Analisis Risiko Gempabumi di Sumatera dengan Cauchy Cluster Process

Abstract

Sumatra is one of the prone areas in Indonesia to earthquakes. This condition is due to its geography which is traversed by active faults, subduction zones, and volcanoes. In this study, the distribution of earthquakes occurrences in Sumatra is modeled by considering the effects of spatial trends due to subduction zones, active faults, and volcanoes and also considering the cluster effects caused by mainshock and aftershock activities using the inhomogeneous Cauchy cluster process. In spatial trend modeling, there are indications that there is multicollinearity issue characterized by a high correlation among geographical factors, so this study considers ridge regularization to overcome this problem. The results of data exploration show that the earthquakes in Sumatra are not homogeneous and form clusters due to geological factors such as the presence of volcanoes, subduction zones, and active faults. Earthquake intensity modeling with ridge regularization produces an AIC value of -2280648 which is smaller than the model without regularization. The Cauchy cluster model by considering ridge regularization resulted in an estimated number of 63 mainshocks with a standard deviation of aftershocks around the mainshocks of 17.685 km. The closer a location to a fault, the risk of an earthquake occurring at that location increases by 1.6972 times. The closer a location to a subduction zone, the risk of an earthquake at that location increases by 1,25899 times, and the closer a location is to a volcano, the risk of an earthquake at that location increases by 1.55910 times.