Modelling seismic rupture scenario by means of a Fiber Bundle Model

The rupture processes of any heterogeneous material constitute a complex physical problem. Earthquakes are the result of rupture in the Earth's crust. This process is difficult to model deterministically due to the number of parameters and physical conditions, which are mostly unknown. Computational physics offers alternative ways to study the rupture process in the Earth's crust by generating synthetic seismic data using physical and statistical complied models. The Fiber Bundle model (FBM) is a cellular automata model, which can help describe the complex rupture processes in heterogeneous materials in a wide range of phenomena. In this work, we propose an alternative method to model earthquake rupture that depicts the main statistical characteristics of real seismicity based on FBM. In particular, we study the rupture stage related to the mainshock and the stress relaxation stage which produces the aftershocks. The mainshock simulator is developed to study the dynamical rupture and the magnitude behavior. This model requires few input parameters coming from observational source inversion to model dynamic characteristics. The aftershock simulator depicts the main statistical patterns in time, magnitude and space domains which are studied via parametric and statistical analyses. The novelty of this work lies in the study of the Earth dynamical rupture processes by using a cellular automata model which describes the general rupture of heterogeneous materials with simple constraints. This stochastic model requires a large enough number of numerical experiments to reduce uncertainties and to give robustness to the method. Therefore, we exploit High-Performance Computing (HPC), to enhance the number of numerical experiments. Furthermore, we introduce a Machine Learning (ML) approach to parameter screening to better approximate real earthquake statistical features.