EARTHQUAKE LOCATION ACCURACY IN THE ARABIAN-EURASIAN COLLISION ZONEEric A. Bergman1, Eric R. Engdahl1, Michael H. Ritzwoller1, and Stephen C. Myers2University of Colorado1 and Lawrence Livermore National Laboratory2Sponsored by Air Force Research Laboratory1Sponsored by National Nuclear Security Administration2Contract No. FA8718-08-C-00201 and LL08-BAA08-69-NDD032ABSTRACTAn extensive and high-quality data set of earthquake locations, many of them calibrated to GT590 levels of accuracy, is being compiled for the region that expresses the continental collision between the Arabian and Eurasian plates, with the primary goal of obtaining improved resolution of the crustal and upper mantle velocity structure through tomographic imaging. A specialized multiple event relocation algorithm is used to calibrate clusters of earthquakes, using several kinds of calibration data, including near-source seismic observations, InSAR data on source location and mechanism, and mapped faulting. Although our data set of calibrated locations (880 events in 21 calibrated clusters) is substantial, compared to what is available for most other regions of the Earth, it is still small in comparison with the full catalog of earthquake locations (~26,000 events) in the region which can be used for tomographic studies. Our strategy is to use the observed arrival times from GT5 events set to study regions in which the ray path coverage is dense enough, and to use these calibrated travel times for validation in regions for which it is necessary to use arrival time data from uncalibrated sources for tomography. In addition to providing data sets for tomography, the calibrated earthquake locations provide insight into the performance of permanent seismic networks that monitor the region. The location accuracy of two global networks and two regional networks, is investigated by comparing their locations with the calibrated locations (the subset of 684 events meeting GT590 criteria) determined in this project. Earthquake locations in this region which are based on regional and teleseismic arrival time data are systematically biased to the southwest and have a 90% location accuracy of 18-23 km, with the lower value achievable by applying limits on secondary azimuth gap. The two regional seismograph networks achieve a 90% accuracy level of 22 and 36 km, respectively. One regional network actually operated as a set of independent subnetworks until 2006 and suffers from a very imbalanced distribution of stations, such that the 90% level of accuracy exceeds 100 km during that time. In broad terms, location accuracy in the region using 1-D travel time models and single event location procedures, is presently limited to about 20 km at the 90% level of accuracy (i.e., GT2090). The tendency for routinely-determined epicenters to be biased to the south can be explained as a consequence of the fact that there are far more seismic stations in the northern hemisphere, relative to the study region, and that the true travel times are larger than the theoretical ones, yielding positive travel-time residuals. The inversion for location tends to push an event further away from the bulk of seismic stations, i.e., to the south.OBJECTIVESThis research has the goal of developing in-country datasets that can be used to improve ground-based monitoring capabilities in the study region, in particular by providing information needed to develop and test more accurate travel time models for seismic phases that propagate in the crust and upper mantle in the region that expresses the continental collision between the Arabian and Eurasian plates. The main thrust of this project is developing high quality data sets for tomography. The research presented here, however, deals with an assessment of the current level of location accuracy in the region of interest from several permanent seismic networks.As part of this project we are conducting detailed analyses of clusters of earthquakes with a multiple event relocation method that has been developed to determine calibrated locations that meet or exceed GT590 levels of accuracy. The success of this analysis usually depends on the availability of near-source data, such as temporary deployments of seismometers for aftershock studies and geological field work after large earthquakes. In-country data sets therefore play a critical role in this kind of analysis. In the region of interest we have developed 21 calibrated earthquake clusters, containing 880 events; 684 of these events qualify as GT590 locations. We compare the epicenters (we do not consider focal depth or origin time) reported in two global seismic catalogs and two regional seismic catalogs with the GT5 locations to gain an understanding of the average level of location accuracy of different networks and also to reveal any evidence for systematic mis-location. All locations in this study, for the multiple event relocation analysis used to determine GT5 locations or the catalogs used for comparison, are based on 1-D Earth models, although the models may differ from cluster to cluster and from catalog to catalog.The global catalogs considered in this study are those of the International Seismological Centre (ISC) and a version of the well-known EHB catalog (Engdahl et al., 1998) that has been carefully reviewed for the region of interest. The two regional catalogs are those published by the International Institute for Earthquake Engineering and Seismology (IIEES) and the Iranian Seismological Centre (IRSC). The IRSC is contained within the Institute of Geophysics of the University of Tehran. The IIEES catalog is based on data from the Iranian National Seismic Network (INSN), consisting of 16 broadband (Güralp CMG-3T seismometers, Güralp digitizers), real-time satellite-telemetered stations. Most stations have been installed since 2000. The IRSC catalog is based on data from the Iranian Seismic Telemetry Network (ISTN). The ITSN consists of 73 three-component short-period (Kinemetrics SS-1), digital, telemetered stations organized in 10 subnetworks. Timing corrections, based on GPS, are applied at the subnetwork recording centers. The earliest installations, for the Tehran and Tabriz subnetworks, occurred in 1995. Until 2006 the subnetworks operated largely autonomously, producing their own bulletins. Since then, all subnetwork data are transmitted to the IRSC where a national bulletin is prepared. The IRSC and IIEES catalogs used for this