2026, Vol.29, No.3, pp.263 - 279

Due to the non-stationary and complex nonlinear dependency of seismic displacement data, classical cross-correlation methods are often inadequate for characterizing their relationships. This study applies a copula-based statistical framework to investigate linear and nonlinear dependence structures in seismic displacement time series recorded during the 6 February 2023 Gaziantep earthquake in Turkey. Using data from multiple IRIS seismic stations including Ankara, Kiev, and Flin Flon, we analyze vertical and horizontal ground displacement components over a 2.5-hour window. Clayton, Frank, and Gumbel copulas are employed to characterize lower-, symmetric-, and upper-tail dependence structures. Copula parameters are estimated within a Bayesian framework using the Metropolis-Hastings Markov Chain Monte Carlo (MCMC) method. The Kolmogorov-Smirnov statistic is used to evaluate goodness-of-fit. We examine both time-synchronized and time-lagged segments to identify optimal delays associated with maximum dependence. Results indicate strong dependence for nearby stations (Ankara--Kiev) and weaker but persistent long-range dependence for distant stations (Ankara-Flin Flon). Furthermore, lower- and upper-tail dependencies correspond to small- and large-scale ground motion features. Time delay analysis shows that horizontal components consistently reach peak dependence earlier than vertical components.

Key words: copula model, Markov Chain Monte Carlo algorithm, Bayesian framework, seismic time series

DOI: https://doi.org/10.5281/zenodo.23278101

Full text:  Acrobat PDF  (1815 KB)   Open Access   

Copyright © Nonlinear Phenomena in Complex Systems. Last updated: October 14, 2026