The emergence of collective behaviours in fault systems. From crustal mechanics to statistical seismology

Earthquakes are likely the most puzzling natural phenomena on our planet. Although the basic physical mechanisms producing crustal instabilities are fundamentally well-known, our scientific approach to seismicity is still unsatisfying and the far-reaching goal of skillful forecasts has been missed till now. Even though essential knowledge has been achieved in related fields from statistical physics to observational seismology and rock mechanics, a comprehensive view embracing key results from the parameters of the seismogenic source to the statistical features of sequences is yet to come. Therefore, the integration of different perspectives of earthquake science may greatly contribute to advance our ability to understand seismicity. My research activity aims to improve our comprehension of the physical processes leading to the occurrence of large earthquakes. As a young earthquake physicist, I have been devoting my efforts to explain how major events emerge in previously stable crustal volumes from small magnitude ones, and their mutual relationships. This thesis is a re-elaborated and harmonized report of a three years-long formative experience as a PhD student at the Department of Earth Sciences of Sapienza University and two visiting periods at the National and Kapodistrian University of Athens in Greece and Caltech, Pasadena, California. The final outcome is itself an original piece of research in the field of seismotectonics and statistical seismology reporting a series of coherent analyses and results. Its structure is not to be intended as a concluded work, but as an ongoing travel diary towards an increasingly confident view of the physical foundations for the next generations of seismic hazard models. My investigations focus on the mechanical processes enabling the emergence of seismicity in fault systems and its statistical properties. Especially, I analyze the connection between physical conditions promoting the occurrence of major quakes, background dynamics bringing them into being and seismic forecasting. My work provides evidence that large events tend to occur under somehow favorable circumstances; moreover, they are forewarned by long-lasting, progressive preparatory phases leading to the large-scale destabilization of crustal volumes. Although often silent, they can be highlighted using appropriate techniques such as the monitoring of the seismic response to stress perturbations and clustering analysis. This thesis consists of two main parts: in the first section, the physical origins of some key features of seismicity and their mutual relationships are discussed, while the second one is devoted to physics-based techniques for the identification of unstable crustal regions. The introduction provides a bird’s-eye view of earthquake physics and its connection to seismic hazard from the viewpoint of complex systems. I strongly emphasize the crucial role of collective behaviors of fault networks in shaping seismic activity. Furthermore, I advocate a cross-scale modeling of earthquake occurrence to grasp the essential traits determining the spatio-temporal evolution of seismicity. In the second chapter, I discuss the physical meaning of some fundamental laws of statistical seismology in the peculiar framework of optimization problems with implications for the relationships between their parameters and the tectonic environment. The following chapters show how the slip behavior of faults is mainly controlled by a few physical parameters through the action of feedback mechanisms. They do not only seem to tune the mechanics of faulting, but they also shape the key characteristics of telluric events from the composition of moment tensors to the temporal development of seismic sequences. Retracing the common thread of universal mechanisms as rulers of seismicity, the last chapters of the first part deal with the physical processes of faulting and their connection with rheology and fault styles, seismic paradoxes and the statistical properties of seismic sequences. Implications for the multiscale modeling from laboratory experiments to natural faults are also described. The second part of my thesis is devoted to the physical characterization of preparatory phases preceding large earthquakes. The sixth and seventh chapters show how the responsiveness of fault systems to stress perturbations increases before the mainshocks. They support the idea that major failures may be predated by the development of long-range interactions within crustal volumes ultimately resulting in large scale instabilities. I also demonstrate that big events tend to occur where minor and moderate seismicity shows a locally Poissonian and globallyclustered behavior (chapter 8). This phenomenon is explained in the light of the ability of crustal volumes to accommodate additional strain and fault systems to deplete it while preserving overall stability. At last, I investigate the seismogenic potential of earthquake clusters showing that a continuum exists within a wide fan of fault slip behaviors ranging from aseismic creep to cascading “foreshocks”. Here, for the first time, a physical explanation to justify why large earthquakes can be preceded by very different seismic patterns is proposed. Due to the development of long-range interactions while approaching instability, earthquake dynamics becomes intrinsically non-local. Consequently, faults show memory effects and strong sensitivity to the stress conditions of nearby crustal volumes.