Reconciling Conflicting Hazard Maps: A Methodological Reappraisal of Poisson and McGuire-Style PSHA Models in Urban Taiwan

This study advances seismic forecasting and probabilistic hazard analysis in tectonically active regions by systematically evaluating how alternative model assumptions influence risk estimates. Taiwan, as a tectonically complex region, serves as the basis for comparing Poisson-based and McGuire-style probabilistic seismic hazard assessment (PSHA) models through comprehensive simulations and in-depth evaluation of six major urban centers. The McGuire-style model, which integrates active fault geometry, recurrence intervals, and magnitude–distance distributions, consistently predicts higher peak ground acceleration (PGA) values, especially in eastern and southern Taiwan, where seismic hazard is high but often underestimated by time-independent approaches. In contrast, the Poisson model, grounded in historical catalogs, tends to smooth hazard estimates and overlook localized risk. Deaggregation analyses demonstrate that the McGuire model more accurately captures the dominant near-field sources, thereby enhancing site-specific hazard attribution. To reconcile these divergences, we propose a hybrid PSHA framework that combines empirical coverage with fault-based specificity, enabling spatially explicit and source-sensitive hazard forecasting. This dual-model approach improves predictive accuracy and supports resilience-oriented seismic planning. The findings provide a robust scientific foundation for earthquake risk governance in complex tectonic settings and contribute to global efforts in disaster risk reduction, aligning with the Sendai Framework.