HVSR-Based Assessment of Soil Nonlinear Behavior During Strong Ground Motions: Insights from the 3 April 2024 Hualien Earthquake Sequence and Regional Comparisons

ABSTRACT In this study, a horizontal-to-vertical spectral ratio (HVSR)–based degree of nonlinearity (DNL) analysis was applied to investigate nonlinear soil behavior during the 2024 Hualien earthquake sequence in eastern Taiwan, including the 3 April mainshock (Mw 7.3) and two significant aftershocks (Mw 6.6 on 3 April and Mw 6.0 on 23 April). By comparing HVSRs from weak and strong motions, DNL values were computed for 18 stations and regressed against peak ground acceleration (PGA), peak ground velocity (PGV), and a strain proxy (PGV/VS30). All three regressions showed clear piecewise linear trends with identified nonlinear thresholds: 111 cm/s2 (PGA), 9.77 cm/s (PGV), and 0.0283% (strain proxy), with the last one showing the strongest correlation with DNL (correlation coefficient = 0.81). DNL–strain relationships from the 2018 Mw 6.4 Hualien earthquake generally fall within the regression model derived from the 2024 Hualien earthquake sequence, indicating similar nonlinear site behavior across events through a period of six years. In contrast, sites on the ChiaNan Plain, which are underlain by fine-grained sediments, exhibited more gradual nonlinear trends without a clear threshold, highlighting the influence of sediment composition. Variations in the vertical-to-horizontal ratios of the PGA and PGV further underscore the role of site conditions. These results offer empirical constraints that can improve the modeling of nonlinear site responses and contribute to applications in seismic microzonation, hazard assessment, and ground-motion prediction.