17: Drivers and History of Late Fluid Flow: Impact on Midcontinent Reservoir Rocks
Published:January 01, 2019
- PDF LinkChapter PDF
Robert H. Goldstein, Bradley D. King, W. Lynn Watney, Tony M. Pugliano, 2019. "Drivers and History of Late Fluid Flow: Impact on Midcontinent Reservoir Rocks", 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:
Late diagenesis records a common history of fluid flow in sub-Permian strata in the midcontinent, where fluid inclusion Th are higher than burial temperatures and Tmice show evolving salinity. Most negative δ18Odolomite and highest Th are at the top of the Mississippian. Fluid inclusion and geochemical data point to advective fluid flow out of basins utilizing Cambrian–Ordovician–Mississippian strata as an aquifer for hydrothermal fluids. The Pennsylvanian was a leaky confining unit. This system evolved from: Stage 1 Pennsylvanian–early Permian pulsed hydrothermal migration of connate brine and gas; between Stages 1 and 2, low-temperature Permian brine reflux; Stage 2 mixing between high-temperature and low-temperature brines during the Permian; and Stage 3 large-scale migration of hydrothermal brines and oil later during the Permian or after. Stages 1–3 were the most important late processes affecting Mississippian reservoirs, and record an inverted thermal structure with most impact of hot fluids at the top of the Mississippian. Stage 4 shows radiogenic 87Sr/86Sr in calcite, supporting a transition to localized fault pumping from basement, likely driven by Laramide fault reactivation. Stage 5 is the current system, with Ozark and Front Range uplift-driven fluid flow and potential for small-scale sporadic fault pumping.
Understanding porosity distribution, thermal history, and fluid migration in Mississippian and other reservoirs of the midcontinent depends on evaluating the controls and impact of late-stage fluid flow. A recent study has focused on late-stage diagenesis in the Ordovician Arbuckle Group in southern Kansas, and hypothesized that late-stage (Pennsylvanian and later) fluid migration events were important in porosity evolution and thermal history (King and Goldstein, 2016). This paper evaluates hypotheses for late-stage fluid flow by completing a diagenetic analysis of much of the Ordovician, Mississippian, and Pennsylvanian section in southern Kansas, concentrating particularly on Mississippian strata, which were impacted greatly by hydrothermal processes.
Hydrothermal fluid migration has affected the study area (e.g., Wojcik et al., 1992, 1994, 1997; Young, 2010; King and Goldstein, 2016). Similarities in late-stage mineral paragenesis, fluid inclusion microthermometric data, and geochemical data in stratigraphic units ranging from the Arbuckle Group to Pennsylvanian strata support common histories of fluid flow (Sharp, 1978; Cathles and Smith, 1983; Garven and Freeze, 1984a, b; Leach and Rowan, 1986; Oliver, 1986; Sverjensky, 1986; Garven et al., 1993; Leach and Sangster, 1995; Coveney et al., 2000). The Berexco Wellington KGS 1-32 core in Kansas provided the opportunity for new core observations and petrographic analysis (Figure 1) to integrate data through a large stratigraphic interval. Data include transmitted-light, reflected light, and UV epifluorescence petrography, back-scattered electron microscopy (BSE), and fluid inclusion analysis. These data are paired with carbon, oxygen, and strontium isotope analyses to aid in understanding the history of fluid flow. The new dataset is augmented by complementary data from the Arbuckle Group, recently published by King and Goldstein (2016); Mississippian, recently published by Ramaker et al. (2014); and Pennsylvanian, published by Walton et al. (1995), and Wojcik et al. (1992, 1994, 1997). Together, they provide an improved understanding of the controls on hydrothermal fluid flow in the midcontinent and its impact on Mississippian and other reservoir rocks.
The Paleozoic section in southern Kansas, U.S.A., resides in an area of shallow burial, no more than 1 or 2 kilometers (3281–6562 ft), laterally adjacent to two deep basins (8–12 kilometers [26,247–32,808 ft]) to the south, the Arkoma and Anadarko basins. The area remained tectonically stable until late in the Mississippian, when Ouachita–Marathon deformation was responsible for folds and faults that were active in the Pennsylvanian and likely early into the Permian. Afterward, the area was tectonically relatively stable until the area was structurally rejuvenated in association with the Late Cretaceous to Paleogene Laramide orogeny. Although most hydrocarbon source rocks in the area are immature, the area is well known for oil and gas production. There is also historical production of lead and zinc associated with Mississippi Valley Type ore deposits of the Tri-State district. On this basis, hydrothermal fluids and hydrocarbons must have had a history of migration through these rocks.
New data generated in this study came from the Wellington KGS 1-32 core, where samples were studied from the Pennsylvanian section down to the base of the Ordovician Arbuckle Group (Figure 2). Stratigraphically, the area sampled by the Wellington KGS 1-32 core starts at granitic basement at a depth of approximately 1577 m (5174 ft). The Arbuckle Group is overlain by a thin section of Reagan Sandstone, which is overlain by approximately 305 m (999 ft) of mostly Lower Ordovician Arbuckle Group dolomite. This is unconformably overlain by the Middle Ordovician Simpson Group (31 m [101 ft]), a package with abundant sandstone with smaller amounts of limestone, dolomite, and shale becoming more prevalent toward the top of the unit (Merriam, 1963; Zeller, 1968; KGS, 2012). The Devonian–Mississippian Chattanooga Shale rests unconformably on underlying strata. The Chattanooga is a black to dark-gray shale and can be silty, dolomitic and pyrite-rich (Zeller, 1968; KGS, 2012; Scheffer, 2012). In the Wellington KGS 1-32 core it is 0.15 m (0.5 ft) thick, but just 2/3 mile (1 km) away, in the Wellington KGS 1-28 well, it is 15 m (50 ft) thick. It is unconformably overlain by 93 m (305 ft) of Mississippian limestone, shale, dolomite, and chert. The upper Mississippian interval is the oil-producing unit in the Wellington oilfield and it is currently being piloted for enhanced oil recovery via injection of CO2 (Scheffer, 2012; http://www.kgs.ku.edu/PRS/Ozark/). The Mississippian is capped by a regional paleokarst surface and overlain by a thick (locally 762 m [2500 ft]); regionally 1326 m (4350 ft) Pennsylvanian to Lower Permian section of cyclothemically alternating shales, limestones and sandstones. The cyclothemic interval is overlain by a section that is rich in Permian marine and continental evaporites.
Although early diagenetic events have clearly affected units in the study area (e.g., Figure 3A, B), only diagenetic events that have been interpreted as occurring late in the paragenesis are addressed here (Figures 3A, B; 4). A late-stage interpretation is primarily based on timing that postdates stylolitization. Late-stage cements are described for Middle Ordovician, Mississippian, and Pennsylvanian strata and compared to already published results from the Arbuckle Group (Figure 3A; King and Goldstein, 2016): the Mississippian (Figure 3B; Ramaker et al., 2014) and the Pennsylvanian (Wojcik et al., 1992, 1994, 1997; and Walton et al., 1995). Pore types are classified using Choquette and Pray (1970). Folk (1965) was used to describe calcite textures and Gregg and Sibley (1984) was used to describe dolomite textures. For each stratigraphic unit, the generalized late paragenesis is illustrated in Figures 3A, B, and Figure 4.
Late-stage diagenesis in the Arbuckle Group was documented by King and Goldstein (2016). Their paragenesis (Figure 3A) showed that after stylolitization, there was fracturing, which was followed by dissolution of quartz-silica phases and carbonate, and precipitation of megaquartz, baroque dolomite, galena, sphalerite, and calcite. Based on timing after the onset of stylolitization, King and Goldstein (2016) argued that the petrographic evidence supported that the Arbuckle’s late-stage diagenesis was during the Pennsylvanian or later. This was argued on the basis of burial history. The Arbuckle would have only been buried under about 0.5 km (1640 ft) of section at the onset of Pennsylvanian deposition (Figure 2). Pennsylvanian deposition likely would have been necessary to achieve the depths necessary for the onset of large-scale through-going stylolitization.
Late diagenesis in the Simpson Group (Figure 4) is similar to that of the Arbuckle Group. A simplified paragenesis shows late-stage fracturing (SF1); the fractures go around grains rather than through them and do not cut across quartz overgrowths (Figure 5A). Some of the original carbonate grains are preserved as molds; timing of this dissolution relative to fracturing is not fully clear. Quartz overgrowths (SMQ; Figure 5B) precipitate after the earlier fractures and before baroque dolomite (Figure 5A). Baroque dolomite (SBD) displays a xenotopic-C texture with pore- and fracture-reducing, saddle-shaped crystals characterized by curved crystal terminations and sweeping extinction under crossed polars. Dolomite is followed by another fracturing event (SF2) that can be observed as nearly vertical fractures cross-cutting detrital quartz grains and quartz overgrowths (Figure 5C). The lack of baroque dolomite in the late fractures suggests fracturing was after SBD precipitation.
Late diagenesis in the Mississippian is similar to that of the Arbuckle Group. The simplified late-stage paragenesis reported here (Figure 4) is consistent with the more detailed late-stage paragenesis reported by Ramaker et al. (2014; Figure 3B) for nearby areas of southeastern Kansas. In the late paragenesis, a significant amount (<5%) of carbonate dissolution creates vuggy pores followed by brecciation from collapse of those pores (Figure 6A, B). Megaquartz cement (MMQ) and chalcedony (MCh) follows this (Figures 6A, C, D; 7C).
Baroque dolomite (MBD) is a relatively scarce cement that follows megaquartz and chalcedony. It has xenotopic-C texture (Figures 6D, 7A) and contains primary petroleum fluid inclusions. Nearly vertical fractures are later (Figure 7B) and these contain calcite cement (MCC), the most common late-stage precipitate in the Mississippian strata (Figures 6B, C; 7A–C). There is some ambiguity in the timing of its precipitation, but anhydrite (MA) appears to be the last major phase (Figure 7C). Montalvo-Lliteras (2015) identified an early form of anhydrite consisting of lath-like inclusions in silica nodules, where sediment was displaced around the nodules. The anhydrite observed here is clearly a late phase and is not Montalvo’s early anhydrite.
Pennsylvanian (Cherokee Group)
Pennsylvanian strata show a similar late-stage diagenetic history to that of the Arbuckle Group, Simpson Group, and Mississippian. Our observations in the Wellington core (Figure 4) are consistent with those of Wojcik et al. (1992, 1994, 1997) and Walton et al. (1995), who published a highly detailed late-stage paragenesis for Pennsylvanian strata in this region. The late paragenesis shows carbonate dissolution, baroque dolomite and ankerite precipitation, petroleum migration, fracturing, and late calcite cementation (Figure 7D).
New fluid inclusion data are presented for the Middle Ordovician Simpson Group and Mississippian strata. These data are discussed in comparison to recently published data on comparable phases from the Arbuckle Group (King and Goldstein, 2016), Mississippian (Ramaker et al., 2014), and Pennsylvanian (Wojcik et al., 1997). In analysis of fluid inclusion data, the term fluid inclusion assemblage (FIA) is used as “the most finely discriminated, petrographically distinguishable group of inclusions” (Goldstein and Reynolds, 1994; Goldstein, 2003). Goldstein and Reynolds (1994) defined a consistent FIA as one in which ≥90% of homogenization temperature (Th) data fall within a range of 10°C–15°C and an inconsistent FIA as one that yields more variable Th data. Given variable sizes and shapes of inclusions, consistent Th data are indications that thermal reequilibration of fluid inclusions has not occurred (e.g., Goldstein and Reynolds, 1994). Microthermometric data consist of homogenization temperatures (Th) and final melting temperatures of ice (Tmice). Tmice values are converted to salinity in weight percent NaCl equivalent (wt. % NaCl eq.). Salinity was calculated from Tmice measurements using the AqSoVir program from Bakker (2008) and the equation from Bodnar (1993). Tmice data were only collected from fluid inclusions with gas bubbles present before final melting.
Simpson Group quartz overgrowths appear clear in transmitted light and typically have no fluid inclusions (Figure 5B). The few fluid inclusions that are present are found originating along the dust rim, boundary between detrital quartz and overgrowth, and extending into the overgrowth from there. Inclusions have irregular walls, are approximately 15–30 μm, two phases with consistent vapor:liquid ratio, and typically are isolated as the only fluid inclusion in an entire quartz overgrowth. There is no petrographic evidence for necking down after a phase change. Only one FIA was found containing multiple fluid inclusions (Figure 8). On the basis of distribution in association with the dust rim, fluid inclusion origin is interpreted as primary.
Homogenization temperatures range from about 60°C to 116°C (Figure 9A). One FIA yielded variable Th values, with one inclusion yielding a particularly low temperature (60.4°C). The variation in Th either indicates thermal reequilibration of the FIA or wide variation in temperature during precipitation. Eutectic melting temperatures range from −21°C to −24°C, with the exception of one inclusion showing initial melting at −35°C. All Tmice values are between −21°C and −22°C (Figure 9B), indicating salinities of 23–24 wt. % NaCl eq. There appears to be a correlation between Tmice and Th (Figure 9B inset).
Baroque dolomite (SBD) typically appears clear under transmitted light with patches of cloudiness defined by the high abundance of aqueous fluid inclusions (Figure 8B). The general shape of larger inclusions is elongate in the direction of crystal growth, which suggests a primary origin. Some FIAs have fluid inclusions that appear flat, indicating a secondary origin. Most primary fluid inclusions are too small for microthermometry. All-liquid inclusions were observed, but only among the smaller size inclusions. All of the microthermometry on primary fluid inclusions was conducted on the two-phase inclusions, which showed consistent vapor:liquid ratios among inclusions of various sizes and shapes. Secondary fluid inclusions had variable vapor:liquid ratios and had petrographic evidence suggesting necking down after a phase change.
A total of three FIAs, two consistent primary and one inconsistent secondary, were subjected to heating for the purpose of obtaining Th data. The values collected from primary FIAs ranged from 119°C to 132°C (Figure 9A). In inner-to-outer transects across crystals, Th values repeatedly increase and decrease by approximately 15°C, similar to the observations in the Arbuckle Group in this well (Goldstein and King, 2015) and in the Mississippian in nearby wells (Ramaker et al., 2014). Th values collected from secondary FIAs ranged from 117°C to 125°C, with some inclusions retaining bubbles well above these temperatures (Figure 9A). Tmice values in primary inclusions are between −21.3 and −21.8°C (Figure 9B), indicating salinities ranging from 23.2 to 23.7 wt. % NaCl eq.
Megaquartz cement (MMQ) contains abundant FIAs that are distributed in a feather-like pattern, oriented in the direction of crystal growth (Figure 10A). Fluid inclusions are typically small in size (<5 μm) and display a round shape, with slight elongation in the direction of crystal growth (Figure 10B). Distribution and orientation relative to crystal growth direction indicates a primary origin. Most fluid inclusions have two phases with a small gas bubble. Many of the smallest fluid inclusions are a single liquid phase. There are a small percentage of gas-dominated inclusions, but these are not paired with the all-liquid inclusions, indicating necking down after a phase change has not caused the variability in liquid–gas ratios (Goldstein and Reynolds, 1994), but that heterogeneous entrapment sampled a high-temperature liquid and separate gas phase.
A total of eight inconsistent primary FIAs were subjected to heating for the purpose of obtaining Th data. The resulting Th data vary widely, with values ranging from 86°C to 161°C (Figure 11A). A total of four inconsistent primary FIAs were subjected to freezing for the purpose of obtaining Tmice values. Tmice was between −1.9 and −2.3°C (Figure 11B), indicating salinities ranging from 3.2 to 3.9 wt. % NaCl eq.
Baroque dolomite (MBD) typically contains cloudy cores and growth bands defined by high abundance of fluid inclusions (Figure 12A). Many of the fluid inclusions are elongate in the direction of crystal growth (Figure 12B). This relationship to crystal growth indicates a primary origin. Fluid inclusions within primary FIAs are typically two-phase inclusions with small bubbles in an aqueous liquid, and they vary in size and shape. Liquid:gas ratios among inclusions in FIAs appear to be consistent (Figure 12B). All-liquid inclusions are only in the smallest size ranges (<5 μm). Additionally, MBD contains primary hydrocarbon inclusions, revealing petroleum migration at the same time as fluid responsible for baroque dolomite precipitation. Hydrocarbon inclusions occur as two-phase fluid inclusions in primary FIAs that appear to have been entrapped during early crystal growth of baroque dolomite (Figure 12C, D). These inclusions appear clear to light yellow under transmitted light and display a bright blue color when exposed to UV epifluorescence illumination (Figure 12C, D). SEM-BSE imaging from these samples (King, 2013) aids in determining the mechanism and timing of entrapment of primary fluid inclusions. Initial dolomite precipitation was in a relatively uniform, medium gray BSE phase (MGP). Later, extensive microfracturing was followed by a dark-gray BSE phase (DGP) responsible for recrystallization of the MGP. Although some primary fluid inclusions can be observed in the earlier phase, the majority rest within DGP. This has likely resulted in the majority of the fluid inclusion data being derived from the latest stage of crystal growth and recrystallization, rather than spanning the entire history of crystal growth.
A total of 15 aqueous primary FIAs were subjected to heating for the purpose of obtaining Th data. The values collected from primary FIAs range from 100°C to 144°C (Figure 11C). A total of eight consistent primary FIAs were subjected to freezing for the purpose of obtaining Tmice. Values are between −11°C and −21°C (Figure 11D), indicating salinities ranging from 15 to 23 wt. % NaCl eq. Tmice data gathered from MBD produced two distributions, with one producing values ranging from −11°C to −12.5°C (15–16.4 wt. % NaCl eq.) and the other producing values ranging from −17.5°C to −21°C (20.6–23 wt. %NaCl eq.).
Calcite cement (MCC) is light yellow in color under transmitted light. Fluid inclusions are clustered in the centers of individual crystals and form three-dimensional arrays (Figure 13A), an indication of primary origin. Curviplanar distributions of fluid inclusions, cutting across entire calcite crystals, indicate some FIAs have a secondary origin (Figure 13C). Liquid:gas ratios among inclusions in each primary and secondary FIA appear consistent (Figure 13B, D). All-liquid inclusions are only in the smallest size ranges (<5 μm).
Seven consistent FIAs, four primary, and three secondary, were subjected to heating for the purpose of obtaining Th data. Primary FIAs produced the higher Th values, ranging from 93°C to 103.5°C, and secondary FIAs produce slightly lower Th values, ranging from 67°C to 83°C (Figure 11E). Five consistent FIAs, three primary and two secondary, were subjected to freezing for the purpose of obtaining Tmice values. Tmice results are between −10.5°C and −16.6°C (Figure 11F), indicating salinities ranging from 14.5 to 19.9 wt. % NaCl eq. Similar to the Th data, Tmice data produced distinct data sets for primary and secondary FIAs, with primary FIAs producing values ranging from −10.5°C to −11.5°C (14.5–15.5 wt. % NaCl eq.) and secondary FIAs producing values ranging from −16.2 to −16.6°C (19.6–19.9 wt. % NaCl eq.).
Fluid Inclusion Discussion
The current subsurface temperature at the top of the Arbuckle Group is approximately 44°C in the Wellington 1-32 well, and overlying units are slightly cooler. These low modern formation temperatures are in stark contrast to the Th values reported here from the Wellington 1-32 core: Simpson Group quartz overgrowths and baroque dolomite (60.4°C–132°C); Mississippian megaquartz cement (interpreted to be ≥86°C); Mississippian baroque dolomite (100°C–144°C); primary FIAs in Mississippian calcite cement (93°C–103.5°C); and secondary FIAs in Mississippian calcite cement (67°C–83°C; Figure 11). Data from the Arbuckle Group in this area (King and Goldstein, 2016) and data from nearby wells in Mississippian strata (Ramaker et al., 2014) show similarly high Th values. Hypotheses that could explain the high paleotemperatures include: (1) much greater burial than present; (2) elevated geothermal gradient compared to present; or (3) injection of hydrothermal fluids. Hypotheses (1) and (2) are easily disproven. Newell’s (1997) burial history analysis in the region argues for a maximum temperature of approximately 73°C–74°C at the base of the Arbuckle Group, assuming geothermal gradients of 25–30°C/km and geologically reasonable stratigraphic thicknesses. Given the most extreme possible assumptions of burial depth and most extreme possible assumptions of geothermal gradients, only 109°C–113°C could be achieved at the base of the Phanerozoic section in Newell’s (1997) analysis. Th values measured in baroque dolomite remain higher than can be explained by even the highest temperature estimates for the underlying Arbuckle Group, rendering hypothesis (1) and (2) unlikely. Moreover, Th data that repeatedly increase and decrease cannot be explained by burial or higher geothermal gradient.
Injection of hydrothermal fluids (hypothesis 3) would produce the elevated Th data, and if migration of fluids occurred in multiple pulses, it would also account for fluctuating Th values in transects, and evolution of the fluid composition (Figure 11). The late mineral paragenesis and fluid inclusion data are similar to the paragenesis and fluid inclusion data from MVT deposits in the midcontinent (e.g., Gregg, 1985, Oliver, 1986; Sverjensky, 1986; Bethke and Marshak, 1990; Leach and Sangster, 1995; Wenz et al., 2012). Thus, the interpretation of hydrothermal fluid flow is not unreasonable.
Fluid and Thermal Evolution
As presented above, the late-stage paragenesis of the Arbuckle Group, Simpson Group, Mississippian strata, and Pennsylvanian strata are similar (Figures 3, 4). Timing after the onset of stylolitization supports the idea that this shared late-stage paragenesis occurred in the Pennsylvanian or later, after significant burial. Fluid inclusions in Pennsylvanian strata were well studied by Wojcik et al. (1994). Their results (Figure 14) showed that after an event of calcite cementation at low temperature from highly saline brines (intermediate calcite), baroque dolomite, and ankerite precipitated at a similar high salinity, and at high temperatures, consistent with a hydrothermal system. Thus, it is not surprising that all of these stratigraphic units have had a “shared experience” of hydrothermal fluid flow. The newly expanded record is synthesized in Figure 15, which places the fluid inclusion data from this study in the context of other recently produced data (Ramaker et al., 2014; King and Goldstein, 2016) from the same area. The new results demonstrate that the late history can be represented in five simplified stages of fluid and thermal evolution.
The record of Stage 1 is preserved in the Arbuckle Group and Mississippian strata in megaquartz cements (Figure 15). Aqueous fluid salinity was relatively low (~3–5 wt. % NaCl eq.) and increasing. At times, a separate gas phase was present. Temperatures are high (~90s–120s°C) and show successive rises and falls. In Mississippian strata, there is a record of long-term rise in temperature. Notably among the consistent FIAs, higher temperatures are preserved in Mississippian strata in comparison to the Arbuckle Group.
The record of Stage 2 is preserved in Simpson Group and Mississippian strata in megaquartz and quartz overgrowth cements (Figure 15). Aqueous fluid salinity is high (~18–24 wt. % NaCl eq.). Fluid inclusion Th values are highly variable in some FIAs lacking evidence for heterogeneous entrapment; some FIAs yield consistent Th values; and some FIAs in Mississippian strata contain large all-liquid inclusions in the same size range as two-phase inclusions (Ramaker et al., 2014). This suggests that the great variability of Th values reflects actual paleotemperature fluctuations rather than thermal reequilibration or heterogeneous entrapment. Temperatures range from about 60°C to 115°C in the Simpson Group and less than 50°C to about 150°C in Mississippian strata. Notably, the highest Th values are recorded from Mississippian strata rather than from the Simpson Group. Aqueous fluid salinities are high. A correlation between Tmice and Th in one FIA in the Simpson Group (Figure 9B inset) and the variability of Th within FIAs show that the low Th values are consistent with normal burial temperatures and the high values are consistent with the temperatures of hydrothermal fluids. The correlation between Tmice and Th indicates a mixing relationship between a low temperature, 23 wt. % NaCl eq. end member and a high temperature, 24 wt. % NaCl eq. end member. Thus, stage 2 preserves an event in which hydrothermal fluid flow decreased for a long enough period of time for temperatures to return to the ambient burial temperature. This was followed by another pulse of hydrothermal fluid flow.
The record of Stage 3 is preserved in the Arbuckle Group, Simpson Group, and Mississippian strata in baroque dolomite (Figure 15). Early baroque dolomite in the Arbuckle Group is placed in Stage 3, but its position is somewhat ambiguous; it may represent an intermediate stage between stages 1 and 3. The rest of the baroque dolomite clearly rests within Stage 3. Aqueous fluid salinity was high (~15–23 wt. % NaCl eq.). At times, a separate oil phase was present. Temperatures are among the highest recorded (~90°C–140°C) and show successive rises and falls. Notably among the consistent FIAs, higher temperatures are preserved in Mississippian strata in comparison to the Arbuckle and Simpson groups.
The record of Stage 4 is preserved in calcite in the Arbuckle Group and Mississippian strata (Figure 15). Primary fluid inclusions have Th values that are high (~90°C–105°C), but lower than those of Stage 3. Secondary fluid inclusions show high Th values as well (~65°C–85°C), but yield data that are lower than those of primary fluid inclusions. Fluid salinity remains high (~15–20 wt. % NaCl eq.), but there is some variability.
The record of Stage 5 is taken from extant conditions downhole in the Arbuckle Group and Mississippian depths in the Wellington 1-32 well (Figure 15). Temperatures are uniformly low. Salinity in the Mississippian remains high, but that in the Arbuckle Group is low. In comparison to conditions toward the end of Stage 4, secondary fluid inclusions in calcite of both Arbuckle Group and Mississippian strata record moderately high temperatures and high salinities. The secondary fluid inclusions likely represent a transition period when temperatures were decreasing. In comparison to the modern system, the salinities in secondary fluid inclusions are similar to the high-salinity values obtained from modern Mississippian fluids (<20% KGS, 2010), but much higher than modern salinities observed in the Arbuckle Group (3–11%; Scheffer, 2012). This could mean that the modern Mississippian reservoir has retained fluids related to ancient fluid flow, whereas the Arbuckle Group more recently acted as an aquifer, flushing ancient fluids and recharging the reservoir with more recent fluids. Alternatively, proximity to Permian evaporites may be responsible for higher salinities in the modern Mississippian reservoir.
CARBON, OXYGEN, AND STRONTIUM ISOTOPES
Oxygen (δ18O) and carbon (δ13C) isotopic data (‰ VPDB) and 87Sr/86Sr from baroque dolomite and calcite help to elucidate the conditions during the late stages of diagenesis. New data are presented for the Simpson Group, Mississippian strata, and Pennsylvanian Cherokee Group. Published data from the Arbuckle Group (King and Goldstein, 2016) and Pennsylvanian (Wojcik et al., 1997) are integrated into its discussion. For comparison purposes, all 87Sr/86Sr values discussed from previous authors’ work have been normalized to NBS 987 at 0.710250.
Simpson Group baroque dolomite δ18O is −9.0‰ and δ13C is −2.4‰ (Figure 16A). Both δ18O and δ13C values fall within the range observed in Arbuckle Group baroque dolomite (−6.6‰ to −10.2‰ and −2.4‰ to −3.6‰; Figure 16A). Mississippian strata baroque dolomite δ18O values range from −10.3‰ to −8.3‰, with an average of −9.4‰ (Figure 16A). The δ13C values range from 1.6‰ to 2.0‰, with an average of 1.9‰. Mississippian δ18O values are near the most depleted of those in the Arbuckle Group (−6.6 to −10.2‰), whereas δ13C values are more enriched than Arbuckle Group samples (−2.4 to −3.6‰). From the top of the Mississippian to the base of the Arbuckle Group, δ18O values become less depleted with increasing depth (Figure 16B).
Dolomite 87Sr/86Sr values in Mississippian strata range from 0.70848 to 0.70976 with an average of 0.70912 (Figure 16D). The most radiogenic 87Sr/86Sr value falls within the range of Arbuckle Group baroque dolomite (0.70908–0.71019; King and Goldstein, 2016). Strontium concentration and 87Sr/86Sr cross-plots show that Sr concentration decreases as isotopic ratios become more radiogenic (King and Goldstein, 2016). Paired 87Sr/86Sr and δ18O analyses show radiogenic values are associated with depleted δ18O, with the most radiogenic value (0.70976) being linked with the most depleted value (−10.11‰ VPDB). Paired δ13C and 87Sr/86Sr analyses display the opposite relationship, with the most radiogenic value being associated with the most enriched δ13C value (for detailed plots, see King, 2013).
The δ18O value of Simpson Group baroque dolomite is −9.0‰. Homogenization temperatures and δ18Odolomite data allow for an approximation of δ18Owater (Land, 1985 published equation). Using the highest and lowest Th values results in δ18Owater values of +2.0‰ and +3.2‰ SMOW. The enrichment of δ18Owater values is common in residual evaporite brines and could also be produced by long-term rock–water interaction at high temperatures (Epstein and Mayeda, 1953; Heimstra, 2003).
The Simpson Group dolomite δ13C value is −2.4‰. The slight depletion of δ13C values may be associated with thermochemical sulphate reduction (TSR). The Th values from fluid inclusions (119°C–132°C) fall within the range proposed by Machel (2001; 100°C–140°C) for this mechanism of carbon depletion to play an important role.
The δ18O values in Mississippian strata baroque dolomite range from −10.3‰ to −8.3‰. Calculating δ18Owater using the ranges of Th and δ18Odolomite from Mississippian strata yields a fluid composition of −1.2‰ to +4.9‰ SMOW. These values are similar to those interpreted for baroque dolomite in the Arbuckle Group (−1.9‰ to +5.6‰ King and Goldstein, 2016), again suggesting values common in residual evaporite brines or long-term rock–water interaction during heating in the basin (Epstein and Mayeda, 1953). If one assumes a consistent isotopic composition of the fluid during baroque dolomite precipitation (+2.0‰), the range in isotopic composition yields a temperature range of 22°C, a range that is about half of that observed in fluid inclusion data (44°C). A decrease in δ18Odolomite from the base of the Arbuckle Group to the top of the Mississippian indicates that there were higher temperatures at the top of the Mississippian section compared to those at the base of the Arbuckle Group.
The Mississippian strata baroque dolomite δ13C values range from 1.6‰ to 2.0‰. According to Lohmann and Walker (1989), δ13C for Mississippian marine calcite was around 4‰. Baroque dolomite is 2.0‰–2.4‰ more depleted than what would be expected from Mississippian seawater. The most likely explanation for the depletion of carbon is that high temperatures (100°C–144°C) allowed for TSR to be an active mechanism during the time of fluid migration, and that rock–water interaction also played a role.
The 87Sr/86Sr values of baroque dolomite in Mississippian strata are more radiogenic than unmodified Mississippian seawater (Denison et al., 1994). The most depleted sample (0.70976) is more radiogenic than any seawater values from the Cambrian (0.70904) to the modern (0.70896; Denison et al., 1998). The radiogenic nature of the baroque dolomite samples suggests rock–water interaction with siliciclastic rocks or basement rock at some point during the fluid migration history (Banner, 1995). The 87Sr/86Sr values provide additional support to interpretations that fluids were hydrothermal in origin and from a basinal source. One of the dolomite 87Sr/86Sr values from Mississippian strata falls within the range of values in the Arbuckle Group; this supports the idea that the dolomites in Mississippian strata are genetically related to that in the Arbuckle Group.
Although the 87Sr/86Sr dataset is not extensive enough to make definitive interpretations, it is notable that the most depleted δ18O value (−10.3‰) also has the most radiogenic 87Sr/86Sr value (0.70976), suggesting that the warmest fluids may have undergone the least amount of rock–water interaction with host carbonates. In addition, as the more enriched δ13C are also associated with the more radiogenic 87Sr/86Sr values, depletion of carbon isotopic composition is not solely controlled by high temperatures; localization of organic matter in Mississippian strata may be a control.
The oxygen, carbon and strontium isotope data from the baroque dolomite support a system of advective hydrothermal fluid flow sourced from an adjacent basin. The hydrothermal fluid had an enriched oxygen isotopic composition and radiogenic 87Sr/86Sr, consistent with residual evaporite brine that had interacted with siliciclastics and basement in a deep sedimentary basin. Along its flow path, the brine interacted with organic matter and host rock to evolve in composition.
For calcite, the δ18O values in Mississippian strata range from the most negative value of −10.3‰ to the most positive value of −8.4‰, with an average of −8.9‰ (Figure 16C). The δ13C values range from the most negative value of −0.4‰ to the most positive value of 1.7‰, with an average of 0.6‰. δ18O values are slightly more depleted than values observed in Arbuckle Group calcite (−9.8‰ to −7.0‰), whereas δ13C values are much more enriched than the Arbuckle Group calcite (−18.5‰ to −3.7‰). For the Pennsylvanian, only one new data point was provided because of the extensive previous work of Wojcik et al. (1992, 1994, 1997). The δ18O value is −9.1‰ and the δ13C value is −1.5‰ (Figure 16C).
The 87Sr/86Sr values range from 0.70842 to 0.70930 with an average of 0.70869 (Figure 16D; King, 2013). These values are significantly less radiogenic than the range observed in Arbuckle Group calcite (0.71257–0.71716). Strontium concentration decreases as 87Sr/86Sr becomes more radiogenic (King, 2013). In one well, the 87Sr/86Sr becomes more radiogenic with increasing depth (Figure 16D).
One can calculate the δ18Owater of fluids responsible for calcite precipitation using the Th range and equation of O’Neil et al. (1969). The δ18Owater values range from +5.0‰ to +1.8‰ SMOW, a range that is consistent with evaporative brines or long-term rock–water interaction (Epstein and Mayeda, 1953). If one assumes a consistent isotopic composition of the fluid during calcite precipitation (+3.0‰), the range in isotopic composition yields a temperature range of 17°C, a range that is similar to that observed in fluid inclusion data (10.5°C).
The δ13C values range from the most negative value of −0.4‰ to the most positive value of 1.7‰. Similar to Mississippian baroque dolomite, the slight depletion of carbon in the calcite sample could be attributed to TSR. Primary fluid inclusions within calcite give temperatures ranging from 93.0°C to 103.5°C with some values just within Machel’s (2001) estimated range for TSR (100°C–140°C).
There is a significant difference between δ13C values of Mississippian calcite (−0.4‰ to 1.7‰) and Arbuckle Group calcite (−18.5‰ to −3.7‰). The difference in δ13C values may represent a differing degree of rock–water interaction in the two units. If so, the fluids could have the same origin. Similar δ18Owater support the idea of similar fluids and differing degrees of rock–water interaction. As is the case with other cements discussed, depletion of δ18Ocalcite may result from thermal fractionation and depleted δ13C may result from TSR.
Wojcik et al. (1992, 1994, 1997) conducted a diagenetic study on Pennsylvanian strata in and near the study area. They found evidence for early-, intermediate-, and late-stage calcite cementation events. In Wojcik et al.’s (1994, 1997) study, they stated that they observed a rare nonferroan calcite cement as the latest event of precipitation, but no further analysis was provided.
The 87Sr/86Sr of analyzed calcite is more radiogenic than unmodified Mississippian seawater (0.70755–0.70812; Denison et al., 1994). The most radiogenic sample (0.70930) is more radiogenic than any seawater values from the Cambrian (0.70904) to the modern (0.70896; Denison et al., 1998). The radiogenic nature of the calcite samples suggests rock–water interaction with siliciclastic material or basement rock at some point during the fluid migration history (e.g., Davies and Smith, 2006; Banner, 1995).
One possible explanation for the wide range in 87Sr/86Sr is that the calcite samples could reflect more than one cementation event. Previous authors have found evidence for multiple events of calcite precipitation in nearby systems, believed to be related to events of hydrothermal fluid flow (Coveney et al., 2000).
Another explanation is that the calcite mostly records a single diagenetic system of approximately the same age. The 87Sr/86Sr values appear to vary from well to well, indicating that rock–water interaction is highly localized. Arbuckle Group calcite 87Sr/86Sr values (Figure 16D; 0.71257–0.71716) are much more radiogenic than the range of values observed in Mississippian calcite (0.70842–0.70930). The 87Sr/86Sr values from one well increase with depth. This supports the idea of near vertical fluid flow in that well. Highly radiogenic brines would have been sourced from below the base of the section, and achieved less radiogenic values as they flowed vertically and interacted with less radiogenic carbonates. The highly radiogenic end-member values, well-to-well variation in 87Sr/86Sr, and decreasing values vertically all suggest that precipitation of Mississippian calcite may have resulted from fracture-controlled vertical migration of fluids.
DISCUSSION: HISTORY OF LATE FLUID FLOW
Similarities in late-stage mineral assemblages between the Arbuckle Group and overlying units suggests there is a genetic link between fluids migrating through the entire stratigraphic package, which ranges from Cambrian–Ordovician to Pennsylvanian strata in this study. Petrographic observations in the various units show a similar paragenesis for the late stages that postdate stylolitization, and therefore, represent burial deep enough to generate pressure solution features. This requires Pennsylvanian or later timing for the late quartz cements, baroque dolomite, oil and gas migration, some silica-phase dissolution, some carbonate dissolution, MVT minerals, and late calcite.
The late-stage fluid and thermal histories, that have affected these Cambrian–Ordovician through Pennsylvanian rocks, have been classified into five successive stages separated by one event of low-temperature brine reflux (Figure 15). Observations associated with each of these stages allow for interpretation of timing of hydrothermal fluid flow, origin of the flow system, structure of the aquifer, and impact on Cambrian–Ordovician, Mississippian, and Pennsylvanian strata.
The first stage left a record of megaquartz cement in the Arbuckle Group and Mississippian strata (Figure 15). Homogenization temperatures show repeated rises and falls in temperature superimposed on an overall increase in Th from the 80s°C to the 130s°C. Even with variable size and shape, consistent Th values in FIAs are common, so thermal reequilibration does not appear to have altered the Th values. There is evidence for a separate gas phase, so inclusions were likely trapped on or around the bubble point. Therefore, no pressure correction need be applied, and Th can be used directly as a record of entrapment temperature (e.g. Goldstein, 2012). The highest consistent Th FIAs come from Mississippian strata, not the stratigraphically deeper Arbuckle Group. For the most part, fluid salinities are relatively low, 3.1–6.0 wt. % NaCl eq., with a slightly higher range of salinities (4.7–12.5 wt. % NaCl eq.) appearing late in Stage 1.
Widespread distribution of Stage 1 megaquartz and a similar record of fluid inclusion temperature and low salinities in Mississippian strata, as in the Arbuckle Group, suggest that the Mississippian through Cambrian–Ordovician section was behaving as a single, vertically connected aquifer that received its fluids from high-temperature, more deeply buried areas of basins to the south. This would indicate that pulses of advective fluid flow out of basins to the south were the first major phases of hydrothermal fluid flow recorded.
The low salinities of the fluids are consistent with seawater derivation with little or no interaction with evaporites. The most likely source would be connate fluids from basins to the south (e.g., Ramaker et al., 2014; King and Goldstein, 2016). Pennsylvanian to early Permian timing can be argued for Stage 1 because of these low salinities. Permian evaporites are well known in the region, and other studies (Goldstein et al., 1991; Wojcik et al., 1994; Luczaj and Goldstein, 2000) demonstrated a regional event of low-temperature reflux of high-salinity brines that Wojcik et al. (1994) postulated to represent an event of charging of basins with residual evaporite brine during the Permian. Musgrove and Banner (1993) demonstrated that such brines still reside in the subsurface of Kansas in places. These fluids would have replaced the low-salinity connate fluids with high-salinity brines. Thus, the low salinities of the fluids from Stage 1 indicate timing before (or perhaps during) charging of the adjacent basins with these high-salinity brines. Stage 1 megaquartz precipitation, therefore, occurred largely prior to Permian reflux, likely sometime in the Pennsylvanian and possibly extending into the Permian (Figure 17A).
This timing coincides with Ouachita–Marathon deformation, which may have generated conditions conducive to advective migration of hot fluids out of the nearby Anadarko or Arkoma basins, using Cambrian–Ordovician through Mississippian strata as the aquifers for conduction of fluids. Tectonically valved, episodic dewatering of multiple zones in basins is thought to be capable of expelling hot fluids (Sharp, 1978; Cathles and Smith, 1983; Oliver, 1986; Kupecz and Land, 1991), but it is also believed to produce too little fluid volume to account for high homogenization temperatures in regionally extensive hydrothermal deposits (Leach and Rowan, 1986; Garven et al., 1993). The evidence we have provided for regional late-stage hydrothermal fluid flow would likely rule this mechanism out as the sole driver for hydrothermal fluid flow. Even if episodic dewatering is not the primary driver for fluid flow, tectonic valving by faulting and fracturing could episodically provide fluid pathways and improved permeability.
Alternatively, gravity-driven fluid flow resulting from topographic uplift could have been a driver (Garven and Freeze, 1984a, b; Bethke and Marshak, 1990; Garven et al., 1993; Garven, 1995; Appold and Garven, 1999). Tectonic uplift associated with the Ouachita orogeny caused deepening of the Anadarko and Arkoma basins while generating uplifts that border the basins (Burgess, 1976; Ye et al., 1996; Gallardo and Blackwell, 1999). In this scenario, the Anadarko or Arkoma basins (source of heat) and the surrounding uplifts (hydraulic head source) would have set the stage for regional hydrothermal fluid migration across the study area (e.g., Leach and Rowan, 1986; Garven et al., 1993; Garven, 1995; Leach and Sangster, 1995). This process could produce initial expulsion of relatively low-salinity connate fluids from the basin during the precipitation of megaquartz. Ouachita deformation and uplift began before Permian reflux, and highly saline brines might not have been available until during or after the Permian.
The second stage left a record of quartz overgrowths in the Simpson Group and megaquartz in Mississippian strata. This stage has not yet been recognized in the Arbuckle Group. Homogenization temperatures show a high degree of variability. In Mississippian strata, there are all-liquid inclusions and two-phase inclusions (in the same size range) with Th in the 150s°C. In the Simpson Group, there are Th values in the low 60s°C up to the 110s°C. Because of the evidence against thermal reequilibration of inclusions in Stage 1, this Th variation is interpreted as a record of temperature fluctuation during precipitation of Stage 2 quartz. Tmice data indicate salinities are much higher than those in Stage 1, yielding salinities of 18–24 wt. % NaCl eq. In the Simpson Group, there is a correlation between Th and salinity (Figure 9B inset). This is interpreted as a mixing relationship between a lower salinity, low-temperature fluid and a high-salinity, high-temperature fluid. The low temperatures are consistent with those expected from normal burial conditions (50°C or less in Mississippian strata and about 60°C in the Simpson Group; Newell, 1997) and the high temperatures indicate hydrothermal fluids.
As with Stage 1, the widespread nature of the record of stage 2 supports regional advective fluid flow out of basins to the south, using the Simpson Group and Mississippian strata as aquifers. The extreme variation in temperature indicates a stop and start of hydrothermal fluid flow, which may represent an important, and not yet understood, punctuation in the drivers for fluid flow. With hydrothermal fluid flow temporarily at a standstill, the temperature of the system could have returned to normal burial temperatures; then, slightly higher salinity hydrothermal fluids were injected into the system at a higher temperature. An event of cooling of preexisting hydrothermal fluids could have led to supersaturation with respect to quartz, encouraging precipitation of quartz in a shallow burial setting (e.g., Rossi et al., 2002).
As previously discussed, Goldstein et al. (1991), Wojcik et al. (1994), and Luczaj and Goldstein (2000) demonstrated an event of low-temperature reflux of Permian age, charging deeper areas with highly saline residual evaporite brines. Salinities of Stage 2 are higher than those of Stage 1 (expulsion of connate fluids) and consistent with those generated during that brine reflux, yet showing hydrothermal temperatures. This indicates that the timing of Stage 2 is either during or after the Permian reflux (Figure 17B, C) that charged adjacent basins with highly saline fluids. These findings are consistent with those of Shelton et al. (2009) who demonstrated a seawater derivation for the chemistry of fluid inclusions associated with ore deposits in Missouri.
The cause for hydrothermal fluid flow in Stage 2 must be consistent with generating a pulse of flow, and thus, a tectonic driver is necessary (e.g., Sibson, 1981, 1987, 2003; Sibson et al., 1988; Cox, 2010; Lupi et al., 2011). Episodic expulsion of fluids released from compaction or episodic fracturing (tectonic valving) to allow flow of fluids driven by hydraulic head in adjacent mountains remain possible. Hydrothermal fluid migration out of adjacent basins could be considered to be driven simply by reflux and its displacement of deep fluids, but this mechanism appears unlikely to be capable of generating the high hydrothermal temperatures and regionally advective flow observed (Buijs and Goldstein, 2003).
Stage 3 preserves a record of baroque dolomite in the Arbuckle and Simpson Groups, Mississippian strata, and Pennsylvanian strata. Fluid inclusion data indicate high-salinity, high-temperature fluids as well as the presence of oil. Th of baroque dolomite in the Arbuckle Group, Simpson Group, and Mississippian strata fall within the range of 93°C–160°C. This is comparable to published data from similar baroque dolomite from Pennsylvanian strata (Wojcik et al., 1994). Among the new analyses of Ordovician through Mississippian strata reported herein, the lowest Th values are in the Arbuckle Group (93°C–131°C) and the highest are in Mississippian strata (100°C–160°C). Salinities range from 16.3 to 20.4% for the Arbuckle Group and 15% to 23% for the Mississippian, closely comparable to published data from Pennsylvanian baroque dolomite and ankerite (Wojcik, et al., 1994). Similar temperature ranges and salinity data among stratigraphic units suggest late-stage cements were precipitated during the same events of high-temperature brines migrating through the entire stratigraphic interval, Cambrian–Ordovician through Pennsylvanian.
Isotopic analysis further supports that baroque dolomite for all units precipitated at approximately the same time from the same system of fluid flow, with depleted δ18O and δ13C values resulting from thermal fractionation and thermochemical sulphate reduction associated with elevated temperatures. Repeated rises and falls in fluid inclusion Th values indicate hydrothermal fluids were pulsed. Radiogenic 87Sr/86Sr values of Arbuckle Group baroque dolomite (0.70908–0.71019) and Mississippian baroque dolomite (0.70848–0.70976) have about the same range, suggesting that fluids responsible for cementation came from the same fluid source with a radiogenic Sr composition, as well as having gone through rock–water interaction with marine carbonates. These observations argue for advective flow of hydrothermal brines, out of basins to the south; those brines took advantage of the entire Cambrian–Ordovician through Pennsylvanian section as an aquifer for transmission of hydrothermal fluids.
Baroque dolomite of Stage 3 preserves a similar high salinity to that of Stage 2 quartz, evidence that hydrothermal fluid migration took place after Permian reflux. Fluid flow is advective and affected a large region and stratigraphic interval. A timing late in the Permian (or early in the Triassic) is reasonable for baroque dolomite based on its paragenetic position after Stage 1 (before Permian reflux), after Stage 2 (after or during Permian reflux), and common Permian or Permo–Triassic timing for many of the MVTs in the region (Wisniowiecki et al., 1983; Pan et al., 1990; Brannon et al., 1996b; Symons et al., 1998; Coveney et al., 2000; Leach et al., 2001).
For the most part, Ouachita deformation ended by Early Permian time (Flawn, 1961; Frezon and Dixon, 1975; Kluth and Coney, 1981; Marshak et al., 2003), but the hydrologic system of head in uplifts adjacent to basins still existed late into the Permian, based on the record of subsidence and siliciclastic sedimentation in the Anadarko basin (Johnson et al., 1988; Johnson, 1989). Thus, gravity driven hydrologic head in mountains to the south was the most likely driver for hydrothermal fluid flow (Figure 17A, C). Additionally, the occurrence of slightly lower salinity fluid inclusions in calcite in the Mississippian, and a progression to lower salinity fluid inclusions in hydrothermal minerals from the Pennsylvanian section (Figure 14; Wojcik et al., 1992, 1994, 1997), may represent a freshwater influence as fluid flow out of the basin progressed.
Thermal Inversion in Stages 1–3
Fluid inclusion microthermometric data and δ18O data produced from baroque dolomite indicate that there was a temperature inversion in the Cambrian–Ordovician through Mississippian hydrothermal aquifer (Figures 15, 16B). Homogenization temperatures measured in Arbuckle Group baroque dolomite (Stage 3) range from 93.0°C to 131.0°C, whereas Mississippian baroque dolomite range from 100°C to 160°C, evidence for higher temperatures in the Mississippian during the precipitation of baroque dolomite (Figure 15). Similarly, Th values for both Stage 1 and Stage 2 quartz show higher maximum temperatures in Mississippian strata in comparison to Ordovician strata (Figure 15).
Additionally, Arbuckle Group baroque dolomite produces δ18O data ranging from −10.2‰ to −6.6‰ and Mississippian baroque dolomite produces a range from −10.3‰ to −8.4‰, with a progressive decrease in values upward toward the top of the Mississippian (Figure 16B). Higher Th and more negative δ18O in the Mississippian indicate high temperatures were more consistently high in Mississippian strata, and that temperatures were lower in the Ordovician (Figure 16B; King and Goldstein, 2016). This is consistent with high conodont alteration indices and fluid inclusion Th data near the top of the Mississippian, from rocks of the Tri-State MVT district (Goebel, 1996). The Mississippian through Ordovician baroque dolomite preserves δ18O values indicating a downward decrease in paleotemperature of approximately 21°C, yielding a paleogeothermal gradient of approximately −49°C/km. It is hypothesized that the temperature trend represents upward migration of the hottest fluids and preferential flow of those fluids higher in the Ordovician–Mississippian aquifer. A plausible physical explanation for this is that higher fluid temperatures lead to lower density, and lower density fluids float upward toward the top of the aquifer. An alternative hypothesis based on density differences from fluid salinity can be disproven, because salinity data are not lower higher in the section.
Thus, during advective fluid migration, the hottest fluids, derived from the deepest depths in the basins to the south, are likely to have encountered permeability pathways that allowed them to float to the top of the aquifer (Figure 18). These pathways could be hypothesized as stratigraphic windows through shales, but evidence points strongly to faulting and fracturing: including the close paragenetic relationship between fracturing and mineralization, and the evidence for repeated pulses of hydrothermal fluid flow explainable by tectonic valving. Faulting and fracturing undoubtedly played a role in enhancing both vertical as well as lateral fluid migration pathways. Deformation synchronous with fluid flow likely led to the hydrothermal pulses interpreted from the fluid inclusion data. In the region, numerous complex fault systems have been attributed to deformation with the same timing as the Ouachita–Marathon orogeny. These include strike-slip faults and relay ramps (Hedke and Watney, 2016) and have many of the structural attributes common in modern-day geothermal systems (Faulds et al., 2011). Given vertical pathways, the lowest density (warmest) fluids would have preferentially migrated vertically until they cooled, or encountered a baffle to fluid flow. Also, some convection in the aquifer may have been possible. In the modern aquifer, the shale-rich Pennsylvanian section serves as such a baffle (Jorgensen et al., 1993). Its effect on ancient thermal structure is immediately apparent from the δ18O data from baroque dolomite (Figure 16B).
In the basal part of the Pennsylvanian, δ18O values are 2.1‰ more enriched than the equivalent baroque dolomite in underlying Mississippian strata, representing a temperature decrease above the Mississippian–Pennsylvanian unconformity of approximately 20°C. Moving up stratigraphically through Pennsylvanian strata, δ18O values are progressively more enriched, evidence for decreasing paleotemperatures upward. In the lower 800 ft (244 m), there was a geothermal gradient of approximately 8°C/km; above it was an upper interval with a gradient of approximately 60°C/km (Figure 16B). These abnormal paleogeothermal gradients are nearly identical to those determined from fluid inclusions in the same cements (see figure 17 in Wojcik et al., 1994).
Although this thermal structure shows that the Pennsylvanian section was not openly connected to the underlying aquifer, the petrographic, fluid inclusion and isotopic record in the Pennsylvanian section supports that it experienced the same late fluid history as the Ordovician–Mississippian section. Thus, the Pennsylvanian section can best be characterized as a leaky confining unit that allowed episodic upward and focused escape of fluids from the Mississippian into the Pennsylvanian. These observations are consistent with results on thermal maturation in the area (Wojcik et al., 1991), which show a suppressed geothermal gradient immediately above the Mississippian–Pennsylvanian boundary and isolated areas of higher thermal maturity. The behavior of the Pennsylvanian, acting as a regional confining unit, is consistent with observations on the extant system (Jorgensen et al., 1993), where the shale-rich Pennsylvanian strata and evaporite-rich Permian strata restrict fluid flow.
The record of Stage 4 is best preserved in scattered distributions of calcite cement in the Arbuckle Group, Mississippian, and Pennsylvanian strata. The sparse distribution and poorly defined cathodoluminescence banding makes it difficult to determine if all calcites analyzed are of the same age and origin. For the sake of discussion, the late calcites at various depths and localities will be treated as being related to one another, rather than being of different ages and origins. This assumption is reasonable, given the calcite’s similar late position in the paragenesis, after an event of fracturing (Figures 3, 4). The ranges of Th data in primary fluid inclusions are from 89.9°C to 131.0°C for primary inclusions and 70.0°C to 89.9°C for secondary inclusions in the Arbuckle Group. In the Mississippian, there are consistent primary FIAs with Th ranging from 93.0°C to 103.5°C, and secondary FIAs with Th values ranging from 67.0°C to 83.0°C. Overall, Th values are higher than would be predicted from maximum burial heating (Newell, 1997), and thus, primary fluid inclusions in calcite appear related to hydrothermal fluid flow, and secondary inclusions may record the transition between hydrothermal fluid flow and more normal burial conditions. Primary FIAs produced salinities ranging from 14.5 to 15.5 wt. % NaCl eq. and secondary FIAs produced values ranging from 19.6 to 19.9 wt. % NaCl eq. These data show that although salinities remain relatively high, there were events of slightly lower salinity fluids, which contrast the consistent high salinities associated with precipitation of the baroque dolomite.
The calculation of the oxygen isotopic composition of fluid responsible for calcite precipitation is produced by using the fluid inclusion homogenization temperatures. It yields an enriched range of δ18Owater values (+5.0‰ to +1.8‰). The values differ from those calculated for baroque dolomite, indicating a different fluid source.
Arbuckle Group calcite displays more depleted δ13C values (−18.5‰ to −3.7‰; Figure 16C) and more radiogenic 87Sr/86Sr values (0.71257–0.71716) than Mississippian calcite, which has values (−0.4‰ to 1.7‰ and 0.70842–0.70930; Figure 16D) that fall near the same range as baroque dolomite for both units. Different wells show different 87Sr/86Sr values, and in one well, there is progressive decrease in 87Sr/86Sr upward (Figure 16D). These observations suggest that the hydrothermal system responsible for calcite was not a regional system associated with advective fluid flow. Rather, hydrothermal fluid flow was more localized and associated with vertical fluid flow.
Pumping of fluids associated with strike-slip faulting provides a reasonable explanation for the observations from calcite (e.g. Davies and Smith, 2006). This would allow for vertical migration of high-temperature highly radiogenic fluids sourced from underlying sandstone or basement. The less radiogenic values observed in Mississippian calcite could be explained by progressive rock–water interaction as fluids migrated vertically. Different geochemical data in different wells provides evidence that flow was highly localized. The similarities in secondary fluid inclusion data in Arbuckle Group and Mississippian calcite may represent continued fluid connectivity between the two units via the same fracture-controlled system.
The late position of calcite in the paragenesis implies that it could have formed well after drivers associated with the Ouachita Orogeny. Coveney et al. (2000) summarized radiometric dates of various calcite cements in the Tri-State area that included 251 ± 11 Ma, 137 ± 3 Ma, 66 ± 2 Ma, and 39 ± 2 Ma (Brannon et al., 1996a, b; Blasch and Coveney, 1988). The 251 ± 11 Ma and 137 ± 3 Ma dates fall in between the timing of the Ouachita and Laramide orogenies, and thus, these dates are not well tied to known tectonic drivers. The two youngest dates (66 ± 2 Ma and 39 ± 2 Ma), however, coincide with the estimated timing of the Laramide orogeny (Tweto, 1980; Coveney et al., 2000). New U/Pb dates on cave-filling thermal calcite in northern Arkansas also yield Laramide ages of 52 ± 2 Ma (Tennyson et al. 2017). Laramide structural reactivation is well known in the midcontinent (Bunker et al., 1988; Berendsen and Blair, 1995; Luza, 1995; Tickoff and Maxson, 2001; Ohlmacher and Berendsen, 2005). If the calcites observed in Mississippian strata and the Arbuckle Group are associated with the younger dates, then faulting during the Laramide orogeny could be a plausible mechanism for fluid-flow (Figure 17D; Marshak et al., 2003; Davies and Smith, 2006). Further research on the calcite cement is required to obtain a more complete understanding of its timing and origin.
After calcite precipitation, moderate-temperature, high-salinity fluids in secondary fluid inclusions in calcite in Mississippian strata and the Arbuckle Group provide evidence of a continuation of a hydrothermal system. Similar salinities to those in primary inclusions suggest the same fracture-controlled system may have been playing a role in connecting the Mississippian and Arbuckle Group. Temperatures are closer to the range estimated by Newell’s (1997) burial history model, suggesting temperatures were transitioning toward normal burial conditions when inclusions were entrapped.
Stage 5 is the extant system in the subsurface, which differs greatly from those that existed during Stages 1–4. Salinities in wells from south–central Kansas (Wellington 1-32) show that since the last phase of entrapment of fluid inclusions in Stage 4, fluid salinities in the Arbuckle Group dropped significantly, whereas those in the Mississippian remain high (Figure 15). Temperatures in both units have dropped to levels that are approximately consistent with the currently shallow burial depth (Figure 15).
Regional studies of fluid chemistry and thermal structure observed in modern wells in the region show that some wells preserve high-salinity fluids with chemistries consistent with Permian residual evaporite brines that have been modified by rock–water interaction (Musgrove and Banner, 1993; KGS, 2010, 2012). There are also regional deep aquifer systems flowing both west (Ozark Plateau aquifer system) and east (Western Interior Plains aquifer system), and possibly meeting in an area of mixing near the study area (Figure 17D; Jorgensen et al., 1993; Musgrove and Banner, 1993; Macfarlane, 2000; Carr et al., 2005). The Arbuckle Group of the Ozark Plateau aquifer system hosts westward fluid flow, and in the study area, it is at a low pressure that is consistent with being open to a recharge zone in the Ozarks (Watney and Rush, 2015).
This Ozark Plateau aquifer system may have flushed the earlier high-salinity fluids that were in the Arbuckle Group and partially replaced them with the fresher waters in south–central Kansas. In contrast, the Mississippian strata still host high-salinity fluids that are similar or the same as those injected during prior diagenetic stages (Scheffer, 2012; Watney and Rush, 2015). In current wells, microbial populations, oxygen and deuterium isotopes, Br−/Cl− and support the idea that there is little vertical connection across internal confining units, including a low-permeability zone in the Arbuckle Group and the Simpson Group–Chattanooga Shale confining units (Scheffer, 2012). Whereas parts of the Arbuckle Group strata have been flushed as part of the Ozark Plateau aquifer system, the Mississippian strata are clearly separated from that aquifer. Thus, the vertical hydrologic connections that existed in the geologic past are no longer as important as they once were. This is explained by the ephemeral nature of repeated fracturing and repeated healing with diagenetic minerals; it is consistent with the paragenesis observed in this study. Many of the vertical connections created by fracturing have now largely been sealed, leading to more compartmentalization.
There is also regional evidence for fluid flow to the east associated with the Western Interior Plains aquifer system. There is likely a zone of mixing where fluids from these two aquifer systems interact (Jorgensen et al., 1993; Musgrove and Banner, 1993; Carr et al., 2005). Furthermore, it is hypothesized that other forms of fluid flow may have an impact in this region. Fluid flow from the south remains possible and fluid flow out of the midcontinent rift remains viable. Moreover, fault pumping, similar to that interpreted during Stage 4, is expected to have a minor impact. Recent studies of induced seismicity in the region indicate that a hydrologic connection to a basement aquifer exists in the region (Bigdoli et al., 2015a, b; Holubnyak et al., 2016). Each time there is movement on a fault that involves basement, fluid pumping associated with that event is possible. This could produce a small-scale analog to the large-scale events that occurred during Stage 4 hydrothermal fluid flow. Intensified deformation during the Laramide orogeny (Tickoff and Maxson, 2001) is interpreted to be responsible for large-scale fault pumping to cause Stage 4 hydrothermal fluid flow directly out of basement. As a hydrologic connection to basement can still be demonstrated in association with seismicity induced from present-day fluid disposal in the region (Bidgoli et al., 2015a, b; Holubnyak et al., 2016), it is expected that less active, but still viable faulting events could lead to localized injection of warm basement fluids in the subsurface. The processes associated with Stage 4 may still be sporadically active.
Fluid inclusion, petrographic and geochemical analysis of late-stage cements in the Cambrian–Ordovician Arbuckle Group, Middle Ordovician Simpson Group, Mississippian strata, and Pennsylvanian strata support a common history of late-stage fluid migration. After burial, fracturing and dissolution events appear to have affected most units, and there was a late-stage paragenesis consisting of megaquartz, baroque dolomite, MVT ore minerals, hydrocarbon migration, and calcite.
Similarities in mineral paragenesis, fluid inclusion microthermometric data, and geochemical data support a genetic link between fluids precipitating megaquartz and baroque dolomite in the study area, supporting the concept that fluids were able to migrate across modern-day baffles or seals via connected pore systems. All late-stage cements produce homogenization temperatures higher than can be explained by normal burial conditions or geothermal gradients, and the geochemical data of the cements support an external source of fluids responsible for precipitation. These data indicate that there was a complex history of late fluid flow with differing drivers, timing, and chemistry, each leading to different impacts on reservoir rocks in the region.
Stage 1 is represented by megaquartz cement, low-salinity connate fluids, and gas. Temperatures are high, and rise and fall through time, during an overall temperature increase. Paleotemperatures are highest at the top of the Mississippian section. These data indicate that tectonic activity associated with the Ouachita orogeny drove pulses of regionally advective hydrothermal fluid migration out of basins to the south. Timing is likely Pennsylvanian and may have extended into early Permian.
Between Stages 1 and 2, salinities increased greatly. This is attributed to a time of regional reflux of residual evaporite brines. During the Permian, brines were generated at the surface and penetrated deeply into the subsurface because of density inversion. This brought low-temperature brines into deeper intervals, and may have charged basins with highly saline fluids.
Stage 2 is represented by megaquartz and quartz overgrowths that precipitated from concentrated brines at both low and high temperature. The highest temperatures are preserved near the top of Mississippian strata. Data indicate mixing between high-temperature hydrothermal brines and low-temperature brines. This event could have been caused by a single tectonically driven pulse of hydrothermal fluid migration. The lower ambient temperature could have been achieved before or after the hydrothermal pulse. Alternatively, the temperature fluctuation could have been caused by interaction between hydrothermal flow and low-temperature Permian brine reflux.
Stage 3 is represented by baroque dolomite, concentrated brine, and oil. Data show pulses of high-temperature migration of hydrothermal brine that affected the Arbuckle Group through Pennsylvanian strata, evidence for regionally advective fluid flow out of basins to the south, likely driven by Ouachita tectonism or uplifts. Highest temperatures are at the top of the Mississippian, and the Ordovician through Mississippian section had an inverted geothermal gradient of −49°C/km. There was a decrease in temperature at the base of the Pennsylvanian. The Pennsylvanian section had a lower interval with suppressed geothermal gradient of 8°C/km and an upper interval with elevated geothermal gradient of 60°C/km. These data are consistent with regionally advective hydrothermal fluid flow, later in the Permian, where Mississippian through Cambrian–Ordovician strata acted as a single vertically connected aquifer, and Pennsylvanian strata acted as a leaky confining unit.
Taken together, Stages 1–3 provide evidence for a hydrothermal system in which Cambrian–Ordovician through Mississippian strata acted as a vertically connected, regionally advective aquifer. Ouachita tectonism was the likely cause of vertical connectivity, and drivers of hydrothermal fluid flow. Highest temperature hydrothermal fluids had the most impact in Mississippian strata. This is likely the result of the structure of the aquifer, where the highest temperature brines, being the lowest in density, floated to the top of the aquifer until they reached the Pennsylvanian leaky confining unit. This structure of hydrothermal fluid flow intensified thermal and chemical alteration toward the top of the hydrothermal aquifer. It indicates that Mississippian reservoirs have been impacted greatly by tectonically driven late hydrothermal fluid flow, which could have concentrated alteration along the top of the Mississippian and where fracture intensity was highest. The hydrothermal fluid flow could have led to thermal maturation of source rocks, which could have contributed charge to reservoirs in the area. This type of thermal structure may be predictable in other areas of hydrothermal fluid flow and may offer an explanation for enhanced reservoir quality immediately below baffles to fluid flow.
Stage 4 is represented by scattered distribution of calcite from high-salinity brines. Temperatures are high but lower than those of Stage 3. Geochemical data indicate different amounts of rock–water interaction in different wells. In one well, that preserves calcite with highly radiogenic Sr isotopes, data show progressive rock–water interaction vertically. These data indicate fault- and fracture-driven fluid flow that tapped radiogenic fluids, likely from the basement. The Stage 4 hydrothermal system stands in stark contrast to those of Stages 1–3. Hydrothermal fluid flow appears highly localized and to involve fault pumping from basement, rather than regionally advective fluid flow out of basins to the south. It is proposed that the deformation that led to this fluid flow could have been associated with Late Cretaceous–Paleogene reactivation of fault systems in the region during the Laramide orogeny. This reactivation could have led to localized breaching of deeper reservoirs and rejuvenation of vertical migration of hydrocarbons and other as yet unknown effects on reservoir quality.
Modern reservoir conditions suggest that stratigraphic units in this study have returned to more normal burial temperatures, with low-temperature fluids recharging the study area from the Front Range to the west and from the Ozark Plateau to the east. Minor hydrothermal fluid flow is still possible in the region given the regional aquifers and sporadic reactivation of basement faults.
Midcontinent reservoirs have been subject to a complex late history of fluid flow. This well-studied region serves as an excellent example of the impact of the evolution of hydrothermal flow on reservoir systems. Tectonically driven, regionally advective hydrothermal fluid flow is dominant (Stages 1–3) and can result in an inverted thermal structure with predictable impact on reservoir quality and thermal maturity. Hydrothermal fluid flow from fault pumping (Stage 4) also had an impact, but that impact is as yet unknown. Finally, the modern system (Stage 5), although relatively tectonically quiescent, cannot be considered fully stagnant. Either regionally advective subsurface fluid flow or fault pumping may still have a minor impact on current subsurface conditions.
This project was supported by the U.S. Department of Energy (DOE) National Energy Technology Laboratory (NETL) under Grant No. DEFE0000002056. The project was managed and administered by the Kansas Geological Survey/KUCR, W. L. Watney, PI, and funded by DOE/NETL and cost-sharing partners. Additional support was provided by the sponsors of KICC. This paper benefitted from reviews by Dennis Prezbindowski, Martin Appold, and Jay Gregg. Their contributions are greatly appreciated.