The Bulletin of Canadian Petroleum Geology is a peer-reviewed scientific journal published four times a year. Founded in 1953, the BCPG aims to be the journal of record for papers dealing with all aspects of petroleum geology, broadly conceived, with a particularly (though not exclusively) Canadian focus. International submissions are encouraged, especially where a connection can be made to Canadian examples.
Below is an extended abstract from a paper in our most recent Bulletin journal. 
To view the full article you must be a member of CSPG. 
Join Today
Pushing the envelope of seismic stratigraphic interpretation: 
a case-study from the Mannville Group using small 3-D surveys

By: M. Smaili & B.S. Hart | Earth and Planetary Sciences Department McGill University

December 2018 -  Bulletin of Canadian Petroleum Geology

In this paper, we integrate 3-D seismic and wireline log data to illustrate some modern techniques of seismic stratigraphy. The imaging target is the Lower Cretaceous Mannville Group, traditionally one of the main targets for hydrocarbon exploration and production in Western Canada. The depositional environments for the Mannville Group in our study area were diverse, ranging from fluvial to shallow marine. Mannville Group strata overlie a major unconformity that separates them from predominantly carbonate rocks of the Paleozoic  (Devonian).

As a first step, we integrated wireline logs and seismic amplitude data in a qualitative way and gained more stratigraphic insights than could be obtained with either data set alone. We then used seismic inversion and a seismic attribute study to make quantitative lithology predictions. Acoustic impedance inversion proved to be an excellent tool for mapping the unconformity, clearly distinguishing the clastic rocks above from the carbonate-dominated units below it. However, the inversion result did not clearly define stratigraphic features within the Mannville Group. To that end, we generated a pseudo-lithology (gamma-ray) volume for the Mannville Group by integrating wireline logs and seismic attributes using a neural network. This pseudo-lithology volume identified stratigraphic features that were not apparent in either the original seismic amplitude or the inversion volume. The results of this study show how integrating the three different 3-D seismic versions was useful for understanding the stratigraphic complexity of the Mannville Group. The approach presented here can be used for similar purposes in other geological settings.

Seismic stratigraphy is an approach to seismic interpretation that is based on principles of stratigraphy. Papers (e.g. Mitchum et al., 1977) published in the American Association  of Petroleum Geologists Memoir 26 first showed how reflection terminations, continuity, and other qualitative “geometry- based” analyses of reflections could be used to help to infer depositional histories and predict  distribution  of  lithologies (at least qualitatively) in undrilled areas. This seismic-based approach, originally based on 2-D seismic data, became an invaluable tool in the petroleum industry’s exploration and development efforts. New data visualization methods became available as interpreters began working with 3-D seismic data (e.g. Brown, 2004; Hart, 2011).

The application of this traditional seismic stratigraphic approach can be challenging in some cases. For example, in low-accommodation settings there might be parasequences and sequences that are too thin for internal geometries to be recognized seismically. In other cases, 3-D seismic surveys cover insufficient area to capture diagnostic reflection terminations that allow for recognition of sequence boundaries, maximum flooding surfaces, and more (Hart et al., 2007). Many onshore 3-D surveys from North America face both of these challenges.

For these and other reasons, Hart (2013) advocated extending the field of seismic stratigraphy to incorporate geophysics-based approaches that allow better definition of stratigraphic features than can originally be obtained by viewing the original amplitude version of the data. These geophysics-based methods include seismic inversion and seismic attribute studies, both of which can be used to make quantitative lithology predictions and to enhance the visibility of thin or otherwise poorly-resolved stratigraphic features.

In this paper, we illustrate how these advanced seismic methods can be used to provide enhanced stratigraphic understanding and prediction using small 3-D seismic datasets from Western Canada. Our work focuses on defining stratigraphic features in the Lower Cretaceous Mannville Group, a major target for hydrocarbon exploration and development. The goal of the stratigraphic analyses was primarily focused on aiding development drilling; hence lithology prediction was a key outcome for this study. A variety of depositional features compartmentalize reservoirs in the fluvial to shallow marine deposits of the Mannville Group in our study area, and the physical properties volumes we generated better image those features than the original seismic volume. Integration of well data, in various ways, formed a key part of our stratigraphic interpretation of the 3-D seismic datasets.

Figure 1. a) Map location of the study area in Western Canada. b) The study area is divided into two 3-D seismic surveys (A and B) slightly removed from one another. Bold dotted lines indicate the approximate location of the seismic surveys. Locations of the wells drilled within the two surveys are denoted by the solid black dots. Cross-sections shown throughout the text are shown with their respective labels.
Geological Setting
The study area is situated approximately 100 km to the east of the city of Edmonton (Fig. 1a). The stratigraphic interval of interest includes clastic rocks of the Mannville Group as well as the underlying, predominantly carbonate rocks of the Devonian Woodbend and Winterburn groups. The stratigraphy is highly variable in such a complex setting, with multiple sea level and sediment supply changes. Putnam (1982) divided the general depositional history of the Mannville Group in eastern Alberta into three main stages: 1) development of fluvial networks draining along structural lows and thereby helping to form the sub-Mannville   paleo-topography   (lower   Mannville   Group); 2) marine transgression followed by progradation in wave- dominated settings (informal “middle” Mannville Group); and 3) continuous regressive sedimentation in continental to shallow marine environments dominated by northward flowing fluvial systems (upper Mannville Group).

The stratigraphic nomenclature used to define the Mannville Group is complex and varies between different regions of the Western Canada Sedimentary Basin. This variability in terminology is due, at least in part, to significant lateral facies variability (Putnam and Oliver, 1980; Hayes, 1986; Wood and Hopkins, 1992; Cant, 1996; Leckie et al., 1997; Karavas et al., 1998). In the current study area of east-central Alberta, the stratigraphy is based primarily on a framework commonly employed in western Saskatchewan. From youngest to oldest (top to bottom), the stratigraphic units include the Colony (corresponding to the top of the Mannville Group), McLaren, Waseca, Sparky, General Petroleum, Rex, Lloydminster, Cummings, Dina, and the informal “Detrital zone” of Williams (1963) and others (Fig. 2).

The Mannville Group is bounded at its base and top by unconformities. Clastic rocks of the Colorado Group uncon- formably overlie the Mannville (Karavas et al., 1998). At its base, a significant regional unconformity juxtaposes Lower Cretaceous clastic rocks of the Mannville and older strata that range in age from Devonian in the area of study through Jurassic further to the west in the Western Canada Sedimentary Basin. This surface is informally referred to as the sub-Mannville unconformity. Relief on the sub-Mannville unconformity was controlled by tectonic factors (Williams, 1963; Christopher, 1984) and has been linked to the formation of incised valleys that affected the deposition of the overlying clastic units and had a direct impact on the early Mannville drainage networks (Jackson, 1984). Other interpretations emphasize the influence of differential erosion of the dipping Paleozoic/Jurassic units on incised valley formation
(e.g. Cant and Stockmal, 1989). Therefore, mapping this un- conformity can be useful for identifying Lower Cretaceous incised valleys and fluvial channels, as well as for identifying and mapping isolated structural highs below the unconformity (“erosional remnants”) which, when porous, have the potential to host hydrocarbons.
Four different Devonian units underlie the sub-Mannville unconformity in the study area; the Nisku, Ireton, Camrose, and Lower Ireton (Fig. 2). These units dip to the southwest, and consist of limestone, dolomite, and carbonaceous shale (Switzer et al., 1994). These predominantly carbonate rocks were exposed at the surface for millions of years (post-Devonian to Jurassic/Early Cretaceous), and were subject to high degrees of weathering, erosion, and dissolution. As a result, the unconformity surface is locally mantled with weathered and erosional debris that are collectively referred to as the “Detrital zone”.

Our confidentiality agreement with the data owner prevents us from identifying the exact location of the seismic surveys, and hence well names used in this study.

Database and Methodology
The dataset for this study consisted of two small 3-D seismic surveys1 and wireline logs for 37 wells. Survey A covers an area of approximately 20.2 km2 and contains 25 wells (23 of which are  vertical)  and  Survey  B  covers  an  area  of  approximately 10.4 km2  and contains 12 wells (8 vertical) (Fig. 1b). Digital logs were available for 30 of the wells, and paper copies of logs were available for the remainder. We focus here on the wells containing digital logs, due to the nature of the digital analyses we describe, but all logs were used to define the stratigraphic framework in our area. Gamma-ray (GR) logs were available for almost all the wells, but other log types (e.g. sonic, density, resistivity) had different availability depending on the well.

The 3-D seismic surveys in both areas have a sample rate of one millisecond (ms), a trace length of 2000 ms two-way travel time (TWT), and a bin size of 25 x 25 meters. The data were processed to be zero-phase according to the SEG-Y file headers. Time-migrated versions of the seismic data were used for this study. Both seismic surveys were acquired using a dynamite source. The seismic bandwidth ranges from 8 to 110 Hz (Survey A) and from 8 to 100 Hz (Survey B), with dominant frequencies of approximately 60 Hz and 55 Hz, respectively. The zone of interest is located between 600 and 750 m in depth, approximately corresponding to 620–780 milliseconds TWT. The log-derived average velocity in the zone of interest for Survey A and Survey B of approximately 2780 m/s and 2900 m/s, yielded a vertical seismic resolution (defined as ¼ of the wavelength) of about 12 and 13 m, respectively. Seismic detection limits, defined using methods outlined by Hart (2011), could be as low as 3 m.

The workflow employed in this study is summarized in Figure 3. The work began with log-based correlations in order to build an initial stratigraphic framework for the Mannville and adjacent units. The well logs used for correlation included gamma-ray, resistivity, photo-electric factor (PEF), density, and sonic logs (as available). Linking the wireline logs to core facies would have provided added knowledge to this “seismic stratigraphy” project (e.g. McCullagh and Hart, 2010) but was not possible due to the unavailability of core in our area. The log-based stratigraphy was transferred to the seismic data using synthetic seismograms. This was followed by a detailed stratigraphic interpretation of the seismic data that merged the well-based stratigraphy and seismic reflection patterns. Sarzalejo and Hart (2006) previously described and used this methodology to study the Mannville Group in southeastern Saskatchewan.

Figure 2. Stratigraphic column of the study area and neighboring regions. The Cretaceous zones follow a scheme based on the Lloydminster area while the Devonian zones follow a scheme based on the Central Plains. The unconformity places any of the four Devonian units against the overlying Cretaceous rocks. (Modified from Hayes et al., 1994; Switzer et al., 1994).
                       2. Acoustic impedance is defined as the product of a rock’s velocity and density, and is there- fore related to lithology, porosity and other properties.

Following this conventional seismic stratigraphic interpretation, we extended our stratigraphic interpretations using two different methods for predicting rock properties. First, we used model-based inversion (e.g. Veeken and Da Silva, 2004) to convert the 3-D seismic volumes to  acoustic  impedance2. This approach integrates the wireline log data, the seismic data, the seismic wavelet, and the interpreted seismic horizons. Next, a seismic attribute study was used to predict “pseudo-lithology” (gamma-ray) following the approach described by Hampson et al. (2001) and used by Leiphart and Hart (2001), Tonn (2002), Tebo and Hart (2005), Sagan and Hart (2006), Sarzalejo Silva and Hart (2013) and others. Finally, we compared the utility of the original seismic data, the inversion results, and the attribute- based prediction of rock properties for identifying stratigraphic features in the Mannville Group.

Figure 3. General workflow displaying the different types of data and analyses employed in this study.