High quality parametric seismic catalogues from several sources around the Dead Sea Fault Zone and in Turkey

Large earthquakes pose a serious threat to humanity, especially since potentially active faults are hidden under the surface and have left no obvious traces of rupture, and because fault interaction and multiple rupture mechanisms are poorly understood. To study said big events, and to improve the forecasting capabilities of current statistical models, we need to study past seismicity. In recent years there has been a large effort towards making seismic data available to the geosciences research communities. Open-access initiatives like the European Integrated Data Archive and the datasets built by the Global Earthquake Model combine and provide earthquake information for regions beyond geopolitical borders, for extended periods of time and magnitude ranges. Users can compare and combine information to create custom earthquake catalogues. The result is a compendium of several sources, or a preliminary meta-analysis on singular events. As a consequence, single events may be reported more than once: in historical times with different sources reporting effects of ground shaking; in instrumental catalogues multiple reporting may also happen when earthquakes are detected by more than one seismic network. Because each catalogue source has different procedures to calculate locations and magnitudes, the resulting parameters may be different and it is not straightforward to extract useful information or to quantify their error. There has also been much progress in models that use seismic catalogues, and their statistical byproducts, for seismic hazard and forecasting. These include the 2020 European Seismic Hazard Model and all recent developments on epidemic type aftershock sequence (ETAS) simulations. The problem is that hazard and forecasting experts use seismic catalogues without knowing the details about their limitations. Possible artifacts in depth, epicentral location and magnitude undetected by the user can negatively impact forecasting and hazard. Earthquake catalogues represent an opportunity for observational seismologists, especially if they cover large areas or long time periods. There is great potential in historical earthquake records in seismically active regions that have been populated for thousands of years, and also in regions that have today a seismic station coverage that records all moderate to large earthquakes. Both historical and instrumental catalogues when combined have the potential to reveal pivotal information for hazard analysis if they are carefully built and analysed. The initial goal of this thesis was to tackle the problem of fault segments interaction allowing an extreme release of seismic moment either within a single event, or within a more erratic sequence. For this ambitious goal, we accomplished the first step: generating high-quality seismic catalogues of earthquake location and magnitude. I focused on two strike-slip fault systems where earthquakes were documented by different sources, using pre-instrumental records in the Dead Sea Transform Fault Zone (DSTFZ), and a decade of instrumental records in Turkey. We did it following a consistent and repeatable procedure in the parameter assessment (historical) and calculation (instrumental), providing uncertainties for the location and magnitude, a known completeness level, and ensuring the catalogue is free of repeated or fake events. Quality of historical earthquake catalogues needs to be revised, in particular in regions with a vast written registry of earthquakes. While instrumental seismicity has only recorded one earthquake of moment magnitude equal or larger than 7 in the DSTFZ, tectonic forces in the region have caused centuries of documented incidence of large earthquakes, and remote sensing supports the idea that the area is still in motion and could host large events in the near future. In chapter 2, we build a high-quality historical catalogue for the region between 27-36N and 31-39 E. Historical records in the DSTFZ come from a multitude of sources that don’t share the spatiotemporal location, the magnitude, and sometimes neither the certitude of occurrence of all events. We obtain parametric earthquake data from previous studies, validate the historical records with field investigations (geological and archaeological studies) and analyse several events in depth with regards to their macroseismic intensities. We further identify how does the magnitude and location correlate with the nearby identified faults, and assess their relation to soft-sedimentary deformation structures in the Dead Sea. We convert all magnitudes to Mw and class the magnitude and location uncertainty among three categories. The result is a new historical parametric catalogue from between 31 BCE and 1900 that contains 93 earthquakes of magnitude equal or larger than 5 in moment magnitude scale, complete down to Mw=7. To the North of the DSTFZ, Turkey is routinely surveyed by two agencies: The Kandilli Observatory (BOUN KOERI Regional Earthquake-Tsunami Monitoring Center) and the Turkish National Network (Disaster and Emergency Management Authority Earthquake Department; AFAD). Both have installed new broadband stations in recent years and nowadays cover the country with an average station spacing of about 60 km. We create a consistent M4+ high-quality catalogue for Turkey within 34° –43°N and 25°–46°E for the time period 2007–2016, with relocated hypocentres that combines observations from KOERI and AFAD using a repeatable procedure, providing quantitative location uncertainties and a magnitude estimate (the average of the magnitudes reported by each network). We simultaneously invert for a new 1D velocity model using a set free from outlier observations. The resulting seismic catalogue has 1645 earthquakes of magnitude equal or larger than 4, with an epicentral precision (for the well- locatable events) of 3 km and a depth precision of 4 km. Both AFAD and KOERI report routine magnitudes in several types, requiring to propose a new local magnitude (Ml) formula for Turkey using vertical amplitudes. This led to a set of consistent magnitudes for all earthquakes in the 10-year instrumental catalogue we relocated before. To calculate the magnitudes, we iteratively inverted for two parameters related to the distance between the source and the station, for individual station effects and for magnitudes themselves, after converting each seismogram to its Wood-Anderson equivalent. The final value of Ml is then scaled to fit with the moment magnitude. Our resulting catalogue with recalculated magnitudes is now complete down to magnitude 4.2 with a b-value of 0.97