A Bayesian Updating Applied to Earthquake Ground-Motion Prediction Equations for Iran

Abstract

This paper presents new ground-motion prediction equations for three distinct seismic regions of Iran via updating the previous global model using observed data for each region by means of Bayesian updating. The Bayesian theory has the advantage that it results in more accurate results even in situations when little data is available. This leads the way for updating global models to obtain new local models for seismotectonic regions with little available data like Iran. The proposed updated model was compared against currently available models for Iran and the results reveal the overall stability and quality performance of the proposed model.

Merging data and experts’ knowledge-based weights for ranking GMPEs

“…A probability distribution is required to consider likeliness of different values of σ (Equation 4). In regard to the probability distribution function of σ , an informative prior in the form of a gamma-normal distribution was considered (Shahidzadeh and Yazdani, 2017; Wang and Takada, 2009). Experience shows that for log 10 -based GMPEs the value of σ is of order 0.3 (0.7 for natural logarithm).…”