A consistent and high-quality M4+ earthquake catalogue for Turkey 2007–2016 from two independent catalogues

SUMMARY We describe the process of pairing M4+ earthquakes reported in the routine catalogues of two seismic networks in Turkey and present the resulting products: (i) hypocentre locations and magnitudes and (ii) minimum 1-D velocity model with station delays for the combined station network. The two networks are the Kandilli Observatory (KOERI) and the National Earthquake Department (AFAD) and the data cover the decade 2007–2016. The seismic stations of these two networks spatially overlap within Turkey, and with their average station spacing of about 60 km, they potentially detect all events of magnitude ≥4 independently. We first obtain separate AFAD and KOERI catalogues using a series of joint hypocentre inversions that allow us to detect outliers and problematic events. We establish separate minimum 1-D velocity models and station delays for each catalogue and reduce the average inversion residuals by 75 per cent compared to our first joint hypocentre inversion using the original catalogue event locations. We then pair M4+ events from AFAD (1219) and KOERI (1271) catalogues using a spatiotemporal window and test if the combined observations indeed belong to the same event. In these tests, we compare the hypocentre location residuals of the combined observations to those of the two individual events, apply epicentral distance limits between the individual and the paired events and assess qualitatively a simple probabilistic solution of the events. As a result, we get 1145 paired events. We finally obtain a minimum 1-D velocity model with delays for all seismic stations of AFAD and KOERI combined and relocate the entire set of events. These include all paired events plus M4+ earthquakes reported by AFAD or KOERI that could not be paired (500). The velocity model has a well-defined Moho at a depth of 32 km. The station delays for both networks regionally correspond very well and exhibit some correlation with the regional subsurface temperature field. The well-locatable hypocentres of the new M4+ catalogue have an epicentre uncertainty of 3 km and a depth uncertainty of 4 km. We ascribe to each paired earthquake a magnitude equal to the average of the reported magnitudes by AFAD and KOERI. The new magnitude frequency distribution is similar to those of AFAD and KOERI for M4+ events. Except for events of magnitude larger than 5.5, there is no systematic bias in the magnitudes reported by AFAD and by KOERI, and the standard deviation is 0.3.