8: High-Resolution Sequence Stratigraphic Architecture and Reservoir Characterization of the Mississippian Burlington-Keokuk Formation, Northwestern Arkansas
Published:January 01, 2019
- PDF LinkChapter PDF
Buddy J. Price, G. Michael Grammer, 2019. "High-Resolution Sequence Stratigraphic Architecture and Reservoir Characterization of the Mississippian Burlington-Keokuk Formation, Northwestern Arkansas", Mississippian Reservoirs of the Midcontinent, G. Michael Grammer, Jay M. Gregg, James Puckette, Priyank Jaiswal, S. J. Mazzullo, Matthew J. Pranter, Robert H. Goldstein
Download citation file:
The Mississippian system in the midcontinent of the United States is a complex carbonate- and chert-dominated system with a large degree of reservoir variability and heterogeneity. An outcrop study was done in the state of Arkansas on the Middle Mississippian (Visean) Burlington-Keokuk Formation to analyze the depositional setting and high-resolution sequence stratigraphic architecture to better understand the reservoir distribution of similar units in the subsurface.
The outcrop location, in the northwestern portion of the state of Arkansas, was studied using an integrated sequence stratigraphic approach, combining high-resolution photography for tracing bed boundaries and lithologic contacts along with facies determination from outcrop and thin section analysis. A range of skeletal packstones to grainstones dominated by crinoidal fragments and an abundance of void-filling syntaxial calcite cements comprised the majority of the outcrop facies. Nodular to bedded siliceous limestone to carbonate-rich chert facies were observed containing up to approximately 50% microporosity. Based upon facies assemblages and the presence of meter-scale sand waves with faint cross bedding on outcrop, these units were likely deposited in a high-energy sand shoal or sand bar in a proximal position on a distally steepened ramp.
Within the outcrop, multiple shoaling upward packages were observed, consisting of siliceous limestones and cherts at the bases overlain by coarsening and thickening upward grainstone bodies. This stacking pattern was observed at two different scales. Larger-scale packages 15 to 35 feet (5–10 m) thick were mappable and continuous across the entire outcrop (1320 ft [400 m]), and are inferred to be controlled by eustatic sea-level change. A smaller-scale stacking pattern was observed on the meter (several feet) scale and were mappable for 165–500 ft (50–150 m) laterally. The lack of limited lateral correlation is inferred to be due to autocyclic controls within the active sand body. The observed shoaling upward patterns create a hierarchy of stacked reservoir and seal units with superimposed variability. These findings illustrate the potential for high-frequency sea-level change and autocyclic control on facies and reservoir distribution that may be seen in the subsurface. Two-dimensional geostatistical modeling further illustrates the need for this level of characterization, as variogram inputs are biased significantly by the segregation of high-frequency sequences and dominant eustatic or autocyclic controls on deposition.
The Lower to Middle Mississippian rocks of the southern midcontinent of the United States contain significant hydrocarbon reservoirs in the states of Oklahoma and Kansas. The Mississippian rocks in the region comprise a mixed carbonate and siliciclastic system of limestone, dolomite, shale, and chert. Many reservoirs are predominantly porous chert facies, including chert breccias, tripolite, spiculite, and fractured cherts (Rogers, 1996; Montgomery et al., 1998; Rogers and Longman, 2001; Watney et al., 2001; Mazzullo et al., 2009). The interplay and distribution of chert has become an important issue as operators target the siliceous reservoirs within this carbonate-dominated system. To characterize the rocks, previous research was done ranging from interpretation of the regional-scale paleodepositional environment (Lane, 1978; Lane and DeKyser, 1980; Gutschick and Sandberg, 1983) to localized outcrop investigations in northwest Arkansas, southwest Missouri, northeast Oklahoma, and southeast Kansas (Cline, 1934; Huffman, 1958, 1960; Thompson and Fellows, 1970; Manger and Shanks, 1976; Thompson, 1986; Choquette et al., 1992; Evans et al., 2010; Mazzullo, 2011; Mazzullo et al., 2011a; Shoeia, 2012; Boardman et al., 2013; Mazzullo et al., 2013; Morris et al., 2013; Childress and Grammer, 2015). Much of the previous work focused on evaluating the macroscale variations in lithofacies, rock fabric, and biostratigraphic zonation within the system. Although the previous work has led to a better understanding of the general depositional history of the Lower to Middle Mississippian in the southern midcontinent, little detailed work has been done in an effort to explain the fine-scale heterogeneity that is observed in both outcrop and subsurface that potentially controls reservoir distribution and architecture.
Previous work on the Lower to Middle Mississippian can be tied to potential third-order eustatic sea-level fluctuations (1–6 Myr; Haq and Schutter, 2008). In the midcontinent, these eustatic sea-level changes have been constrained biostratigraphically in outcrop through the analysis of conodonts (Thompson and Fellows, 1970; Boardman et al., 2010). Analyses of these larger-scale third-order depositional packages do not, however, adequately describe the internal complexity observed in these units, which are made up of smaller-scale sequences and cycles stacked within the third-order sequences. Studies have suggested that these higher-frequency fourth- and fifth-order sea-level changes not only control facies stacking patterns in carbonate systems but also the lateral distribution of facies and often the reservoir architecture on a much finer scale (Goldhammer et al., 1993; Grammer et al., 1996; Kerans and Tinker, 1997; French and Kerans, 2004). Third-order sea-level change constrained by conodont biostratigraphy provides an excellent framework to build upon, but no studies to date have addressed the fine-scale facies heterogeneity and resulting reservoir-scale architecture in the rocks resulting from Milankovitch-scale sea-level changes.
In the subsurface, the Mississippian system of the midcontinent is characterized by a large degree of heterogeneity and uncertainty in terms of predicting reservoir distribution (CoreLab, 2015). Assessing whether high-frequency sea-level fluctuations are controlling facies distribution in the Mississippian units in the midcontinent should elucidate some of the complexity currently observed in the system. The goal of this study is to illustrate that the incorporation of high-frequency cyclicity within a sequence stratigraphic hierarchy, in combination with detailed depositional interpretations, can explain facies distribution and increase the predictability of reservoir and nonreservoir intervals both vertically and laterally. These findings may be utilized to enhance the understanding of similar Mississippian units in the midcontinent of the United States, as well as similar carbonates elsewhere in the world.
Early to Middle Mississippian sediments in the midcontinent were deposited in relatively shallow tropical to subtropical conditions 20–30° south of the paleoequator (Gutschick and Sandberg, 1983; Witzke, 1990). General models suggest that deposition occurred on a distally steepened ramp, with strike trending roughly east–west (Franseen, 2006; Mazzullo et al., 2009, 2011a; Childress and Grammer, 2015, 2019). The ramp was regionally extensive, covering portions of Oklahoma, Kansas, Colorado, Nebraska, Missouri, and Arkansas with proximal shallow water conditions to the north and deeper, sediment starved conditions to the south (Lane and DeKyser, 1980; Gutschick and Sandberg, 1983; Figure 1). The ramp was bounded to the east by the Ozark uplift, to the north and west by the Transcontinental arch, and to the south by the ancestral Anadarko and Arkoma basins (Lane and DeKyser, 1980). Uplift of the Ozark region occurred during the Pennsylvanian until the end of the Paleozoic, and possibly into the Paleogene (Simms et al., 1995), exposing Mississippian age rocks. The resulting outcrop belt is present around the periphery of the Ozark uplift (Figure 2). Older strata are present to the east as units are truncated by uplift and erosion, whereas younger Pennsylvanian strata are present to the west and south as the Mississippian dips into the subsurface.
In the midcontinent, Lower to Middle Mississippian generally refers to Kinderhookian through Osagean-aged strata (Tournasian to Visean). The Late Devonian to Earliest Mississippian Woodford Shale provides the lower boundary for the section while the uppermost units are overlain by Meramecian-age rocks (upper Visean) in outcrop. Between the Woodford and Meramec, a number of formations have been identified in outcrop (Figure 3).
The focus of this work is on the middle to upper Osagean intervals, namely the Burlington-Keokuk Formation. Underlying the Burlington-Keokuk is the Reeds Spring Formation. Throughout the outcrop region, it is typically 70–200 feet (20–60 m) in thickness (Mazzullo et al., 2011a). It is composed of lime mudstone to wackestone with pervasive interbedded to nodular chert. It has been interpreted to represent relatively deep water deposition of transported proximal carbonate muds (Mazzullo et al., 2011a). Locally the Reeds Spring is capped by the Pineville tripolite, which is a white to yellowish gray tripolitic chert interval. The Pineville tripolite ranges in thickness from 0 to 80 feet (0–25 m) and is extremely porous, with porosities reaching 50% in some instances.
Overlying the Reeds Spring and Pineville tripolite is the Burlington-Keokuk Formation, also referred to variously as Upper Boone, Burlington, Keokuk, and Bentonville formations. It was deposited during the Middle Mississippian, upper Osagean (Visean). The unit is a combination of the individual Burlington and Keokuk formations. The two formations generally have a gradational, poorly recognized contact, resulting in the formations being grouped together (Choquette et al., 1992). Within the Mississippian outcrop belt in northern Arkansas and southern Missouri, the interval is typically 100–150 feet (30–45 m) thick (Choquette et al., 1992; Mazzullo et al., 2013) and is dominated by fine to coarse grain skeletal packstones to grainstones with interbedded lenses of mudstone and wackestone and nodular to bedded chert (Mazzullo et al., 2013). The rocks are interpreted to have been deposited in relatively shallow, moderate- to high-energy settings (Mazzullo et al., 2011b).
The Short Creek Oolite caps the Burlington-Keokuk in portions of the outcrop and comprises the uppermost portion of the Osagean. It is typically 0–10 feet (0–3 m) thick, but locally can reach 25 feet (7 m; Greenberg, 1981; Lisle, 1983). The interval consists of oolitic packstones to grainstones with variable amounts of skeletal fragments and carbonate muds in some instances (Lisle, 1983). The rock may exhibit variable degrees of cross-bedding, but may also be void of clear bedding structures (Lisle, 1983; Ritter and Goldstein, 2012). The occurrence of ooids along with key sedimentary structures has led to the interpretation of deposition in high-energy shoal settings as well as downdip transport and deposition of these facies to more distal positions (Ritter and Goldstein, 2012).
During the Mississippian, the world was in a transitional period from predominantly greenhouse conditions in the Devonian to icehouse conditions that prevailed in the Pennsylvanian (Read, 1995). No work has been done to quantify the amplitude of sea-level fluctuation in the midcontinent during the Mississippian, and large-scale swings in sea level during deposition of the Osagean intervals have not been previously identified. Some workers have argued for exposure events potentially related to a combination of tectonic and eustatic change (Mazzullo et al., 2011a), but a significant number of exposure events related to icehouse-scale sea-level fluctuations (50–100 m; Read, 1995) have not been documented. These observations suggest sea-level fluctuations were closer to expected greenhouse-scale amplitudes (5–15 m). Even with low-amplitude sea-level change associated with greenhouse-type conditions, these small eustatic fluctuations impacting a low-inclination carbonate ramp would potentially cause extensive migration of facies belts on short geologic timescales.
DATA AND METHODS
Detailed outcrop analysis was done to constrain the depositional environment, probable effects of high-frequency eustatic sea-level cyclicity, potential reservoir heterogeneity, and possible correlations to subsurface data within the Burlington-Keokuk Formation. The primary study site was an outcrop locality on Highway 71-US 540 approximately three miles north of Bentonville, Arkansas (Figure 4). The trend of the roadcut is northwest to southeast, which roughly parallels the depositional strike of the Burlington-Keokuk system as proposed by Choquette et al. (1992). The total outcrop exposure is approximately 1320 feet (403 m) long and 80 feet (24.5 m) high at its thickest section. At this locality, the Burlington-Keokuk Formation overlies the Pineville tripolite facies of the Reeds Spring Formation, whereas the uppermost section is exposed to present-day subaerial conditions. The capping Short Creek Oolite Member is not present at this location, either due to nondeposition or erosional processes.
The Highway 71 location was selected for study because it is one of the largest outcrop exposures in the area and presented a large percentage of the formation. The comparatively extensive lateral extent of the roadcut allowed for tracing of beds, surfaces, and facies over a significant distance. The extent of the outcrop exposure also allows for identification of potential reservoir heterogeneity on the production scale.
Sampling was done along three vertical transects of the outcrop. Sections were spaced 330 ft and 385 ft (100 m and 118 m) apart to provide control points when interpolating and extrapolating facies and boundaries across the outcrop (Figure 5). Ninety total samples for thin section analysis were collected from the vertical sections, at 1–3 ft (0.3–1 m) spacing. Thin sections were analyzed using PPL and XPL microscopy to identify facies type, grain size trends, and dominant allochem types. Facies identified from thin sections were analyzed to identify repeatable vertical patterns within the outcrop. Repetition of facies stacking was then interpreted to create a representative idealized facies succession for the Burlington-Keokuk as exposed in this outcrop.
Multiple high-resolution images of the outcrop were taken with a GigaPan® Epic Pro, Nikon® D7000 DSLR camera, and Nikon® AF-S Nikkor 300 mm lens. The GigaPan® is an automated device that captured hundreds of images of the outcrop at preset intervals in a large photo montage. GigaPan Stitch™ software seamlessly merged the photos into a gigapixel image. The resulting outcrop photo captured fine details of the rocks, bedding geometry and surfaces, and also provides a holistic view of the outcrop that can’t be examined or analyzed in the field or by traditional photography (Figure 6).
Since the outcrop does not provide a three-dimensional view of the facies geometries present in the system, comparative analysis was done between the Burlington-Keokuk and modern analogs. The comparisons allow for interpretation of sedimentary geometries and relationships in plan view. Combining outcrop-constrained two-dimensional rock data with spatial data from modern analogs can aid in decreasing uncertainty when interpreting facies and reservoir distribution in three dimensions.
Data from the outcrop were input into Petrel to test different two-dimensional realizations tied to facies distributions and the sequence stratigraphic framework determined from the outcrop. The tops of high-frequency sequences were used to constrain layering within the model, whereas facies data from measured sections were distributed within horizons. Horizontal variogram values were changed to create differing views of lateral continuity, and then were compared to observed facies continuity in outcrop. Once adequate matches were chosen, the variograms were related to the identified sequence and cycle hierarchy and dominant controls on deposition. This was done to qualitatively predict variogram values based on an identified sequence stratigraphic hierarchy and to illustrate the importance of fine-scale changes in rock fabric and reservoir that may be overlooked during upscaling processes.
The outcrop is dominated by grain-rich carbonate intervals. Most of the rocks in outcrop are very similar, which add some difficulty in choosing discrete facies. Therefore, facies zones were picked based upon variations in grain-size distribution and relative percentages of given grain sizes. Variations in facies were then used to interpret slight changes in depositional environment and energy within a ramp-type setting (Figure 7). Including the Pineville tripolite at the base of the outcrop, six facies were identified based on relative size and distribution of the dominant allochems.
The Pineville tripolite is a siliceous interval within the upper portion of the Reeds Spring Formation with abundant amounts of interparticle, moldic, and fracture porosity that together result in porosity values up to 50% (Figure 8A). It exhibits a recessive weathering profile and shows a distinct contact with the overlying Burlington-Keokuk Formation. The rock is white to light gray in hand sample and has an abrasive texture. The tripolite is also relatively lightweight due to the abundant amount of pore space present in the samples, which has led to these rocks being referred to as “cotton rock” (McKnight and Fischer, 1970). The rock is composed of a mix of microcrystalline quartz and silt to very fine sand-sized carbonate debris. Crinoid fragments represent the dominant carbonate components (Figure 8A), and elongate pores are present that are interpreted to be sponge spicule molds (Figure 8B), which provides an in situ source of biogenic silica. The primary depositional facies of the tripolite prior to alteration and porosity evolution is inferred to be a spiculitic skeletal wackestone to packstone. Lenses of silt-sized packstone and grainstone facies are interspersed within the porous silica-rich sections.
Facies 1 consists of carbonate-rich chert to siliceous limestone with minor to abundant amounts of micro- to nanoscale interparticle and intercrystalline porosity as defined by Loucks et al. (2012; Figure 8C, D). Total porosity observed in thin section ranges from 3% to 30%. The unit contains 30–80% microcrystalline quartz, with the remaining volume composed of silt to fine-grained carbonate debris and syntaxial calcite cement. Visual identification of constituent grains indicates a dominant contribution from crinoidal material accounting for between 80% and 100% of the skeletal material. Lesser amounts of brachiopods and bryozoans are also present, ranging 0–10% and 0–5%, respectively. Scattered sponge spicules have also been identified in some samples (Figure 8D). Prior to alteration and porosity development, the primary depositional facies of the siliceous zone was a silt to fine-grained skeletal wackestone to packstone with some minor occurrences of coarser-grained skeletal material.
Facies 1 is white to very pale yellow in hand sample. The unit is highly fractured and exhibits a recessive weathering profile. Due to the highly differing characteristics from the rest of the limestones present, the chert is the only facies in the Burlington-Keokuk Formation that can be easily and consistently identified without thin section. Within this particular outcrop, Facies 1 is the only facies in the Burlington-Keokuk with observed reservoir potential.
Facies 2 consists of silt to fine-grained skeletal packstone to grainstone with minor occurrences of medium- to coarse-grained skeletal debris (Figure 9A). The primary depositional fabric is coarser than that of Facies 1, and there is no occurrence of chert or silica within Facies 2. Mud content is generally low or absent but may reach up to 40% in some samples. Skeletal constituents within the unit are composed of 75–100% crinoidal debris, 0–25% brachipod fragments, and 0–5% bryozoans. Small sutured seam stylolites occur within this interval. Little to no porosity was observed in the facies (Figure 9A, B).
Facies 3 is a grainstone composed of silt to very coarse-grained sand-sized skeletal material (Figure 9C, D). Some granule- to pebble-sized skeletal fragments may also be present. Facies 3 is characterized by greater than 50% of the constituent grains being of medium sand size or smaller. Sorting of grain sizes within this unit is poor. Crinoids are the dominant skeletal component and comprise between 50% and 100% of the facies. Bryozoans and brachiopods are minor constituents in Facies 3 (less than 10% combined), but in some samples, they may account for as much as 50% and 15% of the skeletal grains, respectively. Scarce trilobites are also present, making up less than 1% of the skeletal components. Stylolites are observed within the rock, in many instances containing remnant dead oil. There is little to no visible porosity observed within this facies. Original pore space has been largely occluded by syntaxial calcite cements.
Facies 4 is a grainstone composed of silt to very coarse-grained sand-sized skeletal material with some grains reaching granule to pebble size (Figure 9E, F). Like Facies 3, the rocks exhibit poor sorting, but Facies 4 samples contain greater than 50% coarse- to very coarse-grained material. The rocks are dominated by crinoids, accounting for 65–100% of the skeletal debris. Brachiopods are minimal throughout the samples, with visual estimations of 0–5%. Bryozoans are also generally scarce (less than 5%), but may constitute as much as 35% of the total skeletal volume locally. Some minor trilobites are present, but account for less than 1% of the skeletal material. Sutured seam stylolites containing dead oil are present within the facies throughout the outcrop. Overall Facies 4 has little to no visible porosity, with original pore space filled by syntaxial calcite cement.
Facies 5 is composed of medium- to very coarse-grained skeletal fragments with minor amounts of fine sand-sized grains or smaller (<5%; Figure 10A, B). Some granule-sized grains are also present in some samples (<5%). Overall the facies exhibits moderately good to good sorting and is dominated by coarse to very coarse sand-sized grains. Crinoids account for 95–100% of the skeletal components in the facies. Bryozoans constitute 0–5% of the skeletal material, whereas brachiopods contribute 0–2%. Stylolites are common within Facies 5. Little to no visible porosity is observed in the samples at both the macroscales (Figure 10A, B) or microscales (Figure 10C, D). Abundant amounts of clear syntaxial calcite cement is present within the facies.
Facies Stacking Patterns
An idealized facies stacking pattern was identified to develop the sequence stratigraphic framework present in the outcrop by viewing the vertical relationships of facies observed from thin section analysis (Figure 11). These vertical trends in facies are inferred to reflect the stacking pattern of sediments in response to a single rise and fall of sea level. In outcrop, an overall coarsening upward succession was identified with cherty facies (Facies 1) in beds and nodules 4-8 in (10–20 cm) at the base overlain by thickening upward crinoidal facies (Facies 2–5).
In carbonate systems, parasequences are bounded by flooding intervals indicating the start of the next transgressive cycle and exposure surfaces marking the maximum regressive phase (Grammer et al., 1996; Kerans and Tinker, 1997). Within the studied outcrop, only a few key surfaces indicating sequence or cycle boundaries were identified. Due to the paucity of clear bounding surfaces, packages were often defined based upon contacts between coarsening upward successions present in outcrop. Using high-resolution Giga-Pan® images, boundaries were traced along bedding surfaces and correlated between vertically sampled transects to assess their sequence stratigraphic significance.
Sequence Stratigraphic Interpretation
The Burlington-Keokuk Formation overlies the comparatively deep-water Reeds Spring Formation and is regionally capped by the Short Creek Oolite. This stacking pattern represents a gradational shoaling upward succession from mudstone, wackestone, and chert, into crinoidal packstones and grainstones, with overlying oolitic grainstones. This overall stacking is inferred to relate to a lower order composite shallowing-upward sequence (probable third order) controlled by a fall in eustatic sea level. Overall facies changes through this order of cyclicity are relatively subtle, with only few identified exposure-related events from the Reeds Spring to the Short Creek Oolite. Within the Burlington-Keokuk, exposure surfaces are not present, but evidence for subaerial exposure (solution pipes often filled with collapse breccia) occur in the lower part of the section in the tripolite facies. The absence of these features in the grain-dominated facies potentially could be due to relatively low-amplitude sea-level fluctuations that never subjected the formation to subaerial conditions. It is also possible that there are sequence or cycle bounding flooding or exposure surfaces in the system, but the view presented by the road cut isn’t optimal for their identification. Narrow ranges in facies types present in the system suggest low-amplitude sea-level change causing minor shifts in environmental energy. In response to these changes, two potential orders of cyclicity were identified in outcrop superimposed on the overall shoaling upward trend (Figure 12).
Medium-Scale High-Frequency Sequences
Overall the outcrop exhibits three high-frequency sequences. These sequences exhibit a marked change at interpreted boundaries from coarse-grained crinoidal grainstone at the tops of sequences, abruptly changing to fine-grained packstones and grainstones marking the beginning of the next depositional sequence (Figure 12). The upper portion of Sequence 3 has been removed by erosion due to prolonged exposure of the outcrop. This scale of sequence is correlative across the entire outcrop (Figure 13). There are clear differences observed in outcrop between the lower two sequences and the uppermost sequence. Sequences 1 and 2 exhibit definitive shoaling upward successions with cherts at the base and massively bedded zones of Facies 4 and 5. The two sequences also display thickening upward trends in bedding, with beds increasing in thickness from 4 to 8 in. (10–20 cm) at the base of the sequence upward into beds 3 ft (2.5 m) thick. Sequence 3 does not illustrate the same thickening upward trends seen in the lower sections. Also within Sequence 3 there is more abundant interbedded chert, lesser amounts of Facies 4 and 5, and deviations from the idealized facies succession. Sequence 3 may have shown the same trends identified in Sequences 1 and 2, but the upper massively bedded, coarse-grained intervals could have been removed by erosion. It is also possible that the environment or controls on deposition may have varied during deposition of Sequence 3, resulting in slightly different vertical successions.
Small-Scale High-Frequency Cycles
Superimposed within each sequence are a number of higher-frequency shoaling-upward successions. These packages are typically marked a vertical increase in the volume of chert through each high-frequency sequence. The cycles don’t have a complete idealized succession of facies represented in the vertical successions. While some high-frequency cycles are similar in thickness and facies distribution, and their relative position within larger sequences, the cycles cannot be adequately correlated along bedding planes or surfaces to offsetting measured sections (Figure 13). The high-frequency cycles vary laterally between the control sections over only a few hundred feet.
The Burlington-Keokuk Formation in the Ozark region was previously interpreted to have been deposited in high-energy settings, but little work has been done to describe the details of the depositional system. Integration of data and observations from this study, incorporating facies types, grain size, overall lack of mud, bedding features, and interpreted stratigraphic position relative to the Short Creek Oolite within a larger-scale shoaling upward succession, suggests that the Burlington-Keokuk was deposited in a high-energy skeletal sand body or skeletal shoal system in a proximal position on a carbonate ramp. Lack of burrows and trace fossils indicate conditions unsuitable for organisms to colonize, characterized by relatively high current velocities and resulting shifting of sediments and migration of the active sand bodies.
Due to the lateral continuity of beds and relatively small changes in thickness within a given bed across the entire outcrop, the section has been interpreted as representing a parallel cut of the sand body. Most sedimentary features such as sand waves and cross-bedding would be more apparent in a dip-oriented section and may not be apparent in a strike-parallel view of the rocks. While definitive sedimentary structures are difficult to identify in the study interval, some key features have been observed due to differential weathering. Some massive grainstones at the tops of high-frequency sequences illustrate a near sinusoidal weathering profile that may be indicative of higher-velocity current emplacement (Figure 14A). Rocks at these localities preferentially weather and break at slightly inclined angles (approximately 30°) that may represent inclined troughs of sand waves (Figure 14B). Weathered outcrop exposures in other localities representing differing orientations in the Burlington-Keokuk illustrate more apparent cross bedding (Figure 14C). Differential weathering of skeletal grainstones exposes trough cross-stratification and potential wave ripples in the strata, indicative of high-energy deposition.
The facies identified in outcrop can be related to positioning within the sand system in terms of relative current speeds and the effects energy would have on grain size and sorting. The primary spiculitic wackestone and packstone facies of the Pineville tripolite is interpreted to have been deposited in lower-energy conditions below fair weather wave base that were favorable for sedimentation of mixed spiculitic and carbonate material. Deposition is inferred to have been in a position more distal from the sand shoal complex, but could also form within interbar channels. Siliceous carbonates and carbonate-rich cherts of Facies 1 are interpreted to have been deposited near fair weather wave base, outside of the active sand body. These facies would have been deposited in slightly higher-energy conditions than the tripolite due to the overall increase in grain size, but still in relatively low-energy conditions compared to the coarser-grained grainstone facies. Facies 2–5 show a gradational increase in grain size resulting from increasing environmental energy within an active sand body in the inner ramp to ramp crest zone. Facies 5 is interpreted to have been deposited in the highest energy settings and may have had some tidal influence that increased the sorting of the carbonate grains.
Modern analogs are useful in that they allow for an understanding of the facies geometries in plan-view and aid in obtaining an approximation of the geometrical attributes of a given facies (Grammer et al., 2004). The Lily Bank shoal complex in the Bahamas was chosen as a potential modern analog for this study based upon the interpretation of much of the Burlington-Keokuk being a high-energy sand complex. The Lily Bank complex is an ooid-dominated shoal located on the northern edge of Little Bahama Bank, Bahamas (Figure 15). A large amount of research has been done on Lily Bank quantifying spatial variability, facies distribution, and geometrical attributes within the shoal complex (Hine, 1977; Hine and Neumann, 1977; Rankey et al., 2006; Rankey and Reeder, 2011; Sparks and Rankey, 2013), which provide significant insight into the processes and resulting depositional features to be used for comparison. The shoal complex reaches over 2.5 mi (4 km) in width and extends for 17 mi (27 km) in length (Rankey and Reeder, 2011). It was selected as an analog for two main reasons. (1) The shoal complex is located approximately 7.5 mi (12 km) from the platform edge, while many other shoals form at the margin in direct relation to the channeling of energy at the abrupt change in water depth. The setting of the platform, inboard from the platform margin, may be slightly more indicative of a ramp type setting due to the dampening of energy across the platform top. (2) A number of different bar forms are present at Lily Bank representing varying energy regimes similar to what is interpreted to have existed during deposition of the Burlington-Keokuk in this study.
Comparing results from this study with those of Rankey et al. (2006), the outcrop locality shows several similarities to linear shoulder bar deposits (Figure 16). Linear or transverse shoulder bars are asymmetrical sand bodies that are elongated perpendicular or highly oblique to the dominant flow direction (Rankey et al., 2006; Rankey and Reeder, 2011). Shoulder bars on Lily Bank are up to 1.25 miles (2 km) in length and may reach a kilometer or more in width. The bars exhibit a gradational change from underlying packstone facies into thickening upward grainstone bodies (Rankey et al., 2006). Sedimentary structures are also less apparent within the strike dimension in comparison to the dip dimension. Similar trends are exhibited within the Burlington-Keokuk. The individual sequences are laterally extensive for over a kilometer (0.6 mi) in the strike direction. Packages of rocks also exhibit shoaling and thickening upward trends toward high-frequency sequence tops similar to what is seen in the modern linear shoulder bars.
It has been shown that current speeds across the shoulder bars are highest along the bar crest and lowest within the channels between the bars (Rankey et al., 2006). This results in better sorting of grains in the relatively higher-energy portions of the system. The end-member grain types present in the Lily Bank system range from silty, mud-lean, poorly sorted packstones in the deeper, lower-energy channels and lagoonal settings between bars to well-sorted oolitic grainstones along the shallow bar crests (Rankey et al., 2006). This creates a general coarsening upward trend in grain size. Bar forms also show a thickening upward trend in bedding (Figure 17A). These characteristics are very similar to the observations seen in outcrops of the Burlington-Keokuk. A similar coarsening upward trend is observed from silicified packstone to well-sorted crinoidal grainstone. Both settings also illustrate that sorting of grainstone facies aids in the interpretation of the depositional environment. The vertical trends in bed thickness are also illustrated in outcrop. Within the lower two sequences, siliceous wackestone to packstone intervals show a gradational contact with overlying thickening upward beds of massive crinoidal grainstones (Figure 17B).
The Lily Bank shoal complex is present on the Bahamian Platform, which differs greatly from the ramp or distally steepened ramp-type system present during deposition of the Burlington-Keokuk. While the shoal is set back several kilometers from the platform margin, there could still be a significant amount of variability in energy settings due to differences in bathymetric profiles between ramps and platforms. Resulting geometries within the shoal may be variable due to these differences in setting. Lily Bank is also flanked by three isolated reef complexes at the platform margin (Rankey et al., 2006). These reefs can channelize flow and cause variations in direction and intensity of currents responsible for the formation of variable bar forms. Within the upper Osagean, no large-scale shallow water reefs have been identified within the system. Therefore, it is likely that no impediments or barriers to currents existed in the depositional environment to potentially disrupt or alter current orientation. The presence of reefs in association with Lily Bank and the absence of barriers during deposition of the Burlington-Keokuk could create deviations in the bar types present at the two localities. While the differences present between the two systems could result in varying geometries and facies distribution, correlations can still be interpreted between the two systems due to the similar trends in grain size and sorting, as well as trends in bed thickness, despite the difference in grain types. As such, Lily Bank is a viable analog that provides a first-order approximation of the geometries and sedimentary trends that could be possible, as well as insight into the maximum level of facies heterogeneity that may be present in the Mississippian section analyzed in this study.
Controls of Reservoir Distribution
Facies analysis from thin sections and stacking patterns observed in outcrop led to the identification of an idealized stacking of facies deposited during one rise and fall of sea level. Based on the idealized facies succession, the lower-energy cherty facies are overlain by more carbonate-rich grainstones in the shoaling upward package. Cherty facies contain minor to abundant amounts of porosity and are capped by tightly cemented grainstones, a strong depositional control on reservoir facies and potential for stacked reservoir and seal units in response to multiple sea-level fluctuations.
Within the outcrop, high-frequency shoaling-upward sequences are laterally continuous and show repetitive shallowing-upward packages. Due to the continuity of the packages across the outcrop, it is inferred that the vertical variations in facies are most likely driven by Milankovitch-scale eustatic sea-level change. As such, higher-frequency sea-level fluctuations are interpreted to be a controlling factor of fine-scale reservoir distribution, a finding not previously identified in the Mississippian within the midcontinent.
Higher-frequency events superimposed on the eustatically driven packages were not continuous across the outcrop or even between the vertical measured sections. Given the depositional interpretation of the system being a high-energy sand shoal, it is likely that these smaller packages, and the presence of scattered porous chert in the outcrop, is primarily driven by autocyclic events and cannot be tied to eustatic changes. High-energy conditions and the presence of potential migrating sand bodies driven by tide and wave energy have been shown in other environments to produce shallowing-upward successions, like those derived solely from sea-level change (Eberli et al., 2005; Harris, 2010; Rankey and Reeder, 2011). The combination of allocyclic and autocyclic controls adds insight into both the predictive nature of the porous reservoir facies stratigraphically, as well as the level of heterogeneity that may be expected due to in situ processes.
The identification of stacked reservoir and seal units as a function of eustatic change provides a level of predictability in reservoir distribution in both the lateral extent and the vertical segregation. Superimposed autocyclic heterogeneity adds a level of complexity that can inhibit reservoir characterization. This stresses the need to understand the sequence stratigraphy at multiple scales along with an accurate depositional interpretation to understand both allocyclic and autocyclic controls on reservoir distribution.
The distribution of facies is easily observed in the outcrop in two-dimensional space. Interpretation of both allocyclic and autocyclic processes are inferred to be controlling depositional stacking and lateral facies extent across the section. Both of the lower two sequences illustrate continuity, whereas the uppermost sequence shows a high degree of variability. Due to this, various 2-D facies models were created to identify which variable parameters were required to model the dominant allocyclic and autocyclic controls on the 1–10 meter scale and to match the facies distribution observed in outcrop.
Numerical inputs were assigned to facies from each of the three measured sections and modeled in Petrel. The numerical data used included spatial data for each vertically sampled section, upscaled facies logs, and surfaces representing the various sequence tops present in the outcrop. Multiple variograms were tested using the three inputs and observed in comparison to the actual facies distribution in outcrop.
The cell size for this project was 1 ft × 1 ft × 0.17 ft (0.3 × 0.3 × 0.05 m). A total of 492 layers were used to construct the grid, with a total of 643,365 cells used for model generation. The relatively small cell size was utilized for the model because it allowed for smoother, more gradational representations of the facies distribution observed in outcrop. While the small cell size increases resolution, the two-dimensional nature of the model had a relatively small total number of cells, which still allowed for rapid computation.
Five surfaces were input into the model to create the stratigraphic and structural framework. The surfaces included the road surface, top of tripolite, the tops of HF sequences 1 and 2, and the top of the outcrop. The facies model was created from inputs observed along the three vertically sampled transects on the outcrop. Each facies was given a designated number from one through five that corresponded with its position within the idealized facies succession. Transects were first upscaled and then input into the model as facies logs (Figure 18).
Since understanding variations in lateral distribution of facies was the primary test, vertical variograms were held constant at 1 ft (0.3 m) for all facies in each sequence. Lateral variograms were tested to identify which parameters best suited a given sequence. Differing realizations are presented in Figure 19 using 1000, 500, 300, and 100 ft inputs (305, 152, 92, and 30 m, respectively). A strong correlation was observed between variogram settings and interpreted depositional processes. Sequences 1 and 2 show coarsening and thickening-upward trends in response to high-frequency sea-level fluctuation. The dominant eustatic control on these sequences created relatively uniform deposits with relatively low degrees of lateral heterogeneity within the outcrop. Variogram values were therefore higher due to the continuity of facies. Sequence 3 shows a much higher degree of variability in facies both laterally and vertically due to higher degrees of inferred autocyclic processes. This created discontinuous facies bodies that vary on a much finer scale than those present in Sequences 1 and 2. Therefore, Sequence 3 required lower variogram values to more accurately model the heterogeneity within the units. Due to the similar depositional processes and controls and resulting lateral distribution and vertical stacking patterns observed in Sequences 1 and 2, both were assigned matching variograms for each facies. Sequence 3 was assigned relatively lower variogram values.
The variograms of 500 and 1000 ft (152 and 305 m) produced nearly identical outputs within Sequences 1 and 2 (Figure 19). Due to the correlative nature of high-frequency sequences over a length of 1320 ft (403 m) and the inability to correlate autocyclic packages, the variogram of 500 ft (152 m) was chosen for the lower two sequences. Given the 500 ft (152 m) value, the range of influence from one control section would encompass most of the outcrop while also retaining some variability derived from autocyclic processes. A value of 150 ft (46 m) was assigned to the uppermost horizon as it produced the best representation of facies distribution. The resulting model best fits the outcrop distribution of facies and illustrates the importance depositional controls have on modeling parameters (Figure 20).
The combination of significant heterogeneity in high-energy sand bodies coupled with the outcrop dataset being only two-dimensional dataset may result in a number of factors that could alter or impede the modeling process. (1) The inability to correlate autocyclic packages of rock doesn’t allow for maximum control when modeling facies distribution. Due to the complexity created by autocyclicity, high-frequency boundaries cannot be adequately input into the model as parameters controlling the facies distribution. (2) This outcrop provides one “snapshot” of the entire Burlington-Keokuk system. The modeling parameters may not be ideal for other sections within the formation. Varying degrees of autocyclicity or alterations in the depositional environments from what is observed in the outcrop may require different variogram values to adequately model facies distribution. (3) In the two-dimensional outcrop model, the minor variogram was ignored. If the model was carried into the third dimension, an understanding of the degree of minor continuity would be critical, given the interpretation as a high-energy sand systems. Shoulder bars are more continuous along strike, which is the interpreted orientation in the outcrop and would require higher variogram values. Bar forms are less continuous in the dip dimension, have asymmetrical geometries, and exhibit higher degrees of facies variability due to slight changes in environmental energy and water depth (Rankey and Reeder, 2011). It should be expected that variogram inputs would be lower in the dip direction than those along strike. (4) The limit of the outcrop is equivalent to the spacing of approximately 350–400 ft (1150–1320 m). Although carbonate reservoirs may be produced at a well spacing of 10–40 acres, many reservoirs are developed at larger spacing. Larger variograms could be used to capture the overall geometries present in the system, but may not adequately illustrate the finer-scale heterogeneity in the rocks that may control reservoir distribution. An adequate understanding of the depositional environment and sequence stratigraphic architecture can provide insight on whether modeling outputs are geologically reasonable, or if a large degree of oversimplification is inherent in the model. Integration of datasets (outcrop, core, seismic, logs, etc.) can be used to place constraints on the model to maintain its geologic integrity.
This study used an integrated sequence stratigraphic and modeling approach combining multiple datasets to better understand the controls on deposition, resulting facies distribution, and potential reservoir implications of the Mississippian Burlington-Keokuk Formation and similar depositional environments found in the subsurface. The outcrop study location was composed of interbedded carbonate and chert intervals, with reservoir quality present in microporous carbonate-rich chert and siliceous limestone facies. Carbonate facies with little to no porosity varying from crinoidal packstones to grainstones comprise most the formation. A range of grain size and sorting was observed ranging from fine-grained packstone and grainstone to coarse- and very coarse-grained grainstone. Skeletal packstone and grainstone facies all contain a large volume of syntaxial calcite cement, occluding all effective primary pore space in the carbonate facies. The facies assemblage dominated by larger grain sizes and general lack of mud, along with identification of meter-scale sand waves and faint cross bedding, led to the interpretation of the Burlington-Keokuk in the study area as being deposited along an inner ramp within a skeletal sand shoal or sand bar.
Shallowing-upward successions were observed at multiple scales, with porous chert at the base overlain by coarsening and thickening upward grainstone bodies. On the larger scale, 15–35 ft (5–10 m) thick packages correlate over distances greater than 1320 ft (400 m) and are interpreted to be controlled by high-frequency eustatic change due to the level of continuity. Smaller-scale 3–6 ft (1–2 m) thick shoaling upward successions were observed superimposed within the larger-scale packages. These high-frequency packages only correlate over distances of 165–500 ft (50–100 m) and are inferred to be controlled by autocyclic processes within the active shoal system.
The hierarchy of stacking by both allocyclic and autocyclic mechanisms creates a series of stacked reservoir and seal units with superimposed heterogeneity. These observations have not been previously described in the Mississippian of the midcontinent but may lead to significantly better understanding of the reservoir architecture. This study illustrates the potential for explaining variability in the Mississippian reservoirs by integration and characterization of the depositional environment and high-frequency sequence stratigraphic architecture. Two-dimensional geostatistical modeling further illustrates the need for accurate identification and classification of the sequence hierarchy and depositional controls. Differing input parameters need to be varied within individual sequences based on the dominant control on deposition (i.e., allocyclic vs. autocyclic), as well as the resulting variability observed in the lateral and vertical distribution of facies.
This work was completed as part of Oklahoma State University’s Mississippian Consortium, funded by Devon Energy, Chesapeake Energy, Marathon Oil, Sinopec, Maverick Brothers Energy, SM Energy, Newfield Exploration, Redfork Energy, Unit Petroleum, Chaparral Energy, Longfellow Energy, Samson Resources, American Energy Partners, and Trey Industries. Many thanks to the OSU students who assisted in field work and data collect: Miranda Childress, Stephanie Leblanc, Taylor Thompson, and Beth Vanden Berg. We would like to thank James Puckette and Darwin Boardman for their assistance and mentorship on this project, and Matt Pranter for his assistance in geomodeling. We are grateful for the donation of the Petrel Modeling Software from Schlumberger. We would also like to recognize Bing for use of aerial maps and photography. The contents of this chapter were derived primarily from the first author’s master’s thesis.