The thickness of the seismogenic crust (Ts) controls crustal earthquakes. Its upper limit is the seismicity onset depth (SOD) while its base corresponds to the seismicity cutoff depth (SCD) that correlates with the brittle-ductile transition. Thus, it influences the magnitude and location of crustal earthquakes and knowledge of its geometry may aid in seismic hazard assessment. Here we present the first Ts map of the Andean margin. We follow the standard methodology using the statistical depth distribution of events on a grid of equally size square cells. However, we find it has flaws and develop a new approach, based on circular cells of variable radius that changes according to earthquake density. Our results indicate that Ts is heterogenous, showing three controls: thermal structure, subduction geometry and crustal thickness.

We investigate the relationship between the seismogenic behavior of global megathrusts and various subduction parameters. We performed a parametric approach by implementing three decision tree-based Machine Learning (ML) algorithms to predict the b-value of the frequency-magnitude relationship of seismicity as a non-linear combination of subduction variables (subducting plate age and roughness, slab dip, convergence speed and azimuth, distance to closest ridge and plate boundary). Using the Shapley Additive exPlanations (SHAP) to interpret the ML results, we observe that plate age and subduction dip are the most influential variables. The results suggest that older, shallow-dipping plates contribute to low b-values, indicating higher megathrust stress. This pattern is attributed to the higher rigidity of older plates, increasing flexural strength, and generating a shallow penetration angle, increasing the frictional interplate area and intensifying the megathrust stress. These findings offer new insights into the non-linear complexity of seismic behaviour at global scale.

Great subduction earthquakes rupture similar regions of the megathrust during successive events separated by centuries. Mapping these regions (asperities) and their boundaries (barriers) is important for our understanding of processes controlling megathrust segmentation and for seismic hazard assessment. We present a formal seismic segmentation model of the Chile Margin derived from a multivariate analysis of gravity anomalies, basal friction from critical taper and interplate locking from GPS velocities. These proxies integrate over timescales of 10 to 10 years and we analyze their combined spatial covariance using Principal Component Analysis (PCA), from which the Empirical Orthogonal Functions (EOFs) and Principal Components (PCs) are extracted. We find that the 1 and 1+2 PCA modes explain 44% and 67% of the total variance, respectively, suggesting a high degree of spatial correlation. Comparing our results with rupture areas of earthquakes and using the rate-and-state friction theory, we infer that trench-perpendicular PCs reflect the limits of the velocity-weakening (VW) seismogenic zone. Polarity changes in along-trench EOFs (mostly for gravity) are related to changes between unstable and conditionally-stable frictional regimes inside the VW zone and we used them to define the boundaries of 17 unitary segments. These segments correlate with first-order tectonic features of the Andes at >10-km scales as well as with zones of multi-segment ruptures at 10-km scale. We analyze the combined influence of subducting and upper-plate geologic features on the nature of persistent seismic barriers. Our results have implications for understanding seismotectonic processes along the Andean margin and elsewhere