Smoothed Seismicity vs Traditional Source Models: A Comparative PSHA of the Western Gulf of Corinth relative to the ESHM20 source model benchmark

AbstractSource characterization is a fundamental component of Probabilistic Seismic Hazard Assessment (PSHA). Traditional models rely on crustal area source zones, which can introduce artificial hazard boundaries or excessively smooth out the spatial distribution of maximum expected ground motions, particularly within large zones that have highly localized seismicity. Conversely, grid-based smoothed seismicity models offer a purely data-driven alternative; yet their direct impact on PSHA models requires evaluation against benchmarks. In this study, we develop a comparative PSHA for the Western Gulf of Corinth (Central Greece), an active half-graben system characterized by high deformation rates and a well-documented history of strong earthquakes. We quantify the differences between data-driven spatial seismicity models and the traditional source zones defined in the latest European Seismic Hazard Model (ESHM20). Specifically, we implement an optimized grid-based model derived from a novel, non-declustered smoothed seismicity approach, designed to effectively represent the high spatial heterogeneity of epicentral density in the region. To isolate the epistemic uncertainty originating exclusively from the seismic source characterization, the Ground Motion Model logic tree is kept strictly identical across all tested models. We compute hazard metrics, including site-specific hazard curves for key settlements and regional seismic hazard maps. Through statistical comparisons, i.e. incorporating Peak Ground Acceleration spatial ratio maps and boxplot distributions at standard return periods, we evaluate how empirical spatial rates influence hazard estimates compared to zones. We expect these comparisons to highlight how effectively grid-based models can capture abrupt seismic hazard level transitions without imposing artificial boundaries through area source modeling. The anticipated results will provide critical insights into the viability of integrating purely empirical, data-driven seismicity frameworks into formal hazard assessments for high-seismicity regions.