Broadband Strong Ground Motion Modeling Using Planar Dynamic Rupture With Fractal Parameters

Dynamic rupture modeling represents a promising physics-based approach to strong ground motion simulations. However, its application in a broad frequency range (0-10 Hz), interesting for engineering studies, is challenging. The main reason is that widely used and relatively simple planar fault models with smooth distributions of initial stress and frictional parameters, or even self-similar initial stress, result in ground motions depleted in high-frequency content. Here we propose an efficient approach for the linear slip-weakening friction model on a planar fault based on the Ide and Aochi (2005, ) multiscale model with a small-scale random fractal distribution of the slip-weakening distance D-c. We propose a way to combine these variations with a large-scale deterministic dynamic model. We illustrate the approach on an elliptical model and a smooth model of the 2016 M-w 6.2 Amatrice, Italy, earthquake from low-frequency dynamic rupture inversion. To intensify the incoherence of the rupture propagation, we also include a variation of the strength and initial stress, both correlated with D-c. These additions result in sustained high-frequency radiation during the whole rupture propagation and omega-square source time functions. The new model of the Amatrice earthquake generates synthetics that agree with the local ground motion model up to 5 Hz in terms of spectral accelerations while preserving the average and integral dynamic rupture parameters (e.g., stress drop, fracture and radiated energy). The fractal dynamic model can be easily implemented in any dynamic rupture propagation code and is thus readily applicable in broadband physics-based ground motion predictions for earthquake scenarios in seismic hazard assessment.

Automated summary

Dynamic rupture modeling is a promising approach for simulating strong ground motion based on physics. However, its application in the frequency range of interest for engineering studies is challenging due to the limitations of existing fault models. In this study, we propose an efficient approach that combines the linear slip-weakening friction model with a small-scale random fractal distribution to generate high-frequency content in ground motions. We demonstrate the effectiveness of the approach using models of the 2016 Amatrice earthquake.