Three-dimensional full wave seismic modelling versus one-dimensional convolution: the seismic appearance of the Grès d’Annot turbidite system

Abstract

Abstract The increasing use of three-dimensional seismic data for reservoir characterization justifies the need for an accurate and fast seismic modelling tool. Synthetic seismic data sets, where the geological model is controlled at each location, provide crucial information for the understanding of the seismic response in connection with various geological features. The method used here involves a wave equation approximation based on ray tracing and Born formalism in the context of a one-dimensional background, thereby providing economic computation of a very useful subset of geological models. The software enables a high level of flexibility to be applied to a range of realistic experiments. In particular the software provides pre-stack multi-component seismic response of a three-dimensional heterogeneous reservoir, including both reflected and converted waves, at the level of the target. In order to analyse and reconstruct the architecture of deep-water gravity deposits, we present a realistic synthetic seismic experiment of one Grès d’Annot outcrop. A geological interpretation of the outcrop has been based on photo panoramas calibrated with vertical sedimentological sections. The internal distribution of facies has been interpolated using geostatistical simulation, with one of the stochastic realizations of the reservoir architecture serving as the three-dimensional cube of lithofacies corresponding to the reservoir model. These lithofacies were assigned elastic parameters of an analogue Brazilian field. The reservoir model is then buried within a smooth background model, and a seismic model generated. The purpose of generating three-dimensional seismic models is to analyse the seismic signature of the different sedimentary architectures, in order to provide guidelines for the seismic interpretation of comparable subsurface reservoirs. Compared with convolution results, the computed synthetic data are more realistic and closer to the actual seismic data. This is mainly due to the fact that the method takes into account the lateral heterogeneity included in the vicinity of each reflecting point.