Forecasting seismicity rates in western Turkey as inferred from earthquake catalog and stressing history

The spatio-temporal variation in seismicity in western Turkey since the late 1970s is investigated through a rate/state model, which considers the stressing history to forecast the reference seismicity rate evolution. The basic catalog was divided according to specific criteria into four subsets, which correspond to areas exhibiting almost identical seismotectonic features. Completeness magnitude and reference seismicity rates are individually calculated for each subset. The forecasting periods are selected to be the inter-seismic time intervals between successive strong (M ≥ 5.8) earthquakes. The Coulomb stress changes associated with their coseismic slip are considered, along with the constant stressing rate to alter the rates of earthquake production. These rates are expressed by a probability density function and smoothed over the study area with different degrees of smoothing. The influence of the rate/state parameters in the model efficiency is explored by evaluating the Pearson linear correlation coefficient between simulated and observed earthquake occurrence rates along with its 95 % confidence limits. Application of different parameter values is attempted for the sensitivity of the calculated seismicity rates and their fit to the real data to be tested. Despite the ambiguities and the difficulties involved in the experimental parameter value determination, the results demonstrate that the present formulation and the available datasets are sufficient enough to contribute to seismic hazard assessment starting from a point such far back in time.