Scaling and Variability of Broadband Earthquake Source Spectra in Japan

Accurate evaluation of the scaling and variability of earthquake source spectra is essential for reliable ground-motion prediction. This study presents a comprehensive analysis of the broadband source spectral characteristics of crustal, interplate, and intraslab earthquakes in Japan (2004–2024, Mw≤7.5), aiming to improve the predictability of the scaling and variability of ground motions. A multistage spectral ratio method was applied to estimate source spectra and source parameters, such as the apparent stress and the stress parameter, which represents a form of spectral stress drop that characterizes the high-frequency level. This method enables the estimation of double-corner-frequency (DCF) spectra in contrast to the standard approach that assumes single-corner-frequency (SCF) spectra. A systematic transition from SCF to DCF spectra with increasing magnitude was observed across all earthquake types, suggesting a fundamental difference in rupture characteristics between small and large earthquakes. The variability of the stress parameter is consistent with the between-event variability of peak ground acceleration (PGA), suggesting that the long-standing variability discrepancy between spectral stress drop and PGA has been resolved for all earthquake types. Moreover, PGAs for crustal earthquakes were successfully predicted from the estimated spectral parameters, as indicated by their strong correlation with observations. These findings highlight the effectiveness of the DCF model for improving ground-motion predictions. The study also offers broader implications for source parameter estimation, scaling relations, rupture dynamics, and ground-motion prediction. In conclusion, greater attention to the potential of DCF spectra may yield deeper insights into earthquake rupture physics and contribute to more accurate and robust ground-motion prediction.