GNGTS 2013 - Atti del 32° Convegno Nazionale

SEPARATION AND IMAGING OF SEISMIC DIFFRACTIONS D. Urbano, V. Lipari Dipartimento di Elettronica, Informazione e Bioingegneria, Politecnico di Milano, Italy Introduction. Diffracted and reflected seismic waves are different physical phenomena originating from different kinds of subsurface features. Usually, in seismic exploration, the focus is on seismic reflections because they carry most of the information about subsurface. However, sometimes the goal of seismic processing consists in identifying small subsurface features (e.g. faults, fractures and rough edges of salt bodies) or small changes in reflectivity. In all these cases it is diffracted waves which contain the most valuable information (Kanasewich, and Phadke, 1988; Landa and Keydar, 1998). Over the last decade there has been an increasing interest in using diffractions as a direct indicator of different kinds of discontinuities. The energy carried by diffractions can be also used to refine velocity models (Harlan et al. , 1984; Landa et al. , 2008; Reshef and Landa, 2009). Tsingas et al. (2012) proposed to use diffraction imaging as a tool for helping interpretation in fractured reservoirs. Typically diffracted energy is one or even two order of magnitude weaker than reflected one and often it is not easy to distinguish the diffracted events in a full dataset or to identify the image of diffractors in a full seismic migrated image. Therefore, diffractions have to be separated from reflections for imaging or other kind of processing and this still represents one of the main issues when handling diffractions. Several approaches were proposed for the separation of reflections and diffractions, and several different domains were used for this purpose. Khaidukov et al. (2004) take advantage of the different propagation properties of the waves. The recorded wavefield is focused back to the imaginary source location in a pseudo-depth domain and then the reflections are muted out. After defocusing of this muted data they obtain a wavefield where reflection events have been suppressed. Taner et al. (2006) show how to separate reflections and diffractions using plane-wave constant p sections. In this domain reflection energy can be filtered out by the method of plane-wave destructors (PWD) (Claerbout, 1992; Fomel, 2002). Separation of diffractions and reflections and imaging of diffractions by dip-filtering via the same PWD operator, but this time in the post-stack migrated domain, is discussed by Fomel et al. (2007). Separation in the post-migration dip-angle gathers, where reflections always have a concave shape (Audebert et al. , 2002) and diffraction have a different shape (horizontal if migration was performed with the correct velocity) has been proposed by Landa et al. (2008) and Reshef and Landa (2009). In this work we analyze separation of diffractions in the migrated image domain via two algorithms belonging to the last two mentioned methodologies. First we analyze the peculiarities of the separation via dip-filtering of the migrated image. We introduce a different filtering technique based on the local image gradient. Then we illustrate the properties of the dip-angle common image gathers and we analyze the behavior of this domain in the context of separation of diffractions. Finally, we propose to combine these methodologies in an original technique that aims to take advantage of both. The effectiveness of the method has been tested on both synthetic and field datasets. Here we show an example on the synthetic Sigsbee dataset. Separation of diffractions via dip-filtering in the depth image domain. The first studied technique is a variation of the method proposed by Fomel et al. (2007) which performs separation of diffractions by dip filtering in the post-stack domain. The use of dip-filtering in the depth migrated image domain as a method for separating diffractions is based on the underlying assumption that, in a migrated section, reflections are imaged as strong coherent events with a dip that changes with continuity. On the contrary diffractions are imaged as very localized coherent events that do not show any identifiable dip. Thus, once the events that show clearly identifiable and slowly variable dip are removed, the remaining coherent events are interpreted as diffractions. 71 GNGTS 2013 S essione 3.1