A Bayesian Approach to Regional Phase Blockage Imputation at the Iranian Plateau

Date: 4/25/2019

Time: 06:00 PM

Room: Grand Ballroom

The presence of systematically censoring data can seriously undermine the ability to correctly predict model parameters. Sn is partially or completely blocked (left censored) when travelling through hot or absent lithospheric mantle, which is observed frequently at Iranian Plateau. To estimate the left censored amplitude in the Iranian Plateau, we have investigated the problem of measuring Sn Q in regions with large amounts of Sn blockage. We first developed two codes to simulate a left censored Sn amplitude reduction dataset and to impute the left censored amplitude reduction and Q, respectively. This simulation generates a synthetic dataset with both blocked Sn amplitude reduction and efficient Sn amplitude reduction. Then we take a Bayesian approach to impute the censored amplitude reduction data. We also applied our approach on real Sn amplitude data from stations throughout the northern Middle East where Sn blockage is very commonly observed. Comparing Sn Q models with a censored dataset and models that use both observed and imputed Sn amplitudes, we find the inclusion of imputed amplitude reduction results in a Q model with increased accuracy and resolution. To better understanding high frequency pahse blockage, we are also working on different dataset with a more physically realistic model generated by the software SW4. We have generated a large number of synthetic regional seismograms that we then used to estimate Sn Q as well as data blockage. We will then use these simulated seismograms to better understand the process and effect of phase blockage.

Presenting Author: Hongjun Hui

Authors