The story of the Mississippian lime play began over a century ago and captures the evolution of oil and gas exploration in Kansas and Oklahoma, from early drilling of simple traps to the current development of high fluid volume plays and their associated intensive surface and subsurface plumbing infrastructure. The chapters that follow in this volume give detailed accounts of particular reservoirs, fields, processes, and problems in the Mississippian of the midcontinent of the United States. The goal of this overview is to provide a regional and historical framework, especially in Kansas and Oklahoma, within which to consider the observations and interpretations of the authors.

Straddling the Kansas–Oklahoma border, the play fairway stretches in the subsurface from Grady County, Oklahoma, at its southern tip, northeastward into Osage County, Oklahoma, out to the northwestern corner of Kansas, and also includes scattered drilling throughout western Kansas and northwestern Oklahoma (Figure 1). The petroleum system can be most simply described as Mississippian reservoirs charged with hydrocarbons largely migrated from mature Woodford source basins (Johnson and Cardott, 1992) and trapped by the overlying blanket of Pennsylvanian shales. The subsurface reservoirs that have been drilled and exploited represent the stratigraphic range of the Mississippian from Kinderhookian to Chesterian and produce from depths as shallow as 1500 feet (500 m) to as deep as 12,000 feet (3600 m). Their stratigraphic equivalents are exposed in spectacular outcrops in Arkansas, Missouri, and Oklahoma (Mazzullo et al., 2013). As discussed in detail by the authors of this volume, subsurface reservoir types range from transported carbonate ramp and platform deposits to dolomitized clinoforms to primary, diagenetic, and hydrothermal cherts and are ubiquitously fractured to some degree. Reservoir fabric depends not only on depositional and early diagenetic environment but also on the history of tectonism and fluid movement through the basin. The resulting hydrocarbon traps are complex and include a range of structural and stratigraphic configurations overprinted with highly variable distribution of reservoir fluids. Most notable is the proliferation of high water cut throughout the play fairway, which drives not only the economics but also the necessity for subsurface fluid lifting and the development of vast surface infrastructure to power lifting and disposal of produced fluids.

Oil from a Mississippian reservoir was first produced in 1904 in Osage County, Oklahoma (IHS Global, 2016). Subsequent drilling of the pre-Pennsylvanian section in the El Dorado field in 1914 in Butler County, Kansas, extended the growing trend across state lines (Jewett, 1954). Early exploration and development of surface structures was followed by the use of 2-D seismic in the 1950s to define subsurface traps along the Nemaha Ridge in Oklahoma (Clinton, 1959) and the Spivey trend in Kansas (Harper and Kingman counties). In the 1960s, the play moved to the giant stratigraphic trap in central Oklahoma that became known as the Sooner trend. The application of hydraulic fracturing to stimulate lower permeability reservoirs in the 1970s enhanced production from this portion of the play (Harris, 1975). After commodity prices fell in the 1980s, production from all parts of the play declined.

Rising oil prices coupled with the adoption of horizontal drilling technology opened a new phase of development of Mississippian reservoirs in the early part of the 21st century. The impact of this horizontal “Miss Lime” play on the northern midcontinent has been similar to that of its cousins the Barnett Shale in Texas and the Bakken Formation in North Dakota, as it resulted in the single largest addition of oil production to both Oklahoma and Kansas since the onset of oil and gas exploration early in the 20th century (IHS Global,^*2016). The prevailing philosophy during this development phase was that the majority of the reservoir, regardless of its rock fabric or position in the giant system of traps, could be exploited using the same model applied to unconventionally trapped gas in shales—that is, pattern drilling of horizontal wells completed using some form of multistage, high-pressure, high-volume hydraulic fracture stimulation. The early stages of horizontal exploration relied on reservoir evaluation and 3-D seismic to optimize the choice of drilling location and completion practice with promising results. But the capital rush came too hard on the heels of good exploration and quickly outran the ability of operators to analyze results prior to additional investment. Development moved faster than sound geologic and engineering evaluation in many areas, and capital was expended with negative return.

Production from the horizontal drilling boom reversed the 20-year production decline from Mississippian reservoirs in Kansas and Oklahoma. However, the resulting large number of horizontal wells with high water cut—greater than 90% throughout large regions—necessitated installation of significant subsurface and surface infrastructure, including electricity and produced fluid disposal facilities. Associated high operating expenses put a strain on commerciality, and as commodity prices fell, marginal portions of the play were abandoned. In addition, the prevalence of high-volume disposal wells in the play has been increasingly linked to induced seismicity (Boak et al., 2016). Mississippian production in Kansas and Oklahoma peaked in 2015 at almost 500,000 barrels of oil a day and just under four billion cubic feet of gas per day (IHS Global, 2016).

Several excellent overview papers address drilling history in Kansas and Oklahoma during the first half of the 20th century (Jewett, 1954; Clinton, 1957, 1959; Beebe, 1959). Later years are summarized for Kansas by Oros (1979) and for Oklahoma by Boyd (2002). To illustrate how the development of this play tucks into historical boom and bust cycles, we have analyzed well spot and producing formation data through time (Figure 1). These drilling density maps provide a framework for understanding exploration trends.

The earliest exploration for oil and gas in both Kansas and Oklahoma dates back to the 1850s (Jewett, 1954; Boyd, 2002), and both states had established successful production as early as 1890. Mississippian reservoirs were slow to come to the party, as prolific production from shallow Pennsylvanian “shoestring” sands, like the Bartlesville, drove the first boom. For Kansas in particular, the first two decades of the 20th century saw little exploration and production from anything deeper than the Pennsylvanian, although Clinton (1959) refers to chat production in southeastern Kansas around the turn of the century from the Virgil field. In fact, all-time Oklahoma production peaked in 1927 at 762,000 barrels of oil per day (Boyd, 2002) with very little contribution from Mississippian rocks.

Production from the Mississippian was initially established in and around Osage County in Oklahoma, where the first well of record to produce from Mississippian rocks was drilled by Roxana Petroleum in what is now the Domes–Pond Creek field in 1904. By 1920, there were just under 300 wells producing from the Mississippian in Oklahoma and Kansas (Figure 1A). Over the next three decades, drilling of Mississippian reservoirs continued to add production from the northeast-southwest trending structural traps of the Osage and central Kansas, along the shallow shelf of the Anadarko Basin, and in western Kansas. Over 7500 wells producing from the Mississippian were drilled in Kansas and Oklahoma from 1921 through 1960 (Figure 1B).

The postwar years in the 1950s saw the addition of rudimentary 2-D seismic to the tested methodology of drilling surface structures, as well as a concerted effort to drill, complete and produce lower permeability Mississippian reservoirs with the use of “sand-frac” technology (Clinton, 1957, 1959). Early wells indicated that production from Osage County reservoirs in northeastern Oklahoma could be double or triple what prior drilling had indicated, and a bidding war for acreage ensued.

Results were highly variable, and geologists of the day expressed frustration that will sound woefully familiar to today’s operators. Clinton (1959) refers to “a special problem we have been dealing with… is… the variation in fluid levels within what we identify as a single reservoir… we can have the following fluid levels in a particular well—Gas, Oil, Water; then just Oil; then Gas, Water; then just Gas; then Oil, Water, and then just Water” (p. 163). Both Clinton papers lament that wells in zones with high apparent oil saturation refuse to flow, whereas others that penetrate zones indicating zero saturation are prolific producers. In a nice play on the concept of variable oil saturations due to heterogeneity of rock fabric, Clinton (1959) calls this “a differential entrapment problem” (p. 163). Twenty years later, Oros (1979), in her update to Jewett’s overview paper, expands on the difficulty in exploring for oil in Mississippian reservoirs and the lack of significant production to date in Kansas.

Mid-century natural gas drilling was successful in south–central Kansas around the southern edge of the Central Kansas uplift, especially near the Pratt anticline in the Spivey field (Figure 1b). During the 1950s, operators in western Oklahoma and southwestern Kansas drilled their first wells producing oil and gas from the Mississippian in the giant Hugoton field (Bennett, 1960).

New drilling and development of Mississippian reservoirs in the Sooner trend was the highlight of oil and gas activity in Oklahoma in the 1960s (Figure 1C). In a prelude to changes in exploration dynamics that would blossom after the turn of the 21st century, oil was first discovered by exploration drilling decades earlier, but commercial production was not feasible until the application of new technology, hydraulic fracturing, in the Dover–Hennessey trend. As Harris (1975) describes in his comprehensive overview, operators also had to contend with the early flush production followed by rapid decline that is typical of fractured and stimulated tight carbonates. With the oil boom in full swing in the 1970s, operators in north–central Oklahoma and Kansas were drilling at a frantic pace—over 17,500 wells spud in the Mississippian from 1971 through 1986 would produce hydrocarbons. That number decreased drastically after the price dropped, and fewer than 4000 wells were drilled that produced from Mississippian targets over the next 13 years. The first decade of the century saw a return to gas exploration and production in other formations as the price per unit rose by almost an order of magnitude. From the beginning of 2001 until March of 2009, only about 3200 Mississippian producers were drilled, including 60 or so early horizontal wells. The past 7 years have seen a re-institution of the oil boom, this time via the horizontal drill-bit and aided by the addition of 3-D seismic as an exploration tool. Of the over 7400 Mississippian producing wells drilled since March 2009, over 4000 of those have been horizontal wells, located for the most part in north–central Oklahoma and south–central Kansas (Figure 1D). Concurrently, Kansas has continued active development of older fields in the six counties in south–central Kansas bordering Oklahoma (Newell et al., 2014; Newell and Brady, 2015). In this area, over half of the wells drilled have been vertical rather than horizontal.

No historical perspective on activity in the Mississippian would be complete without remarks on the potential for helium production in association with natural gas. Rogers (1921) presented the earliest compilation of helium content as a percentage of produced natural gas in his comprehensive USGS Professional Paper 121. Although his maps (Figure 1 and Plates I and II) predated the discovery and development of helium in the Hugoton field in western Kansas, which later formed the foundation for the National Helium Reserve, they indicate a bulls-eye in high helium content over Osage County in Oklahoma and northward east of the Nemaha Ridge in southeastern Kansas. The observation of helium in shallow gas fields coincides with the area of recent rejuvenation of deeper drilling, where helium contents as high as half a percent have been reported from the Mississippian and below in the Pearsonia area (S. Matson, personal communication, 2014) and from Pennsylvanian sands comingled with Mississippian production near the Foraker field (M. Righetti, personal communication, 2014).

The exploration for Mississippian targets took a fresh turn at the peak of the price boom in 2014. Operators began to investigate the potential of deeper water facies, including true black shales, in the uppermost Mississippian, probably Meramecian and Chesterian in age, along the eastern rim of the Anadarko Basin. The reduction in commodity price to a third of its peak has not curtailed activity in the deeper, costly arena of these emerging plays, as results to date indicate high-value potential.

The rich and prolific Devonian Woodford Shale lies at the heart of the upper Paleozoic petroleum system in the northern midcontinent (Johnson and Cardott, 1992). Ordovician Simpson shales as well as deeper water facies of the Mississippian have also been demonstrated to source oils in both Kansas and Oklahoma (Burruss and Hatch, 1989). Mississippian reservoirs were mainly deposited in the ramp and shelf environment of the Oklahoma basin, whose axis ran west-northwest to east-southeast across southern Oklahoma. Shallow water deposits accumulated adjacent to the central Kansas high, which separated the deep waters of the Ouachita trough from intracratonic basins to the north. Pennsylvanian tectonism associated with the closing of the ancestral Atlantic and formation of the Ouachita–Marathon fold belt created an angular unconformity along the Mississippian–Pennsylvanian contact and set the stage for later stratigraphic trapping. Mountain building and adjacent foreland basin formation and fill followed. This resulted in the timely deposition of a blanket of lower Pennsylvanian sealing shales and concurrent burial and maturation of the chief source rocks in both the Anadarko and Arkoma basins. The hydrocarbon kitchens passed through peak oil generation before the end of the Permian and began to flood the shelf on both sides of the Nemaha Ridge with hydrocarbons (Johnson and Cardott, 1992; Higley, 2013). At the apex of Atokan tectonism, suturing of the continent was underway. The result in Oklahoma was deformation of the ancestral rift basin material to form the Arbuckle uplift and closing of the Ouachita re-entrant by the mid-Permian (Keller, 2012). Throughout Oklahoma and eastern Kansas, the northwest-directed stress resulted in general reactivation of basement rift structures (Marshak et al., 2000), wrench faulting, and complex uplifts developed along restraining bends (McBee, 2003). An overlay of mapped structures on Bouguer gravity data (USGS Mineral Resources On-Line Spatial Data, 2014) illustrates the basin framework inherited from old geometry of the midcontinent rift system (Figure 2). Also shown are the approximate locations of the hydrocarbon kitchens that reached peak oil maturity during the Pennsylvanian (Comer, 1992; Higley, 2013). The predominance of oil east of the Nemaha ridge, coupled with abundant gas on the Anadarko side of the platform, suggests that fault conduits from both hydrocarbon kitchens were active migration pathways immediately following the oil generative phase, whereas gas accumulations are more prevalent in the traps overlying and up-dip from present-day gas kitchens.

Trends of known oil and gas accumulations (Figure 2), coupled with surface and seismic observations of fault patterns, suggest a long-term relationship between basement configuration, structural grain, and the migration and accumulation of oil and gas (Watney et al., 2001, 2008; Wilson et al., 2014). In concert with the high-frequency sea-level changes discussed in the following section, syndepositional structures such as horsts, grabens, and monoclinal flexure, varied local bathymetry, and in turn, varied sediment accommodation along the shallow shelf and ramp (Watney et al., 2008). This strongly affected depositional trajectories and prevented the evolution of a simple play concept (Watney et al., 2001, 2008; Watney, 2014, 2015). Although it is not the focus of this paper to unravel the detailed impact of structure on each oil field, it is our intent to present several observed patterns that suggest avenues for further interpretation and investigation. As an overview, Figure 2 shows the spatial distribution of vertical oil and gas wells which have produced from Mississippian reservoirs and their alignment with structural trends. We show vertical wells only, because vertical development is typically economically constrained to the most pristine trapping scenarios. That is, these wells typically produce from water-free or lower water-cut zones in trapped position high above the free water level. The relationship of traps to ancient, long-lived structural features, or tectonically driven stratigraphic unconformities, is an old observation. Rich (1933) commented on the association of oil-filled structures with those linked to features in older strata, and many maps such as those shown in Figure 3 have documented the relationship of producing fields to deep, basement-involved features. Figure 3A was taken as a snapshot in time of Mississippian producing fields (Ebanks, 1975) and illustrates the orthogonal alignment along the same fault trends shown in Figure 2. Rogers (2001) included the map depicted in Figure 3B in her discussion of chert reservoirs in north–central Oklahoma. Note how distinctly the curvature of the trend mimics the emplacement of the Choctaw fault to the south (Figure 2). This pattern is repeated to the north in Kansas along the southeastern edge of the Central Kansas uplift.

These maps speak to a general premise that underpins the accumulation of oil in the Mississippian of the midcontinent: systematic reactivation of basement fabric throughout time to the present day is the theme. Reactivated rift structures, including Mississippian and younger strike-slip deformation, cover an extended area of the Kansas and Oklahoma shelf and have influenced sedimentation, fluid flow, and hydrocarbon migration (Watney et al., 1997, 2008; McBee, 2003; Hedke and Watney, 2016). Although governing faults and fractures may be subtly expressed in seismic, they can be inferred from potential field data and drilling patterns of producing wells. Even the shelf to basin framework was inherited from differential subsidence controlled by deep-seated faulting and fracturing of brittle rocks. Concurrent and post-Mississippian structural deformation affected deposition of the reservoir and nonreservoir facies, source maturation, and regional petroleum migration. Angular unconformities reflect changing uplift patterns with time. Subsequently, basement-involved fractures inherited from the ancient rift geometry became wrench fault systems as a result of directed stress on the craton during the late Paleozoic, and today they control the fracture permeability that feeds flush production. The resulting structural and stratigraphic grain of the region is elemental to migration, entrapment, and production of hydrocarbons.

The term “Mississippian limestone” is an informal term that refers to much of the Mississippian-age strata across the midcontinent. Although dominated by carbonate deposits, they can be more accurately described as being composed of a mixture of carbonate including both dolomite and limestone, and siliciclastic rocks and cherts. Stratigraphy of the Mississippian-age deposits is not well-defined and is further complicated by changes in nomenclature across short distances and between outcrop and subsurface units. Generalized lithostratigraphic columns were designed for each state and are used to characterize local outcrop and subsurface strata (Figure 4). Modifications to the Mississippian nomenclature have been recently suggested by Mazzullo et al. (2013) on the premise that the revised lithostratigraphic nomenclature is more useful in correlating strata observed in outcrop to that in the subsurface. A brief description of each state’s generalized lithostratigraphic column is given below as these are most commonly used in literature.

The Mississippian section in northeastern Oklahoma is characterized by Kinderhookian-age through Meramecian-age strata (Figure 4). Although not all sources agree, some state that at the base of the Kinderhookian-age strata there is a thin, shaley, siliciclastic-dominated bed present that is correlative to the Bachelor Formation (Mazzullo et al., 2013). The Bachelor Formation is often correlated to the Sycamore sandstone (late Devonian) due to the presence of reworked Devonian-age clasts, but conodont biostratigraphy indicates that deposition occurred entirely during the Mississippian (Mehl, 1961; Manger and Shanks, 1976). Kinderhookian-age strata are composed of, in ascending order, the Compton and Northview formations. The base of the overlying Osagean strata is represented by the Pierson Formation. The Compton, Northview, and Pierson formations comprise what is collectively referred to as the St. Joe group (Mazzullo et al., 2013). The remainder of the Osagean strata is composed of the Reeds Spring Formation, Keokuk Formation, and locally, Short Creek Oolite Member, which are collectively described as the Boone Group (Mazzullo et al., 2013). Meramecian-age strata in northeast Oklahoma are composed of the Warsaw Formation (Mazzullo et al., 2013). New research suggests that the Mayes Group is entirely Chesterian in age (Godwin and Puckette, 2015; Godwin et al., 2019) varying with the Meramecian assignment in Figure 4.

The Mississippian strata in northwest Arkansas are similar to those in northeastern Oklahoma, but do not include the upper units of the Meramecian (Figure 4; Mazzullo et al., 2013). The Kinderhookian-age strata are defined by the Bachelor, Compton, and Northview. The base of the Osagean is represented by the Pierson, and the Bachelor, Compton, Northview, and Pierson are collectively referred to as the St. Joe Limestone Member or formation. The rest of the Osagean and the base of the Meramecian are referred to as the Lower and Upper Boone formations, respectively. The Boone Formation is overlain by the Meramecian-age Short Creek Oolite Member.

The Mississippian section in southwest Missouri is the most complete and consists of Kinderhookian-age through Chesterian-age strata (Figure 4; Mazzullo et al., 2013). Here, the Bachelor Formation is not considered part of the St. Joe group or formation, rather it is a separate formation that exists below the St. Joe. The St. Joe group/or formation consists of the Kinderhookian-age Compton Limestone, Northview Formation, and Baird Mountain Limestone Member, and the Osagean-age Pierson Limestone. The rest of the Osagean-age strata is made up of the Reeds Spring Limestone, Elsey Formation, Burlington–Keokuk Limestone, and Short Creek Oolite Member. The terms “Boone Group” and “Boone Formation” are not recognized in southwest Missouri. Meramecian-age strata are represented by the Warsaw Formation, and the Chesterian-age strata consist of the Hindsville and Batesville formations.

Deposition of Mississippian carbonates occurred across large portions of the United States, including Colorado, Nebraska, Kansas, Oklahoma, Arkansas, Missouri, Iowa, and Illinois, as part of a regionally extensive carbonate platform referred to as the Burlington shelf (Figure 5; Lane, 1978; Lane and De Keyser, 1980; Gutschick and Sandberg, 1983). The Burlington shelf transitioned abruptly into the deep, starved Illinois Basin to the north, but demonstrated a gradual transition across Missouri, Arkansas, and Oklahoma into the Anadarko Basin and the Ouachita trough to the south (Gutschick and Sandberg, 1983).

The overall depositional system for the Mississippian is still highly debated. Lane (1978) describes the Mississippian sediments as being deposited on a carbonate shelf referred to as the Burlington shelf (Figure 5). Although Gutschick and Sandberg (1983) also use shelf terminology to describe Mississippian carbonate deposits across the midcontinent, they state that there is no true definition of a shelf edge. Presently, the Mississippian carbonates are believed to have been deposited across Missouri, Arkansas, and Oklahoma on a ramp to distally steepened ramp environment, and demonstrating aggradational (Early Mississippian–Kinderhookian) followed by progradational (Middle to Late Mississippian–Osagean) geometries (Wilhite et al., 2011; Childress and Grammer, 2019; Mazzullo et al., 2019). This architecture is typically recognized in the subsurface in north–central Oklahoma and southern Kansas where the focus of the “Mississippian limestone” play exists (Wilhite et al., 2011; Mazzullo et al., 2019). However, based on field observations further east in the outcrop belt, Wilhite et al. (2011) have suggested the presence of a foreland bulge region extending from eastern Oklahoma across northern Arkansas and southern Missouri that existed during the Early Mississippian and likely affected (1) the extent to which Early to Middle Mississippian carbonates were deposited and (2) the architecture of Middle to Late Mississippian deposits in easternmost Oklahoma, southern Missouri, and northern Arkansas. Effects of this Early Mississippian foreland bulge region are recognized in the subsurface and can cause difficulty when correlating (Mazzullo et al., 2019).

A major shift in the way the Mississippian system in the midcontinent has been interpreted has resulted from the evaluation of the sedimentary package from a more dynamic and time transgressive standpoint, in particular by developing a sequence stratigraphic framework. Early work established the time-transgressive nature of the Mississippian shelf strata in Missouri using conodont biostratigraphy (Thompson and Fellows, 1970; Lane, 1978; Witzke et al., 1990; Handford and Manger, 1993). This prior work also demonstrated that common lithofacies did not mean the strata were equivalent, just that similar depositional environments were repeated in clinoforms. The dimension of the clinoforms, much longer than thick, has made it difficult to resolve and trace key stratigraphic surfaces to build the sequence stratigraphic framework in the subsurface. Third-order eustatic sea-level changes in the Mississippian are well-defined globally (Haq and Shutter, 2008) and have been identified from various midcontinent outcrop studies (Grammer et al., 2013; Childress and Grammer, 2015, 2019; Mazzullo et al., 2019). One of the key findings from recent work has been the recognition of a threefold hierarchy that exists in the subsurface sedimentary record. Recent work by LeBlanc (2015), Flinton (2016), and Jaeckel (2016) have suggested that this hierarchy represents cyclicity at the 2nd-, 3rd- and probable 4th-order scales. From a reservoir standpoint, regional correlation can be done on the “3rd-order” sequences by means of GR correlation (correlation ground-truthed by core analysis). These correlations clearly illustrate the progradational clinoforms in the system distinguished on the lower shelf and basin (Figure 6) as well as the potential lateral and vertical compartmentalization that occurs between sequences. In addition, within certain subregions throughout the basin, individual sequences with high-graded reservoir potential can be characterized and correlated away from the well bore.

The Mississippian carbonates deposited along the inner to outer basin margin in southern Kansas and northern Oklahoma are dominated by siliceous sponge-bearing dolomitic facies varying from bedded spiculite, to lenticular nodular flaser bedded spiculite and silty organic bearing shale (Mazzullo et al., 2009; Montalvo, 2015; Watney, 2015). Varying amounts of benthic, calcitic skeletal debris and the lack of abundance of photozoan fauna typify a heterozoan carbonate system (Franseen, 2006). Deposition along a complex subsiding ramp and interaction between sea-level and local topography led to varying accretion trajectories of carbonate progradation. This produced complex geometries and stacking patterns of lithofacies as confirmed by contrasts in lithofacies encountered in horizontal wells (Costello et al., 2013; Watney, 2014). Locally, thick pelmatozoan carbonate mud mounds and bryozoan-rich carbonate banks attest to the depth and relief of the ramp as observed in outcrop (Mazzullo et al., 2013) and the subsurface (Montalvo, 2015). Cool-water fauna of the Mississippian were recognized by Gutschick and Sandberg (1983) spanning shelf margins, including the Burlington shelf, whereas sponge-microbial mound facies have also been recognized. Similar facies occur in the Mississippian Tuscumbia Limestone along the ramp margin of the Black Warrior foreland basin in Alabama (Kopaska-Merkel et al., 2013).

The lack of significant early lithification in especially mid- and lower-ramp Mississippian deposits, as well as evidence for current and tractive sediment transport and soft sediment deformation, typify slope deposits associated with cool water depositional settings (James, 1997; Schlager, 2010). Early diagenesis that did occur is primarily focused on the inner ramp of the spicule-rich lithofacies. These areas were initially affected by syndepositional carbonate cements and silicified by spicule dissolution and silica introduced from external sources to form coalesced chert and chert nodules. Partial to pervasive dolomitization developed while the sediment was still porous and permeable. Regional scale nodular evaporite precipitation also occurred on the inner ramp and is believed indicative of a low stand in sea level that most affected the inner ramp (Montalvo, 2015; Watney, 2015). Silica nodules containing relict lath-like anhydrite or gypsum inclusions suggest an evaporite precursor precipitated in a bioclastic subtidal carbonate strata formed before the sediment was lithified (Montalvo, 2015). Analogous nodular evaporites in subtidal deposits are previously documented by Chowns and Elkins (1974) and Milliken (1979). The early evaporites suggest concentration of hypersaline brines descended through the sediment, which commonly led to the formation of pervasive, finely crystalline dolomite while the sediment was still relatively porous (Montalvo, 2015). Silica replacement includes megaquartz with fluid inclusions that indicate precipitation from seawater and evaporated seawater. The overall impact of early precipitation of evaporites and their silicification and related dolomitization notably affected reservoir properties (Watney, 2015). Later stages of silicification and dolomitization and porosity formation including hydrothermal processes are also documented (Goldstein et al., 2019). A useful modern analog for the cool-water setting is described along the southern Australian coast by James et al. (2001). Low relief thickening of spiculitic, skeletal packstones and grainstones resemble the biostromal deposits of the Australian shelf and may be an expression of the mound-building siliceous sponge-microbe biotic associations described by Brunton and Dixon (1994). The felted masses of spiculitic packstones and grainstones and lesser calcitic skeletal fragments suggest fabrics of preserved sponge communities (Watney et al., 2001; Watney, 2015). Ancient analogs that exhibit similar lithofacies, pore types, and strata geometries include the Permian Kapp Starostin formation in the Barents Sea (Ehrenberg et al., 2001) and the Devonian Thirtyone Formation (Ruppel and Barnaby, 2001).

Shalier carbonate lithofacies present in greater proportion in the middle and lower ramp are tight reservoirs due to the lack of sufficient dolomitization, primary porosity, and fabric selective dissolution. Reservoirs composed of these tight rocks require fracturing, karstification, brecciation, or hydrothermal alteration to create pores of a size and connectivity to harbor sufficient hydrocarbon in conventional traps that depend on capillary forces and hydrocarbon column to charge the reservoir.

Petrophysical analysis is imperiled by the type and scale of pore types, ranging from micropores to karst vugs to open fractures, which exist in these reservoirs. There is no substitute for running comprehensive log suites and acquiring samples to establish the rock fabric and related pore system, some of which are illustrated in Figure 7. Solid petrophysical analysis lays the foundation for all steps in the exploitation of these reservoirs, from exploration for high oil-cut zones, to pay delineation, estimation of reserves, forecasting reservoir performance, completion design, and the consideration of enhanced recovery options. Application of recently developed and sophisticated logging tools combined with whole core analysis provides important quantitative information that will bring dividends to the future life of the play (Doveton, 2014; Rush et al., 2016).

Chert occurs as layers and lenses within the Mississippian rocks of the midcontinent and is predominantly present in Osagean strata. Geographically, chert reservoirs are most abundant in northeastern Oklahoma and south–central Kansas. As a rule, they are characterized by high porosity and low permeability and thus function as excellent reservoirs if naturally fractured or stimulated. The observed spatial association of chert with the unconformity separating Mississippian carbonates from Pennsylvanian shales creates an ideal trapping configuration, as these reservoirs are commonly overlain by the regional seal.

Terminology describing chert reservoirs is rife with confusion. This stems in part from the mining literature, as chert is also the host mineral for lead–zinc deposits in the area, and tailings are called “chat” piles in the vernacular. Mazzullo and Wilhite (2010) give a good accounting of the origin and many meanings of descriptive and genetic terms for the local cherts, including chat, spiculite, and tripolite. Tripolite or tripolitic chert is used ubiquitously in northeastern Oklahoma to describe chert reservoir fabric that is highly microporous, and hence low in density. Tripolitic cherts in Osage County, for example, have a log-calculated total porosity in excess of 35% whereas porosity measured from core plugs in the same zone averages around 10% (Figure 8). In thin section, it is apparent that much of the pore space consists of micropores. As a result of their microporous nature, these reservoirs record abnormally low resistivity on standard wireline logs due to large amounts of bound water (Watney et al., 2001). Therefore it is difficult to establish water saturation and oil in place without careful integration of petrography and petrophysics coupled with production testing and history matching. Duren (1960, 1967) was among the first to recognize and document this complex pore geometry and its impact on capillary pressure response and electrical properties in the tripolitic chert reservoirs of the Glick field in Kiowa and Comanche counties in Kansas.

The term “tripolitic” was adopted from weathered cherts in outcrop in Tripoli (Tarr, 1938), and while texturally appropriate is genetically ambiguous. Most recent authors agree that the genesis of the chert reservoirs varies regionally and locally, with some being formed penecontemporaneously in facies rich in biogenic silica (Rogers, 2001; Watney et al., 2001; Franseen, 2006; Mazzullo et al., 2019) and others related to meteoric diagenesis, groundwater movement, or hydrothermal alteration (Watney et al., 2001; Zhao, 2011; Manger, 2014; Goldstein et al., 2019). Note that these few citations rely on and reference the stout volume of literature from the past century. Although too numerous to include in an overview, the collected body of work contains excellent descriptions and interpretations of chert reservoir fabrics. Watney et al. (2001) and Wilson et al. (2014) note the alignment of chert reservoirs with structural features, especially basement-involved faults. Most agree that the present rock fabrics are the result of recrystallization or replacement of limestone by silica, which may or may not be accompanied by solution of calcite (Rogers, 2001; Watney et al., 2001; Manger, 2014). In the case of those chert reservoirs in which the mineral content is almost 100% silica, it is apparent that a source of silica other than that remobilized from biogenic clasts must have been carried to the site of replacement in quantities greater than the original pore volumes. Therefore, a fluid circulation system is required to account for the location and volume of limestones altered to chert.

The carbonate and chert reservoir types discussed above are only examples or end members of a continuum of variable rock fabrics and their associated effect on capillarity, through which hydrocarbons have migrated and been differentially entrapped in the giant Mississippian system. Variation in reservoir character occurs regionally, stratigraphically, and in structurally cross-cutting trends. This contributes to the high level of heterogeneity in the play.

The most recent phase of development of Mississippian reservoirs in the midcontinent has focused on horizontal penetration of multiple reservoirs in a variety of trap types. Since the bottom of the economic downturn in 2009, over 4200 horizontal wells have been drilled, approximately 3600 of them in Oklahoma, and the remaining 600 or so in Kansas (Figure 9). Of these, one operator, Sandridge Energy, drilled over one-third of the horizontals. Four thousand of the horizontal producers from Mississippian reservoirs have sold oil or gas, in total 125 million barrels of oil and 950 billion cubic feet of gas, from the end of 2009 through 2015. Neither Oklahoma nor Kansas reports produced-water volumes. Although most of the BTU content is therefore from gas sold (using a six to one ratio for gas to oil), low natural gas prices at the wellhead during the producing period mean that most of the value in the play has come from produced oil.

The area of development is very large geographically, encompassing the better part of 14 counties in Kansas and Oklahoma or over 10,000 square miles. Note that especially in the most densely drilled portion of the horizontal play area in Harper, Woods, Alfalfa, and Grant counties, there are few if any distinct field boundaries. Early operators assumed that the Mississippian reservoirs were similar to unconventional gas shale plays: laterally consistent in trap type and reservoir character. They developed the play using principals that had worked for the Woodford and other shales, in which rapid leasing, pattern drilling and multistage hydraulic fracture completions were the norm. Because of the heterogeneity of the reservoirs drilled, the wide variety of completion practices that were used (Grieser and Pinkerton, 2013) and the complexity of positions within several different trap types, the production results were instead highly variable (Figure 9). We will look briefly at a few of the most defined or successful examples from which we can pull some common threads. The most obvious of these commonalities, which occurs across reservoir rock types, from shelf to basin, and in open-hole, acidized and multistage hybrid fracs, is an abundance of produced water. One positive scientific result from the necessity of drilling many salt water disposal wells immediately adjacent to production is the abundance of data that have been collected from these vertical wellbores. This allows for deep investigation into the reservoir character of the Mississippian, as well as calibration of the 3-D seismic acquired to guide horizontal development.

The first example illustrates the conventional nature of the traps in the Mississippi lime reservoirs. Range Resources drilled almost 100 horizontals on the crest of the Nemaha Ridge in far southwest Kay County and northeast Noble County in Oklahoma. The prospect was apparently located on the basis of structural closure against the east bounding fault of the Ridge (Figure 10). The first horizontal in the area, Mela 2-29H, was drilled and completed in 2009. The initial potential (IP) of the well was measured at 358 barrels of oil per day (BOPD), 329 thousand cubic feet of gas (MCFGPD), and 2239 barrels of water per day (BWPD). As of the end of 2015, the well had produced almost 90,000 barrels of oil and 15 million cubic feet of gas. The target zone in this area lies around 4000 feet (1200 m) measured depth from the surface, and most laterals extend another 5000 feet (1500 m) and are oriented north–south. Through 2015, those wells with recorded production have made on average about 22,000 barrels of oil per well and just under a quarter of a billion cubic feet (BCF) of gas. On initial production, the wells averaged 184 BOPD, 720 MCFGPD, and 3728 BWPD, which translates to a 5% oil cut and most of the value from oil production. Of interest in this accumulation is the apparent improved oil cut with position of increasing height above free water in the fault-bounded structures (Figure 10).

The largest contiguous block of horizontal drilling is concentrated in Woods and Alfalfa County, Oklahoma, and Harper County, Kansas (Figure 11), and in this paper is referred to as the Greater Cherokee West field. These wells are deeper, producing from 5000 foot (1500 m) laterals at about 6000–7000 feet (2000 m) below surface. Of the 2025 wells drilled since 2009, over 90% were drilled by three operators: Sandridge (over 800 wells), Chesapeake (about 600 wells), and Midstates (about 300 wells). The distribution of wells by operator is shown in Figure 11 and production statistics are given in Table 1. On average, each of the producing wells in the play has made about 40,000 barrels of oil and about a third of a BCF of gas through the end of 2015. Based on IP data, the oil cut in the wells averages about 8% on start-up. A quick look at the data across state lines offers an interesting if perhaps not statistically relevant observation—that wells in Kansas diverge from the norm with lower overall production per well (30,000 barrels on average), but higher oil cut (20%). These data are skewed by the fact that far fewer initial production tests are recorded in the IHS Global database (2016) for Kansas on a per unit basis than in Oklahoma. Of the 300 completions reported, only about 11% reported test data, as compared with 80% of the Oklahoma wells.

Evaluation of company by company data from the Greater Cherokee West field illustrates the importance of focus and location with respect to trap configuration in the play. The most scattered drilling, that of Sandridge, brought results below the norm for the trend—35,000 barrels produced per recorded well since 2009 with an oil cut of 6%. This figure is corroborated by specific analysis of Sandridge’s Harper and Sumner County oil production and water disposal in 2014 (Watney, personal communication, 2015). Midstates, which remained tightly focused on one area, possibly with better reservoir fabric or trap position, achieved average production of 46,000 barrels per well since inception with an average initial oil cut almost double that of Sandridge at 11%. Additionally, Midstates wells have a lower average initial gas to oil ratio than those of the other two operators; this results in improved overall value of the production in the concurrent commodity price environment. Duration of production and estimated ultimate recovery have not been included in this rough statistical overview, so time will bring new insights as drilling and production continue.

As most of the larger, water-free or high oil-to-water ratio traps were discovered and exploited in the early and middle 20th century, it is helpful to view the recent regional horizontal development as the exploitation of their associated water-rich transition and waste zones (Figure 9). In scattered instances, pockets of water-free oil or high oil-cut zones have also been discovered with careful mapping and petrophysical work. After Clinton (1959) articulated the frustration of the industry in finding no logical explanation for the range of water and hydrocarbon distributions encountered in the 1950s, petrophysicists began to quantitatively assess the effect of variations in rock fabric on entrapment geometry and the distribution of water and hydrocarbons within a single trap. Berg (1975) published an elegant description of the phenomenon in siliciclastic rocks and documented that calculations of oil column height and saturations using measured rock properties and capillary pressure equations matched observations from real traps. Schowalter (1979) further defined waste zones as those intervals of relatively impermeable rock, still below the effective seal, into which oil could migrate, but not displace enough water to produce water-free and transition zones of varying height and depth according not only to structural dip but also variations in reservoir quality. Finally, Lucia (1995) described the heterogeneity of carbonate rock fabrics and the resulting range of capillary effect in such reservoirs, and many case studies, including Wilson (1991), have documented waste zone and transition zone accumulations associated with structural and stratigraphic traps in carbonates. Microporous tripolitic chert reservoirs and low-permeability carbonate ramp deposits will have long transition zones on the order of hundreds of feet (tens of meters) or more. Adding to the complexity in the subsurface is the almost ubiquitous presence of natural and induced fractures. Although hydraulic fracture stimulation can overcome low permeability, it can also exacerbate high water production by connecting the wellbore to through-going fault zones. The complexity of carbonate pore types, stratigraphy, and structure encountered in a horizontal well warrant undertaking carefully tailored completions.

No one has said it better than Marlan Downey (2014) in his commentary to the AAPG Explorer entitled “Thinking Like Oil.” Downey makes an eloquent case for the fact that oil will move into any space if the buoyancy pressure of the oil column exceeds the entry pressure of the pore throat. Explorers in the Mississippian lime play have learned from experience that in these fractured and breached reservoirs, the subsurface war between entry pressure and pore geometry also dictates how much water will be produced with the oil. Given the naturally low porosity and heterogeneity of the Mississippian carbonate and chert reservoirs, the identification of commercially productive zones requires definition of the overall structural or stratigraphic trapping configuration (Wilson, 2015) as well as intensive exploration for enhanced permeability due to the presence of grainstone pods, dolomitization, or hydrothermal alteration at maximum height above free water.

As of 2016, two important developments are influencing a change in direction for the horizontal play in the Mississippian of the midcontinent. The first is the occurrence of an earthquake swarm that almost directly overlies high water volume disposal in both time and space. The second is the discovery of unconventional, that is, true in situ traps in shales of Meramecian and Chesterian age in the deeper portion of the Anadarko Basin.

Oklahoma and Kansas have seen a two order of magnitude increase in earthquake activity in the area indicated by mapped epicenters of magnitude 2.5 or greater (Figure 12). The onset of earthquakes coincides in time with increased produced fluid disposal activity since 2009 and overlays the area of horizontal drilling of the Mississippian lime play, as well as other high produced-water plays such as the Hunton (Wilson and Wallis, 2015). Although the increase in induced seismicity, together with falling commodity prices, may result in at least a temporary curtailment of production from the play, geomechanical investigations of the effects of brine disposal (Bidgoli et al., 2015a), CO₂ sequestration (Bidgoli et al., 2015b), and CO₂ injection for enhanced oil recovery (Schwab et al., 2015a, b) will help to further resolve the complex geology of the Mississippian system, as well as better manage the risk of induced seismicity in the future.

Over the same time period, operators have discovered potential in sections of the Mississippian that are composed of source rock facies, originally deposited in the deeper waters closer to the axis of the Oklahoma trough (also shown in Figure 12). Promising initial tests in the Meramec silts and marls have prompted a moderate drilling boom in southwestern Kingfisher County, where operators anticipate that reservoir heterogeneity will be lower than in the lime facies up dip. Black shales of the Goddard as well as the overlying Springer shales have yielded some large initial production results, and drilling patterns suggest the presence of unconventionally trapped oil and gas in narrow, deep thermal maturation fairways further south in Grady, Carter, and Stephens counties. Small clusters of horizontals have also tested the Caney Shale of the Mayes Group in the Ardmore and Arkoma basins. In the shale plays, produced water is minimal, so the seismicity risk from disposal is also lower. As the industry shakes out of its recession, it is likely that the focus of horizontal drilling in the Mississippian will shift to these more homogenous, lower water-cut plays.

The regional setting of the giant Mississippian carbonate system is deceptively simple—a blanket of fractured carbonates, exposed and weathered, draped over the regionally high Cherokee platform, charged from the adjacent basins and capped by widespread shale deposition. However, the imposition of foreland basin tectonics on ancestral rift fabrics, coupled with the heterogeneous distribution of reservoir fabrics, both depositional and diagenetic in origin, created the complex mosaic of traps and fluid distributions of the present subsurface landscape. Exploration and production history reflects this diversity of opportunities. Early exploration for small oil-rich structural traps was followed by the application of fracture stimulation technology and ultimately horizontal drilling to exploit increasingly less permeable and more water-rich portions of Mississippian hydrocarbon traps.

Carbonate and chert reservoirs are tricky beasts, and the Mississippian system of the northern midcontinent is certainly no exception. Variable fabrics resulting from deposition and multiple phases of diagenesis are further complicated by complex stratigraphic packaging, tectonic overprint, and migration of multiple fluids in multiple stages across the regional system of traps. It comes as no surprise that the results of the recent horizontal drilling boom have been quite variable, with a wide range of fluid ratios, production rates, and commercial outcomes. Challenges presented by Mississippian reservoirs underscore the need for rejuvenation of geological skills, including carbonate petrophysics and petrography, and application of these techniques to describe reservoir heterogeneity and its effect on the distribution of hydrocarbons and water in traps. Such skills have largely lain fallow in the industry during the gas shale boom, yet they are vital components in determining production potential in the Mississippian play, and indeed all low-permeability carbonate trends.

The papers in this volume capture the geologic learning that was trampled under the feet of those rushing to county courthouses for leases from 2010 until 2014. In the perfect world, these stratigraphic and structural models and reservoir characterization techniques would be applied ahead of the drill-bit to better define trap configuration and fluid distribution. As the play takes a breath before the inevitable return to exploration for conventional traps in Mississippian reservoirs in the northern midcontinent, the geological community can rejoice in the opportunity to analyze, map, and explore in a more prudent and value-driven manner. Studies such as those presented in this volume are critical to the success of the next phase of thoughtful exploration for and exploitation of Mississippian hydrocarbon accumulations.

The authors are grateful to the staffs of Rock Whisperer LLC, the Kansas Geological Survey at the University of Kansas, and the Boone Pickens School of Geology at Oklahoma State University for their tactical and financial support of the research that underpins this paper. Special thanks are due to Kevin Blake at IHS Global who graciously facilitated use of an integrated well and production dataset from the Mississippian plays in Kansas and Oklahoma. The regional and trend maps that we include would not have been possible without this help. The authors are grateful to BGI Resources LLC, who allowed publication of data from the Foraker field in Oklahoma. We also thank the reviewers for their thoughtful and constructive comments, especially Sal Mazzullo, whose in-depth reading and practical advice were critical to refining the manuscript. And finally, we acknowledge the excellent science and persistent work of all the contributing authors in this volume.