visual comparison to Robust Satellite Techniques: 30 years of thermal infrared satellite data analyses for the study of earthquake preparation phases. Bollettino di Geofisica Teorica ed Applicata. 56. 167-202.