Analysis of the earth-ionosphere resonances and its application to the study of other natural phenomena

Resumen

Schumann Resonances (SRs) are electromagnetic waves generated in the Earth-ionosphere cavity in the Extremely Low Frequency (ELF) band. Lightning discharges are the main contributor to this signal. The study of this phenomenon started around 70 years ago. However, the increasing number of Extreme Low Frequency (ELF) observatories and the development of advanced software techniques have increased interest in the research community. In steady conditions, the knowledge about the physics of the phenomena and the relationship with the electromagnetic condition of the cavity is considerably advanced. However, some aspects remain not substantially solved, for example: i) The relationship between Schumann Resonance (SR) signal and other natural phenomena is not thoughtful clarified, with the exception of seasonal variations associated to the displacement of the thunderstorm centers and diurnal with the changes in the ionosphere. ii) Transient events only related to massive lightning discharges. There is not a theoretical model, neither a simulated one about the impact of the natural phenomena to the SR signal and their impact in the global electric circuit. The usage of theoretical models and classical methodology restricts the acquisition of new knowledge in the SR field. New proposal of innovative methodologies are required to describe the behavior of the SR signal. The methodologies developed along with the data and knowledge acquired in this first part of this PhD career were then focused on the application of Artificial Intelligence to increase the understanding of the resonant signal, whose complexity makes it an ideal focus for this kind of technology In this PhD thesis the advances and results in the SR field are exposed. The main objectives of this PhD thesis are i) To study and validate the data obtained in our Observatory of Sierra de los Filabres and compare with the data from other observatories. ii) Study the ELF in the transient domain and their relation with other phenomena. iii) Investigate the relationship between SR spectrum variation and other phenomena, iv) explore the application of artificial intelligence techniques in the SR field. The Observatory of Sierra de los Filabres was set by the research group TIC019 led by Jose Antonio Gazquez Parra with Nuria Novas Castellano, Rosa M. Garcia Salvador and Manuel Fernandez Ros. The observatory has been capturing data continuously since 2015. The observatory is composed of two orthogonal sensors North-South (NS) and EastWest (EW) and a high-performance sampler, being the first ELF observatory in Spain. The workflow of this PhD thesis is separated in two parts. The first part is focused on reviewing the literature about the processes that contribute to SR and applying methodologies previously developed by the scientific community to the SR from Sierra de los Filabres observatory. In this line, particular methodologies have been proposed to extract additional information based on a signal processing approach. This part includes i) Diurnal and seasonal variations of Sierra de los Filabres Observatory. ii) The study of the ELF transient events and iii) Relationship with other natural phenomena. The next part introduces computer intelligence techniques for assisting the SR signal. First this is achieved with an explanatory model, and second it is done by using a Deep Learning (DL) codifier to reduce the complexity of the SR spectrum. The output of the codifier was used for training an Earthqueake (EQ) detector. The most important outcomes of this thesis are: • The "Sierra de los Filabres" ELF observatory data was validated through the relationship between diurnal and seasonal variations of this observatory’s data by finding correspondence between variations from other observatories. • The relationship between the main thunderstorm centres and seasonal SR variations was discussed by selecting the hours when the thunderstorm centres are active. • The suitability of a Narrow Band Sensor centred on the 1st SR mode for the study of the ELF envelope was analyzed. The results show that the ELF transient events can be captured more precisely. • The impact in the ELF band of the low-medium intensity lightning discharges have similar characteristics to the high-intensity lightning discharges presented in the literature. • Five years of "Sierra de Los Filabres" ELF observatory data has been under study. The homogeneity and continuity of the data has been confirmed. • The relationship between particular SR modes at selected hours with the external variables that affect the SR signal has been revealed. Expected relationships are visible in the results, thus validating the method. However, new relationships have emerged from the results, worthy of further analysis. • Machine Learning (ML) techniques have been applied to analyze the intricate relationship between SR signal and other phenomena with promising results. The ML explanatory models point out new relationships, which opens new doors for investigation. • DL techniques have been used in the SR field for data analysis purposes. A DL codifier has been developed to reduce the dimensionality of the SR data. • A new algorithm has been proposed for enhanced extraction of each SR frequency mode without using a mathematical fit procedure. The results show a substantial improvement over the classical methods. • The use of the DL codifier makes data between different observatories compatible, even when the sensor, acquisition stage, and processing algorithm are entirely different. • It has been developed the first SR tool, based on DL to analyze long time series of SR signal, up to 30 days, to find the relationship between the temporal variation of SR spectrum and seismic phenomena. The results are auspicious. SR are an immense field for studying and applying new techniques to further explore the importance of this natural signal. With the correct tools, it is possible to use the signal to detect relevant changes in the ionosphere, predict the occurrence of a near EQ or obtain preliminary data about the composition of ionospheres in other celestials bodies.