Testing global geomorphological model as site proxy to predict ground-shaking amplification

Estimating site amplification of earthquake ground shaking at new sites and sites without any direct geotechnical measurements of site parameters remains a large challenge in seismic hazard assessment. Currently, the standard procedure is to use site proxies inferred from topographic slope from digital elevation models (DEMs). In this study, we test a geomorphological model for inferred regolith, soil and sediment depth by Pelletier et al. (2016). This model was originally developed as input for hydrology and ecosystem models and is based on several global values in addition to the topographic slope, including geological maps and water table data.

To test the suitability of the geomorphological model for ground-shaking prediction we derive the empirical site amplification for sites in Japan, Italy and California using different regional and global seismological datasets. We use the observed shaking amplification to test the correlation between the observed ground-shaking site amplification and the inferred site proxies and test the performance of site amplification models based on geomorphological proxies. We find that the geomorphological model works equally well or slightly better than the traditional inferred proxies. We therefore argue that this model is a promising alternative proxy that can be used for predicting site amplification on new sites and regions for which no geotechnical information exists (i.e. on a global level). This result has important implications for the development of the new generation of ground-shaking models used for shake maps and seismic hazard models.