Improve earthquake kinematic models on finite fault through dynamic constraints

Kinematic source models are essential tools for understanding earthquake rupture history and predicting ground motion. By characterizing the spatial and temporal evolution of slip on fault planes, these models are crucial for seismology and seismic hazard assessments. However, standard kinematic inversions often lack resolution, especially at high frequencies, failing to capture small-scale heterogeneities that significantly affect near-fault ground motion. This study enhances fault heterogeneity modeling and source time functions (STFs) to improve both low and high frequency ground motion simulations. Earthquake rupture models are categorized as point-source or finite-fault models. Point-source models are suitable for lower frequencies, while finite-fault models better capture spatial slip variability at higher frequencies, accurately representing spatial heterogeneities in slip, rupture velocity, and rise time. STFs describe the slip rate evolution across fault planes and can be modeled with either a single-window approach, assuming a fixed functional form, or a multi-window approach, which allows a complex slip history by summing multiple time windows. High-frequency ground motion modeling, crucial for seismic hazard assessment, requires integrating small-scale fault heterogeneities. Traditional kinematic models often omit these finer variations, resulting in missed high-frequency content. Stochastic models based on spatial random fields, such as the Von Karman autocorrelation function, offer a robust framework for introducing variability in slip, rise time, and rupture velocity, enabling realistic high-frequency seismic radiation essential for near-fault ground motion predictions. In the first part of this Thesis, we perform point source inversions for the Gorzano Fault in Central Italy, analyzing 134 events to characterize the fault’s geometry and kinematics. The results reveal a listric fault with frictional heterogeneities, informing a mechanical model that explains segment reactivation during seismic sequences. The second part focuses on implementing kinematic inversions that incorporate dynamically consistent STFs. We evaluated the impact of various STFs, including the Regularized Yoffe function. Our results indicate that when simple rupture scenarios are well-modeled with single-window approaches, STF choice critically shapes results. However for complex ruptures, our models show that it is critical to constrain at least the coarse variations in rupture velocity. The multi-window methods explored here allow for a more precise capture of rupture velocity variations, reducing sensitivity to the specific STF used. This approach achieves strong consistency in kinematic parameters and effectively reproduces observed ground motion, particularly in near-fault areas. Finally, we addressed high-frequency ground motion simulation by incorporating fault heterogeneities using the Von Karman autocorrelation function, generating realistic slip distributions that capture both large- and small-scale variations. By incorporating irregularities both in slip, rupture velocity, rise time, and acceleration time, we simulated ground motion up to 10 Hz for the 2016 Amatrice earthquake. Results demonstrate that accounting for these heterogeneities significantly improves high-frequency ground motion predictions, enhancing seismic hazard assessments and earthquake engineering applications. The findings of this PhD Thesis provide a refined approach to kinematic modeling, enhancing our understanding of both high-frequency ground motion generation and the variability in model outcomes based on assumptions about the source time function (STF), with significant implications for earthquake mechanics and seismic hazard assessment.