GNGTS 2018 - 37° Convegno Nazionale

566 GNGTS 2018 S essione 3.1 The study area (see Fig. 1), which is the Easternmost part of the Mediterranean Sea, is characterized by the presence of two main basins, namely the Herodotus and the Levantine ones. The former presents an oceanic crust, while the nature of the deep crust of the latter is still matter of debate within the geophysical community. In fact, the presence of a very thick sedimentary layer, makes the study of the Levantine Basin deep structure a difficult task. A part from the above mentioned basins, the Levant is characterized by the presence of the Mediterranean Ridge (an accretionary wedge caused by the African Plate subducting under the Eurasian andAnatolian plates), as well as of the Cyprus arc and the Eratosthenes Sea-mount. A proper description of the East Mediterranean geology can be found for instance in Longacre et al. (2007). Method and Theory. The inversion algorithm is based on the solution presented in Sampietro (2015) and Reguzzoni and Sampietro (2015), basically it consists in an iterative inversion which allows, once the gravitational effect of the most superficial layers has been stripped from the observations, to recover the Moho depth as well as the density distribution within the crystalline crust. The algorithm is based on the following steps: 1. Collect the local available information (e.g. seismic profiles, map of the main geological provinces on the area, densities, etc.); 2. Assign a relation that describes the crust density variation as a function of depth for each geological province. In the absence of better information, it can be inferred from the literature; 3. Reduce the gravity data for the effects of the topography, bathymetry, sediments, lateral density variation inside the crystalline crust and the upper mantle. The crystalline crust one taken from the function studied at step 2, the upper mantle taken e.g. from a global model; 4. Invert the residual field for the Moho depth, and a scale factor for each geological province density function; 5. Re-apply step 3 with the densities estimated at point 4 and iterate up to convergence. To deal with the effect of possible errors in the data reduction, as well as, in the uncertainties related to the a-priori information required by the inversion algorithm (e.g. the shape of geological provinces, starting crustal density models, etc.) a Monte Carlo analysis is performed, thus obtaining an estimate of the accuracy of the results. Basically, a random set of Monte Carlo samples (with the same stochastic characteristics of the a-priori information) is created and for each sample the whole inversion procedure is applied thus finding the effect of the specific input uncertainty on the final Moho depth result. Finally a refinement in the density of sediments, crust and upper mantle is performed by means of a Bayesian inversion approach according to Rossi (2017). Data. The starting points of the inversion are gravity disturbances synthesized at 3500 m above sea level (just above the higher mountain in the study area) from the GECO global model (Gilardoni et al. , 2016). GECO is an optimal combination between the EGM2008 model and GOCE observations. Digital elevation model has been taken from Etopo1 (Amante et al., 2009), and the complete terrain correction has been computed by means of the GTE software (Sampietro et al. , 2016). As for the sediments, their thickness Fig. 1 - Gravity disturbances from the GECO model reduced for the effects of Topography, bathymetry, sediments and the subduction plate beneath Cyprus. White line are the limits of the main geological provinces. White dashed lines show the position of the used seismic profiles.