Modeling the maturation history of the stacked petroleum systems of the Williston Basin, USA

Marine and Petroleum Geology 155 (2023) 106390
Contents lists available at ScienceDirect
Marine and Petroleum Geology

journal homepage: www.elsevier.com/locate/marpetgeo
Modeling the maturation history of the stacked petroleum systems of the

Williston Basin, USA
Sarah E. Gelman
United States Geological Survey (USGS), Energy Resources Program, Denver, CO, USA
A R T I C L E I N F O A B S T R A C T
Keywords: A three-dimensional petroleum systems model was built to support U.S. Geological Survey assessments of un
Basin modeling discovered oil and gas resources in the Williston Basin of North Dakota, Montana, and South Dakota. Numerous
Thermal history Paleozoic source rocks have been proven or postulated in the basin, of which five were the focus of maturation
Petroleum system
and migration modeling: the Ordovician Icebox Formation, the kukersite beds of the Ordovician Red River
Williston basin
Oil migration
Formation, the shales of the Devonian–Mississippian Bakken Formation, the Mississippian Madison Group, and
Bakken the Pennsylvanian Tyler Formation. Calibration of the three-dimensional model to present-day temperature data
Petroleum generation indicates the existence of a north-south trend of high heat flow in western North Dakota, along with a region of
Source rock high heat flow in eastern Montana. These high heat flow trends strongly control the maturity of all studied source
intervals. A Bakken-specific hydrocarbon generation kinetic model was developed to match the calibrated time-
temperature history of the basin to spatial trends in hydrogen index from programmed pyrolysis data. Generation
of hydrocarbons occurred in the Cretaceous through Paleogene due to increased burial. Subsequent uplift and
erosion in the Neogene cooled the basin, ending hydrocarbon generation for all source rocks. The cumulative
volume of hydrocarbons generated by each of the source rocks was calculated and used to compare their relative
robustness. The shales of the Bakken Formation are estimated to have generated approximately 460 billion
barrels of oil equivalent (BBOE), while the Red River Formation generated approximately 130 BBOE, the Tyler
Formation 94 BBOE, the Madison Group 44 BBOE, and the Icebox Formation 28 BBOE. Gross migration trends
were analyzed with respect to historical oil and gas production in the basin and generally indicate segregation of
petroleum systems throughout the stratigraphic column. However, most modeled scenarios indicated significant
loss of Bakken oil to the Madison Group, suggesting that mixing of Madison and Bakken oils may be more
prevalent than has recently been recognized in the U.S. portion of the Williston Basin and is particularly likely in
fractured regions of the basin.
1. Introduction This report describes the regional geological inputs and results of that
model; the corresponding input and resultant data are available in a
The Williston Basin is one of the most important petroliferous basins USGS data release (Gelman and Johnson, 2023).
in the onshore United States; the resource play in the Devon The focus of this study is on the construction and results of a 3D
ian–Mississippian Bakken Formation had both the second highest oil petroleum systems model. Input to the model incorporates a regional
production and proven reserves of any tight play in the country, geological description of the basin, including regional maps of major
following the Permian Basin (U.S. Energy Information Administration, source rocks and reservoir units that are integrated with historical hy
2020). In 2020 and 2021, the U.S. Geological Survey (USGS) completed drocarbon production. Previous published basin modeling studies of the
a series of geology-based assessments of undiscovered oil and gas Williston Basin include a series of one-dimensional (1D) models to
resource potential for the Williston Basin province in North Dakota, support a previous vintage of USGS resource assessments (Gaswirth
Montana, and South Dakota (Schenk et al., 2021a, 2021b, 2021c; Marra et al., 2013), a two-dimensional (2D) transect model (Burrus et al.,
et al., 2021). In support of these assessments, a three-dimensional (3D) 1996a, 1996b), and a 3D model focused specifically on the Bakken
petroleum systems model was created to constrain the major source rock Formation (Kuhn et al., 2010, 2012). In addition to the Bakken, this
intervals of the basin, including their presence and maturation history. modeling study investigates source rocks in the Ordovician Icebox and
E-mail address: [email protected].
https://doi.org/10.1016/j.marpetgeo.2023.106390
Received 3 October 2022; Received in revised form 26 April 2023; Accepted 22 June 2023
Available online 24 June 2023
0264-8172/Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
S.E. Gelman Marine and Petroleum Geology 155 (2023) 106390
Red River Formations, the Mississippian Madison Group, and the portion of the basin due to input data limitations.
Pennsylvanian Tyler Formation. Source rock presence maps were The basin is underlain by a major suture zone of two Archean cra
created for these units, where possible, to serve as inputs to the petro tons: the Wyoming craton on the western side of the basin and the Su
leum systems model. The model leverages a vast database of wireline perior craton on the east; the Trans-Hudson orogenic zone separates the
well logs (IHS Markit, 2022) that were the basis for regional geologic two features (e.g., Anderson, 2007; Nesheim, 2012; Bader, 2019). Final
interpretation using IHS Kingdom software version 2020 (IHS Markit, collision of these cratons happened about 1.6 billion years ago, although
2020); the 3D model was built using Schlumberger PetroMod version the accretion of island arcs and oceanic crust fragments that make up a
2020.1 (Schlumberger, 2020). An overview of this geological input is heterogeneous series of belts within the Trans-Hudson orogen occurred
provided in Sections 3.1 and 3.2. Sections 3.3 and 3.4 describe the over a period of hundreds of millions of years leading up to the
thermal calibration of the model and Section 4 provides a description Superior-Wyoming collision (Green et al., 1985). An intense period of
and summary of modeling results and their integration with historical lithospheric warming and subsequent cooling has been proposed as the
production trends. cause of subsidence associated with the formation of the Williston Basin
(Crowley et al., 1985). Apatite fission-track thermochronology of
2. Geologic setting Archean and early Proterozoic basement in eastern North Dakota pro
duced ages ranging from Cambrian to mid-Ordovician, suggesting a
The Williston Basin is an intracratonic basin that formed during the major uplift and erosion event was followed by a prolonged cooling
Cambrian–Ordovician and experienced several episodic cycles of sub period during the time of initial basin formation (Crowley et al., 1985;
sidence and associated sedimentation. It spans North Dakota, Montana, Sturrock et al., 2021).
and northwestern South Dakota in the U.S. and the Saskatchewan and A few major structural features punctuate the otherwise roughly
Manitoba Provinces of Canada; this study focuses only on the U.S. circular basin, including the Cedar Creek anticline, the Nesson anticline,
Fig. 1. Location map of modeled area of interest for the U.S. part of the Williston Basin. Major structural features are annotated on a structural map of the top of the
Mississippian Kibbey Formation. An inset on the central portion of the basin illustrates smaller-scale features visible on a structural map of the upper part of the
Devonian–Mississippian Bakken Formation.
2
and the Poplar dome (Fig. 1). Similar to rates of subsidence and sedi Table 1
mentation, these structural features have been controlled by sporadic Cumulative oil and gas production by formation as of September 2022 (S&P
episodes of structural growth throughout the history of the Williston Global, IHS Markit, 2022). Includes wells in the Williston Basin in North Dakota
Basin and are especially important in controlling distributions of oil and and Montana, USA. MMBO, million barrels of oil; BCFG: billion cubic feet of gas;
gas resources. The Cedar Creek anticline is highly asymmetric; the MMBOE: million barrels of oil equivalent.
southwest flank has a very steep dip, up to 40◦ (Clement, 1986), while Unit Cumulative Oil Cumulative %Total Hydrocarbons
the northeast flank has a very gentle dip down to the center of the (MMBO) Gas (BCFG) (% of Total in
MMBOE)
Williston Basin. Clement (1986) documented several major periods of
structural movement on the Cedar Creek anticline, including alternating Bakken Formationa 4641 7957 67.26
Madison Groupb 1162 1074 15.11
episodes of reverse and normal faulting, starting at least by the Late
Red River 650 1230 9.64
Devonian and ending with renewed movement in the Late Cretaceous to Formationc
late Paleogene. Similarly, Gerhard et al. (1987) describe a prolonged Duperow and 235 177 2.99
structural history and reversals of movement along the Nesson anticline, Birdbear
primarily driven by a master fault on the west side of the structure. Fault Formationsd
Interlake Formatione 84 216 1.35
movement began by the Ordovician and ended with a major reversal
Tyler Formationf 89 11 1.03
during the Late Cretaceous to late Paleogene. In contrast, the Poplar Lodgepole 64 33 0.79
dome appears to be a slightly more recent feature of the basin. Orchard Formation
(1987) describes its initial appearance around Mississippian time, with Spearfish 64 4 0.73
Formationg
major uplift to the present-day configuration after the Paleocene.
Bighorn Group, 27 50 0.39
Collectively, the timing of deformation of these structural features, as upperh
well as periodic reversals of fault movement, is of particular interest to Winnipegosis 25 23 0.33
hydrocarbon migration studies. Formation
The center of the Williston Basin contains over 15,000 feet of Deadwood 0 71 0.14
Formation and
Phanerozoic sedimentary rock. Sloss (1963) identified and named six
Winnipeg Group
continental-scale unconformities and bounded the rocks between them Dawson Bay and 7 31 0.14
into major sequences: the Sauk (Cambrian to Lower Ordovician), Tip Souris River
pecanoe (Middle Ordovician to Silurian), Kaskaskia (Devonian and Formations
Kibbey Formation 10 0 0.11
Mississippian), Absaroka (Pennsylvanian to Triassic), Zuni (Jurassic to
Paleocene), and Tejas (Eocene to latest Holocene). The rocks of the a
Includes query of Bakken and Three Forks Formations, and informal Sanish
lower and middle Paleozoic are predominantly carbonate while those of sand.
b
the upper Paleozoic, Mesozoic, and Cenozoic are siliciclastic. Burial Includes query of Mission Canyon and Charles Formations and sub-intervals.
c
history plots for the Williston Basin clearly indicate several major pe Includes query of Red River Formation (including sub-zones) and Ordovician
riods of subsidence and associated sedimentation (an example is shown rocks.
d
Includes query of Duperow (including sub-zones), Birdbear, and Nisku
in Fig. 14). These include deposition in the Ordovician and Silurian; a
Formations, and Devonian rocks.
major increase in sedimentation in the Upper Devonian through e
Includes query of Interlake Formation and Silurian rocks.
Mississippian; relative quiescence through the Permian through Lower f
Includes query of Tyler and Heath Formations.
Cretaceous; another major increase in sedimentation in the Upper g
Includes query of Spearfish Formation and Spearfish/Madison (co-reported).
Cretaceous; and finally, inversion of the basin, including uplift and an h
Includes query of Stonewall and Stoney Mountain Formations, and Gunton
uncertain amount of erosion associated with the Laramide orogeny in Member of the Stony Mountain Formation.
the Paleogene. This basin history will be discussed further with respect
to estimating erosion in the Laramide (Section 3.2) and its impact on the to support this regional basin modeling study. Utilizing more than
maturation history of source rocks (Section 4.1). 16,000 wells and their associated logs, twenty-nine formation tops were
Hydrocarbons were first produced in the basin in 1912 with gas picked over a broad geographic area of interest (AOI) to characterize the
discovered in Upper Cretaceous sandstones along the Cedar Creek basin’s fill history and are summarized in Fig. 2 and Table 2. Due to the
anticline (Clement, 1986), followed by the first oil discovery along the sparse nature of wells on the periphery of the basin and few penetrations
Nesson anticline in 1951. Since these early discoveries, oil and gas have to the deepest stratigraphic units, there is considerable variation in the
been produced from virtually every formation in the Paleozoic section, uncertainty associated with these formation top picks. Five “anchor”
from the Cambrian–Ordovician Deadwood Formation to the Permian– surfaces define the stratigraphic framework of the basin model. These
Triassic Spearfish Formation. The first Bakken Formation discovery surfaces were selected based on high confidence identification in well
occurred along the Antelope anticline (a structure obliquely intersecting logs despite occasional low quality, partial logging suites. These anchor
the Nesson anticline; see Fig. 1) in the early 1950s, as vertical wells surfaces include (1) surface topography, (2) the top of the Cretaceous
produced oil from a naturally fractured reservoir in the middle informal Mowry Formation, (3) the top of the Mississippian Kibbey Formation,
member of the Bakken (Pollastro et al., 2012). Two subsequent Bakken (4) the top of the Bakken Formation, and (5) the top of the Red River
discoveries, the Elm Coulee field in Richland County, Montana in 2000, Formation. Between these anchor surfaces, formation tops were utilized
and the Parshall field in Mountrail County, North Dakota in 2006, to calculate isochores, which were added or subtracted from the anchor
transformed the Bakken play into the most prolific in the basin, aided by surfaces to generate structural gridded horizons. These formation tops,
advancing technology in horizontal drilling and multistage fracking associated isochores, and interpreted structural grids are provided in a
completions. Table 1 provides cumulative oil and gas production totals USGS data release (Gelman and Johnson, 2023). Additionally, lith
by formation for the U.S. portion of the Williston Basin (IHS Markit, ofacies and physical rock properties assigned to each layer are provided
2022). in Gelman and Johnson (2023). Due to spatial variations in geology
throughout the basin, some stratigraphic layers utilized a map-based
3. Petroleum systems model inputs and calibration definition of facies, particularly units containing evaporite sequences
and significant source or reservoir units, which are summarized in the
3.1. Stratigraphic interpretation, facies maps, and source characterization following sub-sections.
The stratigraphy of the Williston Basin has been mapped extensively
3
Table 2
Stratigraphic horizons and associated formation top picks used in the basin
model. Formation top picks, gridded horizons, lithology mixtures, and associ
ated thermal, compaction, and fluid flow properties are provided in a USGS data
release (Gelman and Johnson, 2023). Fm, formation; Gp, group; Ma, millions of
years before present.
Age Horizon # Notes on Horizon Creation
(Ma) Picks
0 Topography USGS NED 1/3 arc-second

contoursa
0.02 Erosion
20 Neogene Topography minus isochore of
glacial sedimentsb
43 Erosion
50 Top Paleogene Neogene Unconformity minus
Arikaree/Chadron fragments
70 Top Pierre Fm 395 Calculated from Isochore
97 Top Mowry Fm 11603 Anchor Surface
125 Top Inyan Kara Fm 6247 Calculated from Isochore
145 Top lower part Swift Fm 6152 Calculated from Isochore
170 Top Rierdon Fm 7118 Calculated from Isochore
183 Top Piper Fm 6769 Calculated from Isochore
197 Top Dunham Salt Member Isochore from Density Log
201 Top Spearfish Fm 7251 Calculated from Isochore
228 Top Opeche Shale/Pine Isochore from Density Log
Salt Member
307 Top Tyler Fm 5768 Calculated from Isochore
313 Base of upper part Tyler Calculated from Facies Model
Fm source
315 Top of lower part Tyler Fm Calculated from Facies Model
source
323 Top Otter Fm 6618 Calculated from Isochore
330 Top Kibbey Fm 10812 Anchor Surface
335 Top Charles Fm 10557 Calculated from Isochore
338 Base net halite Charles Fm Isochore from Density Log
339 Top net anhydrite Charles Isochore from Density Log
Fm
340 Top Mission Canyon Fm 7788 Calculated from Isochore
344 Base net anhydrite Mission Isochore from Density Log
Canyon Fm
347.5 Top Mission Canyon Fm 5 ft above Top Lodgepole
source
348 Top Lodgepole Fm 7786 Calculated from Isochore
357 Top upper member, 8445 Anchor Surface
Bakken Fm
362 Top middle member, 7079 Calculated from Isochore
Bakken Fm
364 Top lower member, 5128 Calculated from Isochore
Bakken Fm
368 Top Three Forks Fm 5560 Calculated from Isochore
372 Top Duperow Fm 3639 Calculated from Isochore
375 Base net anhydrite Isochore from Density Log
Duperow Fm
377 Top Dawson Bay Fm 2673 Calculated from Isochore
382 Top Prairie Fm (2nd Red 2435 Calculated from Isochore
Fig. 2. Idealized stratigraphic column for the Williston Basin, after Murphy
Bed)
et al. (2009). Formation tops interpreted for this study are labeled on the right,
383 Top evaporite Prairie Fm 2155 Calculated from Isochore
and major source rocks included in the model are indicated by stars in their 388 Top Winnipegosis Fm 2019 Calculated from Isochore
stratigraphic interval. 408 Top Ashern Fm 2271 Calculated from Isochore
419 Top Interlake Fm 2521 Calculated from Isochore
3.1.1. The Ordovician Icebox Formation source rock 445 Top Red River Fm 2487 Anchor Surface
447 Top “C” anhydrite Red Isochore from Density Log
The Ordovician Icebox Formation is rarely penetrated in the basin River Fm
and was mapped in only 255 wells. The unit has a relatively consistent 448 Base “C” anhydrite Red Density Log Aided Pick
thickness, ranging from 120 to 190 ft, and an obvious trend of sand- River Fm
prone facies in the northwest part of the basin grading to a shale- 450 Top kukersite source 10 ft above Mid Red River
452 Mid Red River Fm Calculated from Isochore Split
prone facies to the southeast (LeFever and LeFever, 2011). The forma
458 Top Winnipeg Gp 307 Calculated from Isochore
tion was defined with a map-based mixture of sand and shale in the basin 463 Top informal Government NDGS Isochore above B. Icebox
model. This unit contains an organic-rich source rock. Although total Creek shale member
organic carbon (TOC) content only ranges from 0.3 to 0.5% in North 467 Base Icebox Fm 255 Calculated from Isochore
Dakota, it is commonly enriched in Canadian samples, averaging 2.32% 470 Top Deadwood Fm 153 Calculated from Isochore
542 Top basement 31 Calculated from Isochore
(Nesheim and Nordeng, 2013, and references therein). A study by
a
Nesheim and Nordeng (2013) focused on the informal Government Quadrangles from North Dakota, Montana, South Dakota, and Wyoming
Creek shale member near the base of the Icebox Formation. The Gov from U.S. Geological Survey, National Geospatial Program, USGS NED 1/3
ernment Creek member thickness, ranging from 0 to 15 ft, as well as TOC arc-second contours, 1 × 1◦ shapefile: U.S. Geological Survey.
and hydrogen index (HI) values of ~4% and 600 mg HC/g TOC,
4
b
Quaternary sediment thickness of the glaciated United States by Soller and only 1–2 ft per bed), core analysis was beyond the scope of this study.
Garrity (2018). Instead, wireline log signatures, described by Nesheim (2017), were
utilized to construct a facies model for identifying potential kukersite
respectively, were used to constrain the modeled source rock interval beds, referred to as “kukersite candidates,” throughout the basin model
(Nesheim and Nordeng, 2013). These TOC and HI values are the AOI (Fig. 3).
maximum recorded in a well in the Nesheim and Nordeng (2013) study. Kukersite beds tend to have an elevated deep resistivity signature
This well is located in Stark County, North Dakota, towards the pe and, where sufficiently thick, elevated neutron and density porosity, as
riphery of the basin. Wells towards the center of the basin are projected well as decreased sonic travel time relative to the surrounding dolomite
to be, on average, lower in organic content. This is due to a presumption (Nesheim, 2017). The gamma ray (GR) signature of these thin beds is
of increased maturity towards the center of the basin. In utilizing the generally nondescript. To systematically isolate the stratigraphic inter
highest TOC sample in their study, also located in presumably a less val containing kukersite beds in well logs, the interval top was con
mature part of the basin, to define the entire interval’s organic enrich strained by the “C” anhydrite bed in the Red River, readily identified on
ment, the basin model accounts for the effect of maturation on the density logs, and the base of the interval was either the total depth (TD)
reduction of organic enrichment in programmed pyrolysis data. if the well was terminated in the Red River, or the mapped top of the
Middle Ordovician Winnipeg Group. Calibrating to the BNRR 1–17 well
3.1.2. Kukersite source rock of the late ordovician Red River formation described by Nesheim (2017), a facies model was created utilizing a
The Late Ordovician Red River Formation is the third most petro qualitative Δ log R cross plotting technique (Bowman, 2010; Passey
liferous unit in the Williston Basin (after the Bakken Formation and et al., 1990). The two kukersite beds in this well with the highest TOC
Mississippian Madison Group; Table 1). Red River rocks have been (>10%) separated from non-kukersite rocks on a logarithmic deep re
previously summarized by Nesheim (2017) and subdivided into A, B, C, sistivity (log10 (LLD)) versus sonic (DT) cross plot (Fig. 3A). The region
and D zones, which contain dolomite and dolomitic mudstone reser of elevated resistivity and sonic on the cross plot was expanded to
voirs. Of particular interest to this study is the proposed source rock for identify kukersite candidates in other wells that lacked the confirmation
Red River oils: a series of thin and organically rich kukersite beds in the of kukersite from core analysis throughout the basin in both North
D zone, first observed in the Canadian part of the basin (Osadetz and Dakota and Montana. All wells with at least one foot of potential
Snowdon, 1995) and more recently in North Dakota (Nesheim, 2017). kukersite (334 total) were examined. Some wells contained “kukersite
The present study expands the work of Nesheim (2017) into Montana. candidates” in unreasonable stratigraphic intervals. For example, in
While detailed core observations would be ideal due to the thin nature of tervals very close to the “C” anhydrite, or in the lowermost section above
the kukersite beds in the Williston Basin (less than 1–8 ft per bed, most the Top Winnipeg pick, were considered to be unrealistic. Other wells
Fig. 3. Overview of the facies model utilized for

mapping source rock presence in the Upper Ordovi
cian Red River Formation. A) Cross-plot of sonic (DT)
and logarithmic deep resistivity (LLD) wireline logs
for the C and D zones of the Red River in the BNRR
1–17 well described by Nesheim (2017). Color illus
trates the difference between log10 (LLD) and DT in
resistivity units and highlights the significant sepa
ration of the kukersite beds from other lithofacies in
section. B) Wireline logs of the BNRR 1–17 well,
including marker beds for the top of the Red River
and base of the “C” zone anhydrite. Thin shaded
sections indicate kukersite beds highlighted in the
cross-plot in A) and confirmed in core by Nesheim
(2017). C) Map of wells investigated for presence of
kukersite beds using the cross-plot facies model.
Green symbols indicate wells in which “kukersite
candidates” were flagged, while small grey symbols
indicate wells in which kukersites are interpreted to
be absent. The green shaded region is the mapped
extent of kukersite beds utilized in the basin model.
GR, gamma ray log; DT, sonic velocity log; log10LLD,
logarithm of deep resistivity log; ft, feet.
5
displayed elevated sonic log signatures through a thick stratigraphic

interval, suggesting that much of the Red River Formation contained
potential kukersite beds; sonic log normalization could be used to
analyze these wells but was not performed in this study. Stratigraphic
and log-based outliers were deleted from the dataset and the remaining
284 wells, shown in green symbols in Fig. 3C, were utilized to create a
regional kukersite presence map. The North Dakota portion of the
kukersite presence map compares favorably with a similar map pub
lished by Nesheim (2017). The regional kukersite map for North Dakota
and Montana was utilized as a self-consistent source presence map for
the Red River Formation in the basin model (Fig. 3C).
3.1.3. The Late Devonian–early Mississippian Bakken Formation

The tops of the upper, middle, and lower informal members of the
Bakken Formation (e.g., as defined in Pollastro et al., 2012) and the top
of the underlying Devonian Three Forks Formation were mapped in
detail for this study. The isochores of each of the Bakken members are
shown in Fig. 4. A significant depocenter is observed in the isochores of
the lower and middle members of the Bakken east of the Nesson anti
cline. This depocenter is also observed in the isochore map of the Three
Forks to the Late Devonian Duperow Formation and in the isochore map
of the Duperow to Middle Devonian Dawson Bay Formation, but it is not
present in the isochore maps of the Dawson Bay or Middle Devonian
Prairie Formations. The upper member of the Bakken lacks this depo
center, suggesting that accommodation, likely created as a result of
structural movement along the Nesson anticline, was filled by the lower
and middle members of the Bakken. Therefore, structural movement
was apparently active throughout most of the Devonian but paused by
the end of the Devonian, and the upper member was deposited as a more
isopachous drape throughout the basin center. All three members of the
Bakken thin and pinch-out to the southwest, with the lower member
pinching out first, followed by the middle, and finally the upper mem
ber, which has the largest areal extent (LeFever, 2008). These pinch-outs
define the stratigraphic trap that controls the Elm Coulee field.
The upper and lower members of the Bakken are well documented
“world-class” source rocks (e.g., Dow, 1974; Williams, 1974; Meissner,
1978; Lillis, 2013; Jin and Sonnenberg, 2013). A compilation of pro
grammed pyrolysis data (Fig. 5; data obtained from the USGS Energy
Geochemistry Database, or EGDB) was used to evaluate organic richness
and provide constraints for Bakken hydrocarbon generation modeling
(discussed in Section 3.4). These data were filtered to eliminate poor
quality and incomplete data, as well as data in which the production
index and Tmax are off the main trend, suggesting sample contamina
tion or staining (filtering details, as well as before-and-after dataset
plots, are provided in the Supplementary Material). While many of these
data come from partially mature samples, the highest TOC, highest HI,
and lowest Tmax values indicate an initial TOC of approximately 20%
and an HI of 700 mg HC/g TOC. Some lean, immature samples may
indicate a propensity towards terrigenous facies to the east; however,
this hypothesis has not been implemented in the current study.
The middle member of the Bakken consists of a heterogeneous and
highly complex mixed carbonate-siliciclastic facies assemblage,
including extensively dolomitized siltstones, sandstones, and carbonate
bound stones and grainstones (Pitman et al., 2001; Canter et al., 2011;
Egenhoff et al., 2011). The diagenetic history of the middle member
rocks includes cementation of calcite, which reduced porosity, and
dolomitization, which increased porosity (Pitman et al., 2001; Brodie
et al., 2018). To recognize the significant impact of these diagenetic
processes on middle Bakken member reservoir rocks, simple diagenetic
models were initially incorporated in the assignment of rock properties
throughout this model layer. Cementation occurred soon after deposi Fig. 4. Isochores of the A) upper, B) middle, and C) lower informal members of
tion (Brodie et al., 2018) and was incorporated into all modeled middle the Bakken Formation. Axes of the Nesson and Antelope anticlines are shown in
Bakken facies, reducing porosity relative to standard mechanical bold dashed lines. Important fields in the Bakken are labeled on B) for spatial
compaction curves. The degree of this porosity reduction was allowed to reference. Mi, miles.
vary spatially according to interpretation of reservoir quality variations
within the middle Bakken member.
6
Fig. 5. Compiled and filtered programmed pyrolysis data for the upper and lower shale members of the Bakken (U.S. Geological Survey Energy Geochemistry
Database). A) Source rock rating chart showing total organic carbon (TOC) versus S2, colored by Tmax. B) Hydrogen index (HI) versus Tmax, colored by TOC. Both
plots illustrate a strong maturation trend to the data in which the highest TOC, highest HI, and lowest Tmax correspond to the most immature samples in the dataset.
The immature end-member was used to constrain original TOC and HI for basin modeling input. S2, height second peak of programmed pyrolysis data; TOC, total
organic carbon; HI, hydrogen index; Tmax, temperature of S2 peak.
To better understand the geologic context of Bakken production well-cemented and not the main target for production. O’Connell et al.
trends spatially, a simplified regional facies map of the middle Bakken (2012) noted the identification of a clean gamma ray, high resistivity
member was constructed for this study and used to implement this limestone during drilling, and utilized the presence of facies D to guide
variable cementation model (Fig. 6). Previous investigations have geosteering efforts to the main target facies beneath the unit. Further,
categorized several sub-facies in the middle Bakken member based on Grau and Sterling (2011) noted that facies D in the vicinity of Parshall
drill core and wireline log interpretations (Sonnenberg et al., 2011 and field is an ooid shoal (as opposed to sandstone in other parts of the
references therein). In general, the middle Bakken member has a basin), and that its presence inhibits beneficial dolomitization of the
non-descript wireline log signature, despite great lithologic complexity surrounding facies, which they term a “shoal shadow.” Indeed, the best
reported in these studies. A regional core-based analysis of the middle production in the basin comes from the area of Parshall field on the
Bakken member was beyond the scope of this project. Acknowledging eastern side of the basin where facies D is absent. This region was
this shortcoming, limited wireline log signatures, documented previ coarsely mapped with the intention of identifying areas with intense
ously in Sonnenberg et al., (2011), were mapped throughout the basin, dolomitization and, therefore, probable high reservoir quality. The area
including the presence of a “clean gamma” unit and general stacking with “absent D″ is represented graphically in Fig. 6B and is shown as
pattern trends. “clean GR absent” in the cross section in Fig. 6D.
The “clean gamma” unit of the middle Bakken member, referred to as The middle Bakken member facies in the far southwestern part of the
facies D in Sonnenberg et al. (2017), consists of variable limestone and basin comprise the unique reservoir of the Elm Coulee field. As discussed
sandstone. It is the coarsest-grained subunit of the middle Bakken above, this region received the least siliciclastic input and favored car
member and is interpreted to be the most basin-ward depositional facies bonate deposition. An elongated northwest-southeast trending carbon
in the AOI. Facies D is analogous to lithofacies 3 in Pitman et al. (2001) ate shoal is interpreted as the cause of the thickening observed in the
and facies B in Canter et al. (2011). Within the middle Bakken member, middle Bakken member isochore (Fig. 4B; Theloy, 2013). Wireline log
facies D is anomalous due to its low gamma ray and elevated resistivity signatures exhibit changes in the stacking pattern between the main
signatures and therefore is mappable using wireline logs. The presence thickening of the Elm Coulee trend and the more central part of the
(or absence) of facies D is interpreted in this study to be a proxy for two basin. While a symmetric hourglass signature is noted in gamma ray logs
important but opposing controls on reservoir quality in the middle northeast of the Elm Coulee thickening (denoted “Elm Coulee Hour
Bakken member: 1) its presence serves as a proxy for coarser siliciclastic glass” in Fig. 6A), a coarsening upward trend is observed at Elm Coulee
influx to the basin, and 2) it inhibits improved secondary porosity. field itself (denoted “Elm Coulee Coarsening Up” in Fig. 6A). The
In wireline log-based mapping, facies D is absent in the southwestern boundary between these two stacking patterns was the subject of
portion of the Williston Basin. In Fig. 6A, where facies D is absent from focused mapping along the southwestern edge of the middle Bakken
the wireline logs, it is referred to as “clean GR absent” and is shown as a member but the hour-glass signature continues throughout most of the
black dot. Based on detrital zircon analysis of middle Bakken member central part of the basin. Based on the synthesis of previous workers (e.
samples from the Canadian part of the Williston Basin, Mohamed (2015) g., Egenhoff et al., 2011; Sonnenberg et al., 2017), the coarsening up
identified three main sources of siliciclastic sediment: two in northern ward stacking pattern is interpreted to represent the lower portion of the
Canada (the Franklinian belt and Canadian shield) and one from the east middle Bakken member and an overall basin-ward progression of
of the Williston Basin (Mazatzal-Yavapai and Grenville provinces). The deposition (denoted “Carb. Shoal w/o UMB” in Fig. 6B). The hourglass
southwestern portion of the Williston Basin is thus interpreted to be the stacking pattern depicts both the progradational (basin-ward) and ret
most distal based on siliciclastic input. Where facies D is present, sili rogradational (land-ward) depositional systems (denoted “Carb. Shoal
ciclastic deposition is interpreted, suggesting relatively good reservoir w/UMB” in Fig. 6B), prior to the maximum flood event represented by
quality, in contrast to the fine-grained shaley facies in the center to the upper shale member of the Bakken Formation. The coarsening
southwestern portion of the basin which is expected to have poorer upward-only stacking pattern at Elm Coulee field implies either
reservoir quality. These facies are distinguished graphically in Fig. 6B as non-deposition or extremely thin deposition of the retrograding upper
“distal intermittent facies D″ and “outer ramp silt/shale,” respectively. part of the middle Bakken member. Reservoir quality is expected to be
Facies D is typically characterized by its coarse-grained sediments good in this region and high porosity and permeability and less
and how it affects diagenesis of underlying rocks. Facies D is commonly cementation were assigned to these rocks in the basin model (detailed
7
(caption on next page)
8
Fig. 6. Overview of the methodology for the regional-scale, wireline log-based facies map of the middle Bakken member. A) Spatial distribution of log in
terpretations, including the presence (brown-to-yellow symbols) or absence (black dots) of facies D. The size of the “presence” symbols is a rough indicator of the
thickness of the unit, with large yellow symbols indicating greater thickness and lesser thickness for small brown symbols. Isochores of the middle Bakken member
are depicted in grey contours. The Elm Coulee region was interpreted in terms of stacking pattern: light blue symbols indicate a coarsening upwards pattern while
dark blue symbols indicate an hourglass pattern. B) Regional map of bulk facies distribution. The Elm Coulee area is broken into two segments depending on whether
the upper middle Bakken member (UMB, indicated by an hourglass stacking pattern on cross section C–C′) is present or not. The presence or absence of facies D is
indicated in the rest of the map, noting that in much of the region, facies D is periodically either subtle, thin, or absent (“intermittent”). The Nesson anticline is dashed
in bold for reference. The locations of cross sections shown in C and D are shown as C–C′ and D-D′ on the map. C) Cross section C–C′ illustrates the interpretive criteria
for the Elm Coulee stacking patterns. D) Cross section D-D′ illustrates the interpretive criteria for the presence of facies D in the Parshall area. LP: Lodgepole For
mation; UB: upper Bakken member; MB: middle Bakken member; LB: lower Bakken member; 3F: Three Forks Formation. GR, gamma ray; LLD, HLLD, and AT90, deep
resistivity.
property assignments are provided in Gelman and Johnson, 2023). member of the Bakken in Stark County, North Dakota (Jarvie, 2001;
The area east of Parshall field was also mapped utilizing wireline log Gaswirth et al., 2013); Lodgepole oil is volumetrically minor in terms of
signatures of the middle Bakken member. Numerous workers have basinwide production and not the subject of further attention in this
proposed various mechanisms to account for the development of the study. In contrast, the Mission Canyon and Charles Formations contain
trap at Parshall field—it is not a straightforward pinch-out like Elm reservoirs which have cumulatively produced the most hydrocarbons in
Coulee. This study explores two possible mechanisms that may result in the basin, except for the Bakken (Table 1). The oil produced out of the
a severe reduction in reservoir quality associated with this trap: 1) a Mission Canyon and Charles Formations has generally been typed to a
pinch-out or facies change from the prime target facies (Sonnenberg different source than the Bakken in the U.S. portion of the basin. This
et al., 2011, 2017), and 2) a maturation-associated diagenetic effect necessity for a unique non-Bakken source rock has led many workers to
originating from organic acids causing dissolution. While the former suggest that the Madison is dominantly self-sourcing (Price and LeFever,
mechanism is mapped using facies-based observations, the latter effect is 1994; Jarvie, 2001; Lillis, 2013), thus requiring identification of an
driven by maturation and signifies a loss of enhanced porosity in undiscovered robust source rock within the Madison Group, as discussed
immature areas just east of Parshall (Johnson, 2009; Theloy, 2013). below. The implications of a self-sourcing Madison petroleum system
Both potential mechanisms were tested in the basin model. In an initial will be discussed at length in Section 5.
model scenario, a reduction in reservoir quality was assigned to the Limited data exist on the stratigraphic interval, spatial distribution,
region in which facies D re-emerges (albeit intermittently) east of Par and thicknesses of proposed Madison source beds in North Dakota and
shall, termed “proximal intermittent facies D″ in Fig. 6B; this attempts to Montana. Nesheim et al. (2021) reviewed previously published infor
address a possible pinch-out or lithologic change as the trapping mation and provided programmed pyrolysis data supporting a potential
mechanism. In a second model scenario, an Arrhenius-based secondary source interval in the Bluell subinterval of the Madison in western
porosity model was created to generate additional porosity where the McKenzie County, North Dakota. They suggest the presence of a 20- to
shales of the Bakken Formation are mature. This model utilized the 30-foot gross interval of organic-rich carbonate rock, identified with
single activation energy hydrocarbon generation kinetic parameters of core data and mapped throughout the western half of McKenzie County.
Jin and Lewan (2017), but instead of generating hydrocarbons, it gen In an effort to estimate a net thickness of source rock for the basin model,
erates secondary porosity where mature. The effects of these two the results of their study were qualitatively filtered for continuous
modeled mechanisms are discussed in the Results (Section 4.3). stratigraphic intervals that had greater than 1% TOC. One borehole
As mentioned earlier in this section, a generalized cementation (Mitten 12–10, Fig. 4 in Nesheim et al., 2021) contained approximately
model was implemented for all the middle Bakken member facies, albeit 5 feet of continuous samples with over 1% TOC (up to 5% TOC,
this effect varies in magnitude according to interpreted reservoir qual 9350–9355 ft MD). Most of the remaining samples from that well, as
ity. To integrate the regional log-based facies observations described well as samples from most of the other wells in their report, had less than
above with isolated reports of porosity and permeability, initial 1% TOC or contained isolated ~1–2% TOC samples and generally had
compaction curves and cementation were scaled to assign the highest high production indices. This suggests staining, contamination, or
reservoir quality to regions with “absent facies D,” and “carb. shoal w/o otherwise generated oil affected the programmed pyrolysis results. An
UMB” facies (see Fig. 6B). A second tier of moderate reservoir quality attempt was made to identify a log-based signature of the
was assigned to areas with “continuous facies D,” “distal intermittent organically-enriched net section employing the methods used for the
facies D,” and “outer shoal w/UMB” regions. A third tier of poor porosity kukersites of the Red River Formation described in Section 3.1.2.
and permeability was assigned to the “outer ramp silt/shale” facies and Although subtle increases were observed in gamma ray and sonic travel
the region of “proximal intermittent facies D,” although the latter was time logs, these signatures were quickly lost when examining sur
balanced with the alternative secondary porosity model that tested the rounding wells, and an attempted facies model failed to produce a
maturity-associated trapping mechanism at Parshall field, described in reliable identification of high TOC samples.
the previous paragraph. Detailed properties for each of these facies, Rather than create a quantitative source rock map for the Madison,
including their diagenetic models as implemented, are provided in the source rock of the Bluell subinterval described by Nesheim et al.
Gelman and Johnson (2023). (2021) is instead used in this study as a proxy for spatially discrete, thin,
but moderately enriched source beds (visualized here as “wisps” or
3.1.4. The Mississippian Madison Group source rocks and evaporites discrete horizons) within and throughout the Mission Canyon and
The Mississippian Madison Group consists of a mixture of limestones, Charles Formations. The sum of these beds is an approximated 5-foot
dolostones, thin shales, and evaporites. Three major subdividing for interval of postulated source rock within the Mission Canyon that ex
mations were mapped within the Madison Group, from oldest to youn tends throughout the modeled AOI. Table 3 lists assigned source rock
gest: the Lower Mississippian Lodgepole Formation, the Mississippian TOC, HI, and kinetics, which are based on the net values from Nesheim
Mission Canyon Formation, and the Middle Mississippian Charles For et al. (2021) and the description of a moderately sulfur-rich marine
mation. The Lodgepole acts as a seal to the Bakken petroleum system carbonate source rock based on Madison oil geochemical data (Chen
and was assigned low porosity and permeability rock properties with a et al., 2009; Lillis, 2013). Additional published programmed pyrolysis
very high capillary entry pressure. The oil produced from the Lodgepole data supporting and describing potential Madison source beds, espe
historically has been geochemically typed to the Bakken petroleum cially for net thickness, would be invaluable for improving this repre
system and is found in carbonate mounds directly overlying upper shale sentation of Madison source rock.
9
Table 3
Source rock richness and kinetics assigned to source layers in the basin model.
Fm, formation; Gp, group; GCS, Government Creek shale; TOC, total organic
carbon; HI, hydrogen index; TII(B), type 2 organic matter type or type B in
Pepper and Corvi (1995); TII-S(A), sulfur-rich type 2 organic matter type or type
A in Pepper and Corvi (1995).
Source Layer TOC HI (mg HC/ Kinetics
(wt.%) g TOC)
Tyler Fm, upper part 15 800 TII(B), Pepper and Corvi

(1995)
Tyler Fm, lower part 30 400 TII(B), Pepper and Corvi
(1995)
Madison Gp 5 400 TII-S(A), Pepper and Corvi
(1995)
Bakken Fm 20 700 Modified “North Dakota,”
Tegelaar and Noble (1994)
Red River Fm 10 850 Kukersite (Canada), Sample 1,
Chen et al. (2019)
Icebox Fm, informal 4 600 TII(B), Pepper and Corvi
GCS member (1995)
In contrast to source rock intervals, evaporites of the Madison are

easily identified on wireline logs and are numerous within the upper
part of the Mission Canyon and Charles Formations. Utilizing formation
tops for the Charles, Mission Canyon, and Lodgepole, well log properties
were extracted in each interval to identify and calculate net thicknesses
of the evaporite beds. Density log values above 2,850 kg/m3 were
considered likely anhydrite, and below 2,300 kg/m3 were considered
likely halite. A summation of the results of these log extractions was
used to create net isochores of halite and anhydrite, shown in Fig. 7.
Anhydrite is the predominant evaporite observed in the Mission Canyon;
it thickens in an arcuate pattern to the east and to the south and likely
was deposited along the southeastern margin of the Mississippian-aged
basin. Both anhydrite and halite are found in the Charles. The anhydrite
covers most of the basin and forms a thick north-south trend in North
Dakota. Halite of the Charles Formation is concentrated in the center of
the basin and reaches net thicknesses of up to 450 feet.
3.1.5. The Pennsylvanian Tyler Formation source rocks

The Pennsylvanian Tyler Formation consists of a highly heteroge
neous mixture of shales, sandstones, and shaley limestones (Nesheim
and Nordeng, 2014a). Production in the Tyler in the Williston Basin has
primarily been limited to sandstone reservoirs in stratigraphic traps in
Stark and Billings Counties, North Dakota. Oil geochemistry has
demonstrated that oils produced from the Tyler were not derived from
Bakken, Madison, or Red River sources (Dow, 1974; Williams, 1974;
Lillis, 2013). Two potential source rock intervals have been described in
detail: 1) a shale in the lower part of the Tyler Formation, located in the
center of the basin and easily identified on well logs with an anoma
lously elevated gamma ray signature; and 2) a carbonate mudstone in
the upper part of the Tyler that is much more cryptic in well logs and
interbedded with the sandstone reservoirs in Stark and Billings counties
(Nesheim and Nordeng, 2012, 2014a, 2014b). Source presence maps
and net thickness estimates, based on detailed core and well log in
terpretations together with programmed pyrolysis data, are available for
the Tyler for the North Dakota portion of the basin (Nesheim and Nor
deng, 2014a, 2014b). Uncertainty remains regarding the extent of these
sources in Montana; this work expands on the North Dakota descriptions
to extend these maps into Montana to cover the full basin model AOI.
Fig. 7. Isochores of net thickness of evaporites in the Mississippian Madison
The source in the lower part of the Tyler is characterized by elevated
Group. A) Net thickness of halite in the Charles Formation; B) Net thickness of
GR peaks and associated high deep resistivity (LLD) as well as high sonic anhydrite in the Charles Formation; and C) Net thickness of anhydrite in the
travel time (DT) wireline log signatures (Curl well; Fig. 8A). These GR Mission Canyon Formation. Halite in the Mission Canyon was found to be
spikes, persistent in the lower half of the Tyler Formation throughout negligible. Black dots indicate well control, where the tops of the Charles,
much of North Dakota and Montana, are limited by an unconformity to Mission Canyon, and Lodgepole Formations were interpreted, and RHOB logs
the north and a reduction in thickness with an associated diminishing cover both the Charles and Mission Canyon intervals. Ft, feet.
log characteristics to the south. The source in the upper part of the Tyler
is less easily identified on well logs; it is characterized by a low,
10
Fig. 8. Overview of the facies model utilized for mapping source rock presence in the Tyler Formation. A) Selected logs are shown from the Curl well which was
described in detail by Nesheim and Nordeng (2014a, 2014b). This well illustrates the source rocks in the lower part of the Tyler, where organic-rich shales are
identified in green on the right-side track. B) Selected logs are shown from the Govt-Taylor well which was described in detail by Nesheim and Nordeng (2014a,
2014b). This well illustrates the source rocks in the upper part of the Tyler, where organic-rich carbonate mudstones are identified in blue on the right-side track. C) A
facies model based on sonic (DT) and logarithmic deep resistivity (LLD) wireline logs calibrated to the Curl and Govt-Taylor wells. Symbols in the cross plot are
colored by gamma ray (GR); circles correspond to data from the Curl well and triangles to the Govt-Taylor well. A grey dashed region indicates the DT-LLD space
corresponding to organic-enrichment (labeled “ΔlogR outliers” in log plots). Within that region, data with high GR (>100 a.p.i.) indicate “lower part of Tyler source,”
while low GR (<100 a.p.i.) indicate “upper part of Tyler source.” D) Map distribution of the net thickness of source rock in the lower part of the Tyler. The yellow
circle locates the Curl well used for facies calibration. E) Map distribution of the net thickness of source rock in the upper part of the Tyler. The yellow triangle locates
the Govt-Taylor well used for facies calibration. GR, gamma ray log; DT, sonic velocity log; log10LLD, logarithm of deep resistivity log; ft, feet; mi, mile.
nondescript GR and elevated LLD and DT signatures (Govt-Taylor well; or lower parts of the Tyler source rocks, calibrated to the type wells
Fig. 8B). In contrast to the lower part of the Tyler source, the source in described in core studies by Nesheim and Nordeng (2014a, 2014b).
the upper part of the Tyler has a GR signature that is non-unique and Bounded stratigraphically by the tops of the Upper Mississippian Otter
blends with the log signature of the sandy reservoirs above and below it. Formation and the Tyler Formation, the region of an DT-LLD cross-plot
Similar to the methods utilized for characterizing the kukersite beds in in which organically-enriched source rocks in this interval are observed
the Red River Formation (Section 3.1.2), a facies model was constructed is shown in grey in Fig. 8C. The two source intervals are differentiated by
using DT and LLD wireline logs to broadly identify wells with the upper a GR cutoff of 100 API: intervals with higher GR are considered to be
11
source rock in the lower part of the Tyler and intervals with lower GR are this study utilized similar estimates. However, while calibrating this
considered source rock in the upper part of the Tyler. Net thickness maps study’s model to maturity data, it became clear that these erosion esti
were then constructed from wells with each source interval identified in mates produced model results that were over-mature, especially when
the DT-LLD facies model (Fig. 8D and E). These maps compare well with utilizing the Sweeney and Burnham (1990) EASY%Ro kinetic model.
those published for North Dakota (Nesheim and Nordeng, 2014a, Incorporating the Nielsen et al. (2017) BASIN%Ro kinetic model for
2014b), both in spatial extent and maximum net thickness. While a maturity only partially alleviated this over-maturity. A more complex
primary goal in this effort was to extend these maps into Montana, the multi-stage erosion model was then implemented, guided by the
results indicate that little source rock from the Tyler is present in following principles: 1) Laramide tectonism is responsible for final
Montana. An exception is a pod of source rock from the lower part of the modification of the structural shape of the modern basin (i.e., the final
Tyler along the Cedar Creek anticline, which should be observed in core sense of offset on major faults, and relief from the center to the borders
to confirm its presence. Each net-thickness map was utilized to create of the basin were established in the Paleogene); and 2) approximately
layers in the basin model for the respective Tyler source rocks, with 700 feet of deposition and subsequent erosion of late Eocene and
TOC, HI, and kinetics set defined in Table 3 and based on reported Neogene sedimentary rocks are suggested based on erosional remnants
programmed pyrolysis data. of Eocene to Oligocene White River Group and early Miocene Arikaree
Formation at Kildeer Mountain, North Dakota. The interpreted modern
3.1.6. The Permian–Triassic–Jurassic evaporites structure map of the Cretaceous Mowry Formation was utilized to
Prominent evaporate sequences exist in Permian, Triassic, and restore the shape of the same surface in the Paleogene and thus estimate
Jurassic stratigraphic sections within the basin. These include the the amount of missing upper Cretaceous and Paleogene rocks. This
Permian Opeche Shale, the Pine Member of the Permian–Triassic resulted in an average deposition between the Upper Cretaceous and the
Spearfish Formation, and the Dunham Salt Member of the Middle Eocene of 1700 additional feet of presently missing sedimentary rock
Jurassic Piper Formation. Similar to the evaporites of the Madison (ranging from 1000 to 2500 feet). During the Eocene, 1200 feet of this
Group, net thickness maps were created using density logs. Halite was section was eroded. Another 700 feet of rock was deposited in the late
identified in these units using a density log cutoff of 2,300 kg/m3 be Eocene and Neogene. After this, the remaining ~1200 feet were eroded
tween the tops of the Tyler and Spearfish Formations (quantifying the from 20 Ma to the end of the last Quaternary glacial maximum. This
combined salts of the Opeche Shale and Pine Member) and in the Piper more complex erosion model achieved a close match with the measured
Formation (quantifying the Dunham Salt Member). Anhydrite was maturity data, which is discussed in the next section.
negligible in these units.
3.3. Thermal model calibration
3.2. Characterizing erosion events
Heat flow from the crystalline basement and lithospheric mantle into
Erosive events have had an impact on the maturation history of the sedimentary column, termed basal heat flow in basin modeling
proven and postulated source rocks in the Williston Basin. While studies, must be estimated to calibrate a 3D thermal model to modern
numerous erosive events have occurred, only those where the estimated subsurface temperature data. This study utilized a variety of borehole
eroded or missing section is greater than 500 ft and where the age of the temperature measurements for this purpose. The highest confidence
erosive event is close to that of maximum burial will be considered. dataset was provided by the North Dakota Geologic Survey and includes
Several continental-scale sequence boundaries were identified by Sloss 24 high resolution static temperature logs spread across North Dakota
(1963), as well as other regional or local unconformities. These un (Nesheim and Stolldorf, written communication; Stolldorf, 2019, 2020).
conformities, while stratigraphically significant, have missing sections These 24 wells include not just temperatures at great depths, but
of less than 500 ft or do not occur close to the time of maximum burial. continuous shallow temperature gradients as well. Due to compaction of
Unconformities in the Cenozoic, associated with Laramide tectonism sediments, geothermal gradients will commonly increase with depth,
and Quaternary glacial erosion, do have missing sections that are likely leading to a concave-down curve of temperature versus depth. The
greater than 500 ft and occur at or near maximum burial in the Late North Dakota temperature logs generally display this trend, but curi
Cretaceous and Cenozoic. ously, the shallowest section has an almost linear, or even inverted
Throughout the Cretaceous, an interior seaway spanning North (concave-up), temperature gradient that extends approximately to the
America covered the Williston Basin and was the setting for deposition depth of the Lower Cretaceous Inyan Kara Formation. A thick Cretaceous
of a thick succession of marine shales that compose the first few thou shale package overlies the Inyan Kara and low thermal conductivity
sand feet of stratigraphic section in the basin (Blakey, 2014). In Late values for these shales have been previously required to match the very
Cretaceous to Eocene time, the Laramide orogeny uplifted mountain shallow geothermal gradient observed in this section (Gosnold et al.,
ranges throughout the North American Cordillera and created basins 2012). The geothermal gradient deeper than the Inyan Kara is steeper
with thick accumulations of sediments, including the Powder River and more easily matched with default thermal conductivity properties
Basin southwest of the Williston, which is partly bounded by the Black included in standard basin modeling software. An alternative explana
Hills uplift (e.g., DeCelles, 2004). The Williston Basin remained part of tion for this shallow linear geothermal gradient may be the effect of
the last reaches of the interior seaway into the early Paleogene before surface temperature cycling during Quaternary glaciation and inter
Laramide tectonism finally affected the basin in the Eocene (Blakey, glacial periods. The timescales of this glaciation and resultant surface
2014). The Eocene unconformity is related to a significant increase in temperature cycling are typically much shorter than basin modeling
elevation in response to structural activation of the basin. Subsequent timescales (tens of thousands of years versus millions of years). To
Neogene sedimentation is rarely preserved, although fluvial and lacus address this, an average low surface temperature (38 ◦ F, defined as an
trine upper Eocene and Oligocene units are occasionally observed as intermediate value between freezing/glacial and
erosional remnants (Li and Fan, 2018). The amount of missing section present-day/interglacial mean annual surface temperature) was incor
associated with the Eocene unconformity is unknown in the Williston porated into the basin model throughout the Quaternary until the end of
Basin, but it plays a significant role in the maturation history of all the last glacial maximum at 0.02 Ma, in an attempt to replicate the
modeled source rock units due to its influence during peak burial. thermal conduction effect of glaciation on the geothermal gradient.
In a prior 3D basin modeling study, Kuhn et al. (2012) estimated 2, Upon assigning Cretaceous shales a low grain thermal conductivity and
000 feet of missing section associated with Paleogene erosion in their incorporating a low surface temperature throughout the Quaternary, an
base case model, along with another 300 feet of missing section asso acceptable match between modeled and measured temperature was
ciated with Quaternary glaciation. Early simulations of the 3D model in obtained; results of the temperature log thermal calibration curves are
12
available in the Supplementary Material. from points to a map, heat flow values were compared with several data-
To supplement and infill the temperature log data, a large database based or interpretation-based maps related to basement geology,
was compiled with drill stem test (DST) temperatures and bottom hole including gravity maps, magnetic anomaly maps, as well as basement
temperatures (BHT) from boreholes throughout Montana, North Dakota, composition and terrane maps. Following Bader (2019), lineaments
and South Dakota (IHS Markit, 2022). Basin modeling software is typi mapped on a magnetic anomaly map (Bankey et al., 2002) provided an
cally designed to calibrate temperature profiles on a well-by-well basis, additional platform to interpret the underlying cratonic suture zone
yet the large DST and BHT databases contained >7000 and > 14,000 beneath the Williston Basin (Fig. 11B). These lineaments were particu
wells, respectively, posing a challenge to incorporate such a dataset. The larly insightful in geologically guiding interpolation of these basal heat
DST data were first plotted with depth to estimate a basinwide flow values into a map, shown in Fig. 11C.
geothermal gradient, shown in Fig. 9A. Most basins have geothermal Previous work in the Williston Basin recognized a high heat flow
gradients that vary spatially; to better understand these spatial varia anomaly along a north-south trend in western North Dakota that extends
tions, the DST data were colored according to the difference between into Canada, coinciding with the North American Central Plains Con
each temperature point and the basin average (Fig. 9B), and then plotted ductivity Anomaly (NACPCA) (Jones and Savage, 1986; Majorowicz
spatially (Fig. 9C). While considerable scatter exists amongst these data, et al., 1986, 1988; Jones and Craven, 1990). However, this feature has
the following spatial trends were identified during the calibration pro been controversial historically. Notably, Gosnold (1990) suggested that
cess: 1) geothermal gradients higher than average are observed in a this heat flow anomaly should be explained by lower-than-presumed
north-south trend in western North Dakota, and 2) a point feature of thermal conductivity of the thick Cretaceous shales in the basin rather
extremely high temperature gradient is observed in eastern Montana. than basal heat flow, with a further temperature gradient anomaly in the
To better manage these datasets, the original well data (Fig. 10A) shallowest 1 km of sediment caused by post-glacial warming (Gosnold
were grouped into pseudo-wells on an evenly spaced grid throughout the et al., 2012). Nevertheless, previous basin modeling studies have
basin (Fig. 10B). This reduced the number of wells for analysis and required high basal heat flow for thermal calibration along the NACPCA,
amplified the geothermal trends in areas with dense well spacing. typically ~65–70 mW/m2 (Burrus et al., 1996a; Kuhn et al., 2012). This
However, even with this amplification of geothermal gradient, scatter study assigned low thermal conductivity to the Cretaceous shales which
remained in both the BHT and DST datasets. BHT data often underes did reduce the magnitude of basal heat flow overall (by ~5 mW/m2) but
timate formation temperatures and generally must be corrected (e.g., did not change the spatial distribution of these heat flow trends. The
Bullard, 1947). Unfortunately, necessary data for these corrections, elevated heat flow trend along the NACPCA and an additional zone of
including temperatures recorded from multiple logging runs at similar high heat flow near Poplar dome persist in this study’s basal heat flow
depths and times since circulation stopped, were inconsistent and often map (Fig. 11C).
unavailable for this dataset. DST data typically are expected to be closer While the present-day calibrated heat flow map in Fig. 11C reflects
to formation temperature, although it is assumed that inaccurate low thousands of modern temperature data, it may not be an appropriate
measurements are far more common than inaccurate high readings. To map for the full geologic history of the Williston Basin. Early versions of
reconcile these short-comings with such a large dataset, rather than fit a the basin model utilized a trend of constant heat flow through geologic
geothermal gradient through the center of the data, a high-side fit was time, set to the map in Fig. 11C, coupled with a very low estimate for
estimated for both the BHT and DST pseudo-wells. An example of a Paleogene erosion (500 feet). Upon increasing the estimate of both
calibrated pseudo-well near the Nesson anticline is shown in Fig. 10C. Paleogene and Neogene erosion as discussed in the previous section, a
The calibrated basal heat flow at each well (or pseudo-well) was variable heat flow history was also introduced, particularly focusing on
plotted spatially to construct a basal heat flow map applicable for the the effect of the Laramide on lithospheric heat flow. The Laramide is
entire AOI (Fig. 11A). Several spatial trends in heat flow were observed, presumed to be associated with a period of low-angle or flat slab sub
including a north-south, high heat flow trend in western North Dakota duction along the western margin of North America, which may imply a
(~65 mW per meter squared, mW/m2), an area of high heat flow in “refrigeration” or cooling effect on the continental lithosphere of the
eastern Montana (~70 mW/m2), low heat flow values in the far north North American cordillera (Dumitru et al., 1991). While very poorly
eastern corner of Montana (50 mW/m2), and generally low heat flow in constrained, this motif was incorporated into the basin model by shifting
central North Dakota (~45 mW/m2). To guide interpolation of heat flow the present-day heat flow map lower by 10 mW/m2 from 90 to 50 Ma. A
Fig. 9. Compilation of drill stem test temperatures basinwide. A: All data plotted with depth, and a representative geothermal gradient for the basin is shown in
black. B) The same data as A, but each temperature data point is colored according to how much hotter (red) or cooler (blue) it is compared to the representative
gradient for the basin. C) A map distribution of the data in B. While scattered, some warm and cool geothermal gradient trends are observed (for example, a higher
gradient trending north-south in western North Dakota).
13
Fig. 10. Conversion of individual wells with drill stem test data, distributed in over 7000 wells basinwide, to a series of pseudo-wells on an even grid. A) Spatial
distribution of temperature data available for calibration from individual wells. B) Spatial distribution of aggregated temperature data as pseudo-wells. This reduced
the number of wells for analysis during thermal calibration and amplified geothermal gradient trends in data-dense areas. C) The lumped temperature data from one
pseudo-well near the Nesson anticline, with the calibrated model temperature gradient shown in as a black line. Mission C., Mission Canyon Formation. DegF, degrees
Fahrenheit; ft, feet.
linear interpolation between this lower basal heat flow and the present determined a “North Dakota” model, 2) Jin and Lewan (2017) deter
basal day heat flow map was then assigned from 50 to 43 Ma, based on mined a Bakken-specific model, and 3) Liu et al. (2020) utilized kerogen
relatively early retreat of the Laramide in this very distal position of the pyrogram information to infer kinetic models for both the upper and
orogeny (Copeland et al., 2017). lower Bakken shale members. A comparison between these models is
While wellbore temperatures are used to constrain the modern shown in Fig. 12. The Pepper and Corvi (1995) reaction cracks kerogen
thermal regime in the Williston Basin, they do not inform paleo- to hydrocarbon over a broad temperature window, whereas the Jin and
geothermal gradients. Maturity data is typically used for this purpose, Lewan (2017) model reacts over a very short temperature window—the
as it is sensitive to a time-integrated temperature history. Unfortunately, result of a reaction composed of a single activation energy. Intermediate
vitrinite reflectance data throughout much of the Paleozoic section in between these is the model of Tegelaar and Noble (1994). Comparing
the Williston Basin have experienced vitrinite suppression (Price and the temperatures at which the reactions occur, the Pepper and Corvi
Barker, 1985; Price et al., 1986). To work around this limitation, a (1995) model reaches a 50% transformation ratio at the lowest tem
recent study by Abarghani et al. (2019) has determined solid bitumen peratures, the Tegelaar and Noble (1994) model at intermediate tem
reflectance values in the Bakken, which were converted to an equivalent peratures, and the Jin and Lewan (2017) model generates hydrocarbons
vitrinite reflectance (%Roe). These data can be compared with various at the highest temperatures. With an externally constrained
vitrinite reflectance kinetic models to evaluate the modeled temperature time-temperature history for the Bakken (via calibrated temperature
history of the basin. Assuming the more complex Paleogene and and solid bitumen reflectance), each of these kinetic models was simu
Neogene erosion model described in Section 3.2 and the variable heat lated to determine the best match with the spatial distribution of HI for
flow history throughout the Laramide described above, the Nielsen et al. the Bakken, which was used as a proxy for the progress of catagenesis.
(2017) BASIN%Ro kinetic model provided a very close match to the data Individual results of each of these kinetic models are provided in the
from Abarghani et al. (2019), except for the Abarghani et al. (2019) Supplementary Material and are discussed here in terms of the differ
Wells 1, 2, and 3, all located in the far northeastern corner of the AOI. ence between modeled and measured HI. If this difference is positive,
These are also the most immature wells in their study. While an increase then modeled HI exceeds measured HI, and the model is considered
in eroded section in the far northeastern area of the model was briefly immature relative to the data; if the difference is negative, modeled HI is
considered, ultimately, the discrepancy between the immature model too low, the reaction has thus progressed too far, and the model is
and the very early mature data was not deemed robust enough to alter considered over-mature relative to the data. The Pepper and Corvi
the model inputs. The final calibration plots of these data to the model (1995) model was severely overmature in comparison with measured
are provided in the Supplementary Material. HI. The Tegelaar and Noble (1994) model was also overmature, but the
difference was lower in magnitude. The Jin and Lewan (2017) model
was both over-mature and under-mature—in central portions of the
3.4. Calibration of Bakken hydrocarbon generation using programmed
basin the reaction progressed too quickly and seemingly outpaced the
pyrolysis data
data, while in more peripheral areas of the basin it progressed too slowly
and was immature. The difference between modeled and measured HI
Programmed pyrolysis data have been used previously to constrain
for the Jin and Lewan (2017) model was bimodal, indicating that the
the maturity of the shales of the Bakken Formation (Kuhn et al., 2012;
timing of the reaction matched the data, but reacted over too narrow of a
Jin and Sonnenberg, 2013). With the time-temperature and burial his
temperature range.
tory of the Bakken constrained, this study uses the programmed pyrol
A hybrid kinetic model was created from a combination of the timing
ysis data (obtained from the USGS EGDB), initially presented in Section
of the Jin and Lewan (2017) model and the width of the Tegelaar and
3.1.3 and Fig. 5, to constrain the kinetic model for hydrocarbon gener
Noble (1994) model. This hybrid is referred to as “Tegelaar-100” and it
ation in the Bakken. In addition to generic generation kinetics for Type II
uses the activation energy distribution from the Tegelaar and Noble
marine source rocks (Pepper and Corvi, 1995), three Bakken-specific
(1994) model with a frequency factor of 1027 myr− 1, calibrated to pass a
kinetic models have been reported: 1) Tegelaar and Noble (1994)
14
Fig. 11. Overview of creation of the gridded basal heat flow calibration map.
A) Basal heat flow interpretation for all temperature datasets utilized for
thermal model calibration, including North Dakota Geologic Survey static
temperature logs (NDGS Temp. Log), and basinwide drill stem test (IHS DST)
and bottomhole temperatures (IHS BHT) in pseudo-wells. Colors indicate the
heat flow pick while size of the symbol indicates the confidence in the pick
(larger indicates higher confidence). B) Magnetic anomaly map of North
America (Bankey et al., 2002) with lineaments interpreted after Bader (2019).
GD: Glennie domain; TFFZ: Tabbernor fold and fault zone; FFB: Flin Flon belt;
SBZ: Superior boundary zone. C) Final calibrated basal heat flow map utilized
in the basin model, which was gridded using the heat flow picks from A with
guidance from the magnetic lineaments in B. High heat flow regions include a
north-south trend in western North Dakota and around the Poplar dome area of
eastern Montana. Unit mW/m2 denotes milliwatts per meter squared.
50% transformation ratio at a similar thermal stress as the Jin and

Lewan (2017) model. This hybrid model is compared with the other
kinetic models in Fig. 12. The comparison between modeled and
measured HI values using this hybrid model is shown in Fig. 13A, while
the difference between modeled and measured HI is shown in Fig. 13B.
The hybrid kinetic model predicts HI to within 200 mg HC/g TOC, but
still exhibits some discrepancy. In addition to the hydrocarbon genera
tion kinetics, variation in initial HI may exist throughout the basin and
uncertainty in paleo heat flow or spatial variations in eroded section
may partly (or wholly) be responsible for this discrepancy in predicted
HI. Future work may attempt to follow the method of Chen et al. (2017)
and quantitatively invert the kinetics from the programmed pyrolysis
data. Any remaining discrepancies after performing this procedure
would likely be attributable to these other factors rather than purely the
generation kinetic model. For this 3D modeling study, however, the final
kinetic model for the Bakken source rock is the hybrid “Tegelaar-100.”
4. Model results and implications for the stacked petroleum

systems of the Williston Basin
While the Williston Basin has experienced a long sedimentary history

punctuated by several episodes of subsidence, it was the increased
sedimentation in the Cretaceous western interior seaway and subse
quent Laramide uplift and erosion that most dramatically influenced the
maturation history of source rocks in the basin. A characteristic burial
history plot in Fig. 14 illustrates periods of increased subsidence in the
Ordovician–Silurian (corresponding to deposition of the Red River and
Interlake Formations), the Late Devonian–Mississippian (the Madison
Group), and in the Late Cretaceous (Mowry and Pierre Formations). The
following sub-sections will describe basin model results underpinned by
this burial history, including: 1) time-temperature paths and maturation
histories of major source rock intervals, 2) a comparison of modeled
timing and extent of hydrocarbon generation between these source in
tervals, and finally 3) integration of maturation and migration modeling
with observed spatial production trends in the basin. Present-day sub
surface temperature maps for all formations and maturity maps for all
source rock intervals are provided in the USGS data release (Gelman and
Johnson, 2023).
4.1. Time-temperature and maturation history
All the Williston Basin source rocks experienced a time-temperature

history dominated by Late Cretaceous and Paleogene burial and subse
quent Neogene cooling. In Fig. 15, the time-temperature history of the
Bakken source rocks near the basin center clearly shows steep increases
in temperature during deposition in the Mississippian, Late Cretaceous,
and Paleogene, followed by a temperature decline throughout the
Neogene. This cooling trend was driven by three contributing factors: 1)
(caption on next column) Eocene erosion, 2) surface uplift from sea level to the present-day
elevation (with a corresponding decrease in surface temperature), and
3) general cooling of mean Earth surface temperatures from the
15
This genetic time-temperature history results in a maturation history

characterized by hydrocarbon generation concentrated in the Creta
ceous to Paleogene, followed by a sharp decline in generation during the
Neogene to present. Illustrating this, modeled vitrinite reflectance
through time in the Bakken shale members increases steeply during
burial in the Late Cretaceous and Paleogene, then levels off during uplift
and erosion in the Neogene (Fig. 15). Modeled hydrocarbon generation
tracks this history, signifying that if any hydrocarbons are being
generated today, it is at a much slower rate than in the past.
Maturity of all source intervals increases with depth and is strongly
influenced by areas of high basal heat flow (Fig. 16). The deepest source
rock investigated in this study, the informal Government Creek shale
within the Icebox Formation (Fig. 16A), ranges from peak oil to dry gas
maturity. It is most mature in Billings County, North Dakota, directly
over the region of highest basal heat flow in the North Dakota portion of
the AOI, although the unit is thin in this location. At its depocenter, on
the eastern flank of the Nesson anticline, vitrinite reflectance (%Ro)
ranges from 1.2 to 2%, corresponding to a transitional wet gas to dry gas
maturity window. Due to their younger age and slightly shallower cover,
Fig. 12. Compilation of hydrocarbon generation kinetics for the Bakken shale the kukersite beds of the Red River Formation (Fig. 16B) are slightly less
members. Tegelaar-100 indicates the final, hybrid kinetic model used in this mature, with maturity ranging from the early oil window (0.67 %Ro) on
study (described in Section 3.4). Tegelaar denotes the “North Dakota” model of the northern and southern edges of the AOI to the dry gas window (2.2 %
Tegelaar and Noble (1994); Pepper and Corvi TII(B) denotes the generic Type II
Ro) in the deepest portion of the basin. At shallower burial depths, the
model of Pepper and Corvi (1995); Jin and Lewan denotes the Bakken-specific
shales of the Bakken (Fig. 16C) are generally in the oil window in much
model of Jin and Lewan (2017); and Liu Upper Bakken and Liu Lower Bakken
indicate the respective models of Liu et al. (2020).
of the basin; the highest maturity regions overlie areas with high heat
flow along the north-south NACPCA trend in North Dakota and near
Poplar dome in Montana (up to 1.4 %Ro). At Parshall field, modeled
Paleocene-Eocene Thermal Maximum through the Holocene. The Icebox
maturity is only 0.65 %Ro, decreasing to the east, which corresponds to
Formation source rock, kukersites in the Red River Formation, source
the very early oil window. This is consistent with the programmed py
horizons within the Madison Group, and Tyler Formation source rocks
rolysis data in Fig. 13. The Madison Group (Fig. 16D) source rock was
all experience similar time-temperature paths, albeit shifted to higher or
presumed to be present throughout the AOI as disparate and heteroge
lower temperatures depending on their relative stratigraphic position.
neous thin intervals and is in the oil window in the center of the basin.
Fig. 13. Calibration of hydrocarbon generation ki

netics for the Bakken shale members to programmed
pyrolysis data (U.S. Geological Survey Energy
Geochemistry Database). A) Modeled (gridded)
versus measured (data points) hydrogen index (HI)
values, illustrating the good spatial match between HI
predicted by the basin model (using the hybrid
Tegelaar-100 kinetics) and measured data. White
outline (annotated “P”) indicates outline of Parshall
field. B) Map of the difference between modeled and
measured HI. Positive values indicate that the model
is less mature than the data, while negative values
indicate that the model is more mature than the data.
A histogram of the difference for each data point is
shown on the right of the color bar. This difference
was analyzed for each kinetic model tested in this
study to determine an optimal kinetic model for the
Bakken. Unit mg HC/g TOC, milligrams of hydro
carbon per gram of total organic carbon.
16
Fig. 14. Typical burial history plot for the Williston

Basin, taken from the center of the basin. Maturity is
calculated using the Basin%Ro kinetic model of
Nielsen et al. (2017). Each of the five source rock
intervals investigated in this study is highlighted
with a bold black line. The deepest sources, the
Icebox and Red River Formations, enter an early oil
window during the late Paleozoic, but only reach the
peak oil window and the wet gas window in the
Cretaceous and Paleogene. The shallower source
rock intervals in the Bakken Formation, Madison
Group, and Tyler Formation, enter the oil window in
the Cretaceous, and reach peak generation in the
Paleogene before uplift and erosion halt maturation
in the late Neogene. Ordov., Ordovician; Sil., Silu
rian; Miss., Mississippian; Pen., Pennsylvanian;
Paleog., Paleogene; Neo., Neogene; src, source rock
interval; Ma, millions of years before present; ft, feet.
4.2. Comparative generative volumes of the Williston source rocks
Combining source presence mapping, source rock thickness, rich

ness, and modeled maturity, a best estimate of the hydrocarbon gener
ative volume for each source rock was calculated. Like many estimates
of generative hydrocarbon volumes, these estimates remain uncertain.
Thicker or thinner organic-rich intervals than modeled, under or over-
estimates of TOC or HI, and, of particular significance, mapping of the
geographic extent of source rock intervals can all lead to significant
uncertainty. Nevertheless, these estimates provide a relative sense of the
robustness of each of the modeled source intervals and give context to
historical production volumes, as well as potential future producible
volumes, of hydrocarbons in each respective petroleum system.
By far, the largest volume of hydrocarbons generated in the Williston
Basin is attributed to the upper and lower shale members of the Bakken.
Calculations from the basin model indicate 240 billion barrels of oil
equivalent (BBOE) were generated from the lower shale of the Bakken,
while the upper shale of the Bakken generated 220 BBOE, for a com
bined total of 460 BBOE. This calculation is in general agreement with
the volume of oil previously estimated for the Bakken of 413 BBOE by
Fig. 15. Time-temperature, maturation, and hydrocarbon generation history of Leigh Price (written communication, 2000). However, cumulative hy
the Bakken Formation for a typical well in the center of the Williston Basin. drocarbon generation volumes based on numerous model simulations
Source rock maturation, displayed as vitrinite reflectance (shown as solid black ranged from ~350 to ~475 BBOE, so while the Bakken is likely the best
curve), dramatically increases from the Late Cretaceous through the early characterized of the source rocks considered in this study, there remains
Paleogene in response to rapid burial. Hydrocarbon (HC) generation mass significant uncertainty in these results. Published estimates of generated
(shown as dashed green curve) follows a similar trend, while hydrocarbon
Bakken oil during the past five decades have typically increased with
generation rate (shown as dotted green curve) shows peaks both immediately
time, although significant variability exists due to improved data,
prior to Eocene erosion and prior to Neogene erosion, then finally declining to
near-zero at present-day due to continued uplift and erosion. The green shaded
calculation techniques, and understanding of the Bakken petroleum
region indicates the timeframe of hydrocarbon generation of the Bakken shale system. Dow (1974) estimated 10 BBOE was generated, Schmoker and
members. Basin%Ro refers to the kinetic maturation model of Nielsen et al. Hester (1983) estimated 132 BBOE, Webster (1984) estimated 92 BBOE,
(2017). Ma denotes millions of years before present; Mtons/cell denotes Meissner and Banks (2005) estimated 32 BBOE, Flannery and Kraus
megatons per modeled cell as a unit of hydrocarbon generation mass; and (2006) estimated 300 and Kuhn et al. (2012) estimated 320 BBOE. It is
mgHC/gTOC/Ma denotes milligrams of hydrocarbon per gram of total organic important to emphasize that all these estimates, including the model
carbon per million years as a unit of hydrocarbon generation rate. Carbonif., results reported in this study (460 BBOE), represent original hydrocar
Carboniferous; Paleog., Paleogene; Neo., Neogene. bon generation—not in place oil, nor reserves, nor undiscovered hy
drocarbon potential. In addition to the amount generated in the 3D
Maximum maturity in the modeled Madison source rock reaches 1.15 % model, an additional 720 BBOE was never generated in the Bakken
Ro and is focused over regions of high basal heat flow. Finally, the lower source rocks due to insufficient maturity. Obviously, the shale members
and upper source intervals of the Tyler Formation (Fig. 16E) are in the of the Bakken are “world class” source rocks.
early oil window in much of the area where they were mapped. Peak oil The second-most prolific source rock in the model is the Red River,
window maturity in the lower part of the Tyler source is observed in the which generated an estimated 130 BBOE. This large volume is heavily
deepest portions of the basin in McKenzie County, North Dakota, where influenced by its richness, both in TOC and HI, despite the limited
it reaches a maximum of 0.73 %Ro. The highest maturity in the upper thickness of the kukersite beds. Nevertheless, significant uncertainty in
part of the Tyler source is situated over the region of high basal heat flow net thickness and geographic extent remains. Based on the 3D modeling,
in Billings County, North Dakota, where it reaches a maximum of 0.78 % the source rocks of the Tyler, Madison, and Icebox were significantly less
Ro. prolific than the Bakken and Red River. The source in the upper part of
the Tyler generated 64 BBOE, the source in the lower part of the Tyler
17
Fig. 16. Compilation of maturity maps at present-day

for the five source rock intervals investigated in this
study. Regions in which source rocks are not inter
preted to be present are partially covered, indicating
that hydrocarbons are unlikely to be generated in
these areas despite sufficient maturity. A) Maturity of
the informal Government Creek Shale of the Icebox
Formation. B) Maturity of the kukersite beds of the
Red River Formation. C) Maturity of the upper and
lower Bakken shale members. D) Maturity of the
source horizon in the Madison Group. E) Maturity of
the sources in the upper and lower parts of the Tyler
Formation. Basin%Ro denotes the kinetic maturity
model of Nielsen et al. (2017). Ma denotes millions of
years before present.
generated 30 BBOE, the Madison generated 44 BBOE, and the Icebox 4.3. Migration modeling: integrating model results with observed
generated 28 BBOE. The Tyler source rocks are relatively rich (Nesheim production
and Nordeng, 2014a, 2014b), despite their modest net thickness and
limited geographic extent. The model indicates they have significant To relate modeled hydrocarbon generation with observed spatial
generative potential. The estimate for the source rocks of the Madison is production trends, both a Darcy’s Law hydrocarbon migration simula
highly uncertain because it is the only source rock in this study not based tion, which focused on modeling the continuous Bakken Petroleum
on a regionally extensive published database of programmed pyrolysis System, and map-based migration, which focused on modeling con
data, with only the limited data of the Bluell source rock of Nesheim ventional migration pathways, were incorporated into the 3D model.
et al. (2021) available. This study did not presume that every thin shale While hydrocarbon migration processes are governed by assigned
in the Madison is an organically enriched source rock. Although porosity, permeability, and capillary entry pressure of lithofacies in the
numerous thin shaley beds were observed in wireline logs (Peterson, Darcy migration model, these parameters are poorly constrained, so this
1984), the model was constrained using one shale with associated data study’s migration modeling results are qualitative. Map-based migration
establishing some organic enrichment—albeit, extrapolated basinwide, is also subject to uncertainty since the source rocks in the Williston Basin
far beyond its reported extent. As mentioned in Section 3.1.4, future charged reservoirs predominantly during the Cretaceous and Paleogene.
work to describe these thin shale beds can help to improve the charac Therefore, any subsequent alteration to basin geometry would alter
terization of the cumulative Madison source rock generation potential. migration flow-path directions. Despite these limitations, some inter
esting broad-scale trends were observed and are described as follows.
18
The Winnipeg Petroleum System, as defined in the USGS 2020 “Eastern Margin” assessment units in the USGS 2020 Assessment
assessment of undiscovered resources, includes reservoirs in the Win (Schenk et al., 2021a). Some production is observed outside the extent of
nipeg Group and Deadwood Formation (Schenk et al., 2021a; Anna, the mapped kukersite beds south of the Nesson anticline, in Dunn,
2013). In the 3D basin model, the only source rock supplying hydro Billings, and McKenzie Counties, North Dakota. While these hydrocar
carbons to these reservoirs was the Icebox, consistent with oil bons might be supplied by the Icebox source, it should be noted that
geochemistry (Smith and Bend, 2004). While this source rock generated during the facies modeling described in Section 3.1.2, a very slight
only a modest volume of hydrocarbons in this study, Darcy migration broadening of the DT-LLD space considered for “kukersite candidacy”
modeling suggested that oil and gas were expelled from the source both would flag some wells with potential source presence in this region. This
downward and upward into the underlying and overlying Winnipeg was not considered in this study because Nesheim (2017) reported an
strata. A very small volume of Icebox hydrocarbons migrated into res absence of kukersite beds in three wells examined in core in this area;
ervoirs in the Red River in the model but is insignificant compared to the however, future core studies may help solidify whether intermittent
contribution of Red River-sourced hydrocarbons to that formation. kukersite beds are present to charge these reservoirs. Finally, production
Maturity of the Icebox at present-day is in the wet gas and dry gas GOR trends indicate more gas-rich fluids in the center of the basin than
windows in the center of the basin, which aligns with the high at the basin margin or at the Cedar Creek anticline (Fig. 17B), which
production-based gas-to-oil ratios (GOR) observed (Fig. 17A). Produc matches the modeled maturity for the Red River.
tion in the Winnipeg and Deadwood spatially overlaps with the source The Bakken Petroleum System includes reservoirs in the Bakken and
presence mapping from Nesheim and Nordeng (2013), indicating only Three Forks Formations, and carbonate mounds in the lowermost part of
short distance lateral or vertical migration from source to reservoir. the Lodgepole Formation, sourced by the upper and lower shales of the
The Red River Petroleum System includes carbonate reservoirs in the Bakken (Gaswirth and Marra, 2015; Marra et al., 2021). While Bakken
A through D zones, as well as in the overlying Ordovician Stony source rock is present and mature throughout the center of the basin,
Mountain and Stonewall Formations of the Bighorn Group and Interlake implying migration pathways to virtually the entire basin, the low
Formation (Schenk et al., 2021a; Anna, 2013). This model produced porosity and permeability of the Lodgepole effectively seals and isolates
abundant hydrocarbons in Red River reservoirs overwhelmingly sourced the petroleum system. However, forcing the Lodgepole to be such a
by the kukersite beds of the Red River Formation. Darcy migration competent seal was a challenge in the basin model. Since default car
modeling suggested that hydrocarbons migrated upward and downward bonate compaction curves predicted relatively high porosity (~10%) at
within the Red River, but the model failed to migrate significant hy depths of 10,000 feet, a cementation model was included to reduce
drocarbons into the overlying upper Bighorn Group and Interlake res porosity to 5% at that depth. Additionally, although detailed
ervoirs. However, lithologic properties were not a major focus for the porosity-permeability data were unavailable for calibration, perme
Interlake and Bighorn in this study, so this apparent lack of migration ability was reduced relative to default porosity-permeability trends and
may be more a limitation of the basin modeling inputs than a real feature the sealing capacity, governed by an assigned capillary entry pressure
of these petroleum systems. Notably, in this study and in Nesheim relationship with porosity, was increased. Some model simulations even
(2017), kukersite beds are not interpreted to be present east of the experimented with setting the Lodgepole as a numerical “perfect seal.”
Nesson anticline. In Fig. 17B, migration flow-paths from mature regions Even with these artificial parameters, invasion percolation and
where kukersite beds are present indicate hydrocarbon migration is low-resolution Darcy flow migration solvers allowed substantial Bakken
possible to southern, northern, and western reservoirs in the basin, but oil to migrate into the Lodgepole. A higher resolution Darcy flow
the region east of the Nesson anticline appears to be in a migration migration solver reduced this effect, but at the expense of significantly
shadow. This interpretation is similar to that of Khan et al. (2006) and is increased computational time. Finally, setting a value of two for Petro
consistent with the distinction between “Fairway” and much more risky Mod’s Petroleum Mobility Factor resulted in minimal Bakken oil
Fig. 17. Model results with production-based gas-to-oil ratio (GOR, with units of scf/bbl). A) The Winnipeg Petroleum System. Expelled hydrocarbons from the
informal Government Creek shale member of the Icebox Formation is shown on the map. Well production from the Winnipeg system overlays regions of interpreted
source presence. B) The Red River Petroleum System. Maturity where the kukersite beds are present is shown, with flow lines indicating possible migration directions.
The absence of source east of the Nesson anticline suggests a possible migration shadow on the east side of the basin. Well production GOR generally agrees with
maturity of the kukersite beds, with higher GOR corresponding to the center of the basin where higher maturity is predicted. Mtons, megatons; scf/bbl, standard
cubic feet of gas per barrel of oil.
19
migrating into the Lodgepole. This has the effect of increasing the The basin model ultimately matched several regional-scale criteria
effective permeability for Darcy migration to all lithologies by two log regarding the distribution of oil in the Bakken Petroleum System: 1) the
units. While changing capillary seal and permeability-porosity re model saturated all three members of the Bakken, up to a maximum of
lationships of the Bakken and Lodgepole were unable to effectively stop 65% oil saturation in the upper and lower shales, and ~70% saturation
Bakken oil from migrating into the Lodgepole, numerical settings such in the middle Bakken member along the Nesson anticline (Fig. 18C). 2)
as increasing the time resolution of the Darcy solver and this bulk shift Declines in modeled oil saturation and modeled oil-in-place were
towards increasing permeability suggest that numerical effects play an observed running northwest–southeast in McKenzie and Williams
important role in modeling the enormous volume of oil migrating within Counties, North Dakota, which coincide with a decrease in cumulative
the Bakken system. Comparing the effect of including versus excluding oil production in the area (Fig. 18A). In the basin model, this decline in
this numerical Mobility Factor, the final versions of the model migrated saturation and in-place oil is caused by a reduction in reservoir quality
120 (Mobility Factor set to two) versus 5,000 (Mobility Factor set to associated with the transition from “distal intermittent facies D″ to the
zero) million barrels of oil into the Lodgepole. region where facies D is consistently absent (“silt/shale” in Fig. 6),
To better understand the trapping mechanism at Parshall field, a suggesting there may be a relationship between broad-scale reservoir
numerical experiment was conducted to evaluate two competing facies distributions and well productivity. 3) Oil was expelled downward
mechanisms. As discussed in Section 3.1.3, the trap on the east side of from the lower Bakken shale member into the underlying layers of the
Parshall field (Fig. 18A) has been suggested to either be due to a pinch- Upper Devonian Birdbear, Three Forks, and Duperow Formations,
out of high-quality reservoir facies within the middle Bakken member consistent with previous observations (Nesheim, 2019). Of the modeled
(Sonnenberg et al., 2011, 2017) or a reduction in secondary porosity, 461 BBOE generated by the shales of the Bakken, 14 BBOE remained as
associated with organic acid dissolution, at the boundary between free oil in the pore space of the Bakken shale members, 15 BBOE was
mature and immature Bakken source rock (Johnson, 2009; Theloy, adsorbed in the shales, 125 BBOE was expelled into the middle Bakken
2013). To test these two hypotheses, the model was first constructed member, 302 BBOE was expelled downward into the underlying Three
with a facies boundary where the middle Bakken member facies D Forks and Birdbear, and another 4 BBOE was expelled further downward
re-emerges east of Parshall field (Fig. 18B), shown as the “proximal into the Duperow. The remainder of 0.2 BBOE was migrated into the
intermittent D″ facies in Fig. 6B. A second alternative scenario was then Lodgepole (for the simulation setting Mobility Factor to two). Of the oil
constructed by incorporating a new diagenetic model conditioned to in the middle Bakken member, ~2 BBOE had migrated out of the model
create secondary porosity once the shale members of the Bakken become boundary into Canada, with this migration rate increasing throughout
mature (Fig. 18C). In this relatively coarse test, the facies-based reduc the Cenozoic until the end of the model simulation. While subject to
tion in reservoir quality produces a sharp boundary at which oil satu much uncertainty, this regional-scale result satisfies the observations of
ration decreases at a discrete edge, whereas in the maturation-based an oil-saturated Bakken Formation, downward expulsion into the Three
model, oil saturation generally decreases more diffusely and broadly. Forks and Birdbear, and a mostly self-contained petroleum system, with
The difference between the facies boundary model and the diagenetic some oil migration into the Canadian portion of the basin, where it is
model is caused by maturation transitioning over a time-temperature produced from both Bakken and Madison reservoirs (Jiang and Li, 2002;
window rather than at a discrete thermal stress. In terms of the phys Chen et al., 2009). Future work documenting porosity-permeability and
ical process this represents, organic acids produced during early matu capillary entry pressure relationships for all the formations involved in
ration are unlikely to discretely disappear and produce a sharp trap the Bakken Petroleum System (source, reservoir, and seal), as well as
boundary. In terms of oil saturation, the sharp trapping boundary in the incorporating over-pressure modeling, fracturing and their effects on
facies-based trap produced higher hydrocarbon saturations (~65%) in permeability and fluid migration, and improved regional-scale facies
the area of Parshall field, while the more diffuse maturation-based trap mapping can help to improve the reasonability and reduce uncertainty
produced only 55–60%. While data are not currently available in this in these calculations.
study to examine the trap edge in detail, the qualitative results of this The Madison Petroleum System includes reservoirs in the Mission
numerical experiment may assist in future interpretations of these Canyon, Charles, and Spearfish Formations (Gaswirth et al., 2013;
processes. Schenk et al., 2021b). The source of oils for these formations has been
Fig. 18. Integration of (A) cumulative oil production in the Bakken with (B and C) modeled oil saturation of the middle Bakken member. Regions of highest cu
mulative oil production include the western side of the Elm Coulee field (labeled “EC”), Parshall field (labeled “P”), and the center of the basin in general. A reduction
in cumulative oil production coincides roughly with the interpreted absence of facies D. This boundary trends approximately NNW-SSE in the center of McKenzie
County, North Dakota (“silt/shale facies” from Fig. 6). Modeled oil saturation in the middle Bakken member is generally between 50 and 75% in regions where the
Bakken shale members are mature. Two scenarios were tested to interrogate the trapping mechanism at Parshall field: a facies-based trap (B) and a maturity-based
trap (C). While a decline in oil saturation occurs east of Parshall field in both scenarios, the facies-based scenario results in a sharper boundary and higher hy
drocarbon saturation in the productive areas of Parshall field. The maturity-based trap results in a more gradational boundary and lower hydrocarbon saturation at
Parshall field. Outlines of the implemented facies model from Fig. 6 are shown in black.
20
debated over several decades. Prior to the 1990s, oil in Madison reser estimated Bakken generative volume down (~400 BBOE), at least 100
voirs was thought to have been generated by Bakken source rocks (Dow, billion barrels of Bakken oil may still remain unaccounted for. This oil
1974; Williams, 1974). More recent studies based on oil geochemistry might have migrated laterally up-dip towards Canada, downward into
have concluded that Madison oils are self-sourced, particularly in the U. the Duperow Formation and other Devonian units, or stratigraphically
S. part of the basin (Price and LeFever, 1994; Jarvie, 2001; Lillis, 2013). upwards into the Madison—or a combination of all three.
Controversy remains, however, since some Bakken-sourced oils have Based on Madison spatial production trends, many fields are found
been identified in Madison reservoirs particularly in fractured areas in near fracture zones. In addition to the Nesson anticline and Poplar dome,
both the U.S., mainly along the Poplar dome and Nesson anticline where mixed Bakken-Madison oil has been confirmed (Jarvie, 2001),
(Jarvie, 2001), and Canada (Chen et al., 2009). Additionally, mixing is Madison production also occurs over the Little Knife anticline, the
particularly difficult to identify because of the significantly different Billings Nose anticline, the Red Wing Creek paleo-impact structure, and
concentrations of biomarkers between the two oil families, obscuring the Cedar Creek anticline—all areas where fractures and vertical
mixing estimates (Jiang and Li, 2002). Oil geochemical analyses on the migration pathways are likely (Fig. 19A). Furthermore, Chen et al.
lighter hydrocarbon fraction, which is abundant in both Bakken and (2009) identified mixed Bakken-Madison oils on the Canadian side of
Madison oils, have shown significant overlap between the highly het the basin along the northeast-southwest striking Torquay-Rocanville
erogeneous Madison oils and oils from the Bakken (Obermajer et al., fracture zone, indicating that even subtle regional lineaments may
2000). provide sufficient fracturing for Bakken oil to migrate into Madison
Prior to the emergence of the Bakken continuous resource play in the reservoirs. In the U.S. side of the basin, the Mondak field is likewise
late 2000s, it was a challenge to imagine that the most prolific source located along a regional lineament, called the Mondak fracture system
rock in the basin had not contributed significant oil to overlying Madi (Gerhard et al., 1990), although on subsurface maps it is spatially distant
son reservoirs, which were responsible for producing the most oil in the from obvious structural features. The reservoir of the Mondak field is
basin at the time. This was especially troubling since the Madison was fractured limestone in the Mission Canyon Formation, despite the trap
considered organic-lean in general, with published reports indicating not being structural (Parker and Hess, 1980). Conspicuously, it overlies
that organic-rich rocks were mostly confined to the lower or middle a poorer-producing section of the Elm Coulee trend, which may suggest
parts of the Lodgepole (Price and LeFever, 1994; Osadetz and Snowdon, hydrocarbon leakage out of the Bakken system within these regional
1995; Burrus et al., 1996a). In fact, this absence of Bakken oil, which was fracture zones, even far from anticlinal structures.
theoretically generated by the Bakken source rocks but seemingly did As described previously, this basin model tended to migrate Bakken
not migrate into the Madison, was the basis for early proposals of the oil into the Lodgepole Formation, despite assigning high capillary entry
Bakken resource play (Price and LeFever, 1994). It has now become pressure and low permeability rock properties to this sealing layer.
clear that the Bakken and underlying Devonian formations indeed store Burrus et al. (1996a, 1996b) also describe a significant portion of
a significant volume of Bakken oil, as evidenced by production from Bakken oil leaking into the Madison in their modeling, diffusing into the
these continuous resource plays (Table 1) and recent assessed undis Lodgepole Formation as a low saturation fluid. The loss of oil to the
covered resource potential (Marra et al., 2021). Gherabati et al. (2019) Lodgepole was also discussed by Kuhn et al. (2012), as it influenced the
estimated original oil in place in the Bakken Petroleum System, distance oil was able to migrate laterally in the middle Bakken member
concluding that the combined Bakken and Three Forks have stored 262 in their model. The complete isolation of Bakken oil from the Madison
billion barrels of oil. Rounding that number up (~300 BBOE), and the seems improbable in the context of these studies. Although none of these
Fig. 19. Model results with production-based gas-to-oil ratio (GOR, with units of scf/bbl). A) The Madison Petroleum System. Source rocks are generally poorly
constrained but likely composed of several heterogeneous discrete horizons, and are simplified as a thin cumulative layer throughout the modeled area of interest.
Major structural features in which Madison production is observed are labeled. PD: Poplar dome; N: Nesson anticline; RW: Red Wing Creek paleo-impact structure;
LK: Little Knife anticline; BN: Billings Nose anticline; CC: Cedar Creek anticline. The Mondak field discussed in Section 4.3 is labeled. The edges of the halite (CH) and
anhydrite (CA) in the Charles Formation are labeled, and the pinch-out of these salt layers may provide migration pathways for oil to reach the Charles and Spearfish
reservoirs in Renville and Bottineau Counties, North Dakota. Note that the far northeastern region of the AOI is generally immature, even for a Type IIS source rock
(0.5–0.55 %Roe), implying lateral migration into the area. Finally, map-based migration drainage areas are shown in black. B) The Tyler Petroleum System. The map
shows expelled hydrocarbons from the combined source intervals in the lower and upper parts of the Tyler Formation. Flow lines indicate possible migration di
rections from mature kitchens. Oil-prone production with low GOR is primarily located in the kitchen for the source in the upper part of the Tyler, in southwestern
North Dakota. Mtons, megatons; scf/bbl, standard cubic feet of gas per barrel of oil.
21
modeling studies have incorporated fractures and faults in the Madison, (Jarvie, 2001). This suggests that high proportions of Bakken oil are
the existence of such vertical migration conduits would more easily required for recognition amongst the more biomarker-concentrated and
allow Bakken oil to migrate into the Mission Canyon and Charles For heterogeneous oils of the Madison.
mations rather than remain diffused in the Lodgepole. The preponder
ance of these observations and modeling results suggests that some 5. Conclusions
Bakken oil has likely made its way into the Madison, not only at the
Nesson anticline and Poplar dome, but perhaps much more broadly A 3D petroleum systems modeling study was undertaken to support
throughout the basin. recent USGS assessments of undiscovered resources in the U.S. portion of
The Tyler Petroleum System includes sandstone reservoirs sourced the Williston Basin. The model, as well as all input and resultant data,
from the Tyler Formation, as described in Section 3.1.5 (Anna, 2013; are available publicly in a USGS data release (Gelman and Johnson,
Schenk et al., 2021c). Production has been focused on sandstone strat 2023). This report provides interpretive details on the input data and
igraphic traps in Billings and Stark Counties, North Dakota, which results of this basin modeling study. Regional geologic interpretation of
overlaps with the region in which the upper part of the Tyler source rock wireline well logs for 29 formation tops formed the basis of the model
is both its thickest and most mature (Fig. 19B). There is relatively little and provided a framework for the creation of various facies maps that
production from the Tyler outside of this region, although a few sporadic documented source presence, reservoir presence, and evaporite thick
wells exist in the central portion of the basin where the lower part of the nesses throughout the basin. While numerous source rocks have been
Tyler source is present and mature. In the basin model migration described throughout the stratigraphic column of the Williston Basin,
simulation, Tyler oils migrated mostly up and down section where the five were investigated in this study: 1) the Icebox Formation, 2)
source rocks were mature; no Tyler oil migrated beyond the sides of the kukersites of the Red River Formation, 3) upper and lower shales of the
AOI. If lateral migration does occur, flow vectors generally indicate Bakken Formation, 4) source horizons in the Madison Group, and 5)
southward and westward migration of oil (Fig. 19B). Although the Tyler sources in the upper and lower parts of the Tyler Formation. Key find
source rocks are rich, they are thin and spatially limited, and may not ings include:
have the generative volume to support long-distance lateral migration.
The most productive wells in the Tyler thus far target reservoirs within a 1) Source rock mapping of the kukersite beds in the Red River
few tens of feet from the mature upper part of the Tyler source rock, confirmed their previously documented absence east of the Nesson
suggesting that migration may be limited in the Tyler petroleum system. anticline, implying a migration shadow for Red River oil in the
Future success in the Tyler likely depends on the discovery of additional eastern side of the basin. The mapping also extended kukersite
stratigraphic traps in geographic locations where the source in the upper presence into eastern Montana, providing a good match to observed
part of the Tyler is present and mature, since very little vertical migra Red River production in that area.
tion is required between source rock and reservoir rock. Produced fluids 2) Source rock mapping of the Tyler generally agreed with previous
in the Tyler are generally very low in gas, with an average cumulative work in North Dakota, but additionally suggested a potential source
production GOR of only 124 scf/bbl (ranging from a few tens to few pod present in the lower part of the Tyler near the Cedar Creek
hundreds of scf/bbl). This GOR is aligned with the early oil to peak oil anticline. Future confirmation of organic enrichment here could
maturity of the source rocks, as well as mixed carbonate Type I/II have important implications for the possibility of Pennsylvanian oil
oil-prone organic matter (Nesheim and Nordeng, 2014a, 2014b). along the anticline, as predicted by the model using this source map.
The stacked nature of the Williston Basin petroleum systems is 3) Regional-scale wireline-log facies mapping of the middle Bakken
probably best exemplified along its few prominent structural features. member suggests that facies trends may correlate with cumulative
While much of the central portion of the basin contains congruent re production in some parts of the basin. These trends include a lack of
gions of stratigraphically stacked mature source rocks, the Cedar Creek coarse-grained “facies D″ deposition in areas with historically poorer
anticline, Nesson anticline, and Poplar dome are all foci for hydrocarbon production (labeled “silt/shale” in Fig. 6). This contrasts with the
migration and provide large-scale structural traps, leading to accumu absence of “facies D″ in the Parshall field area, where production is
lations at multiple stratigraphic levels. Hydrocarbon charge generally highest in the basin. Regional analysis and integration of complex
peaks in the Paleogene, concurrent with final structural activation along core-based facies is necessary for basin-scale models, although even
these major features. Re-adjustment of traps may have occurred as uplift these coarse groupings may provide insight into broad-scale reser
progressed, along with a return from normal to reverse faulting at the voir quality trends, highlighting areas of future exploration
Cedar Creek and Nesson anticlines. While migration modeling in this potential.
study was only qualitative, the relative proportions of oil from various 4) The thermal model was calibrated to a variety of subsurface tem
sources are listed for context on the charge history of these structures. At perature datasets, resulting in a basal heat flow map that validates a
the Cedar Creek anticline, the model predicted the Red River sourced high heat flow anomaly trending north to south in western North
55% of the oil, Tyler 37%, Madison 8%, and Icebox less than 1%, with no Dakota adjacent to the NACPCA. In addition, a high heat flow
contribution from the Bakken. The contribution of oil from the Tyler is anomaly in eastern Montana near Poplar dome was identified, which
dependent on the presence of source rock in the lower part of the Tyler in may be previously unrecognized. These high heat flow regions have
the vicinity of the anticline, which requires confirmation with core data important implications for the extent of source rock maturation.
for higher confidence in this result. In contrast to the Cedar Creek 5) All source rocks investigated were mature in the center of the basin
anticline, Bakken oil dominates the Nesson anticline (83%), followed by and their maturation history is dominantly influenced by Cretaceous
oil from the Red River (10%), Icebox (3%), Tyler (3%) and Madison sedimentation and subsequent Laramide uplift and erosion. Peak
(1%). Migration of oil to the Nesson anticline is likely limited by the generation occurred in the Paleogene, prior to regional uplift, and
small drainage area around the structure—most oil in the center of the relatively little hydrocarbons are still being generated today.
basin migrates up-dip to the periphery of the basin instead (see 6) Based on the basin model’s hydrocarbon generation estimates, the
Fig. 19A). Similarly, Bakken oil dominates at Poplar dome (93%), fol most prolific source rocks in the Williston Basin are the upper and
lowed distantly by Madison oil (7%). While a small fraction of oil from lower Bakken shale members, which is expected and consistent with
the central portion of the basin can migrate to Poplar dome, most of the production data. The kukersite beds of the Red River Formation have
western side of the basin drains instead north towards Canada or south also generated a substantial volume of oil, followed by Tyler and
towards the Cedar Creek anticline (Fig. 19A). Interestingly, the only Madison sources. The Icebox is modeled to have produced the lowest
parts of the basin in which mixed Bakken and Madison oil has been volume of hydrocarbons of the source rocks studied. Madison source
confirmed geochemically are the Nesson anticline and Poplar dome rocks remain uncertain compared with the volume of oil that has
22
already been produced and attributed to this source rock. Future million-year history of calcite cementation and dolomitization in the Devonian
Bakken Formation. J. Sediment. Res. 88, 91–104.
investigations to identify and characterize the richness, net thick
Bullard, E.C., 1947. The time necessary for a bore hole to attain temperature equilibrium.
ness, and geographic distribution of Madison source horizons can Geophys. Supp. Month. Not. Royal Astr. Soci. 5 (5), 127–130.
help to further advance basin modeling studies of the Williston Basin, Burrus, J., Osadetz, K.G., Wolf, S., Doligez, B., Visser, K., Dearborn, D., 1996a. A two-
and they can help elucidate or eliminate the role of mixing of Bakken dimensional regional basin model of Williston Basin hydrocarbon systems. AAPG
(Am. Assoc. Pet. Geol.) Bull. 80 (2), 265–291.
and Madison oils in the basin. Burrus, J., Wolf, S., Osadetz, K.G., Visser, K., 1996b. Physical and numerical modeling
constraints on oil expulsion and accumulation in the Bakken and Lodgepole
petroleum systems of the Williston Basin (Canada-U.S.A.). Bull. Can. Petrol. Geol. 44
Declaration of competing interest (3), 429–445.
Canter, L., Skinner, O., Sonnenfeld, M., Williams, M., 2011. Facies and mechanical
stratigraphy of the middle bakken, Mountrail county, North Dakota [abs.]. In:
The authors declare that they have no known competing financial Recovery CSPG CSEG CWLS Convention. Calgary, Alberta, Canada, May 9–11, 2011,
interests or personal relationships that could have appeared to influence accessed April 5, 2023, at. https://geoconvention.com/wp-content/uploads/abstrac
ts/2011/329-Facies_and_Mechanical_Stratigraphy_of_Middle_Bakken.pdf.
the work reported in this paper. Chen, Z., Osadetz, K.G., Jiang, C., Li, M., 2009. Spatial variation of bakken or Lodgepole
oils in the Canadian Williston Basin. AAPG (Am. Assoc. Pet. Geol.) Bull. 93 (6),
Data availability 829–851.
Chen, Z., Liu, X., Guo, Q., Jiang, C., Mort, A., 2017. Inversion of source rock hydrocarbon
generation kinetics from Rock-Eval data. Fuel 194, 91–101.
Data is available in cited USGS Data Release which is publicly Chen, Z., Liu, X., Osadetz, K.G., 2019. Petroleum generation kinetic models for late
available. ordovician kukersite yeoman Formation source rocks, Williston Basin (southern
saskatchewan), Canada. Fuel 241, 234–246.
Clement, J.H., 1986. Cedar Creek: A Significant Paleotectonic Feature of the Williston
Acknowledgements Basin, vol. 41. AAPG Memoir, pp. 213–240.
Copeland, P., Currie, C.A., Lawton, T.F., Murphy, M.A., 2017. Location, location,
location: the variable lifespan of the Laramide orogeny. Geology 45 (3), 223–226.
I would like to thank Tim Nesheim and Travis Stolldorf from the Crowley, K.D., Ahern, J.L., Naeser, C.W., 1985. Origin and epeirogenic history of the
North Dakota Geological Survey for providing helpful discussions on the Williston Basin: evidence from fission-track analysis of apatite. Geology 13 (9),
Williston Basin and for providing the static temperature logs utilized 620–623.
DeCelles, P.G., 2004. Late Jurassic to Eocene evolution of the cordilleran thrust belt and
during thermal calibration of this model. Discussions with Kristin Marra foreland basin system, western U.S.A. Am. J. Sci. 304, 105–168.
and Chris Schenk on assessments, the Bakken, and the regional geology Dow, W.G., 1974. Application of oil-correlation and source-rock data to exploration in
of the basin were greatly appreciated. I would also like to thank the Williston Basin. AAPG (Am. Assoc. Pet. Geol.) Bull. 58 (7), 1253–1262.
Dumitru, T.A., Gans, P.B., Foster, D.A., Miller, E.L., 1991. Refrigeration of the western
members of the National and Global Assessment (NAGA) team, in
Cordilleran lithosphere during Laramide shallow-angle subduction. Geology 19,
particular Jane Hearon and Ben Johnson, at the U.S. Geological Survey 1145–1148.
(USGS) for their support, camaraderie, and advice throughout this Egenhoff, S., Van Dolah, A., Jaffri, A., Maletz, J., 2011. Facies architecture and sequence
project. Constructive reviews from Brian Shaffer (USGS), Janet Pitman stratigraphy of the Middle Bakken Member, North Dakota. In: Robinson, J.W.,
LeFever, J.A., Gaswirth, S.B. (Eds.), The Bakken-Three Forks Petroleum System in
(USGS Emeritus), Justin Birdwell (USGS), Colin Sturrock (USGS), and the Williston Basin. Rocky Mountain Association of Geologists, Denver, Colo.
two anonymous journal reviewers were greatly appreciated and signif Flannery, J., Kraus, J., 2006. Integrated Analysis of the Bakken Petroleum System: US
icantly improved the clarity, comprehensiveness, and discussion of this Williston Basin. American Association of Petroleum Geologists, Search and
Discovery. Article #10105.
manuscript. Schlumberger is also acknowledged for use of their Petro Gaswirth, S.B., Lillis, P.G., Pollastro, R.M., Anna, L.O., 2013. Geologic Assessment of
Mod basin modeling software. Any use of trade, firm, or product names Undiscovered Oil and Gas Resources in the Madison Group, Williston Basin, North
is for descriptive purposes only and does not imply endorsement by the Dakota and Montana, Chap. 4 of U.S. Geological Survey Williston Basin Province
Assessment Team, Assessment of Undiscovered Oil and Gas Resources of the
U.S. Government. Williston Basin Province of North Dakota, Montana, and South Dakota, 2010, (Ver
1.1, November 2013). U.S. Geological Survey Digital Data Series DDS-69-W, p. 28.
Gaswirth, S.B., Marra, K.R., 2015. U.S. Geological Survey 2013 assessment of
Appendix A. Supplementary data
undiscovered resources in the Bakken and Three Forks Formations of the U.S.
Williston Basin Province. AAPG (Am. Assoc. Pet. Geol.) Bull. 99 (4), 639–660.
Supplementary data to this article can be found online at https://doi. Gelman, S.E., Johnson, B.G., 2023. Data Release for the 3D Petroleum Systems Model of
org/10.1016/j.marpetgeo.2023.106390. the Williston Basin. U.S. Geological Survey data release, USA. https://doi.org/
10.5066/P9N7O1OT.
Gerhard, L.C., Anderson, S.B., LeFever, J.A., 1987. Structural history of the Nesson
References Anticline, North Dakota, Rocky Mountain Association of Geologists 1987. In:
Symposium in Williston Basin: Anatomy of a Cratonic Oil Province, pp. 337–354.
Gerhard, L.C., Fischer, D.W., Anderson, S.B., 1990. Petroleum geology of the Williston
Abarghani, A., Ostadhassan, M., Gentzis, T., Carvajal-Ortiz, H., Ocubalidet, S.,
Basin. In: Leighton, M.W., Kolata, D.R., Oltz, D.F., Eidel, J.J. (Eds.), Interior Cratonic
Bubach, B., Mann, M., Hou, X., 2019. Correlating rock-eval Tmax with bitumen
Basins, vol. 51. AAPG Memoir, pp. 509–559.
reflectance from organic petrology in the Bakken Formation. Int. J. Coal Geol. 205,
Gherabati, S.A., Hamlin, H.S., Smye, K.M., Eastwood, R.L., Male, F.R., McDaid, G., 2019.
87–104.
Evaluating hydrocarbon-in-place and recovery factor in a hybrid petroleum system:
Anderson, F.J., 2007. Precambrian basement drillhole map of North Dakota. North
Case of Bakken and three forks in North Dakota. Interpretation 7 (3), T607–T624.
Dakota Geolog. Surv. Geologic Investig. 43, 1 (sheet).
Gosnold, W.D., 1990. Heat flow in the great plains of the United States. J. Geophys. Res.
Anna, L.O., 2013. Geologic Assessment of Undiscovered Oil and Gas in the Williston
95 (B1), 353–374.
Basin Province, Montana, North Dakota, and South Dakota, Chap. 3 of U.S.
Gosnold, W.D., McDonald, M.R., Klenner, R., Merriam, D., 2012. Thermostratigraphy of
Geological Survey Williston Basin Province Assessment Team, Assessment of
the Williston Basin. GRC Trans. 36, 663–670.
Undiscovered Oil and Gas Resources of the Williston Basin Province of North Dakota,
Grau, A., Sterling, R.H., 2011. Characterization of the Bakken system of the Williston
Montana, and South Dakota, 2010 (Ver. 1.1., November 2013). U.S. Geological
Basin from pores to production: The power of a source rock/unconventional
Survey Digital Data Series 69-W, p. 71.
reservoir couplet, Search and Discovery Article #40847. In: AAPG International
Bader, J.W., 2019. An ancient everest: precambrian basement terranes of the Williston
Conference and Exhibition.
Basin. Geo News 46 (1), 30–34.
Green, A.G., Weber, W., Hajnal, Z., 1985. Evolution of Proterozoic terrains beneath the
Bankey, V., Cuevas, A., Daniels, D., Finn, C.A., Hernandez, I., Hill, P., Kucks, R.,
Williston Basin. Geology 13 (9), 624–628.
Miles, W., Pilkington, M., Roberts, C., Roest, W., Rystrom, V., Shearer, S., Snyder, S.,
IHS Markit, 2020. Kingdom Seismic and Geological Interpretation Software V2020: IHS
Sweeney, R., Velez, J., Phillips, J.D., Ravat, D., 2002. Digital Data Grids for the
Markit, 15 Inverness Way East. Englewood, CO 80112, U.S.A., at. https://kingdom.
Magnetic Anomaly Map of North America, USGS Open-File Report 2002-414.
ihs.com/.
https://doi.org/10.3133/ofr02414.
IHS Markit, 2022. Enerdeq US Well History and Production; Database Available from IHS
Blakey, R.C., 2014. Paleogeography and Paleotectonics of the Western Interior Seaway.
Markit, 15 Inverness Way East. Englewood, CO 80112, U.S.A. https://spglobal.co
Jurassic-Cretaceous of North America. Search and Discovery Article #30392.
m/commodityinsights. (Accessed 30 September 2022).
Bowman, T., 2010. Direct Method for Determining Organic Shale Potential from Porosity
Jarvie, D.M., 2001. Williston Basin petroleum systems: Inferences from oil geochemistry
and Resistivity Logs to Identify Possible Resource Plays. American Association of
and geology. Mt. Geol. 38 (1), 19–41.
Petroleum Geologists Annual Convention. Search and Discovery Article #110128.
Brodie, M.W., Aplin, A.C., Hart, B., Orland, I.J., Valley, J.W., Boyce, A.J., 2018. Oxygen
isotope microanalysis by Secondary Ion Mass Spectrometry suggests continuous 300-
23
Jiang, C., Li, M., 2002. Bakken/Madison petroleum systems in the Canadian Williston Nesheim, T.O., Nordeng, S.H., 2014a. Core based examination of Upper Tyler Formation
Basin. Part 3: geochemical evidence for significant Bakken-derived oils in Madison source rocks within southwestern North Dakota: North Dakota. Geolog. Surv.
Group reservoirs. Org. Geochem. 33, 761–787. Geolog. Investig. 173, 28.
Jin, H., Sonnenberg, S.A., 2013. Characterization for source rock potential of the Bakken Nesheim, T.O., Nordeng, S.H., 2014b. Continued geological and geochemical evaluation
Shales in the Williston Basin, North Dakota and Montana. In: Unconventional of the Tyler Formation, a dual petroleum system. North Dakota Geolog. Surv.
Resources Technology Conference. Denver, Colorado, August 12–14, 2013, Geologic Investig. 177, 43.
Abstracts, article 1581243. https://library.seg.org/doi/10.1190/urtec2013-013. Nesheim, T.O., Nordeng, S.H., Onwumelu, S.J., 2021. Investigations into petroleum
(Accessed 5 April 2023). source rocks within the Mississippian Madison Group: Bluell subinterval of western
Jin, H., Lewan, M., 2017. Oil-generation kinetics for oil-prone Bakken shales and its McKenzie County. North Dakota Geolog. Surv. Geologic Investig. 255, 21.
implication. In: Unconventional Resources Technology Conference. Austin, Texas, Nielsen, S.B., Clausen, O.R., McGregor, E., 2017. basin%Ro: A vitrinite reflectance model
July 24–26, 2017, Abstracts, article 2671492. https://library.seg.org/doi/10.15530 derived from basin and laboratory data. Basin Res. 29 (Suppl. 1), 515–536.
/urtec-2017-2671492. (Accessed 5 April 2023). Obermajer, M., Osadetz, K.G., Fowler, M.G., Snowden, L.R., 2000. Light hydrocarbon
Johnson, M.S., 2009. Parshall Field, North Dakota—Discovery of the Year for the Rockies (gasoline range) parameter refinement of biomarker-based oil-oil correlation studies:
and beyond [abs. June 7–10, 2009, Abstracts. In: American Association of Petroleum an example from Williston Basin. Org. Geochem. 31, 959–976.
Geologists Annual Convention. AAPG Search and Discovery, Denver, Colorado. O’Connell, K., Skari, D., Wheeler, A.J., Rennie, A., 2012. Geosteering improves Bakken
Article #20081. results. Am. Oil Gas Rep. 55 (1), 87–93.
Jones, A.G., Savage, P.J., 1986. North American Central Plains Conductivity Anomaly Orchard, D.M., 1987. Structural history of Poplar Dome and the dissolution of Charles
Goes East. Geophys. Res. Lett. 13 (7), 685–688. Formation salt, Roosevelt County, Montana. In: Fifth International Williston Basin
Jones, A.G., Craven, J.A., 1990. The North American Central Plains conductivity Symposium. Saskatchewan Geological Society, Regina, pp. 169–177.
anomaly and its correlation with gravity, magnetic, seismic, and heat flow data in Osadetz, K.G., Snowdon, L.R., 1995. Significant Paleozoic Petroleum Source Rocks in the
Saskatchewan, Canada. Phys. Earth Planet. In. 60, 169–194. Canadian Williston Basin: Their Distribution, Richness and Thermal Maturity
Khan, D.K., Rostron, B.J., Margitai, Z., Carruthers, D., 2006. Hydrodynamics and (Southeastern Saskatchewan and Southwestern Manitoba), vol. 487. Geological
petroleum migration in the Upper Ordovician Red River Formation of the Williston Survey of Canada, Bulletin.
Basin. J. Geochem. Explor. 89, 179–182. Parker, J.M., Hess, P.D., 1980. MonDak Mississippian oil field, Williston Basin [abs. In:
Kuhn, P.P., di Primio, R., Horsfield, B., 2010. Bulk composition and phase behavior of the American Association of Petroleum Geologists Association Round Table, Abstracts,
petroleum sourced by the Bakken Formation of the Williston Basin. In: Proceedings vol. 64. AAPG Bulletin, pp. 762–763.
of the 7th Petroleum Geology Conference, pp. 1065–1077. Passey, Q.R., Creaney, S., Kulla, J.B., Moretti, F.J., Stroud, J.D., 1990. A practical model
Kuhn, P.P., di Primio, R., Hill, R., Lawrence, J.R., Horsfield, B., 2012. Three-dimensional for organic richness from porosity and resistivity logs. AAPG (Am. Assoc. Pet. Geol.)
modeling study of the low-permeability petroleum system of the Bakken Formation. Bull. 74 (12), 1777–1794.
AAPG (Am. Assoc. Pet. Geol.) Bull. 96 (10), 1867–1897. Peterson, J.A., 1984. Stratigraphy and Sedimentary Facies of the Madison Limestone and
LeFever, J.A., 2008. Structural contour and isopach maps of the Bakken Formation in Associated Rocks in Parts of Montana, Nebraska, North Dakota, South Dakota, and
North Dakota. North Dakota Geolog. Surv. Geologic Investig. 59, 5 (sheets). Wyoming. U.S. Geological Survey Professional Paper, pp. A1–A34, 1273–A.
LeFever, R.D., LeFever, J.A., 2011. Sandy lithofacies within the Icebox Formation Pepper, A.S., Corvi, P.J., 1995. Simple kinetic models of petroleum formation. Part I: oil
(Ordovician), North Dakota and Montana. In: North Dakota Geological Survey and gas from kerogen. Mar. Petrol. Geol. 12 (3), 291–319.
Geological Investigations, 140, p. 1 (sheet). Pitman, J.K., Price, L.C., LeFever, J.A., 2001. Diagenesis and fracture development in the
Li, L., Fan, M., 2018. Cenozoic sediment provenance in the northern Great Plains Bakken Formation, Williston Basin: Implications for reservoir quality in the Middle
corresponds to four episodes of tectonic and magmatic events in the central North Member. U. S. Geol. Surv. Prof. Pap. 1653, 1–19.
American Cordillera. Tectonics 37, 4018–4036. Pollastro, R.M., Roberts, L.N.R., Cook, T.A., 2012. Geologic model for the assessment of
Lillis, P.G., 2013. Review of oil families and their petroleum systems of the Williston technically recoverable oil in the Devonian–Mississippian Bakken Formation,
Basin. Mt. Geol. 50 (1), 5–31. Williston Basin. In: Breyer, J.A. (Ed.), Shale Reservoirs—Giant Resources for the 21st
Liu, X., Nordeng, S.H., Chen, Z., Jiang, C., Mort, A., 2020. Organic matter variation Century, vol. 97. AAPG Memoir, pp. 205–257.
within Upper and Lower Bakken shales of the Williston Basin by extracting kerogen Price, L.C., Barker, C.E., 1985. Suppression of vitrinite reflectance in amorphous rich
pyrogram information. Int. J. Coal Geol. 229, 103574. kerogen—A major unrecognized problem. J. Petrol. Geol. 8 (1), 59–84.
Majorowicz, J.A., Jones, F.W., Jessop, A.M., 1986. Geothermics of the Williston Basin in Price, L.C., Daws, T., Pawlewicz, M., 1986. Organic metamorphism in the Lower
Canada in relation to hydrodynamics and hydrocarbon occurrences. Geophysics 51 Mississippian–Upper Devonian Bakken Shales Part 1: Rock-Eval pyrolysis and
(3), 767–779. vitrinite reflectance. J. Petrol. Geol. 9 (2), 125–162.
Majorowicz, J.A., Jones, F.W., Osadetz, K.G., 1988. Heat flow environment of the Price, L.C., LeFever, J., 1994. Dysfunctionalism in the Williston Basin: the Bakken/mid-
electrical conductivity anomalies in the Williston Basin, and occurrence of Madison petroleum system. Bull. Can. Petrol. Geol. 42 (2), 187–218.
hydrocarbons. Bull. Can. Petrol. Geol. 36 (1), 86–90. Schenk, C.J., Mercier, T.J., Woodall, C.A., Ellis, G.S., Finn, T.M., Le, P.A., Marra, K.R.,
Marra, K.R., Mercier, T.J., Gelman, S.E., Schenk, C.J., Woodall, C.A., Cicero, A.D., Leathers-Miller, H.M., Drake II, R.M., 2021a. Assessment of Undiscovered
Drake II., R.M., Ellis, G.S., Finn, T.M., Gardner, M.H., Hearon, J.S., Johnson, B.G., Conventional Oil and Gas Resources in the Lower Paleozoic of the Williston Basin
Lagesse, J.H., Le, P.A., Leathers-Miller, H.M., Timm, K.K., Young, S.S., 2021. Province, 2020. U.S. Geological Survey Fact, p. 2. https://doi.org/10.3133/
Assessment Of Undiscovered Continuous Oil Resources In The Bakken And Three fs20203063. Sheet 2020-3063.
Forks Formations Of The Williston Basin Province, North Dakota And Montana, Schenk, C.J., Mercier, T.J., Woodall, C.A., Ellis, G.S., Finn, T.M., Le, P.A., Marra, K.R.,
2021: U.S. Geological Survey Fact Sheet 2021-3058, p. 4. https://doi.org/10.3133/ Leathers-Miller, H.M., Drake II, R.M., 2021b. Assessment of Undiscovered
fs20213058. Conventional Oil and Gas Resources of Upper Paleozoic Strata in the Williston Basin
Meissner, F.F., 1978. Petroleum geology of the Bakken Formation, Williston Basin, North Province, 2020. U.S. Geological Survey Fact, p. 4. https://doi.org/10.3133/
Dakota and Montana. In: Rehrig, D. (Ed.), The Economic Geology of the Williston fs20213023. Sheet 2021-3023.
Basin, 24th Annual Conference: Montana. Geological Society, pp. 207–227. Schenk, C.J., Mercier, T.J., Woodall, C.A., Finn, T.M., Le, P.A., Marra, K.R., Ellis, G.S.,
Meissner, F.F., Banks, R.B., 2005. Computer simulation of hydrocarbon generation, Leathers-Miller, H.M., Drake II, R.M., 2021c. Assessment of Undiscovered Oil and
migration, and accumulation under hydrodynamic conditions—Examples from the Gas Resources in the Pennsylvanian Tyler Formation of the Williston Basin Province,
Williston and San Juan Basins, USA [abs. October 15–18, 2000, Abstracts. In: 2020. U.S. Geological Survey Fact, p. 2. https://doi.org/10.3133/fs20203067. Sheet
American Association of Petroleum Geologists International Conference and 2020-3067.
Exhibition, Bali, Indonesia. AAPG Search and Discovery. article #40179. https:// Schlumberger, 2020. PetroMod Basin and Petroleum Systems Modeling Software: IES
www.searchanddiscovery.com/documents/2005/banks/. (Accessed 5 April 2023). GmbH. Ritterstrasser 23, 52072 (Aachen, Germany).
Mohamed, T.I., 2015. Sequence Stratigraphy and Provenance of the Bakken Formation in Schmoker, J.W., Hester, T.C., 1983. Organic carbon in Bakken Formation, United
Southestern Alberta and Southwestern Saskatchewan: Calgary, Alberta. Master’s Stations portion of the Williston Basin. AAPG (Am. Assoc. Pet. Geol.) Bull. 67 (12),
thesis. University of Calgary, p. 106. https://doi.org/10.11575/PRISM/27758, 20 2165–2174.
figs. Sloss, L.L., 1963. Sequences in the cratonic interior of North America. Geol. Soc. Am.
Murphy, E.C., Nordeng, S.H., Juenker, B.J., Hoganson, W., 2009. North Dakota Bull. 74, 93–114.
Stratigraphic Column: North Dakota Geological Survey Miscellaneous Series 91, 1 Smith, M., Bend, S., 2004. Geochemical analysis and familial association of Red River
sheet. and Winnipeg reservoired oils of the Williston Basin, Canada. Org. Geochem. 35 (4),
Nesheim, T.O., 2017. Stratigraphic and geochemical investigation of kukersites 443–452.
(petroleum source beds) within the Ordovician Red River Formation, Williston Soller, D.R., Garrity, C.P., 2018. Quaternary sediment thickness and bedrock topography
Basin. AAPG (Am. Assoc. Pet. Geol.) Bull. 101 (9), 1445–1471. of the glaciated United States east of the Rocky Mountains. U.S. Geolog. Surv. Scien.
Nesheim, T.O., 2012. Review of radiometric ages from North Dakota’s Precambrian Investig. Map 3392. https://doi.org/10.3133/sim3392, 2 sheets, scale 1:5,000,000.
basement. North Dakota Geolog. Surv. Geologic Investig. 160, 1 sheet. Sonnenberg, S.A., Vickery, J., Theloy, C., Sarg, J.F., 2011. Middle Bakken Facies,
Nesheim, T.O., 2019. Examination of downward hydrocarbon charge within the Bakken- Williston Basin, USA: A key to prolific production [abs. April 10–13, 2011, Abstracts.
Three Forks petroleum system – Williston Basin. North Am. Mar. Petr. Geol. 104, In: American Association of Petroleum Geologists Annual Convention and
346–360. Exhibition. AAPG Search and Discovery, Houston, Texas. article #50449.
Nesheim, T.O., Nordeng, S.H., 2012. Source rock intervals within the Tyler Formation, Sonnenberg, S.A., Theloy, C., Jin, H., 2017. The giant continuous oil accumulation in the
North Dakota. North Dakota Geolog. Surv. Geologic Investig. 151, 1 sheet. Bakken petroleum system, U.S. Williston Basin. In: Merrill, R.K., Sternbach, C.A.
Nesheim, T.O., Nordeng, S.H., 2013. Examination of the Icebox Formation’s (Winnipeg (Eds.), Giant Fields of the Decade 2000–2010, vol. 113. AAPG Memoir, pp. 91–120.
Group, Ordovician) source rock potential within North Dakota. North Dakota Stolldorf, T.D., 2019. Williston Basin static temperature logging program continues to
Geolog. Surv. Geolog. Investig. 169, 63. yield results. North Dakota Geolog. Surv. Geo News 46 (1), 14–16.
24
Stolldorf, T.D., 2020. Using static temperature to more accurately predict geothermal U.S. Energy Information Administration [EIA], 2020. Annual Report of Domestic Oil and
gradients in the Williston Basin. North Dakota Geolog. Surv. Geo News 47 (1), Gas Reserves, Form EIA-23L. https://www.eia.gov/naturalgas/crudeoilreserves/
14–16. pdf/usreserves.pdf. (Accessed 26 September 2022).
Sturrock, C.P., Flowers, R.M., Macdonald, F.A., 2021. The late Great Unconformity of the U.S. Geological Survey, [undated] Energy Resources Program Geochemistry Laboratory
central Canadian Shield. G-cubed 22 (6), e2020GC009567. https://doi.org/ Database (EGDB): U.S. Geological Survey database, accessed December 15, 2020, at
10.1029/2020GC009567. https://certmapper.cr.usgs.gov/data/apps/geochem-db/.
Sweeney, J.J., Burnham, A.K., 1990. Evaluation of a simple model of vitrinite reflectance U.S. Geological Survey, National Geospatial Program, 20200826, USGS NED 1/3 Arc-
based on chemical kinetics. AAPG (Am. Assoc. Pet. Geol.) Bull. 74 (10), 1559–1570. Second Contours for Williston W, North Dakota 20200826 1 X 1 Degree Shapefile: U.
Tegelaar, E.W., Noble, R.A., 1994. Kinetics of hydrocarbon generation as a function of S. Geological Survey.
the molecular structure of kerogen as revealed by pyrolysis-gas chromatography. Webster, R.L., 1984. Petroleum source rocks and stratigraphy of the Bakken Formation in
Org. Geochem. 22 (3–5), 543–574. North Dakota. In: Woodward, J., Meissner, F.F., Clayton, J.L. (Eds.), Hydrocarbon
Theloy, C., 2013. Integration of Geological and Technological Factors Influencing Source Rocks of the Greater Rocky Mountain Region: Rocky Mountain Association of
Production in the Bakken Play, Williston Basin, Golden, Colorado. PhD dissertation. Geologists, pp. 57–81. Denver, CO.
Colorado School of Mines, p. 223, 120 figs. Williams, J.A., 1974. Characterization of oil types in Williston Basin. AAPG (Am. Assoc.
Pet. Geol.) Bull. 58 (7), 1243–1252.
25

Modeling the maturation history of the stacked petroleum systems of the Williston Basin, USA

Uploaded by

Copyright:

Available Formats

Modeling the maturation history of the stacked petroleum systems of the Williston Basin, USA

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Modeling the maturation history of the stacked petroleum systems of the Williston Basin, USA

Uploaded by

Copyright:

Available Formats

Marine and Petroleum Geology 155 (2023) 106390

Contents lists available at ScienceDirect

Marine and Petroleum Geology

Modeling the maturation history of the stacked petroleum systems of the

E-mail address: [email protected].

The stratigraphy of the Williston Basin has been mapped extensively

0 Topography USGS NED 1/3 arc-second

Fig. 3. Overview of the facies model utilized for

displayed elevated sonic log signatures through a thick stratigraphic

3.1.3. The Late Devonian–early Mississippian Bakken Formation

(caption on next page)

Tyler Fm, upper part 15 800 TII(B), Pepper and Corvi

In contrast to source rock intervals, evaporites of the Madison are

3.1.5. The Pennsylvanian Tyler Formation source rocks

50% transformation ratio at a similar thermal stress as the Jin and

4. Model results and implications for the stacked petroleum

While the Williston Basin has experienced a long sedimentary history

4.1. Time-temperature and maturation history

All the Williston Basin source rocks experienced a time-temperature

This genetic time-temperature history results in a maturation history

Fig. 13. Calibration of hydrocarbon generation ki­

Fig. 14. Typical burial history plot for the Williston

4.2. Comparative generative volumes of the Williston source rocks

Combining source presence mapping, source rock thickness, rich­

Fig. 16. Compilation of maturity maps at present-day

You might also like

Fig. 13. Calibration of hydrocarbon generation ki

Combining source presence mapping, source rock thickness, rich