APPLICATION AND LIMITATIONS OF SEISMIC ATTRIBUTES ON 2D RECONNAISSANCE SURVEYS

Abstract

Reconnaissance surveys of 2D seismic data are acquired before most 3D seismic surveys to define the area of interest and regional trends. While seismic attributes are routinely applied to 3D data, they are less commonly applied to 2D data. Today’s low commodity prices call for the utilization of all available data. To quantify the value of 2D attribute analysis I use attributes computed from a 3D survey covering the same area as the baseline or ideal result. Using a 2D seismic survey overlying a 3D volume offshore New Zealand in the Taranaki Basin, picked key horizons on 2D and 3D data, which were in turn used to create surfaces and isochrons. Candidate attributes included RMS amplitude, instantaneous frequency, instantaneous phase, similarity, curvature, and spectral components. The Taranaki Basin is a Cretaceous-aged failed rift basin with a complex structural history. A poststack data conditioning workflow is constructed for 2D and 3D to suppress crosscutting migration artifacts, and sharpen discontinuity features of the structurally complicated environment. I used concepts of structural geology to validate these discontinuity features as faults rather than organized artifacts. Structure-oriented filtering deteriorates on 2D lines where out-of-the plane reflectors overprint features of interest. I also noted an underreported observation that edge-preserving structure-oriented filtering can introduce frequencies that fall beyond the limits of the recorded seismic data. On 2D vertical seismic lines, in the dip direction single trace attributes such as spectral components, RMS amplitude and instantaneous attributes provide images comparable to those computed from 3D. Geometric attributes computed from 2D dip lines xx render useful images of apparent dip and apparent (or Euler) curvature while similarity maps the apparent dip of faults and channel edges. Generating useful attribute maps from 2D data is more problematic, with major limitations of both software flexibility and the author skill. While I can generate good quality time structure, isochron, and horizon curvature maps from 2D data, I believe that effective interpolation of shorter wavelength anomalies as mapped by RMS amplitude and spectral components requires Kriging guided by a variogram constructed from the time-structure maps.