3-D P-wave velocity structure beneath eastern Turkey applying local earthquake tomography (LET) method

Abstract

Eastern Turkey is a seismically active region exhibiting a complex structure in terms of both tectonic and geologic· features. A temporary seismic network consisting of 29 broadband three-component broadband stations were installed in the scope of Eastern Turkey Seismic Experiment (ETSE) to address the important questions. Detailed and accurate evidences relevant to the crustal and upper mantle structure of the region were obtained using the seismic data collected in between October 1999 and August 2001. Results of the ETSE project had a considerable contribution to understanding and interpreting the tectonic evolution along with the seismicity of the region. In this study, local earthquake data from the ETSE project were used in order to determine the upper crust seismic P wave velocity structure of Eastern Turkey and the surrounding regions by three-dimens}onal (3-D) Local Earthquake Tomography (LET) method. As far as resolution is concerned, S waves were not included in this study due to strong attenuation, insufficient number of S phase readings and higher picking errors with respect to P phases. LET is the 3-D imaging process ofthe velocity structure of a target volume by using the travel time data of the earthquakes recorded within the seismic network in an active region. To provide accurate tomographic results, a high quality data set, initial hypocentral parameters and a minimum one-dimensional (1-D) velocity model that adequately represents the region are required. Initially, data compilation and earthquake locations were determined. Initial locations of 9*4 events were performed by HYPOCENTER algorithm using an initial 1-D velocity model, which was obtained from the previous studies in the region. Following the earthquake location procedure, higher quality events were selected in order to construct the 1-D minimum velocity model for the region. As a basic data selection criteria, events with azimuthal gap (GAP) ::; 200° and number of observations (P) ~ 8 were selected from the initial data set (6978 P-phase readings). 1-D velocity model was calculated by VELEST algorithm performing a simultaneous and iterative 1- D velocity inversion. New hypocentral parameters and station corrections were also calculated in addition to the minimum velocity model. Three different trial velocity models were used in order to construct an initial velocity model based on the results of a set of inversions with ten iterations and four calculation steps. Each velocity model were constructed in 0-42 depth range in accordance with the crustal structure of the region, however, accurate results were obtained down to~ 15-20 km due to the depth distribution ofthe events. The relocation of the earthquakes was performed by VELEST via the 1-D minimum velocity model and the relevant station corrections. In the final step, various stability tests were applied to check the resolution capability minimum 1-D velocity model. As a result of these tests, it was suggested that 1-D minimum velocity model adequately represented the region. After the derivation of the most appropriate velocity model, 3-D tomographic inversions were applied to the final data set. Events with GAP :::;; 180° and P > 8 were reselected from the data set. Number of events for the tomographic inversion reduced to 504 after applying this criteria (Total number of 6742 P-phase readings). 3-D tomographic inversion was iteratively and simultaneously performed by SIMULPS 14 algorithm using node-grid geometry for model parameterization. Considering the event-station distribution, a horizontal grid with 30x30 km grid spacing was chosen. In vertical direction, depth values taken from the 1-D velocity model were used. Modeling was made down to 42 km. Tomographic inversions with four iterations and two processing steps were initiated after determining the appropriate control parameters and the damping factor. At the end of these processes, 3-D P wave velocity model and the resulting hypocenters were determined. A significant reduction in data variance (- %50) and in residuals (-%50) was observed during these processes. In order to assess the solution quality and the resolving power of the 3-D model, tests with the synthetic data were performed. Critical parameters affecting the resolution estimates were calculated and mapped along with absolute velocities (Vp) and% perturbations relative to the 1-D initial velocity model in both horizontal and vertical cross-sections. Consequently, after the tomographic applications, the compatibility of the results with the tectonic and seismological features of the region were evaluated and also compar~d with the results of the previous studies in the region. The differences between the initial and the final hypocentral parameters were emphasized in various cross-sections. XV