Automated identification and timing of crustal phases for tomographic studies and applications to the Marmara region in NW Turkey
Final Report Abstract
We derive a refined hypocenter catalog for the seismicity M≥2.2 throughout the Sea of Marmara region in northwestern Turkey for the time period 2007-2010. Its unprecedented precision is based on automatically determined P- and S-wave travel times. They replace previously available, non-optimized manually derived travel times which are the major parameters to be improved in the face of an offshore location of the fault with no on- or near-fault stations. The quality of automatically determined travel times is carefully examined by comparing them to manual reference picks which were determined with a scheme emphasizing highest possible consistency and precision. The variance of the time differences between automatically determined and corresponding reference travel times is approximately 0.2s for P- and 1.2s for S-waves. The high accuracy obtained for the travel times results in an improved hypocenter catalog with fewer but well-located events that allow to image the major fault branches of the NAFZ below the Sea of Marmara. Evaluation of manually determined reference onset times of P- and S-waves of the three major crustal phases Pg, Pn, PmP and S wave pendants, shows that the inter arrival times can be very small. On the other hand the higher order statistics (HOS), the Akaike Information Criterion (AIC) as well as autoregressive forward prediction can only work accurately when the portion of a waveform around a phase onset to be timed has a minimum width. We therefor concluded that automatic picking of the secondary phases - one of the principal objectives of this project - cannot be achieved and instead decided to put the focus of the study on fault-zone imaging using well-located seismicity and to also study kinematics by determining composite fault-plane solutions using automatically determined first -motion polarities. First results for the new hypocenter catalog and focal mechanisms are presented here while the completion of the two ’new’ tasks is still ongoing. 75% of automatically determined first-motion polarities of P-wave onsets derived with our new algorithm coincide with manually determined reference values and can be used to calculate earthquake focal mechanisms. The large network aperture with lacking stations immediately above the seismicity and insufficient azimuthal station density prevents inversion for focal mechanisms of single events. Therefore we aim at calculating composite focal mechanisms for spatial event clusters. Despite considerable uncertainties the resulting focal mechanisms are in good agreement with the regional tectonic setting and with results from previous studies analyzing the same area but earlier time intervals.
