GNGTS 2017 - 36° Convegno Nazionale

528 GNGTS 2017 S essione 3.1 le riflessioni inclinate arrivano fino al fondo mare i fino a circa metà della sezione, mentre nella parte più lontana dalla costa sono in discordanza angolare con sedimenti più recenti. Questo confine è stato identificato in tutte le sezioni Boomer. Ringraziamenti. Si ringrazia il gruppo di acquisizione dati sismici dell’OGS e in paticolar modo il capo progetto Daniel Nieto. Si ringrazia anche il personale del comune di Ancona e soprattutto il Dr. Cardellini per il fondamentale supporto logistico. Bibliografia Baradello L., Carcione JM.; 2008: Optimal seismic-data acquisition in very shallow ������ ������� �� ��� ������ water. Surveys in the Venice lagoon. Geophysics, 73 (6), Q59–Q63, DOI 10.1190/1.2976117. Bernardini M., Crescenti U., Rainone M., Sciarra N., Tazioli G.S.; 1986: La campagna geognostica . In: Crescenti U. (eds.), La grande frana di Ancona del 13 Dicembre 1982. Studi Geologici Camerti, Spec. Vol., 95–119. Böhm G., Rossi G., Vesnaver A.; 1999: Minimum time ray-tracing for 3-D irregular grids . J. of Seismic Exploration, 8 , 117–131. Cotecchia V.; 2006: The Second Hans Cloos Lecture. Experience drew from the great Ancona landslide of 1982 . Bull. Eng. Geol. Env., 65 , 1-41, DOI 10.1007/s10064-005-0024-z. Target-oriented, probabilistic seismic-petrophysical inversion with geostatistical and hard data constraints M. Aleardi Earth Sciences Department, University of Pisa, Italy Introduction. The estimation of petrophysical reservoir properties from seismic data is an inverse problem that can be solved by combining geophysical modelling and inverse theory. This inverse problem is ill-posed, non-unique, and is strongly affected by noise and measurements errors. Therefore, it is frequently casted into a statistical framework (Tarantola, 2005), which allows for the estimation of the posterior probability of model parameters given the observed data. A common statistical method is the Bayesian approach that combines the available prior information and the likelihood function linking the unknownmodel parameters and the measured data. Under certain statistical assumptions and assuming a linear forward model, the solution of the Bayesian inverse problem can be analytically formulated. For example, analytical solutions exist if the error distribution is Gaussian and if the prior distribution of the model parameters is Gaussian, Gaussian mixture, or generalized Gaussian. If a linear or linearized rock-physics model is available, the analytical approach can be also extended to the seismic-petrophysical inversion to directly estimate the posterior distribution of reservoir properties from seismic pre- stack data. Amplitude versus angle (AVA) inversion has been widely used in the field of seismic inversion and particularly in seismic-petrophysical inversion (Avseth et al., 2005; Aleardi et al., 2016a , 2016b). However, in most cases, seismic-petrophysical inversion considers each data gather independently, for example assuming a 1D forward model. As a result, the inversion outcomes are sensitive to data noise and when combined to form a 2D or 3D image, important geologic features can be masked. For this reason, lateral constraints have been extensively used in seismic inversion (Buland et al., 2003; Tetyukhina et al., 2010; Bongajum et al., 2013). In this context, also geostatistical approaches are being actively investigated, where a priori knowledge is derived from two-point or multi-point statistics (e.g. Zunino et al., 2015). In this work, we implement a seismic-petrophysical inversion algorithm based on a geostatistical approach in which also hard data (i.e. well log) constraints are taken into account. For computationally feasibility reasons, we limit our attention to a target-oriented inversion