Estimate The Focal Mechanism of Earthquake in Indonesia By Using 1-D Convolutional Neural Network (CNN)

<p>Indonesia is located between three collisions of active plate tectonics (Pacific, Eurasia, and Australia), resulting in a high seismicity zone, especially along the subduction zone. Besides the subduction zone, there are also many faults as a result of these collisions. As the earthquake source, both are controlled by focal mechanisms. Focal mechanism is the geometry of fault movements. Unfortunately, Indonesia's earthquake catalog data is not complete. There is missing information in some focal mechanism data, especially the data with more than 6 Magnitudes between January 1st, 1973, and February 1<sup>st</sup>, 2023. To complete the focal mechanism data, 1-D Convolutional Neural Network (CNN) is applied as the common and powerful method of Machine Learning. Started by grouping the earthquake catalog data with clear focal mechanism information as the training data with its training label and otherwise as the test data with the unknown label, then applied these training and label data to convolutional layer with some neurons, CNN can estimate focal mechanism (label) of the test data. This process is done iteratively, and a good model is observed with little loss value in the L curve.</p>