<i>The Greek Database of Seismogenic Sources: seismotectonic implications for North Greece</i>

Abstract

Although Greece and the broader Aegean Region belong to the most tectonically and seismically active places in the world, a GIS-based database of seismogenic sources was lacking. Data and information concerning the occurrence and behaviour of numerous active faults are usually scattered or hidden in the rich literature, making their acquirement a hard and time-consuming task. On the other hand, such data and information are useful for several purposes: SHA (and especially probabilistic SHA), geodynamic models and seismotectonic interpretations are some of the disciplines for which such a database can be an irreplaceable tool. The core of this research is the development of GreDaSS which was started from North Greece as a pilot area. The compilation procedure includes several stages. The first stage is the recognition of a seismogenic source and the collection of its respective available data and information. The synthesis, critical analysis and homogenization of the collected material follows, in order to define the principal seismotectonic parameters, to plot the source in a geographic coordinate system according to its geometric attributes and to fill in all the corresponding informational fields (comments, open questions, summaries, pictures and references). The final stage includes a processing of all informational levels in order to be linked and interactive. This procedure was followed for 38 CSSs and 58 ISSs that belong to the study area (North Greece). However, some of those steps were followed for the rest sources of the broader Aegean Region in the frame of the SHARE project. During the parameterization process, an important issue came up: given that the majority of the seismogenic sources in Greece is not connected with recently (historically or instrumentally) recorded earthquakes – which is the more hazardous case, the only data to rely on are the ones based on cumulative effects-based investigations (or generally characterized as geological investigations). On the other hand, investigations based on single-event effects (having the seismological investigations as the most prevailing method) sometimes provide wrong information especially when it is about old events. The comparison of these two different approaches is important to define the reliability of the provided data and information. Based on four characteristic case studies which all involve a major earthquake covering a wide period of the historical and instrumental era, it is shown that i) the single-event effects-based investigations provide improved results as technology advances, and ii) the cumulative effects-based investigations steadily provide reliable results regardless if the seismogenic source is associated with recent earthquakes or not. Another important matter that emerged during parameterization is the determination of two important SHA parameters: the maximum fault depth and the slip rate. Defining the maximum fault depth is important because the faults dimensions can be constrained and hence its maximum expected magnitude. For crustal faults, maximum depth is limited within the seismogenic layer, which can be defined by the brittle-ductile transition zone (BDT). Although it is not necessary that a fault will rupture the entire thickness of the seismogenic layer, by estimating the depth of the BDT we can set the maximum depth which a possible rupture can reach. The estimation of the BDT depth relies on the calculation of the strength profiles that reflect the mechanical behaviour of the upper lithosphere. In the current dissertation, the rheological models of three different areas are compared and calibrated based on the local seismic distribution. Since temperature is the key-factor of lithospheric strength, the geothermal gradients are also calculated. However, in the radiogenic heat productive crust, the geotherm equations depend on some thermophysical parameters. For this reason several sensitivity tests were run for various compositions of the upper and lower crust which allowed distinguishing two relative thermal conditions of the crust: the cool and warm crust. Based on these tests and on rheological modelling, the BDT depth of the three case studies is calculated and it is compared with the available seismic distributions. The results are in good match and verify the rheological models of each case study, giving confidence for further application in the surrounding areas. The calculation of the slip rate is based on the geodetic strain rate field that derives from GPS measurements. Given that GPS measurements span only few decades, short-term slip rate is actually calculated by this method. The theoretical approach relies on the fact that the deformation between two GPS stations that move relatively to each other is not uniform, but it is concentrated within a zone around the fault plane. During a seismic event, most of the so far accumulated elastic strain is suddenly released on the fault plane and transformed into relative slip, leaving a small, practically neglectful, amount of plastic deformation. The calculation process is based on a series of strain fundamental formulas that can provide the two slip rate components: the strike-slip component and heave which provides the dip-slip component. The obtained results derive from three different datasets that provide either the principal strain axes, or dilatation and maximum shear strain. The patterns coming from the three results are quite similar, although some differences exist in their absolute values. The geodetically calculated slip rates are also compared with the ones found in the literature (wherever these are available), which are basically obtained by geological methods. The comparison suggests that the former not only show reasonable values, but they are of the same order of magnitude. The usually lower values of the geodetic slip rates can be explained by the facts that GPS stations measure velocities during a very short period of a seismic cycle, and that strain accumulation is not necessarily linear with time, so it could be wrong to infer that its rate is steady. It is important to mention that during the calculation process rake is also calculated as a collateral result. The completeness of GreDaSS for North Greece, including all seismogenic sources that are known so far, allowed the separation of the area into five sectors, each documenting an internal uniform behaviour. Moreover, the pattern, geometry and the geodynamic setting of some regions imply the presence of low-angle normal (detachment) faults defining a seismogenic volume in which faults’ behaviour can change significantly (fault interaction, random rupture path, etc.). Nevertheless, beyond these qualitative observations, the database can offer several other applications especially for the enhancement of SHA. Two examples are given in this thesis. The first one concerns the improvement of the seismic zonation map of Greece by introducing a new seismogenic zonation map. Until now, seismic zonation of Greece was based only on seismological data. This method produces several inconsistencies with major seismotectonic features in the Aegean. The proposed seismogenic zonation map of this thesis relies on other geological criteria and especially the seismogenic sources of GreDaSS. However, the best solution would be a map that would take into account both geological and seismological criteria. The second application example involves earthquake triggering scenarios that are based on Coulomb stress transfer. The Volos and Nea Anchialos ISSs, both lying in the Pagasitikos Gulf, are selected as a case study. Each fault is responsible for one of the two strongest shocks of the 1980 Volos earthquake sequence. Coulomb stress change models based on the seismogenic sources of GreDaSS suggest that earthquake triggering between these two ISSs is a possible scenario. Concluding, the development and the state-of-the-art of GreDaSS for North Greece are presented in this dissertation, along with the calculation of two basic parameters, the observation of some seismotectonic features and the usage of the database as a SHA tool. It is important to mention though, that the database is always updatable, which means that whenever new data and information are available these will be included in GreDaSS.