A Prediction Method of Ground Motion for Regions without Available Observation Data (LGB-FS) and Its Application to both Yangbi and Maduo Earthquakes in 2021

Currently available earthquake attenuation equations are locally applicable, and methods based on observation data are not applicable in areas without available observation data. To solve the above problems and further improve the prediction accuracy of ground motion parameters, we present a prediction model referred to as a light gradient boosting machine with feature selection (LGB-FS). It is based on a light gradient boosting machine (LightGBM) constructed using historical strong motion data from the NGA-west2 database and can quickly simulate the distribution of strong motion near the epicenter after an earthquake. Cases study shows that compared with GMPE methods and those based on real-time observation data, the model has a better prediction effect in areas without available observation data and can be applied to Yangbi Earthquake and Maduo Earthquake. The feature importance evaluation based on both information gains and partial dependence plots (PDPs) reveals the complex relationships between multiple factors and ground motion parameters, allowing us to better understand their mechanisms and connections.