The time‐averaged shear‐wave velocity in the top 30 m, VS30, is used to represent the site condition in many ground‐motion models (GMMs). Regionalized GMMs account for regional differences in the ln(VS30) scaling by including region‐specific coefficients for the site term. For example, recent GMMs developed for subduction zone earthquakes as part of the Next Generation Attenuation‐Subduction (NGA‐Sub) project include region‐specific VS30 scaling for seven regions: South America, Central America, Japan, New Zealand, Taiwan, Cascadia (Pacific Northwest America), and Alaska. Of these seven regions, South America has the largest within‐event standard deviation. One cause for the larger within‐event standard deviation is a weaker relation between VS30 and the site term in South America compared to other regions. A cause for this weak correlation is that most of the VS30 values for the South American region in the NGA‐Sub data set are based on proxies. The relation between VS30 and the site term improves significantly for periods less than 1 s if only stations with measured VS30 are considered, but the correlation is weak for periods greater than 1 s even for stations with measured VS30, indicating that, for the South America region, VS30 is not well correlated with the deeper shear‐wave velocity profile that controls the long‐period amplification. To improve the site terms for South America, we develop a model based on the horizontal‐to‐vertical spectral ratio (HVSR) from microtremors, similar to the approach used by Pinilla‐Ramos et al. (2022) for California. The data set includes 660 recordings from 51 earthquakes recorded at 274 sites. The site‐term model includes the period‐dependent HVSR amplitude and the geometric mean of the average HVSR amplitude over the frequency band of 0.25 to 15 Hz. Including these two parameters reduces the standard deviation of site‐specific site terms over the period range 0.3–3 s, with the largest reduction in the period range 1–2 s. This site‐term model can be incorporated into subduction GMMs and implemented in probabilistic seismic hazard analyses for South America.

You do not have access to this content, please speak to your institutional administrator if you feel you should have access.