Partitioning Site Amplification and κ ₀ in Earthquake Recordings

This study explores site amplification and spectral decay ( κ 0 ) estimates, focusing on the challenge of separating the two in observed Fourier amplitude spectra (FAS). We apply two general approaches: a Bayesian broadband inversion for simultaneous estimation of site amplification and κ 0 , and slope‐based methods for direct κ 0 estimation from the acceleration or displacement FAS with assumptions used to address site amplification. The broadband inversion results indicate that the approach is robust to the initial (prior) site amplification assumptions, with differences in priors mainly affecting high‐frequency amplification and changing κ 0 by less than 10%. Constraining amplification to theoretical values from shear wave velocity measurements influences the inverted κ 0 by a similar amount. The slope‐based methods consistently yield significantly different κ 0 from the acceleration FAS and displacement FAS, with the acceleration κ 0 values 0.02–0.04 s smaller than the displacement κ 0 . The slope‐based methods are strongly influenced by the assumptions associated with site amplification, as well as issues related to the proximity of the source corner frequency. The slope‐based κ 0 estimates show poor agreement and low correlation ( ρ = 0.13) with the broadband inversion κ 0 results, such that the broadband approach is recommended for estimating κ 0 . The observed discrepancies highlight the strong coupling between site amplification and κ 0 , underscoring the importance of using broadband inversion techniques for capturing site‐specific ground motion characteristics.