January 2026
SPARK: Seminar Platform for Adaptive Resilience Knowledge
Join us for our next seminar:
A Regionalised, Partially Non-Ergodic Ground Motion Model for Subduction Earthquakes in South America
Abtract
Ground-motion models (GMMs) are central to seismic hazard and loss assessment, yet traditional global models rely on the ergodic assumption, treating all variability as randomness and assuming transferability across tectonic regions. Growing evidence shows that systematic source, path, and site differences vary regionally and can be modelled explicitly, reducing aleatory variability and motivating partially or fully non-ergodic approaches. Although several global subduction GMMs include regionalised coefficients or empirical site adjustments to capture these differences, they still perform inconsistently in South America—overpredicting shaking in the northern Nazca segment and underpredicting it in the southern Andes. Gaps also remain in representing mainshocks, aftershocks, and foreshocks, and in explaining the elevated short-period between-event variability known as the “bump in tau”. At longer periods, within-event variability remains high because the widely used average shear-wave velocity in the top 30 m (V_s30) correlates poorly with true site response, which is strongly influenced by deeper shear-wave structures. Orientation-independent measures such as RotD50 are widely adopted, though studies indicate they underestimate directionality compared with maximum-component metrics.
To address these limitations, this thesis developed the SISMA database, containing 38,841 high-quality three-component recordings from Peru, Chile, Colombia, Ecuador, and Argentina. All data were processed using harmonised procedures consistent with NGA standards, and comprehensive source, path, and site metadata were assembled. A new classification model for interface and intraslab earthquakes was developed using the Fourier-amplitude-spectrum S-wave method based on horizontal-to-vertical spectral ratios. This approach proved more stable than spectral-acceleration-based alternatives and showed that topographic amplification exerts a stronger influence on site response than deeper velocity structures. The study also quantified regional differences in anelastic attenuation, incorporated foreshock and aftershock effects, and demonstrated that a quadrilinear magnitude-scaling model reduces between-event residuals relative to the conventional bilinear form. Finally, a new orientation-independent intensity measure, MaxRotD50, was introduced and used alongside RotD50 to develop a regionalised, partially nonergodic GMM for subduction earthquakes in South America. Comparisons with regionalised global models identified path-related and depth-scaling effects as primary drivers of the “bump in tau” and demonstrated substantial reductions in long-period within-event variability.
Date/Time: Friday, January 16, 2026, 10:00 AM EST
Speaker: Dr. Jorge Luis Paredes Estacio
