Predicting Lg Blockage in the Middle East Using a Bayesian Lasso Logistic Regression Model

The regional seismic phase Lg is an important tool for investigating bulk crustal property and discriminating seismic sources. In this study, we used Lg efficiency data from the Middle East to develop a Bayesian logistic regression model to predict the probabilities of Lg blockage. This approach provides us with a quantitative way to map the regions of Lg blockage as well as a method to reliably predict the likelihood of blockage. We observe blockage zones in the oceanic crust like the south Caspian Sea. We also observe high probabilities of Lg blockage in the continental orogenic belts, like the western Greater Caucasus, and moderate to high probabilities over most continental collisional boundaries. The high-probability patterns around the continental collision plate boundaries are domi-nated by long-distance Lg waves. The probability tomography model also suggests that the continental collisional processes would not necessarily block Lg but highly affect Lg propagation, especially for longer paths, due to crustal intrinsic attenuation (eastern Anatolia) or the scattering effect from changes in crustal waveguide (Zagros).