Implementation P-Wave Detection in a Seismic Waveform Using PhaseNet

Indonesia is located in an active tectonic zone with a high frequency of earthquakes. One of the factors is due to Indonesia’s location, which is between the meeting of three tectonic plates: the Indo-Australian, Eurasian, and Pacific. The impact results in life loss, infrastructure damage, and other material losses. An earthquake monitoring instrument is essential to reduce these impacts. A seismometer and accelerometer are instruments that are used to monitor earthquakes. In processing wave data by seismic sensors, one is determining the arrival time of P-wave and S-wave from sensor recordings. P-wave is an essential parameter in the Earthquake Early Warning System stage. In the first second after the P wave is detected, the accelerometer can estimate earthquake parameters, assess the danger of earthquake motion, and be used as an alarm in real-time from accelerometer measurements. The picking process using traditional methods results in subjective interpretations and long processing times and tends to be inconsistent for extensive data. This study implemented PhaseNet based on deep learning for the detection of P-wave and S-wave in an earthquake event in the Bayah area on January 3, 2024. The result of this study shows that the phase score of the P -wave value in CSJI, CNJI, and PBJI is 0.858. 0.862 and 0.909, then S-wave values 0.634, 0.558, and 0.536. Model phase net has been successfully implemented using raw data signal accelerometer and is suitable for applying to areas with a specific scope and dynamic seismic.