Parameterisation of auto-spectral density of earthquake induced strong ground motions

This paper explores the possibility of a new parametric representation of auto-spectral density (ASD) that enables extraction of several meaningful insights. Recorded translational ground motions are considered first and the resulting parametric model is capable to address the scenario when seismic energy is distributed over multiple frequency bands. A slightly different form is developed to address the rotational ground motion. Variation of ASD over the footprint of seismic arrays, both large (SMART1) and dense (LSST), is studied. The assessment is based on spectral similitude using spectral contrast angles and distance correlation. Possible correlation of the parameters of the proposed ASD model against site and event characteristics is investigated based on selected ground motions from PEER database. Model parameters thus computed are used in synthetically predicting the ASD of ground motions at the LSST array and that shows good resemblance against the recorded events.