MAGNITUDE SIMULATION USING THE GENERALIZED PARETO DISTRIBUTION

In the seismic hazard assessment, the magnitude of the seismic scenarios is simulated using the Monte Carlo simulation (MCS) approaches. The magnitude simulation is based on recurrence-magnitude relationship or the Gumbel distribution. In the current study, the appropriate distribution of the magnitude is investigated and the extreme value theory (EVT) and the generalized Pareto distribution (GPD) are identified to produce the magnitude of seismic scenarios. The used seismic catalog of Tehran (radius=100 km) is extracted from the International Institute of Earthquake Engineering and Seismology (IIEES) database from 1930-2019. In order to remove the temporally and spatiallydependent events (foreshocks and aftershocks) of the catalog, the Gardner and Knopoff (1974) declustering algorithm is applied. The distribution of the magnitude using the Normal distribution, Lognormal distribution, Generalized Extreme value (GEV), Exponential distribution, Inverse Gaussian distribution and the GPD is examined. The negative of log likelihood, the Akaike information criteria (AIC) and the Bayesian information criteria (BIC) are selected to compare the goodness of fitting of the mentioned distributions (Table 1). It is notable that the lowest values in the different distributions show the best fitting. The results of this comparison show that the GPD has the best fitting on the magnitude data.

Keywords:

Generalized Pareto distribution , Magnitude , Return level , Seismic hazard assessment