Modelling the earthquakes occuring on Turkey by Markov chains

In this study, it is aimed to predict seismic hazard by Markov chains, a stochastic process, by means of earthquake occurence data (magnitude M>=4 and from 1901 to 2006) of Turkey in which between North (36-42)-East (26-45) coordinates. For the purpose, Markov matrices for different ?t is obtained, the Markov matrices?s entropies are computed and the most correct matrix for our purpose is found by Maximum Entropy Principle. By the matrix, both regional transition probability matrix and stationary distribution is obtained, whether first passage time distribution is geometric distribution is tested by Ki-kare analysis. At any time, in the first or second regions, the probability of minimum k (k=1,2,3,4,5) earthquakes is computed. Also, whether the model is correct is tested in a simulation study by Ki-kare analysis. For magnitudes and time intervals ?t =0.07 year, the method yields a 81.125% aftcast success rate for the entire catalog. Finally, step probability distribution is estimated for five periods after 2006.Markov chains, seismic hazard, entropy, earthquake on Turkey