GNGTS 2024 - Atti del 42° Convegno Nazionale

Session 2.1 GNGTS 2024 Defining the conditional probability of a noteworthy event, like an earthquake, involves analysing the variations in a geophysical observable that typically precedes it. In this context, time series representation involves binary events, with '1' indicating instances where the observable exceeds a certain magnitude threshold. For instance, these events could be seismic activities surpassing a specific magnitude or any observable exceeding its defined threshold. Subsequently, the likelihood of an earthquake is assessed by examining the correlation between these binary events. Or also, a probability increase for an earthquake can be defined by the probability gain where P(EQ) represents the frequency of the earthquake event of a magnitude greater than a fixed value, and EA represents the observable A event, being the conditional probability defined by the Pearson Coefficient R = corr(EQ,EA) and the frequency of the event A, P(EA). Such a relation (2) for the probability gain due to the observable A is valid only for binary event series. Let's explore the scenario where we observe two quantities, drawing from existing networks like the electromagnetic network in central Italy and the network comprised of NOAA satellites. The first network records magnetic field pulses, while the second detects electron precipitation. The magnetic pulses have the potential to alter the trajectory of electrons reaching the ionosphere, indicating compatibility and dependence between the two observables. These quantities can be observed concurrently (represented by the symbol ∩ ) or individually, without distinction (represented by the symbol U). The correspondent probability gains (Fidani, 2021) where EB is the electron burst event and MP is the magnetic pulse event, which reduces to G EB G MP if the events were compatible but completely independent, and which reduces to [G EB P(EB) + G MP P(MP) – G EB G MP P(EB)P(MP)]/[P(EB) + P(MP) – P(EB)P(MP)] if the events were compatible but completely independent. In the case of compatible and dependent events it will be necessary to evaluate correlations of the type corr(EQ,EB ∩ MP), i.e. concerning the concurrence of EB and MP events. Ultimately, these probability gains define the increase in conditional probabilities of the occurrence of an earthquake of a certain minimum magnitude compared to the observation of the two observables together or only one of the two, regardless of which. In a model verification scenario, the region where the alarm is activated is specified for each observable, as illustrated in Fig. 2, for instance. If this region is not identical for both observables, then the probability gains of the observables need to be recalculated within the shared area. In the presence of