A probabilistic approach toward earthquake hazard assessment using two first-order Markov models in Northeastern India

Abstract

In this paper, a probabilistic analysis of earthquakes of a seismically active region [Northeast (NE) India] is carried out in the temporal domain. Two models have been used; the first one is for the probability estimation of at least one major earthquake striking the area under consideration within a definite span of time in the future. The second model is used to evaluate the likelihood of occurrences of earthquakes falling in different magnitude ranges, just following an earthquake having a recorded magnitude. Both the models are Markovian in nature and model the occurrences of earthquakes as a first-order Markov process. The first model predicts the long-term risks of the region of experiencing at least one major earthquake, and the second model predicts the immediate short-term risks. NE India, which is one of the seismically most active regions in the world, is chosen for the present study. Earthquake catalog for NE Indian region is prepared from various sources. The geology, tectonic setup and the seismicity of the area are used to classify the study region into three seismotectonic polygons. Both the Markovian models are applied to each of these zones, and their seismic hazards are estimated. Finally, it is also shown that how the accuracy of the results predicted by the two models is affected by incompleteness in the dataset and how it inevitably leads us to the conclusion that while the first model can be used even with a relatively high degree of incomplete data, the second model simply fails if the dataset is incomplete.