Location performance of seismic networks by a numerical simulation : application to BU-KOERI (Turkey)

Abstract

The main function of a seismic network is to provide high quality data for monitoring seismic zones and seismic hazard analysis. In this study, a numerical simulation (SNES, Seismic Network Evaluation through Simulation) technique is utilized for the evaluation of hypocenter location performance of a seismic network. The SNES method gives, as a function of magnitude, hypocentral depth and confidence level, the spatial distribution of the number of active stations in the location routine and their relative azimuthal gaps along with confidence intervals in hypocentral parameters concerning both the geometry of the seismic network and the use of an insufficient velocity model. The application of the SNES method also permits the evaluation of the magnitude of completeness (Mc), the background noise levels at the stations and assessment of appropriateness of the velocity model used in location routine. The SNES method does not take into account location due to systematic model error, so it allows estimating only the precision of the hypocentral location. In this study, SNES method is applied to the BU-KOERI (Boğaziçi University - Kandilli Observatory Earthquake Research Institute) seismic network to evaluate and quantify its performance for locating local and regional seismicity. This application has allowed to identify the background noise levels of the KOERI seismic stations and to evaluate the goodness of the velocity model used in the location routine. Furthermore, the SNES method has allowed identifying some seismogenic areas on the national territory that are currently not enough covered instrumentally. The upgrading of the network in these areas could be optimized using the SNES technique. For each station, a large number of three-hour waveform segments during a 6 year period were selected from continuous digital seismic data. Each year was divided into four seasons and one month represents each season (Winter, Spring, Summer and Fall were represented by January, April, July and October, respectively). Five different days and nights were also picked in each month (day times were preferred from 5 a.m. to 5 p.m. and night times from 5 p.m. to 5 a.m.). For each five nights and five days of a month, the acceleration PSD curves of the vertical component of the noise were generated, related to the period 2005-2010. Then the Turkish territory was gridded with 5 km distances and the mean PSD values (in the frequency range between 1-12 Hz) were assigned to the intersection points of each 5 km distances. The variance and the residual times of P and S phases were determined to evaluate the appropriateness of the velocity model used by BUKOERI. Hereafter, seismic spectra calculation was obtained to evaluate the detection capability of the seismic network. Finally, the earthquake simulation was made by constructing the SNES maps for magnitudes of 2.5, 3.0, 3.5, fixing the hypocentral depth at 10 km and the confidence level at 95 per cent. Through the application of the SNES method, it is showed that the KOERI provides the best monitoring coverage in the southeast of Marmara Sea and Gulf of Gökova with maximum number of twelve stations, for ML 2.5, with errors that are 3 km and 5.5 km for epicenter and hypocentral depth, respectively. At ML 3.0, BU-KOERI seismic network gives the best quality of the epicentral estimate with a maximum number of twenty stations in Marmara Region with errors that are 2 and 4 km for epicenter and hypocentral depth, respectively. At ML 3.5, the seismic network provides the best monitoring coverage in the mid part of Marmara Sea and some south parts of it where the errors in epicentral and hypocentral estimate are 2 km and 3 km, respectively, with at least 28 seismic stations. This seismic network also provides a threshold of completeness down to ML 3.0 for almost all Turkish territory. As a result, location performance of BU-KOERI seismic network should be improved by installing more seismic stations on the Turkish territory where the coverage of the stations is insufficient. In addition, with the construction of OBSs, the location performance of the network is expected to be enhanced.