Earthquake analysis in East Java, Indonesia between 1960 - 2017 using Markov chain model

Earthquake occurrence in East Java has been analyzed using a Markov Chain Model. The catalog data comprised of earthquake events from East Java and its vicinity from 1960 until 2017. Data were classified into subduction earthquake and inland fault earthquake. After that, the data were declustered using Reasenberg (1985) algorithm for removing the aftershock. Spatial analysis was conducted by dividing the research area into nine regions. Probability transition matrices which give information about the highest probability transition of earthquake occurrence in each region were calculated for different magnitude thresholds. Furthermore, the Chi-Square test was applied to examine the independence between earthquakes occurrence. For seismic hazard analysis purpose, the temporal Markov chain analysis was employed by determining the active (1) and inactive (0) period in each region based on the occurrence of earthquakes with M ≥ 5 and depth d < 70 km. From the two-state probability transition, the mean duration of active and inactive states in each region have been obtained. Both spatial and magnitude analysis results inferred that subduction earthquake with M ≥ 4, M ≥ 5, and inland fault earthquake with M ≥ 3 exhibited strong first-order Markov property, i.e. there was a robust dependency between an earthquake occurrence and the successive occurrence. The mean duration of the inactive state in the research area varied from 2.5 until 13.5 years.