Assessment of Liquefaction Effects on Ground Motion Frequency Parameters for Accelerogram-Based Liquefaction Detection

An accelerogram-based method has been recently developed to quickly assess the liquefaction occurrence using only the ground-motion records. In this method, two frequency-related ground-motion parameters, the richness of the low-frequency components (termed RL), and the temporal variation rate of the mean instantaneous frequency (termed MIFr), are extracted from accelerograms and used as explanatory variables of a logistic regression model that outputs a new probability indicator of liquefaction occurrence (termed LQI). This method achieves an overall accuracy of over 90% when classifying the liquefaction occurrence in both the training dataset and the validation dataset, and the method shows promising potential for applications in real-time disaster mitigation systems and rapid postearthquake loss estimations. However, the success of detecting liquefaction from accelerograms relies on the assumption that the liquefaction effect on the site-specific ground motions is more significant than other effects. In this study, we select eight factors representing the source, path, and site effects of a ground-motion record, and conduct correlation analyses between the eight independent factors and the proposed two frequency-related ground-motion parameters (RL and MIFr). The results reveal that, in general, liquefaction has a more significant effect on the site-specific ground-motion frequency parameters compared to other factors such as earthquake source, path, and site effects, which is the basis of the accelerogram-based liquefaction assessment methods. This study will benefit the advancement of the accelerogram-based methods for rapid liquefaction assessment.