Corrado, Paola (2025) Parametrization of seismic sources and quantification of uncertainties in probabilistic seismic and tsunami forecasts. [Tesi di dottorato]

Documento PDF corrado_paola_38.pdf Download (16MB)

Abstract

Earthquakes represent one of the most complex and unpredictable natural hazards, and their statistical characterization is essential for seismic hazard and risk assessment. Despite decades of research, the mechanisms controlling the size distribution of earthquakes and the identification of possible precursory patterns remain debated. This uncertainty stems both from physical complexity and from methodological limitations, such as catalog incompleteness and biases in magnitude estimation. This thesis addresses these limitations from different perspectives. First, it investigates the role of catalog completeness and magnitude correlations in shaping statistical inferences on earthquake occurrence. Using synthetic and real earthquake catalogs, I show that apparent magnitude correlations and anomalously low b-values are often artifacts of short-term incompleteness rather than evidence of precursory processes. Second, the thesis explores how physical and structural factors of the lithosphere control earthquake size distributions. I show that variations in the b-value correlate not only with regional heat flow and tectonic regime, but also with fault geometry and kinematics. These findings indicate that earthquake statistics are not universal but strongly conditioned by both geodynamic environment and local fault architecture. Third, I develop a Bayesian framework for focal mechanism forecasts that explicitly accounts for magnitude dependence. Accurate forecasts of focal mechanisms are fundamental not only for seismic hazard assessment but also for tsunami hazard modeling, which requires rapid knowledge of fault geometry and kinematics immediately after an earthquake. The results show that ignoring magnitude dependence introduces systematic biases in hazard models, whereas it produces more reliable and physically consistent forecasts with direct implications for both seismic and tsunami risk assessment. This approach provides a foundation for incorporating physical constraints into operational forecasting in near-real-time scenarios. Finally, the thesis examines the identification of foreshocks. A comparison between Nearest Neighbor (NN) and ETAS-based declustering reveals that the widely used NN method, although popular, fails to capture the full range of foreshock properties defined by ETAS. This finding suggests that the identification of foreshocks is sensitive to the chosen methodology and that caution is required when interpreting them as genuine physical precursors. In summary, this thesis extends our understanding of how both methodological choices and physical processes shape earthquake statistics. By emphasizing the importance of unbiased magnitudes, consistent catalogs, and physically informed models, it contributes to more reliable earthquake forecasting and hazard assessment. Taken together, the results indicate that progress toward answering the central question—whether earthquakes are governed purely by chance or by systematic physical rules—requires integrating high-quality data, robust statistical frameworks, and geodynamic context.

Downloads

Actions (login required)

Modifica documento