A Machine Learning Approach for Parameter Screening in Earthquake Simulation

Earthquakes are the result of rupture in the Earth's crust. The rupture process is difficult to model deterministically due to the number of unmeasurable parameters involved and poorly constrained physical conditions, as well as the very diverse scales involved in their nucleation (meters) and complete failure (up to hundreds of kilometers). In this research work we focus on synthetic seismic catalogs generated with a stochastic modeling technique called Fiber Bundle Model (FBM). These catalogs can be readily compared with statistical measures computed from real earthquake series, but the link between the FBM parameters and the characteristics of the obtained earthquake series is difficult to assess. Furthermore, the stochastic nature of the model requires a large amount of realizations in order to attain statistical robustness. The aim of this work is to estimate the FBM parameters that generate aftershock sequences that are similar to those generated by real seismic events. In order to estimate the optimal combination of parameters that generate such sequences, we executed a large number of simulations with different combinations of parameters using High-Performance Computing (HPC) resources to reduce compute time. Lastly, the synthetic datasets were used to train a supervised Machine Learning (ML) model that analyzes and extracts statistical patterns that reproduce the observations regarding aftershock occurrence and its spatio-temporal distribution in real seismic events.