Onsite Early Prediction of PGA Using CNN With Multi-Scale and Multi-Domain P-Waves as Input

Although convolutional neural networks (CNN) have been applied successfully to many fields, the onsite earthquake early warning by CNN remains unexplored. This study aims predict peak ground acceleration (PGA) of incoming seismic waves using CNN, which is achieved analyzing first 3 s P-wave data collected from a single site. Because amplitude large and small earthquakes can differ, multi-scale input proposed in this order let observe different scales. Both time frequency domains are combined into multi-domain input, therefore aspects. Since only maximum absolute value history target output used instead raw value. The arrangement shows its superiority one directly inputting CNN. Moreover, predicted PGA accuracy approach higher than support vector regression that employed extracted features as input. also alert performances acceptable based on two independent damaging earthquakes.