Abstract In the inner to middle Sable Subbasin, offshore Nova Scotia, a thick (~100-200 m) and areally-extensive (>20 km x > 100 km) sheet-like fluvial-marine succession has been identified at the top of the Missis-auga Formation using core, well log, and seismic (2D and 3D) data. Despite its relatively planar nature and significant width perpendicular to interpreted paleoflow (at least 20 km), the base of the fluvial sheet is interpreted to be a wide incised valley formed during slow relative sea level fall and lowstand. Deposition of the fluvial-marine succession has occurred during subsequent slow relative sea level rise, which was punctuated by several higher frequency relative sea level falls that episodically caused fluvial systems to bypass the shelf, allowing for continued shelf margin progradation.

Abstract We integrated core, wire-line logs, a three-dimensional (3-D) seismic volume, and a seismic attribute-derived 3-D lithology volume to define the stratigraphy of the Lower Cretaceous Mannville Group for a small area in Saskatchewan. The lithology volume was generated by integrating the seismic data with wire-line logs through the use of a probabilistic neural network. The stratigraphic interpretation was an iterative process: first, based on wire-line logs and cores; then, based on the integration of well and 3-D seismic data; and finally, by integrating the attribute-derived lithology volume with the other data sets. Integration of the lithology volume into our stratigraphic interpretation, along with the exploitation of seismic-based visualization technologies, helped us to construct a better geologic model than what could have been constructed using only well data or conventional seismic-stratigraphic analysis techniques. Unfortunately, despite the high-frequency content (and good to excellent quality of the data), meter-scale variations of lithology in the primary reservoir interval could not be detected seismically because of the low acoustic-impedance contrasts between the various lithologies in this interval. Various types of noise in the seismic data also degraded the attribute-based property prediction.