APPLICATION OF DEEP LEARNING PHASENET, GAMMA EVENT ASSOCIATION, AND HYPOCENTER DETERMINATION WITH NONLINLOC METHOD IN MICROSEISMIC EARTHQUAKES IN GEOTHERMAL AREA

• Main Author: Ihsan Sagara Putera, Muhammad
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/79169
• Institution: Institut Teknologi Bandung

Indonesia is a country with large geothermal potential, therefore it needs to be accompanied by monitoring of the geothermal system in its use. One of the geothermal monitoring carried out is seismicity monitoring in geothermal areas. This monitoring is related to fluid recharge, both from artificial recharge and natural recharge around faults, which can cause pressure and fracturing in rock pores. Therefore, the authors carried out an analysis of the seismicity monitoring of geothermal areas automatically using PhaseNet and compared it with catalog data. PhaseNet is machine learning that can determine the arrival time of P and S waves automatically. Next, phase association is carried out using the Gasussian Mixture Model Associator (GaMMA). The final stage of data processing is determining the location of the earthquake with Nonlinloc. In this research, waveform data from the geothermal area for the period January – March 2022 was used. PhaseNet can carry out picking but still has shortcomings in selecting P and S waves for an event. From the PhaseNet picking results, event association was carried out with GaMMA and quality control with Wadati diagrams which resulted in 18 events. Next, the 18 events are input into NonLinLoc to determine the hypocenter. 15 NonlinLoc events have hypocenter differences of less than one kilometer from the catalog hypocenter and three other events have more than one kilometer.