Clustering of seismicity in the Indonesian Region for the 2018-2020 Period using the DBSCAN Algorithm

Abstract

Indonesia is located at the confluence of 3 large, active plates that are constantly moving. Therefore, Indonesia is one of the countries that has a high level of seismicity risk. This study aims to classify seismicity data in the Indonesian region based on coordinate data which contains variable data on frequency of occurrence, depth, and strength of seismicity. Seismicity data was obtained from the BMKG official website using data for the period 2018 to 2020. The clustering technique used was the DBSCAN algorithm. This algorithm requires epsilon and MinPts input parameters. The results of the cluster formed will then be validated using silhouette coefficients. Based on the coordinate data, 4 clusters were formed with 4 disturbances. Based on the characteristic data, 3 clusters were formed with 5 disturbances. The silhouette coefficient obtained was 0.35 for coordinate data and 0.39 for characteristic data. This research is useful for increasing the use value of abundant seismicity information and can be used as an effort to mitigate seismicity natural disasters.